Allow to build coresight as modules. This gives developers the feasibility to test their code without reboot.
This series is based on below two series.
- "coresight: allow to build components as modules" https://lkml.org/lkml/2018/6/5/989 - "coresight: make drivers modular" https://lkml.org/lkml/2020/1/17/468
Change from v7: Depends on below change for ETM hotplug (Sai) https://lore.kernel.org/linux-arm-kernel/20200729051310.18436-1-saiprakash.r... Add mutex lock to protect etmdrvdata[] (Suzuki) Add helper function coresight_get_ref() (Suzuki) Reorg replicator and funnel change. Use two patches to support these two drivers. (Suzuki) Add fix tag to "coresight: etm: perf: Fix warning caused by etm_setup_aux failure" (Suzuki) Update author of "coresight: cti: Fix bug clearing sysfs links on callback"
Change from v6: Correct module description for CATU (Mike) Check ect_ret equals 0 and set ect_enabled flag (Mike) Add Tested-by and Reviewed-by from Mike
Change from v5: Add below CTI clean up change from Mike into series -https://lists.linaro.org/pipermail/coresight/2020-July/004349.html Increase module reference count when enabling CTI device (Mike)
Change from v4: Fix error handling in coresight_grab_devicei() (Greg) Add coresight: cti: Fix remove sysfs link error from Mike -https://lists.linaro.org/pipermail/coresight/2020-July/004275.html Move cti_remove_conn_xrefs() into cti_remove() (Mike) Align patch subject to coresight: <component>: <description> (Mike)
Change from v3: Rebase to coresight-next (Mike and Mathieu) Reorder try_get_module() (Suzuki) Clean up etmdrvdata[] in device remote path (Mike) Move cti_remove_conn_xrefs to cti_remove (Mike)
Change from v2: Rebase to 5.8-rc5. Export coresight_add_sysfs_link and coresight_remove_sysfs_link Fix one cut and paste error on MODULE_DESCRIPTION of CTI
Change from v1: Use try_module_get() to avoid module to be unloaded when device is used in active trace session. (Mathieu P)
Change from above two series. This series adds the support to dynamically remove module when the device in that module is enabled and used by some trace path. It disables all trace paths with that device and release the trace path. Kim Phillips (8): coresight: use IS_ENABLED for CONFIGs that may be modules coresight: etm3x: allow etm3x to be built as a module coresight: etm4x: allow etm4x to be built as a module coresight: etb: allow etb to be built as a module coresight: tpiu: allow tpiu to be built as a module coresight: tmc: allow tmc to be built as a module coresight: allow funnel driver to be built as module coresight: allow replicator driver to be built as module
Mian Yousaf Kaukab (2): coresight: export global symbols coresight: tmc-etr: add function to register catu ops
Mike Leach (2): coresight: cti: Fix remove sysfs link error coresight: cti: Fix bug clearing sysfs links on callback
Tingwei Zhang (12): coresight: cpu_debug: add module name in Kconfig coresight: cpu_debug: define MODULE_DEVICE_TABLE coresight: add coresight prefix to barrier_pkt coresight: add try_get_module() in coresight_grab_device() coresight: stm: allow to build coresight-stm as a module coresight: etm: perf: Fix warning caused by etm_setup_aux failure coresight: cti: add function to register cti associate ops coresight: cti: don't disable ect device if it's not enabled coresight: cti: increase reference count when enabling cti coresight: cti: allow cti to be built as a module coresight: catu: allow catu drivers to be built as modules coresight: allow the coresight core driver to be built as a module
drivers/hwtracing/coresight/Kconfig | 54 ++++-- drivers/hwtracing/coresight/Makefile | 22 +-- drivers/hwtracing/coresight/coresight-catu.c | 37 ++++- drivers/hwtracing/coresight/coresight-catu.h | 2 - .../{coresight.c => coresight-core.c} | 156 +++++++++++++++--- .../hwtracing/coresight/coresight-cpu-debug.c | 2 + .../{coresight-cti.c => coresight-cti-core.c} | 62 ++++++- drivers/hwtracing/coresight/coresight-etb10.c | 27 ++- .../hwtracing/coresight/coresight-etm-perf.c | 13 +- .../hwtracing/coresight/coresight-etm-perf.h | 5 +- ...resight-etm3x.c => coresight-etm3x-core.c} | 27 ++- ...resight-etm4x.c => coresight-etm4x-core.c} | 118 +++++++++---- .../hwtracing/coresight/coresight-funnel.c | 62 ++++++- .../hwtracing/coresight/coresight-platform.c | 1 + drivers/hwtracing/coresight/coresight-priv.h | 24 ++- .../coresight/coresight-replicator.c | 63 ++++++- drivers/hwtracing/coresight/coresight-stm.c | 20 ++- drivers/hwtracing/coresight/coresight-sysfs.c | 2 + .../{coresight-tmc.c => coresight-tmc-core.c} | 24 ++- .../hwtracing/coresight/coresight-tmc-etf.c | 2 +- .../hwtracing/coresight/coresight-tmc-etr.c | 21 ++- drivers/hwtracing/coresight/coresight-tmc.h | 3 + drivers/hwtracing/coresight/coresight-tpiu.c | 19 ++- include/linux/coresight.h | 3 +- 24 files changed, 650 insertions(+), 119 deletions(-) rename drivers/hwtracing/coresight/{coresight.c => coresight-core.c} (92%) rename drivers/hwtracing/coresight/{coresight-cti.c => coresight-cti-core.c} (94%) rename drivers/hwtracing/coresight/{coresight-etm3x.c => coresight-etm3x-core.c} (97%) rename drivers/hwtracing/coresight/{coresight-etm4x.c => coresight-etm4x-core.c} (95%) rename drivers/hwtracing/coresight/{coresight-tmc.c => coresight-tmc-core.c} (95%)
Provide name of cpu_debug module in Kconfig help section.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/Kconfig | 3 +++ 1 file changed, 3 insertions(+)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 02dbb5ca3bcf..4663fd1bbffc 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -110,6 +110,9 @@ config CORESIGHT_CPU_DEBUG properly, please refer Documentation/trace/coresight/coresight-cpu-debug.rst for detailed description and the example for usage.
+ To compile this driver as a module, choose M here: the + module will be called coresight-cpu-debug. + config CORESIGHT_CTI bool "CoreSight Cross Trigger Interface (CTI) driver" depends on ARM || ARM64
On Fri, Aug 07, 2020 at 07:11:30PM +0800, Tingwei Zhang wrote:
Provide name of cpu_debug module in Kconfig help section.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/Kconfig | 3 +++ 1 file changed, 3 insertions(+)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 02dbb5ca3bcf..4663fd1bbffc 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -110,6 +110,9 @@ config CORESIGHT_CPU_DEBUG properly, please refer Documentation/trace/coresight/coresight-cpu-debug.rst for detailed description and the example for usage.
To compile this driver as a module, choose M here: the
module will be called coresight-cpu-debug.
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
config CORESIGHT_CTI bool "CoreSight Cross Trigger Interface (CTI) driver" depends on ARM || ARM64 -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Define a MODULE_DEVICE_TABLE for cpu_debug so module can be auto loaded on boot.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight-cpu-debug.c | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/drivers/hwtracing/coresight/coresight-cpu-debug.c b/drivers/hwtracing/coresight/coresight-cpu-debug.c index 96544b348c27..1d0880b3764a 100644 --- a/drivers/hwtracing/coresight/coresight-cpu-debug.c +++ b/drivers/hwtracing/coresight/coresight-cpu-debug.c @@ -665,6 +665,8 @@ static const struct amba_id debug_ids[] = { {}, };
+MODULE_DEVICE_TABLE(amba, debug_ids); + static struct amba_driver debug_driver = { .drv = { .name = "coresight-cpu-debug",
From: Kim Phillips kim.phillips@arm.com
Checking for ifdef CONFIG_x fails if CONFIG_x=m. Use IS_ENABLED that is true for both built-ins and modules, instead. Required when building coresight components as modules.
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight-etm-perf.h | 2 +- drivers/hwtracing/coresight/coresight-priv.h | 2 +- include/linux/coresight.h | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.h b/drivers/hwtracing/coresight/coresight-etm-perf.h index 015213abe00a..05f89723e282 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.h +++ b/drivers/hwtracing/coresight/coresight-etm-perf.h @@ -57,7 +57,7 @@ struct etm_event_data { struct list_head * __percpu *path; };
-#ifdef CONFIG_CORESIGHT +#if IS_ENABLED(CONFIG_CORESIGHT) int etm_perf_symlink(struct coresight_device *csdev, bool link); int etm_perf_add_symlink_sink(struct coresight_device *csdev); void etm_perf_del_symlink_sink(struct coresight_device *csdev); diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index f2dc625ea585..d801a2755432 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -165,7 +165,7 @@ int coresight_make_links(struct coresight_device *orig, void coresight_remove_links(struct coresight_device *orig, struct coresight_connection *conn);
-#ifdef CONFIG_CORESIGHT_SOURCE_ETM3X +#if IS_ENABLED(CONFIG_CORESIGHT_SOURCE_ETM3X) extern int etm_readl_cp14(u32 off, unsigned int *val); extern int etm_writel_cp14(u32 off, u32 val); #else diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 58fffdecdbfd..3bb738f9a326 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -324,7 +324,7 @@ struct coresight_ops { const struct coresight_ops_ect *ect_ops; };
-#ifdef CONFIG_CORESIGHT +#if IS_ENABLED(CONFIG_CORESIGHT) extern struct coresight_device * coresight_register(struct coresight_desc *desc); extern void coresight_unregister(struct coresight_device *csdev);
On Fri, Aug 07, 2020 at 07:11:32PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Checking for ifdef CONFIG_x fails if CONFIG_x=m. Use IS_ENABLED that is true for both built-ins and modules, instead. Required when building coresight components as modules.
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/coresight-etm-perf.h | 2 +- drivers/hwtracing/coresight/coresight-priv.h | 2 +- include/linux/coresight.h | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-)
For patches 02 and 03:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.h b/drivers/hwtracing/coresight/coresight-etm-perf.h index 015213abe00a..05f89723e282 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.h +++ b/drivers/hwtracing/coresight/coresight-etm-perf.h @@ -57,7 +57,7 @@ struct etm_event_data { struct list_head * __percpu *path; }; -#ifdef CONFIG_CORESIGHT +#if IS_ENABLED(CONFIG_CORESIGHT) int etm_perf_symlink(struct coresight_device *csdev, bool link); int etm_perf_add_symlink_sink(struct coresight_device *csdev); void etm_perf_del_symlink_sink(struct coresight_device *csdev); diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index f2dc625ea585..d801a2755432 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -165,7 +165,7 @@ int coresight_make_links(struct coresight_device *orig, void coresight_remove_links(struct coresight_device *orig, struct coresight_connection *conn); -#ifdef CONFIG_CORESIGHT_SOURCE_ETM3X +#if IS_ENABLED(CONFIG_CORESIGHT_SOURCE_ETM3X) extern int etm_readl_cp14(u32 off, unsigned int *val); extern int etm_writel_cp14(u32 off, u32 val); #else diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 58fffdecdbfd..3bb738f9a326 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -324,7 +324,7 @@ struct coresight_ops { const struct coresight_ops_ect *ect_ops; }; -#ifdef CONFIG_CORESIGHT +#if IS_ENABLED(CONFIG_CORESIGHT) extern struct coresight_device * coresight_register(struct coresight_desc *desc); extern void coresight_unregister(struct coresight_device *csdev); -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Add coresight prefix to make it specific. It will be a export symbol.
Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight-etb10.c | 2 +- drivers/hwtracing/coresight/coresight-priv.h | 8 ++++---- drivers/hwtracing/coresight/coresight-tmc-etf.c | 2 +- drivers/hwtracing/coresight/coresight.c | 3 ++- 4 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-etb10.c b/drivers/hwtracing/coresight/coresight-etb10.c index 03e3f2590191..04ee9cda988d 100644 --- a/drivers/hwtracing/coresight/coresight-etb10.c +++ b/drivers/hwtracing/coresight/coresight-etb10.c @@ -525,7 +525,7 @@ static unsigned long etb_update_buffer(struct coresight_device *csdev,
cur = buf->cur; offset = buf->offset; - barrier = barrier_pkt; + barrier = coresight_barrier_pkt;
for (i = 0; i < to_read; i += 4) { buf_ptr = buf->data_pages[cur] + offset; diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index d801a2755432..dcb8aeb6af62 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -66,8 +66,8 @@ static DEVICE_ATTR_RO(name) #define coresight_simple_reg64(type, name, lo_off, hi_off) \ __coresight_simple_func(type, NULL, name, lo_off, hi_off)
-extern const u32 barrier_pkt[4]; -#define CORESIGHT_BARRIER_PKT_SIZE (sizeof(barrier_pkt)) +extern const u32 coresight_barrier_pkt[4]; +#define CORESIGHT_BARRIER_PKT_SIZE (sizeof(coresight_barrier_pkt))
enum etm_addr_type { ETM_ADDR_TYPE_NONE, @@ -104,10 +104,10 @@ struct cs_buffers { static inline void coresight_insert_barrier_packet(void *buf) { if (buf) - memcpy(buf, barrier_pkt, CORESIGHT_BARRIER_PKT_SIZE); + memcpy(buf, coresight_barrier_pkt, + CORESIGHT_BARRIER_PKT_SIZE); }
- static inline void CS_LOCK(void __iomem *addr) { do { diff --git a/drivers/hwtracing/coresight/coresight-tmc-etf.c b/drivers/hwtracing/coresight/coresight-tmc-etf.c index 6375504ba8b0..44402d413ebb 100644 --- a/drivers/hwtracing/coresight/coresight-tmc-etf.c +++ b/drivers/hwtracing/coresight/coresight-tmc-etf.c @@ -519,7 +519,7 @@ static unsigned long tmc_update_etf_buffer(struct coresight_device *csdev,
cur = buf->cur; offset = buf->offset; - barrier = barrier_pkt; + barrier = coresight_barrier_pkt;
/* for every byte to read */ for (i = 0; i < to_read; i += 4) { diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index e9c90f2de34a..310b1b825dd1 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -53,7 +53,8 @@ static struct list_head *stm_path; * beginning of the data collected in a buffer. That way the decoder knows that * it needs to look for another sync sequence. */ -const u32 barrier_pkt[4] = {0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff}; +const u32 coresight_barrier_pkt[4] = { + 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff};
static int coresight_id_match(struct device *dev, void *data) {
On Fri, Aug 07, 2020 at 07:11:33PM +0800, Tingwei Zhang wrote:
Add coresight prefix to make it specific. It will be a export symbol.
Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/coresight-etb10.c | 2 +- drivers/hwtracing/coresight/coresight-priv.h | 8 ++++---- drivers/hwtracing/coresight/coresight-tmc-etf.c | 2 +- drivers/hwtracing/coresight/coresight.c | 3 ++- 4 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-etb10.c b/drivers/hwtracing/coresight/coresight-etb10.c index 03e3f2590191..04ee9cda988d 100644 --- a/drivers/hwtracing/coresight/coresight-etb10.c +++ b/drivers/hwtracing/coresight/coresight-etb10.c @@ -525,7 +525,7 @@ static unsigned long etb_update_buffer(struct coresight_device *csdev, cur = buf->cur; offset = buf->offset;
- barrier = barrier_pkt;
- barrier = coresight_barrier_pkt;
for (i = 0; i < to_read; i += 4) { buf_ptr = buf->data_pages[cur] + offset; diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index d801a2755432..dcb8aeb6af62 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -66,8 +66,8 @@ static DEVICE_ATTR_RO(name) #define coresight_simple_reg64(type, name, lo_off, hi_off) \ __coresight_simple_func(type, NULL, name, lo_off, hi_off) -extern const u32 barrier_pkt[4]; -#define CORESIGHT_BARRIER_PKT_SIZE (sizeof(barrier_pkt)) +extern const u32 coresight_barrier_pkt[4]; +#define CORESIGHT_BARRIER_PKT_SIZE (sizeof(coresight_barrier_pkt)) enum etm_addr_type { ETM_ADDR_TYPE_NONE, @@ -104,10 +104,10 @@ struct cs_buffers { static inline void coresight_insert_barrier_packet(void *buf) { if (buf)
memcpy(buf, barrier_pkt, CORESIGHT_BARRIER_PKT_SIZE);
memcpy(buf, coresight_barrier_pkt,
CORESIGHT_BARRIER_PKT_SIZE);
Indentation problem.
}
static inline void CS_LOCK(void __iomem *addr) { do { diff --git a/drivers/hwtracing/coresight/coresight-tmc-etf.c b/drivers/hwtracing/coresight/coresight-tmc-etf.c index 6375504ba8b0..44402d413ebb 100644 --- a/drivers/hwtracing/coresight/coresight-tmc-etf.c +++ b/drivers/hwtracing/coresight/coresight-tmc-etf.c @@ -519,7 +519,7 @@ static unsigned long tmc_update_etf_buffer(struct coresight_device *csdev, cur = buf->cur; offset = buf->offset;
- barrier = barrier_pkt;
- barrier = coresight_barrier_pkt;
/* for every byte to read */ for (i = 0; i < to_read; i += 4) { diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index e9c90f2de34a..310b1b825dd1 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -53,7 +53,8 @@ static struct list_head *stm_path;
- beginning of the data collected in a buffer. That way the decoder knows that
- it needs to look for another sync sequence.
*/ -const u32 barrier_pkt[4] = {0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff}; +const u32 coresight_barrier_pkt[4] = {
0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff};
As of kernel 5.8 lines up to 100 characters are allowed. No need to break this line.
With that:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
static int coresight_id_match(struct device *dev, void *data) { -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
From: Mian Yousaf Kaukab ykaukab@suse.de
Export symbols used among coresight modules.
Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight-etm-perf.c | 1 + drivers/hwtracing/coresight/coresight-sysfs.c | 2 ++ drivers/hwtracing/coresight/coresight-tmc-etr.c | 6 ++++++ drivers/hwtracing/coresight/coresight.c | 8 ++++++++ 4 files changed, 17 insertions(+)
diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index 1a3169e69bb1..dcb0592418ae 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -517,6 +517,7 @@ int etm_perf_symlink(struct coresight_device *csdev, bool link)
return 0; } +EXPORT_SYMBOL_GPL(etm_perf_symlink);
static ssize_t etm_perf_sink_name_show(struct device *dev, struct device_attribute *dattr, diff --git a/drivers/hwtracing/coresight/coresight-sysfs.c b/drivers/hwtracing/coresight/coresight-sysfs.c index 82afeaf2ccc4..34d2a2d31d00 100644 --- a/drivers/hwtracing/coresight/coresight-sysfs.c +++ b/drivers/hwtracing/coresight/coresight-sysfs.c @@ -102,6 +102,7 @@ int coresight_add_sysfs_link(struct coresight_sysfs_link *info)
return ret; } +EXPORT_SYMBOL_GPL(coresight_add_sysfs_link);
void coresight_remove_sysfs_link(struct coresight_sysfs_link *info) { @@ -122,6 +123,7 @@ void coresight_remove_sysfs_link(struct coresight_sysfs_link *info) info->orig->nr_links--; info->target->nr_links--; } +EXPORT_SYMBOL_GPL(coresight_remove_sysfs_link);
/* * coresight_make_links: Make a link for a connection from a @orig diff --git a/drivers/hwtracing/coresight/coresight-tmc-etr.c b/drivers/hwtracing/coresight/coresight-tmc-etr.c index b29c2db94d96..ad991a37e2d2 100644 --- a/drivers/hwtracing/coresight/coresight-tmc-etr.c +++ b/drivers/hwtracing/coresight/coresight-tmc-etr.c @@ -255,6 +255,7 @@ void tmc_free_sg_table(struct tmc_sg_table *sg_table) tmc_free_table_pages(sg_table); tmc_free_data_pages(sg_table); } +EXPORT_SYMBOL_GPL(tmc_free_sg_table);
/* * Alloc pages for the table. Since this will be used by the device, @@ -340,6 +341,7 @@ struct tmc_sg_table *tmc_alloc_sg_table(struct device *dev,
return sg_table; } +EXPORT_SYMBOL_GPL(tmc_alloc_sg_table);
/* * tmc_sg_table_sync_data_range: Sync the data buffer written @@ -360,6 +362,7 @@ void tmc_sg_table_sync_data_range(struct tmc_sg_table *table, PAGE_SIZE, DMA_FROM_DEVICE); } } +EXPORT_SYMBOL_GPL(tmc_sg_table_sync_data_range);
/* tmc_sg_sync_table: Sync the page table */ void tmc_sg_table_sync_table(struct tmc_sg_table *sg_table) @@ -372,6 +375,7 @@ void tmc_sg_table_sync_table(struct tmc_sg_table *sg_table) dma_sync_single_for_device(real_dev, table_pages->daddrs[i], PAGE_SIZE, DMA_TO_DEVICE); } +EXPORT_SYMBOL_GPL(tmc_sg_table_sync_table);
/* * tmc_sg_table_get_data: Get the buffer pointer for data @offset @@ -401,6 +405,7 @@ ssize_t tmc_sg_table_get_data(struct tmc_sg_table *sg_table, *bufpp = page_address(data_pages->pages[pg_idx]) + pg_offset; return len; } +EXPORT_SYMBOL_GPL(tmc_sg_table_get_data);
#ifdef ETR_SG_DEBUG /* Map a dma address to virtual address */ @@ -766,6 +771,7 @@ tmc_etr_get_catu_device(struct tmc_drvdata *drvdata)
return NULL; } +EXPORT_SYMBOL_GPL(tmc_etr_get_catu_device);
static inline int tmc_etr_enable_catu(struct tmc_drvdata *drvdata, struct etr_buf *etr_buf) diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 310b1b825dd1..b7151c5f81b1 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -55,6 +55,7 @@ static struct list_head *stm_path; */ const u32 coresight_barrier_pkt[4] = { 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff}; +EXPORT_SYMBOL_GPL(coresight_barrier_pkt);
static int coresight_id_match(struct device *dev, void *data) { @@ -180,6 +181,7 @@ int coresight_claim_device_unlocked(void __iomem *base) coresight_clear_claim_tags(base); return -EBUSY; } +EXPORT_SYMBOL_GPL(coresight_claim_device_unlocked);
int coresight_claim_device(void __iomem *base) { @@ -191,6 +193,7 @@ int coresight_claim_device(void __iomem *base)
return rc; } +EXPORT_SYMBOL_GPL(coresight_claim_device);
/* * coresight_disclaim_device_unlocked : Clear the claim tags for the device. @@ -209,6 +212,7 @@ void coresight_disclaim_device_unlocked(void __iomem *base) */ WARN_ON_ONCE(1); } +EXPORT_SYMBOL_GPL(coresight_disclaim_device_unlocked);
void coresight_disclaim_device(void __iomem *base) { @@ -216,6 +220,7 @@ void coresight_disclaim_device(void __iomem *base) coresight_disclaim_device_unlocked(base); CS_LOCK(base); } +EXPORT_SYMBOL_GPL(coresight_disclaim_device);
/* enable or disable an associated CTI device of the supplied CS device */ static int @@ -468,6 +473,7 @@ void coresight_disable_path(struct list_head *path) { coresight_disable_path_from(path, NULL); } +EXPORT_SYMBOL_GPL(coresight_disable_path);
int coresight_enable_path(struct list_head *path, u32 mode, void *sink_data) { @@ -1377,6 +1383,7 @@ int coresight_timeout(void __iomem *addr, u32 offset, int position, int value)
return -EAGAIN; } +EXPORT_SYMBOL_GPL(coresight_timeout);
struct bus_type coresight_bustype = { .name = "coresight", @@ -1553,6 +1560,7 @@ bool coresight_loses_context_with_cpu(struct device *dev) return fwnode_property_present(dev_fwnode(dev), "arm,coresight-loses-context-with-cpu"); } +EXPORT_SYMBOL_GPL(coresight_loses_context_with_cpu);
/* * coresight_alloc_device_name - Get an index for a given device in the
When coresight device is in an active session, driver module of that device should not be removed. Use try_get_module() in coresight_grab_device() to prevent module to be unloaded. Use get_device()/put_device() to protect device data in the middle of active session.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Suggested-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight.c | 61 ++++++++++++++++++++++--- 1 file changed, 54 insertions(+), 7 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index b7151c5f81b1..0b0e31577b9b 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -634,13 +634,45 @@ struct coresight_device *coresight_get_sink_by_id(u32 id) return dev ? to_coresight_device(dev) : NULL; }
+/** + * coresight_get_ref- Helper function to increase reference count to module + * and device. + * Return true in successful case and power up the device. + * Return false when failed to get reference of module. + */ +static inline bool coresight_get_ref(struct coresight_device *csdev) +{ + struct device *dev = csdev->dev.parent; + + /* Make sure the driver can't be removed */ + if (!try_module_get(dev->driver->owner)) + return false; + /* Make sure the device can't go away */ + get_device(dev); + pm_runtime_get_sync(dev); + return true; +} + +/** + * coresight_put_ref- Helper function to decrease reference count to module + * and device. Power off the device. + */ +static inline void coresight_put_ref(struct coresight_device *csdev) +{ + struct device *dev = csdev->dev.parent; + + pm_runtime_put(dev); + put_device(dev); + module_put(dev->driver->owner); +} + /* * coresight_grab_device - Power up this device and any of the helper * devices connected to it for trace operation. Since the helper devices * don't appear on the trace path, they should be handled along with the * the master device. */ -static void coresight_grab_device(struct coresight_device *csdev) +static int coresight_grab_device(struct coresight_device *csdev) { int i;
@@ -649,9 +681,20 @@ static void coresight_grab_device(struct coresight_device *csdev)
child = csdev->pdata->conns[i].child_dev; if (child && child->type == CORESIGHT_DEV_TYPE_HELPER) - pm_runtime_get_sync(child->dev.parent); + if (!coresight_get_ref(child)) + goto err; + } + if (coresight_get_ref(csdev)) + return 0; +err: + for (i--; i >= 0; i--) { + struct coresight_device *child; + + child = csdev->pdata->conns[i].child_dev; + if (child && child->type == CORESIGHT_DEV_TYPE_HELPER) + coresight_put_ref(child); } - pm_runtime_get_sync(csdev->dev.parent); + return -ENODEV; }
/* @@ -662,13 +705,13 @@ static void coresight_drop_device(struct coresight_device *csdev) { int i;
- pm_runtime_put(csdev->dev.parent); + coresight_put_ref(csdev); for (i = 0; i < csdev->pdata->nr_outport; i++) { struct coresight_device *child;
child = csdev->pdata->conns[i].child_dev; if (child && child->type == CORESIGHT_DEV_TYPE_HELPER) - pm_runtime_put(child->dev.parent); + coresight_put_ref(child); } }
@@ -687,7 +730,7 @@ static int _coresight_build_path(struct coresight_device *csdev, struct coresight_device *sink, struct list_head *path) { - int i; + int i, ret; bool found = false; struct coresight_node *node;
@@ -721,7 +764,11 @@ static int _coresight_build_path(struct coresight_device *csdev, if (!node) return -ENOMEM;
- coresight_grab_device(csdev); + ret = coresight_grab_device(csdev); + if (ret) { + kfree(node); + return ret; + } node->csdev = csdev; list_add(&node->link, path);
On Fri, Aug 07, 2020 at 07:11:35PM +0800, Tingwei Zhang wrote:
When coresight device is in an active session, driver module of that device should not be removed. Use try_get_module() in coresight_grab_device() to prevent module to be unloaded. Use get_device()/put_device() to protect device data in the middle of active session.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Suggested-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/coresight.c | 61 ++++++++++++++++++++++--- 1 file changed, 54 insertions(+), 7 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index b7151c5f81b1..0b0e31577b9b 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -634,13 +634,45 @@ struct coresight_device *coresight_get_sink_by_id(u32 id) return dev ? to_coresight_device(dev) : NULL; } +/**
- coresight_get_ref- Helper function to increase reference count to module
- and device.
- Return true in successful case and power up the device.
- Return false when failed to get reference of module.
- */
+static inline bool coresight_get_ref(struct coresight_device *csdev) +{
- struct device *dev = csdev->dev.parent;
- /* Make sure the driver can't be removed */
- if (!try_module_get(dev->driver->owner))
return false;
- /* Make sure the device can't go away */
- get_device(dev);
- pm_runtime_get_sync(dev);
- return true;
+}
+/**
- coresight_put_ref- Helper function to decrease reference count to module
- and device. Power off the device.
- */
+static inline void coresight_put_ref(struct coresight_device *csdev) +{
- struct device *dev = csdev->dev.parent;
- pm_runtime_put(dev);
- put_device(dev);
- module_put(dev->driver->owner);
+}
/*
- coresight_grab_device - Power up this device and any of the helper
- devices connected to it for trace operation. Since the helper devices
- don't appear on the trace path, they should be handled along with the
- the master device.
*/ -static void coresight_grab_device(struct coresight_device *csdev) +static int coresight_grab_device(struct coresight_device *csdev) { int i; @@ -649,9 +681,20 @@ static void coresight_grab_device(struct coresight_device *csdev) child = csdev->pdata->conns[i].child_dev; if (child && child->type == CORESIGHT_DEV_TYPE_HELPER)
pm_runtime_get_sync(child->dev.parent);
if (!coresight_get_ref(child))
goto err;
- }
- if (coresight_get_ref(csdev))
return 0;
+err:
- for (i--; i >= 0; i--) {
struct coresight_device *child;
child = csdev->pdata->conns[i].child_dev;
if (child && child->type == CORESIGHT_DEV_TYPE_HELPER)
}coresight_put_ref(child);
- pm_runtime_get_sync(csdev->dev.parent);
- return -ENODEV;
} /* @@ -662,13 +705,13 @@ static void coresight_drop_device(struct coresight_device *csdev) { int i;
- pm_runtime_put(csdev->dev.parent);
- coresight_put_ref(csdev); for (i = 0; i < csdev->pdata->nr_outport; i++) { struct coresight_device *child;
child = csdev->pdata->conns[i].child_dev; if (child && child->type == CORESIGHT_DEV_TYPE_HELPER)
pm_runtime_put(child->dev.parent);
}coresight_put_ref(child);
} @@ -687,7 +730,7 @@ static int _coresight_build_path(struct coresight_device *csdev, struct coresight_device *sink, struct list_head *path) {
- int i;
- int i, ret; bool found = false; struct coresight_node *node;
@@ -721,7 +764,11 @@ static int _coresight_build_path(struct coresight_device *csdev, if (!node) return -ENOMEM;
- coresight_grab_device(csdev);
- ret = coresight_grab_device(csdev);
I suggest trying to grab the device before allocating memory for the node. If grabing the device fails memory doesn't have to be allocated and released needlessly.
I will likely come back to this patch later after reviewing the rest the this set.
Thanks, Mathieu
- if (ret) {
kfree(node);
return ret;
- } node->csdev = csdev; list_add(&node->link, path);
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Thu, Aug 13, 2020 at 03:36:12AM +0800, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:35PM +0800, Tingwei Zhang wrote:
When coresight device is in an active session, driver module of that device should not be removed. Use try_get_module() in coresight_grab_device() to prevent module to be unloaded. Use get_device()/put_device() to protect device data in the middle of active session.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Suggested-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/coresight.c | 61 ++++++++++++++++++++++--- 1 file changed, 54 insertions(+), 7 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c
b/drivers/hwtracing/coresight/coresight.c
index b7151c5f81b1..0b0e31577b9b 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -634,13 +634,45 @@ struct coresight_device
*coresight_get_sink_by_id(u32 id)
return dev ? to_coresight_device(dev) : NULL; } +/**
- coresight_get_ref- Helper function to increase reference count to
module
- and device.
- Return true in successful case and power up the device.
- Return false when failed to get reference of module.
- */
+static inline bool coresight_get_ref(struct coresight_device *csdev) +{
- struct device *dev = csdev->dev.parent;
- /* Make sure the driver can't be removed */
- if (!try_module_get(dev->driver->owner))
return false;
- /* Make sure the device can't go away */
- get_device(dev);
- pm_runtime_get_sync(dev);
- return true;
+}
+/**
- coresight_put_ref- Helper function to decrease reference count to
module
- and device. Power off the device.
- */
+static inline void coresight_put_ref(struct coresight_device *csdev) +{
- struct device *dev = csdev->dev.parent;
- pm_runtime_put(dev);
- put_device(dev);
- module_put(dev->driver->owner);
+}
/*
- coresight_grab_device - Power up this device and any of the helper
- devices connected to it for trace operation. Since the helper
devices
- don't appear on the trace path, they should be handled along with
the
- the master device.
*/ -static void coresight_grab_device(struct coresight_device *csdev) +static int coresight_grab_device(struct coresight_device *csdev) { int i; @@ -649,9 +681,20 @@ static void coresight_grab_device(struct
coresight_device *csdev)
child = csdev->pdata->conns[i].child_dev; if (child && child->type == CORESIGHT_DEV_TYPE_HELPER)
pm_runtime_get_sync(child->dev.parent);
if (!coresight_get_ref(child))
goto err;
- }
- if (coresight_get_ref(csdev))
return 0;
+err:
- for (i--; i >= 0; i--) {
struct coresight_device *child;
child = csdev->pdata->conns[i].child_dev;
if (child && child->type == CORESIGHT_DEV_TYPE_HELPER)
}coresight_put_ref(child);
- pm_runtime_get_sync(csdev->dev.parent);
- return -ENODEV;
} /* @@ -662,13 +705,13 @@ static void coresight_drop_device(struct
coresight_device *csdev)
{ int i;
- pm_runtime_put(csdev->dev.parent);
- coresight_put_ref(csdev); for (i = 0; i < csdev->pdata->nr_outport; i++) { struct coresight_device *child;
child = csdev->pdata->conns[i].child_dev; if (child && child->type == CORESIGHT_DEV_TYPE_HELPER)
pm_runtime_put(child->dev.parent);
}coresight_put_ref(child);
} @@ -687,7 +730,7 @@ static int _coresight_build_path(struct
coresight_device *csdev,
struct coresight_device *sink, struct list_head *path)
{
- int i;
- int i, ret; bool found = false; struct coresight_node *node;
@@ -721,7 +764,11 @@ static int _coresight_build_path(struct
coresight_device *csdev,
if (!node) return -ENOMEM;
- coresight_grab_device(csdev);
- ret = coresight_grab_device(csdev);
I suggest trying to grab the device before allocating memory for the node. If grabing the device fails memory doesn't have to be allocated and released needlessly.
Agree. I'll fix in next revision.
Thanks, Tingwei
I will likely come back to this patch later after reviewing the rest the this set.
Thanks, Mathieu
- if (ret) {
kfree(node);
return ret;
- } node->csdev = csdev; list_add(&node->link, path);
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project
- Kconfig becomes a tristate, to allow =m - add a stm_remove function, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/Kconfig | 5 ++++- drivers/hwtracing/coresight/coresight-stm.c | 20 +++++++++++++++++++- 2 files changed, 23 insertions(+), 2 deletions(-)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 4663fd1bbffc..6433f835fc97 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -86,7 +86,7 @@ config CORESIGHT_SOURCE_ETM4X data tracing may also be available.
config CORESIGHT_STM - bool "CoreSight System Trace Macrocell driver" + tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 select CORESIGHT_LINKS_AND_SINKS select STM @@ -96,6 +96,9 @@ config CORESIGHT_STM logging useful software events or data coming from various entities in the system, possibly running different OSs
+ To compile this driver as a module, choose M here: the + module will be called coresight-stm. + config CORESIGHT_CPU_DEBUG tristate "CoreSight CPU Debug driver" depends on ARM || ARM64 diff --git a/drivers/hwtracing/coresight/coresight-stm.c b/drivers/hwtracing/coresight/coresight-stm.c index 673d2f56ed1e..b74072e78436 100644 --- a/drivers/hwtracing/coresight/coresight-stm.c +++ b/drivers/hwtracing/coresight/coresight-stm.c @@ -948,6 +948,17 @@ static int stm_probe(struct amba_device *adev, const struct amba_id *id) return ret; }
+static int __exit stm_remove(struct amba_device *adev) +{ + struct stm_drvdata *drvdata = dev_get_drvdata(&adev->dev); + + coresight_unregister(drvdata->csdev); + + stm_unregister_device(&drvdata->stm); + + return 0; +} + #ifdef CONFIG_PM static int stm_runtime_suspend(struct device *dev) { @@ -980,6 +991,8 @@ static const struct amba_id stm_ids[] = { { 0, 0}, };
+MODULE_DEVICE_TABLE(amba, stm_ids); + static struct amba_driver stm_driver = { .drv = { .name = "coresight-stm", @@ -988,7 +1001,12 @@ static struct amba_driver stm_driver = { .suppress_bind_attrs = true, }, .probe = stm_probe, + .remove = stm_remove, .id_table = stm_ids, };
-builtin_amba_driver(stm_driver); +module_amba_driver(stm_driver); + +MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_DESCRIPTION("Arm CoreSight System Trace Macrocell driver"); +MODULE_LICENSE("GPL v2");
When coresight_build_path() fails on all the cpus, etm_setup_aux calls etm_free_aux() to free allocated event_data. WARN_ON(cpumask_empty(mask) will be triggered since cpu mask is empty. Check event_data->snk_config is not NULL first to avoid this warning.
Fixes: f5200aa9831f38 ("coresight: perf: Refactor function free_event_data()") Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight-etm-perf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index dcb0592418ae..3728c44e5763 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -126,10 +126,10 @@ static void free_sink_buffer(struct etm_event_data *event_data) cpumask_t *mask = &event_data->mask; struct coresight_device *sink;
- if (WARN_ON(cpumask_empty(mask))) + if (!event_data->snk_config) return;
- if (!event_data->snk_config) + if (WARN_ON(cpumask_empty(mask))) return;
cpu = cpumask_first(mask);
On Fri, Aug 07, 2020 at 07:11:37PM +0800, Tingwei Zhang wrote:
When coresight_build_path() fails on all the cpus, etm_setup_aux calls etm_free_aux() to free allocated event_data. WARN_ON(cpumask_empty(mask) will be triggered since cpu mask is empty. Check event_data->snk_config is not NULL first to avoid this warning.
Fixes: f5200aa9831f38 ("coresight: perf: Refactor function free_event_data()") Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
For patches 07 and 08:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
drivers/hwtracing/coresight/coresight-etm-perf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index dcb0592418ae..3728c44e5763 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -126,10 +126,10 @@ static void free_sink_buffer(struct etm_event_data *event_data) cpumask_t *mask = &event_data->mask; struct coresight_device *sink;
- if (WARN_ON(cpumask_empty(mask)))
- if (!event_data->snk_config) return;
- if (!event_data->snk_config)
- if (WARN_ON(cpumask_empty(mask))) return;
cpu = cpumask_first(mask); -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm3x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m - append -core to source file name to allow module to be called coresight-etm3x by the Makefile - add an etm_remove function, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org --- drivers/hwtracing/coresight/Kconfig | 5 +++- drivers/hwtracing/coresight/Makefile | 3 ++- ...resight-etm3x.c => coresight-etm3x-core.c} | 27 ++++++++++++++++++- 3 files changed, 32 insertions(+), 3 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm3x.c => coresight-etm3x-core.c} (97%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 6433f835fc97..8fd9fd139cf3 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -65,7 +65,7 @@ config CORESIGHT_SINK_ETBV10 special enhancement or added features.
config CORESIGHT_SOURCE_ETM3X - bool "CoreSight Embedded Trace Macrocell 3.x driver" + tristate "CoreSight Embedded Trace Macrocell 3.x driver" depends on !ARM64 select CORESIGHT_LINKS_AND_SINKS help @@ -74,6 +74,9 @@ config CORESIGHT_SOURCE_ETM3X This is primarily useful for instruction level tracing. Depending the ETM version data tracing may also be available.
+ To compile this driver as a module, choose M here: the + module will be called coresight-etm3x. + config CORESIGHT_SOURCE_ETM4X bool "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 19497d1d92bf..d619cfd0abd8 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -11,7 +11,8 @@ obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ coresight-replicator.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm-cp14.o \ +obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o +coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \ coresight-etm4x-sysfs.o diff --git a/drivers/hwtracing/coresight/coresight-etm3x.c b/drivers/hwtracing/coresight/coresight-etm3x-core.c similarity index 97% rename from drivers/hwtracing/coresight/coresight-etm3x.c rename to drivers/hwtracing/coresight/coresight-etm3x-core.c index bf22dcfd3327..82b333c40006 100644 --- a/drivers/hwtracing/coresight/coresight-etm3x.c +++ b/drivers/hwtracing/coresight/coresight-etm3x-core.c @@ -895,6 +895,23 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id) return ret; }
+static int __exit etm_remove(struct amba_device *adev) +{ + struct etm_drvdata *drvdata = dev_get_drvdata(&adev->dev); + + etm_perf_symlink(drvdata->csdev, false); + + if (--etm_count == 0) { + cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING); + if (hp_online) + cpuhp_remove_state_nocalls(hp_online); + } + + coresight_unregister(drvdata->csdev); + + return 0; +} + #ifdef CONFIG_PM static int etm_runtime_suspend(struct device *dev) { @@ -937,6 +954,8 @@ static const struct amba_id etm_ids[] = { { 0, 0}, };
+MODULE_DEVICE_TABLE(amba, etm_ids); + static struct amba_driver etm_driver = { .drv = { .name = "coresight-etm3x", @@ -945,6 +964,12 @@ static struct amba_driver etm_driver = { .suppress_bind_attrs = true, }, .probe = etm_probe, + .remove = etm_remove, .id_table = etm_ids, }; -builtin_amba_driver(etm_driver); +module_amba_driver(etm_driver); + +MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace driver"); +MODULE_LICENSE("GPL v2");
On Fri, Aug 07, 2020 at 07:11:38PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm3x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm3x by the Makefile
- add an etm_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +++- drivers/hwtracing/coresight/Makefile | 3 ++- ...resight-etm3x.c => coresight-etm3x-core.c} | 27 ++++++++++++++++++- 3 files changed, 32 insertions(+), 3 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm3x.c => coresight-etm3x-core.c} (97%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 6433f835fc97..8fd9fd139cf3 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -65,7 +65,7 @@ config CORESIGHT_SINK_ETBV10 special enhancement or added features. config CORESIGHT_SOURCE_ETM3X
- bool "CoreSight Embedded Trace Macrocell 3.x driver"
- tristate "CoreSight Embedded Trace Macrocell 3.x driver" depends on !ARM64 select CORESIGHT_LINKS_AND_SINKS help
@@ -74,6 +74,9 @@ config CORESIGHT_SOURCE_ETM3X This is primarily useful for instruction level tracing. Depending the ETM version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm3x.
config CORESIGHT_SOURCE_ETM4X bool "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 19497d1d92bf..d619cfd0abd8 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -11,7 +11,8 @@ obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ coresight-replicator.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm-cp14.o \ +obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o +coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \ coresight-etm4x-sysfs.o diff --git a/drivers/hwtracing/coresight/coresight-etm3x.c b/drivers/hwtracing/coresight/coresight-etm3x-core.c similarity index 97% rename from drivers/hwtracing/coresight/coresight-etm3x.c rename to drivers/hwtracing/coresight/coresight-etm3x-core.c index bf22dcfd3327..82b333c40006 100644 --- a/drivers/hwtracing/coresight/coresight-etm3x.c +++ b/drivers/hwtracing/coresight/coresight-etm3x-core.c @@ -895,6 +895,23 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id) return ret; } +static int __exit etm_remove(struct amba_device *adev) +{
Not sure adding __exit does anything special here since all the code is removed by the kernel anyway. The same comment goes for the other drivers in this set.
- struct etm_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- if (--etm_count == 0) {
cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING);
if (hp_online)
cpuhp_remove_state_nocalls(hp_online);
- }
- coresight_unregister(drvdata->csdev);
- return 0;
+}
#ifdef CONFIG_PM static int etm_runtime_suspend(struct device *dev) { @@ -937,6 +954,8 @@ static const struct amba_id etm_ids[] = { { 0, 0}, }; +MODULE_DEVICE_TABLE(amba, etm_ids);
static struct amba_driver etm_driver = { .drv = { .name = "coresight-etm3x", @@ -945,6 +964,12 @@ static struct amba_driver etm_driver = { .suppress_bind_attrs = true, }, .probe = etm_probe,
- .remove = etm_remove, .id_table = etm_ids,
}; -builtin_amba_driver(etm_driver); +module_amba_driver(etm_driver);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace driver");
That should be "Arm CoreSight Embedded Trace Macrocell 3.x driver"
With the above:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Thu, Aug 13, 2020 at 04:47:41AM +0800, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:38PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm3x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm3x by the Makefile
- add an etm_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +++- drivers/hwtracing/coresight/Makefile | 3 ++- ...resight-etm3x.c => coresight-etm3x-core.c} | 27 ++++++++++++++++++- 3 files changed, 32 insertions(+), 3 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm3x.c =>
coresight-etm3x-core.c} (97%)
diff --git a/drivers/hwtracing/coresight/Kconfig
b/drivers/hwtracing/coresight/Kconfig
index 6433f835fc97..8fd9fd139cf3 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -65,7 +65,7 @@ config CORESIGHT_SINK_ETBV10 special enhancement or added features. config CORESIGHT_SOURCE_ETM3X
- bool "CoreSight Embedded Trace Macrocell 3.x driver"
- tristate "CoreSight Embedded Trace Macrocell 3.x driver" depends on !ARM64 select CORESIGHT_LINKS_AND_SINKS help
@@ -74,6 +74,9 @@ config CORESIGHT_SOURCE_ETM3X This is primarily useful for instruction level tracing.
Depending
the ETM version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm3x.
config CORESIGHT_SOURCE_ETM4X bool "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 diff --git a/drivers/hwtracing/coresight/Makefile
b/drivers/hwtracing/coresight/Makefile
index 19497d1d92bf..d619cfd0abd8 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -11,7 +11,8 @@ obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ coresight-replicator.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o
coresight-etm-cp14.o \
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o +coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \ coresight-etm4x-sysfs.o diff --git a/drivers/hwtracing/coresight/coresight-etm3x.c
b/drivers/hwtracing/coresight/coresight-etm3x-core.c
similarity index 97% rename from drivers/hwtracing/coresight/coresight-etm3x.c rename to drivers/hwtracing/coresight/coresight-etm3x-core.c index bf22dcfd3327..82b333c40006 100644 --- a/drivers/hwtracing/coresight/coresight-etm3x.c +++ b/drivers/hwtracing/coresight/coresight-etm3x-core.c @@ -895,6 +895,23 @@ static int etm_probe(struct amba_device *adev,
const struct amba_id *id)
return ret; } +static int __exit etm_remove(struct amba_device *adev) +{
Not sure adding __exit does anything special here since all the code is removed by the kernel anyway. The same comment goes for the other drivers in this set.
We still need to remove any reference in other part of Kernel to etm device. For example, we remove ETM module but keep coresight module. ETM-perf and coresight-core still could access ETM device structure if etm_perf_symlink() and coresigh_unregister() are not called in etm_remove. We need clean any reference to ETM device structure up.
Thanks, Tingwei
- struct etm_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- if (--etm_count == 0) {
cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING);
if (hp_online)
cpuhp_remove_state_nocalls(hp_online);
- }
- coresight_unregister(drvdata->csdev);
- return 0;
+}
#ifdef CONFIG_PM static int etm_runtime_suspend(struct device *dev) { @@ -937,6 +954,8 @@ static const struct amba_id etm_ids[] = { { 0, 0}, }; +MODULE_DEVICE_TABLE(amba, etm_ids);
static struct amba_driver etm_driver = { .drv = { .name = "coresight-etm3x", @@ -945,6 +964,12 @@ static struct amba_driver etm_driver = { .suppress_bind_attrs = true, }, .probe = etm_probe,
- .remove = etm_remove, .id_table = etm_ids,
}; -builtin_amba_driver(etm_driver); +module_amba_driver(etm_driver);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace driver");
That should be "Arm CoreSight Embedded Trace Macrocell 3.x driver"
With the above:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project
On 08/07/2020 12:11 PM, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm3x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm3x by the Makefile
- add an etm_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +++- drivers/hwtracing/coresight/Makefile | 3 ++- ...resight-etm3x.c => coresight-etm3x-core.c} | 27 ++++++++++++++++++- 3 files changed, 32 insertions(+), 3 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm3x.c => coresight-etm3x-core.c} (97%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 6433f835fc97..8fd9fd139cf3 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -65,7 +65,7 @@ config CORESIGHT_SINK_ETBV10 special enhancement or added features. config CORESIGHT_SOURCE_ETM3X
- bool "CoreSight Embedded Trace Macrocell 3.x driver"
- tristate "CoreSight Embedded Trace Macrocell 3.x driver" depends on !ARM64 select CORESIGHT_LINKS_AND_SINKS help
@@ -74,6 +74,9 @@ config CORESIGHT_SOURCE_ETM3X This is primarily useful for instruction level tracing. Depending the ETM version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm3x.
- config CORESIGHT_SOURCE_ETM4X bool "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64
diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 19497d1d92bf..d619cfd0abd8 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -11,7 +11,8 @@ obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ coresight-replicator.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm-cp14.o \ +obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o +coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \ coresight-etm4x-sysfs.o diff --git a/drivers/hwtracing/coresight/coresight-etm3x.c b/drivers/hwtracing/coresight/coresight-etm3x-core.c similarity index 97% rename from drivers/hwtracing/coresight/coresight-etm3x.c rename to drivers/hwtracing/coresight/coresight-etm3x-core.c index bf22dcfd3327..82b333c40006 100644 --- a/drivers/hwtracing/coresight/coresight-etm3x.c +++ b/drivers/hwtracing/coresight/coresight-etm3x-core.c @@ -895,6 +895,23 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id) return ret; } +static int __exit etm_remove(struct amba_device *adev) +{
- struct etm_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- if (--etm_count == 0) {
cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING);
if (hp_online)
cpuhp_remove_state_nocalls(hp_online);
- }
As mentioned in the previous version, this needs fixing similar to the etm4x to avoid races with etm_count. I understand that it is an exisiting problem. But this set exposes it more due to etm_remove() calls and is much more probable to hit it.
Suzuki
On Thu, Aug 13, 2020 at 05:47:22PM +0800, Suzuki K Poulose wrote:
On 08/07/2020 12:11 PM, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm3x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm3x by the Makefile
- add an etm_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +++- drivers/hwtracing/coresight/Makefile | 3 ++- ...resight-etm3x.c => coresight-etm3x-core.c} | 27 ++++++++++++++++++- 3 files changed, 32 insertions(+), 3 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm3x.c => coresight-etm3x-core.c} (97%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 6433f835fc97..8fd9fd139cf3 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -65,7 +65,7 @@ config CORESIGHT_SINK_ETBV10 special enhancement or added features.
config CORESIGHT_SOURCE_ETM3X
- bool "CoreSight Embedded Trace Macrocell 3.x driver"
- tristate "CoreSight Embedded Trace Macrocell 3.x driver" depends on !ARM64 select CORESIGHT_LINKS_AND_SINKS help
@@ -74,6 +74,9 @@ config CORESIGHT_SOURCE_ETM3X This is primarily useful for instruction level tracing. Depending the ETM version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm3x.
config CORESIGHT_SOURCE_ETM4X bool "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 19497d1d92bf..d619cfd0abd8 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -11,7 +11,8 @@ obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ coresight-replicator.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm-cp14.o \ +obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o +coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \ coresight-etm4x-sysfs.o diff --git a/drivers/hwtracing/coresight/coresight-etm3x.c b/drivers/hwtracing/coresight/coresight-etm3x-core.c similarity index 97% rename from drivers/hwtracing/coresight/coresight-etm3x.c rename to drivers/hwtracing/coresight/coresight-etm3x-core.c index bf22dcfd3327..82b333c40006 100644 --- a/drivers/hwtracing/coresight/coresight-etm3x.c +++ b/drivers/hwtracing/coresight/coresight-etm3x-core.c @@ -895,6 +895,23 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id) return ret; }
+static int __exit etm_remove(struct amba_device *adev) +{
- struct etm_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- if (--etm_count == 0) {
cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING);
if (hp_online)
cpuhp_remove_state_nocalls(hp_online);
- }
As mentioned in the previous version, this needs fixing similar to the etm4x to avoid races with etm_count. I understand that it is an exisiting problem. But this set exposes it more due to etm_remove() calls and is much more probable to hit it.
OK. I'll fix this. Can you help to review ETM4 fix? If that's fine, I'll make similar fix here.
Unfortunately, I don't have any device with ETM3. Can anyone help to verify the function on ETM3? I'll really appreciate that.
Thanks, Tingwei
Suzuki
On Thu, 13 Aug 2020 at 04:57, Tingwei Zhang tingweiz@codeaurora.org wrote:
On Thu, Aug 13, 2020 at 05:47:22PM +0800, Suzuki K Poulose wrote:
On 08/07/2020 12:11 PM, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm3x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm3x by the Makefile
- add an etm_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +++- drivers/hwtracing/coresight/Makefile | 3 ++- ...resight-etm3x.c => coresight-etm3x-core.c} | 27 ++++++++++++++++++- 3 files changed, 32 insertions(+), 3 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm3x.c => coresight-etm3x-core.c} (97%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 6433f835fc97..8fd9fd139cf3 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -65,7 +65,7 @@ config CORESIGHT_SINK_ETBV10 special enhancement or added features.
config CORESIGHT_SOURCE_ETM3X
- bool "CoreSight Embedded Trace Macrocell 3.x driver"
- tristate "CoreSight Embedded Trace Macrocell 3.x driver" depends on !ARM64 select CORESIGHT_LINKS_AND_SINKS help
@@ -74,6 +74,9 @@ config CORESIGHT_SOURCE_ETM3X This is primarily useful for instruction level tracing. Depending the ETM version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm3x.
config CORESIGHT_SOURCE_ETM4X bool "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 19497d1d92bf..d619cfd0abd8 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -11,7 +11,8 @@ obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ coresight-replicator.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm-cp14.o \ +obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o +coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \ coresight-etm4x-sysfs.o diff --git a/drivers/hwtracing/coresight/coresight-etm3x.c b/drivers/hwtracing/coresight/coresight-etm3x-core.c similarity index 97% rename from drivers/hwtracing/coresight/coresight-etm3x.c rename to drivers/hwtracing/coresight/coresight-etm3x-core.c index bf22dcfd3327..82b333c40006 100644 --- a/drivers/hwtracing/coresight/coresight-etm3x.c +++ b/drivers/hwtracing/coresight/coresight-etm3x-core.c @@ -895,6 +895,23 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id) return ret; }
+static int __exit etm_remove(struct amba_device *adev) +{
- struct etm_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- if (--etm_count == 0) {
cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING);
if (hp_online)
cpuhp_remove_state_nocalls(hp_online);
- }
As mentioned in the previous version, this needs fixing similar to the etm4x to avoid races with etm_count. I understand that it is an exisiting problem. But this set exposes it more due to etm_remove() calls and is much more probable to hit it.
OK. I'll fix this. Can you help to review ETM4 fix? If that's fine, I'll make similar fix here.
Unfortunately, I don't have any device with ETM3. Can anyone help to verify the function on ETM3? I'll really appreciate that.
I'll take care of ETMv3. Please wait before sending a new revision since I am still reviewing the current set. I will let you know when I am done.
Thanks, Tingwei
Suzuki
On Thu, Aug 13, 2020 at 11:00:04AM -0600, Mathieu Poirier wrote:
On Thu, 13 Aug 2020 at 04:57, Tingwei Zhang tingweiz@codeaurora.org wrote:
On Thu, Aug 13, 2020 at 05:47:22PM +0800, Suzuki K Poulose wrote:
On 08/07/2020 12:11 PM, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm3x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm3x by the Makefile
- add an etm_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +++- drivers/hwtracing/coresight/Makefile | 3 ++- ...resight-etm3x.c => coresight-etm3x-core.c} | 27 ++++++++++++++++++- 3 files changed, 32 insertions(+), 3 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm3x.c => coresight-etm3x-core.c} (97%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 6433f835fc97..8fd9fd139cf3 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -65,7 +65,7 @@ config CORESIGHT_SINK_ETBV10 special enhancement or added features.
config CORESIGHT_SOURCE_ETM3X
- bool "CoreSight Embedded Trace Macrocell 3.x driver"
- tristate "CoreSight Embedded Trace Macrocell 3.x driver" depends on !ARM64 select CORESIGHT_LINKS_AND_SINKS help
@@ -74,6 +74,9 @@ config CORESIGHT_SOURCE_ETM3X This is primarily useful for instruction level tracing. Depending the ETM version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm3x.
config CORESIGHT_SOURCE_ETM4X bool "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 19497d1d92bf..d619cfd0abd8 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -11,7 +11,8 @@ obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ coresight-replicator.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm-cp14.o \ +obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o +coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \ coresight-etm4x-sysfs.o diff --git a/drivers/hwtracing/coresight/coresight-etm3x.c b/drivers/hwtracing/coresight/coresight-etm3x-core.c similarity index 97% rename from drivers/hwtracing/coresight/coresight-etm3x.c rename to drivers/hwtracing/coresight/coresight-etm3x-core.c index bf22dcfd3327..82b333c40006 100644 --- a/drivers/hwtracing/coresight/coresight-etm3x.c +++ b/drivers/hwtracing/coresight/coresight-etm3x-core.c @@ -895,6 +895,23 @@ static int etm_probe(struct amba_device *adev, const struct amba_id *id) return ret; }
+static int __exit etm_remove(struct amba_device *adev) +{
- struct etm_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- if (--etm_count == 0) {
cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING);
if (hp_online)
cpuhp_remove_state_nocalls(hp_online);
- }
As mentioned in the previous version, this needs fixing similar to the etm4x to avoid races with etm_count. I understand that it is an exisiting problem. But this set exposes it more due to etm_remove() calls and is much more probable to hit it.
OK. I'll fix this. Can you help to review ETM4 fix? If that's fine, I'll make similar fix here.
Unfortunately, I don't have any device with ETM3. Can anyone help to verify the function on ETM3? I'll really appreciate that.
I'll take care of ETMv3. Please wait before sending a new revision since I am still reviewing the current set. I will let you know when I am done.
Thanks a lot, Mathieu. I'll wait for your comments.
Thanks, Tingwei
Suzuki
linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m - append -core to source file name to allow module to be called coresight-etm4x by the Makefile - add an etm4_remove function, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot - protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org --- drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118 +++++++++++++----- 3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c => coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x.
config CORESIGHT_SOURCE_ETM4X - bool "CoreSight Embedded Trace Macrocell 4.x driver" + tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR @@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented version data tracing may also be available.
+ To compile this driver as a module, choose M here: the + module will be called coresight-etm4x. + config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \ - coresight-etm4x-sysfs.o +obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x-core.c similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted");
+static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config *config)
static int etm4_online_cpu(unsigned int cpu) { - if (!etmdrvdata[cpu]) + mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); + if (!etmdrvdata[cpu]) { + mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0; + }
if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable) coresight_enable(etmdrvdata[cpu]->csdev); + mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0; }
static int etm4_starting_cpu(unsigned int cpu) { - if (!etmdrvdata[cpu]) + mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); + if (!etmdrvdata[cpu]) { + mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0; + }
spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock) @@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock); + mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0; }
static int etm4_dying_cpu(unsigned int cpu) { - if (!etmdrvdata[cpu]) + mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); + if (!etmdrvdata[cpu]) { + mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0; + }
spin_lock(&etmdrvdata[cpu]->spinlock); if (local_read(&etmdrvdata[cpu]->mode)) etm4_disable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock); + mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0; }
@@ -1360,24 +1373,30 @@ static int etm4_cpu_pm_notify(struct notifier_block *nb, unsigned long cmd, { struct etmv4_drvdata *drvdata; unsigned int cpu = smp_processor_id(); + int ret = NOTIFY_OK;
+ mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); if (!etmdrvdata[cpu]) - return NOTIFY_OK; + goto out;
drvdata = etmdrvdata[cpu];
if (!drvdata->save_state) - return NOTIFY_OK; + goto out;
- if (WARN_ON_ONCE(drvdata->cpu != cpu)) - return NOTIFY_BAD; + if (WARN_ON_ONCE(drvdata->cpu != cpu)) { + ret = NOTIFY_BAD; + goto out; + }
switch (cmd) { case CPU_PM_ENTER: /* save the state if self-hosted coresight is in use */ if (local_read(&drvdata->mode)) - if (etm4_cpu_save(drvdata)) - return NOTIFY_BAD; + if (etm4_cpu_save(drvdata)) { + ret = NOTIFY_BAD; + goto out; + } break; case CPU_PM_EXIT: /* fallthrough */ @@ -1386,10 +1405,12 @@ static int etm4_cpu_pm_notify(struct notifier_block *nb, unsigned long cmd, etm4_cpu_restore(drvdata); break; default: - return NOTIFY_DONE; + goto out; }
- return NOTIFY_OK; +out: + mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); + return ret; }
static struct notifier_block etm4_cpu_pm_nb = { @@ -1430,7 +1451,7 @@ static int __init etm4_pm_setup(void) return ret; }
-static void __init etm4_pm_clear(void) +static void etm4_pm_clear(void) { cpu_pm_unregister_notifier(&etm4_cpu_pm_nb); cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING); @@ -1487,25 +1508,20 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) if (!desc.name) return -ENOMEM;
- etmdrvdata[drvdata->cpu] = drvdata; - if (smp_call_function_single(drvdata->cpu, etm4_init_arch_data, drvdata, 1)) dev_err(dev, "ETM arch init failed\n");
- if (etm4_arch_supported(drvdata->arch) == false) { - ret = -EINVAL; - goto err_arch_supported; - } + if (etm4_arch_supported(drvdata->arch) == false) + return -EINVAL;
etm4_init_trace_id(drvdata); etm4_set_default(&drvdata->config);
pdata = coresight_get_platform_data(dev); - if (IS_ERR(pdata)) { - ret = PTR_ERR(pdata); - goto err_arch_supported; - } + if (IS_ERR(pdata)) + return PTR_ERR(pdata); + adev->dev.platform_data = pdata;
desc.type = CORESIGHT_DEV_TYPE_SOURCE; @@ -1515,17 +1531,19 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) desc.dev = dev; desc.groups = coresight_etmv4_groups; drvdata->csdev = coresight_register(&desc); - if (IS_ERR(drvdata->csdev)) { - ret = PTR_ERR(drvdata->csdev); - goto err_arch_supported; - } + if (IS_ERR(drvdata->csdev)) + return PTR_ERR(drvdata->csdev);
ret = etm_perf_symlink(drvdata->csdev, true); if (ret) { coresight_unregister(drvdata->csdev); - goto err_arch_supported; + return ret; }
+ mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu)); + etmdrvdata[drvdata->cpu] = drvdata; + mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu)); + pm_runtime_put(&adev->dev); dev_info(&drvdata->csdev->dev, "CPU%d: ETM v%d.%d initialized\n", drvdata->cpu, drvdata->arch >> 4, drvdata->arch & 0xf); @@ -1536,10 +1554,6 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) }
return 0; - -err_arch_supported: - etmdrvdata[drvdata->cpu] = NULL; - return ret; }
static struct amba_cs_uci_id uci_id_etm4[] = { @@ -1551,6 +1565,22 @@ static struct amba_cs_uci_id uci_id_etm4[] = { } };
+static int __exit etm4_remove(struct amba_device *adev) +{ + struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev); + + etm_perf_symlink(drvdata->csdev, false); + + mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu)); + etmdrvdata[drvdata->cpu] = NULL; + mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu)); + + coresight_unregister(drvdata->csdev); + + return 0; +} + + static const struct amba_id etm4_ids[] = { CS_AMBA_ID(0x000bb95d), /* Cortex-A53 */ CS_AMBA_ID(0x000bb95e), /* Cortex-A57 */ @@ -1568,18 +1598,26 @@ static const struct amba_id etm4_ids[] = { {}, };
+MODULE_DEVICE_TABLE(amba, etm4_ids); + static struct amba_driver etm4x_driver = { .drv = { .name = "coresight-etm4x", + .owner = THIS_MODULE, .suppress_bind_attrs = true, }, .probe = etm4_probe, + .remove = etm4_remove, .id_table = etm4_ids, };
static int __init etm4x_init(void) { int ret; + int cpu; + + for_each_possible_cpu(cpu) + mutex_init(&per_cpu(etmdrvdata_lock, cpu));
ret = etm4_pm_setup();
@@ -1595,4 +1633,20 @@ static int __init etm4x_init(void)
return ret; } -device_initcall(etm4x_init); + +static void __exit etm4x_exit(void) +{ + int cpu; + + amba_driver_unregister(&etm4x_driver); + etm4_pm_clear(); + for_each_possible_cpu(cpu) + mutex_destroy(&per_cpu(etmdrvdata_lock, cpu)); +} +module_init(etm4x_init); +module_exit(etm4x_exit); + +MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace v4 driver"); +MODULE_LICENSE("GPL v2");
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118 +++++++++++++----- 3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c => coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x-core.c similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted"); +static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config *config) static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable) coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock) @@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
Ouch!
A mutex wrapping a spinlock to protect the exact same drvdata structure. Because etmdrvdata_lock[] is a per cpu variable I suggest to get rid of drvdata::spinlock altogether. To do so:
1) In etm4_enable_sysfs(), get rid of the spinlock/unlock(). 2) In etm4_enable_hw_smp_call(), grab the mutex based on the value of drvdata->cpu and set drvdata->stick_enable based on the return value of etm4_enable_hw() while holding the mutex. 3) In etm4_disable_sysfs(), just change spinlock/unlock to mutex operations. 4) Get rid of spinlock operations in etm4_starting/dying_cpu().
As Suzuki pointed out the same kind of gymnastic needs to be done for ETMv3.
return 0; } static int etm4_dying_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (local_read(&etmdrvdata[cpu]->mode)) etm4_disable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} @@ -1360,24 +1373,30 @@ static int etm4_cpu_pm_notify(struct notifier_block *nb, unsigned long cmd, { struct etmv4_drvdata *drvdata; unsigned int cpu = smp_processor_id();
- int ret = NOTIFY_OK;
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); if (!etmdrvdata[cpu])
return NOTIFY_OK;
goto out;
drvdata = etmdrvdata[cpu]; if (!drvdata->save_state)
return NOTIFY_OK;
goto out;
- if (WARN_ON_ONCE(drvdata->cpu != cpu))
return NOTIFY_BAD;
- if (WARN_ON_ONCE(drvdata->cpu != cpu)) {
ret = NOTIFY_BAD;
goto out;
- }
switch (cmd) { case CPU_PM_ENTER: /* save the state if self-hosted coresight is in use */ if (local_read(&drvdata->mode))
if (etm4_cpu_save(drvdata))
return NOTIFY_BAD;
if (etm4_cpu_save(drvdata)) {
ret = NOTIFY_BAD;
goto out;
break; case CPU_PM_EXIT: /* fallthrough */}
@@ -1386,10 +1405,12 @@ static int etm4_cpu_pm_notify(struct notifier_block *nb, unsigned long cmd, etm4_cpu_restore(drvdata); break; default:
return NOTIFY_DONE;
}goto out;
- return NOTIFY_OK;
+out:
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- return ret;
} static struct notifier_block etm4_cpu_pm_nb = { @@ -1430,7 +1451,7 @@ static int __init etm4_pm_setup(void) return ret; } -static void __init etm4_pm_clear(void) +static void etm4_pm_clear(void) { cpu_pm_unregister_notifier(&etm4_cpu_pm_nb); cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING); @@ -1487,25 +1508,20 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) if (!desc.name) return -ENOMEM;
- etmdrvdata[drvdata->cpu] = drvdata;
- if (smp_call_function_single(drvdata->cpu, etm4_init_arch_data, drvdata, 1)) dev_err(dev, "ETM arch init failed\n");
- if (etm4_arch_supported(drvdata->arch) == false) {
ret = -EINVAL;
goto err_arch_supported;
- }
- if (etm4_arch_supported(drvdata->arch) == false)
return -EINVAL;
etm4_init_trace_id(drvdata); etm4_set_default(&drvdata->config); pdata = coresight_get_platform_data(dev);
- if (IS_ERR(pdata)) {
ret = PTR_ERR(pdata);
goto err_arch_supported;
- }
- if (IS_ERR(pdata))
return PTR_ERR(pdata);
- adev->dev.platform_data = pdata;
desc.type = CORESIGHT_DEV_TYPE_SOURCE; @@ -1515,17 +1531,19 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) desc.dev = dev; desc.groups = coresight_etmv4_groups; drvdata->csdev = coresight_register(&desc);
- if (IS_ERR(drvdata->csdev)) {
ret = PTR_ERR(drvdata->csdev);
goto err_arch_supported;
- }
- if (IS_ERR(drvdata->csdev))
return PTR_ERR(drvdata->csdev);
ret = etm_perf_symlink(drvdata->csdev, true); if (ret) { coresight_unregister(drvdata->csdev);
goto err_arch_supported;
}return ret;
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = drvdata;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- pm_runtime_put(&adev->dev); dev_info(&drvdata->csdev->dev, "CPU%d: ETM v%d.%d initialized\n", drvdata->cpu, drvdata->arch >> 4, drvdata->arch & 0xf);
@@ -1536,10 +1554,6 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) } return 0;
-err_arch_supported:
- etmdrvdata[drvdata->cpu] = NULL;
- return ret;
} static struct amba_cs_uci_id uci_id_etm4[] = { @@ -1551,6 +1565,22 @@ static struct amba_cs_uci_id uci_id_etm4[] = { } }; +static int __exit etm4_remove(struct amba_device *adev) +{
- struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = NULL;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- coresight_unregister(drvdata->csdev);
- return 0;
+}
static const struct amba_id etm4_ids[] = { CS_AMBA_ID(0x000bb95d), /* Cortex-A53 */ CS_AMBA_ID(0x000bb95e), /* Cortex-A57 */ @@ -1568,18 +1598,26 @@ static const struct amba_id etm4_ids[] = { {}, }; +MODULE_DEVICE_TABLE(amba, etm4_ids);
static struct amba_driver etm4x_driver = { .drv = { .name = "coresight-etm4x",
.suppress_bind_attrs = true, }, .probe = etm4_probe,.owner = THIS_MODULE,
- .remove = etm4_remove, .id_table = etm4_ids,
}; static int __init etm4x_init(void) { int ret;
- int cpu;
- for_each_possible_cpu(cpu)
mutex_init(&per_cpu(etmdrvdata_lock, cpu));
ret = etm4_pm_setup(); @@ -1595,4 +1633,20 @@ static int __init etm4x_init(void) return ret; } -device_initcall(etm4x_init);
+static void __exit etm4x_exit(void) +{
- int cpu;
- amba_driver_unregister(&etm4x_driver);
- etm4_pm_clear();
- for_each_possible_cpu(cpu)
mutex_destroy(&per_cpu(etmdrvdata_lock, cpu));
+} +module_init(etm4x_init); +module_exit(etm4x_exit);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace v4 driver");
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On 08/13/2020 08:39 PM, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118 +++++++++++++----- 3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c => coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
- config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64
diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x-core.c similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted"); +static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config *config) static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable) coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0; }
static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock) @@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
Ouch!
A mutex wrapping a spinlock to protect the exact same drvdata structure.
Actually this mutex is for "protecting" etmdrvdata[cpu], not the drvdata struct as such. But as you said, it becomes like a double lock.
Having mutex is an overkill for sure and can't be used replace spin_lock, especially if needed from atomic context. Having looked at the code, we only ever access/modify the etmdrvdata[cpu] on a different CPU is when we probe and there are chances of races. One of such is described here
http://lists.infradead.org/pipermail/linux-arm-kernel/2020-July/589941.html
I will see if I can spin a couple of patches to address these. Once we make sure that the etmdrvdata[cpu] is only modified by "cpu" then we don't need this mutex and stick with what we have.
Suzuki
On Fri, Aug 14, 2020 at 07:46:12AM +0800, Suzuki K Poulose wrote:
On 08/13/2020 08:39 PM, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118
+++++++++++++-----
3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c =>
coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig
b/drivers/hwtracing/coresight/Kconfig
index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented
version
data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 diff --git a/drivers/hwtracing/coresight/Makefile
b/drivers/hwtracing/coresight/Makefile
index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) +=
coresight-funnel.o \
obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c
b/drivers/hwtracing/coresight/coresight-etm4x-core.c
similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted"); +static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config
*config)
static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- } if (etmdrvdata[cpu]->boot_enable &&
!etmdrvdata[cpu]->sticky_enable)
coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- } spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock)
@@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
Ouch!
A mutex wrapping a spinlock to protect the exact same drvdata structure.
Actually this mutex is for "protecting" etmdrvdata[cpu], not the drvdata struct as such. But as you said, it becomes like a double lock.
Having mutex is an overkill for sure and can't be used replace spin_lock, especially if needed from atomic context. Having looked at the code, we only ever access/modify the etmdrvdata[cpu] on a different CPU is when we probe and there are chances of races. One of such is described here
http://lists.infradead.org/pipermail/linux-arm-kernel/2020-July/589941.htm l
I will see if I can spin a couple of patches to address these. Once we make sure that the etmdrvdata[cpu] is only modified by "cpu" then we don't need this mutex and stick with what we have.
Suzuki
With Suzuki's idea, I made some change to remove mutex lock and change etmdrvdata[i] on CPU i. I didn't change the part in probe to assign drvdata to etmdrvdata. That's after all drvdata is prepared and coresight device is registered. Once hotplug/PM call back see the value, it can use it directly. If we have racing in probe and these call backs, the worse case is call backs finds out etmdrvdata is NULL and return immediately. Does this make sense?
static void __exit clear_etmdrvdata(void *info) { int cpu = *(int *)info; etmdrvdata[cpu] = NULL; }
static int __exit etm4_remove(struct amba_device *adev) { struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
etm_perf_symlink(drvdata->csdev, false);
/* * Taking hotplug lock here to avoid racing between etm4_remove and * CPU hotplug call backs. */ cpus_read_lock(); if (cpu_online(drvdata->cpu)) /* * The readers for etmdrvdata[] are CPU hotplug call backs * and PM notification call backs. Change etmdrvdata[i] on * CPU i ensures these call backs has consistent view * inside one call back function. */ smp_call_function_single(drvdata->cpu, clear_etmdrvdata, &drvdata->cpu, 1); else etmdrvdata[drvdata->cpu] = NULL;
cpus_read_unlock();
coresight_unregister(drvdata->csdev);
return 0; }
Thanks, Tingwei
linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
On Fri, Aug 14, 2020 at 05:52:24PM +0800, Tingwei Zhang wrote:
On Fri, Aug 14, 2020 at 07:46:12AM +0800, Suzuki K Poulose wrote:
On 08/13/2020 08:39 PM, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118
+++++++++++++-----
3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c =>
coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig
b/drivers/hwtracing/coresight/Kconfig
index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented
version
data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 diff --git a/drivers/hwtracing/coresight/Makefile
b/drivers/hwtracing/coresight/Makefile
index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) +=
coresight-funnel.o \
obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c
b/drivers/hwtracing/coresight/coresight-etm4x-core.c
similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted"); +static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config
*config)
static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- } if (etmdrvdata[cpu]->boot_enable &&
!etmdrvdata[cpu]->sticky_enable)
coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- } spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock)
@@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
Ouch!
A mutex wrapping a spinlock to protect the exact same drvdata structure.
Actually this mutex is for "protecting" etmdrvdata[cpu], not the drvdata struct as such. But as you said, it becomes like a double lock.
Having mutex is an overkill for sure and can't be used replace spin_lock, especially if needed from atomic context. Having looked at the code, we only ever access/modify the etmdrvdata[cpu] on a different CPU is when we probe and there are chances of races. One of such is described here
http://lists.infradead.org/pipermail/linux-arm-kernel/2020-July/589941.htm l
I will see if I can spin a couple of patches to address these. Once we make sure that the etmdrvdata[cpu] is only modified by "cpu" then we don't need this mutex and stick with what we have.
Suzuki
With Suzuki's idea, I made some change to remove mutex lock and change etmdrvdata[i] on CPU i. I didn't change the part in probe to assign drvdata to etmdrvdata. That's after all drvdata is prepared and coresight device is registered. Once hotplug/PM call back see the value, it can use it directly. If we have racing in probe and these call backs, the worse case is call backs finds out etmdrvdata is NULL and return immediately. Does this make sense?
static void __exit clear_etmdrvdata(void *info) { int cpu = *(int *)info; etmdrvdata[cpu] = NULL; }
static int __exit etm4_remove(struct amba_device *adev) { struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
etm_perf_symlink(drvdata->csdev, false); /* * Taking hotplug lock here to avoid racing between etm4_remove and * CPU hotplug call backs. */ cpus_read_lock(); if (cpu_online(drvdata->cpu)) /* * The readers for etmdrvdata[] are CPU hotplug call backs * and PM notification call backs. Change etmdrvdata[i] on * CPU i ensures these call backs has consistent view * inside one call back function. */ smp_call_function_single(drvdata->cpu, clear_etmdrvdata, &drvdata->cpu, 1);
You should check the error here to confirm if this was really complete. Otherwise, fallback to the manual clearing here.
else etmdrvdata[drvdata->cpu] = NULL; cpus_read_unlock(); coresight_unregister(drvdata->csdev); return 0;
}
Yes, this is exactly what I was looking for. Additionally we should fix the races mentioned in the link above. I believe, we can solve that by the following :
---8>---
coresight: etm4x: Delay advertising per-cpu drvdata
Delay advertising the per-cpu etmdrvdata until we have successfully initialised. This is to avoid races and unnecessary save/restore of the ETM state, before the device is properly setup.
Signed-off-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight-etm4x.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x.c index 6d7d2169bfb2..30f7ffa07eb6 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x.c @@ -1499,8 +1499,6 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) return -ENOMEM;
cpus_read_lock(); - etmdrvdata[drvdata->cpu] = drvdata; - if (smp_call_function_single(drvdata->cpu, etm4_init_arch_data, drvdata, 1)) dev_err(dev, "ETM arch init failed\n"); @@ -1509,10 +1507,8 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) cpus_read_unlock();
/* etm4_pm_setup_cpuslocked() does its own cleanup - exit on error */ - if (ret) { - etmdrvdata[drvdata->cpu] = NULL; + if (ret) return ret; - }
if (etm4_arch_supported(drvdata->arch) == false) { ret = -EINVAL; @@ -1547,6 +1543,8 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) goto err_arch_supported; }
+ /* Advertise this after we have successfully initialised */ + etmdrvdata[drvdata->cpu] = drvdata; pm_runtime_put(&adev->dev); dev_info(&drvdata->csdev->dev, "CPU%d: ETM v%d.%d initialized\n", drvdata->cpu, drvdata->arch >> 4, drvdata->arch & 0xf); @@ -1559,7 +1557,6 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) return 0;
err_arch_supported: - etmdrvdata[drvdata->cpu] = NULL; etm4_pm_clear(); return ret; }
On Fri, Aug 14, 2020 at 06:42:01PM +0800, Suzuki K Poulose wrote:
On Fri, Aug 14, 2020 at 05:52:24PM +0800, Tingwei Zhang wrote:
On Fri, Aug 14, 2020 at 07:46:12AM +0800, Suzuki K Poulose wrote:
On 08/13/2020 08:39 PM, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118
+++++++++++++-----
3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c =>
coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig
b/drivers/hwtracing/coresight/Kconfig
index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the
implemented
version
data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T))
|| ARM64
diff --git a/drivers/hwtracing/coresight/Makefile
b/drivers/hwtracing/coresight/Makefile
index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) +=
coresight-funnel.o \
obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c
b/drivers/hwtracing/coresight/coresight-etm4x-core.c
similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 =
self-hosted");
+static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct
etmv4_config
*config)
static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- } if (etmdrvdata[cpu]->boot_enable &&
!etmdrvdata[cpu]->sticky_enable)
coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- } spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock)
@@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int
cpu)
if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
Ouch!
A mutex wrapping a spinlock to protect the exact same drvdata
structure.
Actually this mutex is for "protecting" etmdrvdata[cpu], not the drvdata struct as such. But as you said, it becomes like a double
lock.
Having mutex is an overkill for sure and can't be used replace spin_lock, especially if needed from atomic context. Having looked at the code, we only ever access/modify the etmdrvdata[cpu] on a different CPU is when we probe and there are chances of races. One of such is described here
http://lists.infradead.org/pipermail/linux-arm-kernel/2020-July/589941.htm
l
I will see if I can spin a couple of patches to address these. Once we make sure that the etmdrvdata[cpu] is only modified by "cpu" then we don't need this mutex and stick with what we have.
Suzuki
With Suzuki's idea, I made some change to remove mutex lock and change etmdrvdata[i] on CPU i. I didn't change the part in probe to assign drvdata to etmdrvdata. That's after all drvdata is prepared and coresight device is registered. Once hotplug/PM call back see the value, it can use it directly. If we have racing in probe and these call backs, the worse case is call backs finds out etmdrvdata is NULL and return immediately. Does this make sense?
static void __exit clear_etmdrvdata(void *info) { int cpu = *(int *)info; etmdrvdata[cpu] = NULL; }
static int __exit etm4_remove(struct amba_device *adev) { struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
etm_perf_symlink(drvdata->csdev, false); /* * Taking hotplug lock here to avoid racing between etm4_remove
and
* CPU hotplug call backs. */ cpus_read_lock(); if (cpu_online(drvdata->cpu)) /* * The readers for etmdrvdata[] are CPU hotplug call
backs
* and PM notification call backs. Change etmdrvdata[i]
on
* CPU i ensures these call backs has consistent view * inside one call back function. */ smp_call_function_single(drvdata->cpu, clear_etmdrvdata,
&drvdata->cpu, 1);
You should check the error here to confirm if this was really complete. Otherwise, fallback to the manual clearing here.
Yes. I don't need to check cpu_online since it's checked in smp_call_function_single(). I can just check return value and fallback to manual clearing.
else etmdrvdata[drvdata->cpu] = NULL; cpus_read_unlock(); coresight_unregister(drvdata->csdev); return 0;
}
Yes, this is exactly what I was looking for. Additionally we should fix the races mentioned in the link above. I believe, we can solve that by the following :
I already did the same thing in this patch if you ignore the mutex part. :)
Thanks, Tingwei
---8>---
coresight: etm4x: Delay advertising per-cpu drvdata
Delay advertising the per-cpu etmdrvdata until we have successfully initialised. This is to avoid races and unnecessary save/restore of the ETM state, before the device is properly setup.
Signed-off-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/coresight-etm4x.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x.c index 6d7d2169bfb2..30f7ffa07eb6 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x.c @@ -1499,8 +1499,6 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) return -ENOMEM; cpus_read_lock();
- etmdrvdata[drvdata->cpu] = drvdata;
- if (smp_call_function_single(drvdata->cpu, etm4_init_arch_data, drvdata, 1)) dev_err(dev, "ETM arch init failed\n");
@@ -1509,10 +1507,8 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) cpus_read_unlock(); /* etm4_pm_setup_cpuslocked() does its own cleanup - exit on error */
- if (ret) {
etmdrvdata[drvdata->cpu] = NULL;
- if (ret) return ret;
- }
if (etm4_arch_supported(drvdata->arch) == false) { ret = -EINVAL; @@ -1547,6 +1543,8 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) goto err_arch_supported; }
- /* Advertise this after we have successfully initialised */
- etmdrvdata[drvdata->cpu] = drvdata; pm_runtime_put(&adev->dev); dev_info(&drvdata->csdev->dev, "CPU%d: ETM v%d.%d initialized\n", drvdata->cpu, drvdata->arch >> 4, drvdata->arch & 0xf);
@@ -1559,7 +1557,6 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) return 0; err_arch_supported:
- etmdrvdata[drvdata->cpu] = NULL; etm4_pm_clear(); return ret;
}
2.24.1
On 08/14/2020 12:00 PM, Tingwei Zhang wrote:
On Fri, Aug 14, 2020 at 06:42:01PM +0800, Suzuki K Poulose wrote:
On Fri, Aug 14, 2020 at 05:52:24PM +0800, Tingwei Zhang wrote:
On Fri, Aug 14, 2020 at 07:46:12AM +0800, Suzuki K Poulose wrote:
On 08/13/2020 08:39 PM, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
@@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int
cpu)
if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
Ouch!
A mutex wrapping a spinlock to protect the exact same drvdata
structure.
Actually this mutex is for "protecting" etmdrvdata[cpu], not the drvdata struct as such. But as you said, it becomes like a double
lock.
Having mutex is an overkill for sure and can't be used replace spin_lock, especially if needed from atomic context. Having looked at the code, we only ever access/modify the etmdrvdata[cpu] on a different CPU is when we probe and there are chances of races. One of such is described here
http://lists.infradead.org/pipermail/linux-arm-kernel/2020-July/589941.htm
l
I will see if I can spin a couple of patches to address these. Once we make sure that the etmdrvdata[cpu] is only modified by "cpu" then we don't need this mutex and stick with what we have.
Suzuki
With Suzuki's idea, I made some change to remove mutex lock and change etmdrvdata[i] on CPU i. I didn't change the part in probe to assign drvdata to etmdrvdata. That's after all drvdata is prepared and coresight device is registered. Once hotplug/PM call back see the value, it can use it directly. If we have racing in probe and these call backs, the worse case is call backs finds out etmdrvdata is NULL and return immediately. Does this make sense?
static void __exit clear_etmdrvdata(void *info) { int cpu = *(int *)info; etmdrvdata[cpu] = NULL; }
static int __exit etm4_remove(struct amba_device *adev) { struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
etm_perf_symlink(drvdata->csdev, false); /* * Taking hotplug lock here to avoid racing between etm4_remove
and
* CPU hotplug call backs. */ cpus_read_lock(); if (cpu_online(drvdata->cpu)) /* * The readers for etmdrvdata[] are CPU hotplug call
backs
* and PM notification call backs. Change etmdrvdata[i]
on
* CPU i ensures these call backs has consistent view * inside one call back function. */ smp_call_function_single(drvdata->cpu, clear_etmdrvdata,
&drvdata->cpu, 1);
You should check the error here to confirm if this was really complete. Otherwise, fallback to the manual clearing here.
Yes. I don't need to check cpu_online since it's checked in smp_call_function_single(). I can just check return value and fallback to manual clearing.
else etmdrvdata[drvdata->cpu] = NULL; cpus_read_unlock(); coresight_unregister(drvdata->csdev); return 0;
}
Yes, this is exactly what I was looking for. Additionally we should fix the races mentioned in the link above. I believe, we can solve that by the following :
I already did the same thing in this patch if you ignore the mutex part. :)
Oh, yes. I see that now :-)
Cheers Suzuki
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118 +++++++++++++----- 3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c => coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x-core.c similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted"); +static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config *config) static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable) coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock) @@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_dying_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (local_read(&etmdrvdata[cpu]->mode)) etm4_disable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} @@ -1360,24 +1373,30 @@ static int etm4_cpu_pm_notify(struct notifier_block *nb, unsigned long cmd, { struct etmv4_drvdata *drvdata; unsigned int cpu = smp_processor_id();
- int ret = NOTIFY_OK;
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); if (!etmdrvdata[cpu])
return NOTIFY_OK;
goto out;
drvdata = etmdrvdata[cpu]; if (!drvdata->save_state)
return NOTIFY_OK;
goto out;
- if (WARN_ON_ONCE(drvdata->cpu != cpu))
return NOTIFY_BAD;
- if (WARN_ON_ONCE(drvdata->cpu != cpu)) {
ret = NOTIFY_BAD;
goto out;
- }
switch (cmd) { case CPU_PM_ENTER: /* save the state if self-hosted coresight is in use */ if (local_read(&drvdata->mode))
if (etm4_cpu_save(drvdata))
return NOTIFY_BAD;
if (etm4_cpu_save(drvdata)) {
ret = NOTIFY_BAD;
goto out;
break; case CPU_PM_EXIT: /* fallthrough */}
@@ -1386,10 +1405,12 @@ static int etm4_cpu_pm_notify(struct notifier_block *nb, unsigned long cmd, etm4_cpu_restore(drvdata); break; default:
return NOTIFY_DONE;
}goto out;
- return NOTIFY_OK;
+out:
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- return ret;
} static struct notifier_block etm4_cpu_pm_nb = { @@ -1430,7 +1451,7 @@ static int __init etm4_pm_setup(void) return ret; } -static void __init etm4_pm_clear(void) +static void etm4_pm_clear(void) { cpu_pm_unregister_notifier(&etm4_cpu_pm_nb); cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING); @@ -1487,25 +1508,20 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) if (!desc.name) return -ENOMEM;
- etmdrvdata[drvdata->cpu] = drvdata;
- if (smp_call_function_single(drvdata->cpu, etm4_init_arch_data, drvdata, 1)) dev_err(dev, "ETM arch init failed\n");
- if (etm4_arch_supported(drvdata->arch) == false) {
ret = -EINVAL;
goto err_arch_supported;
- }
- if (etm4_arch_supported(drvdata->arch) == false)
return -EINVAL;
etm4_init_trace_id(drvdata); etm4_set_default(&drvdata->config); pdata = coresight_get_platform_data(dev);
- if (IS_ERR(pdata)) {
ret = PTR_ERR(pdata);
goto err_arch_supported;
- }
- if (IS_ERR(pdata))
return PTR_ERR(pdata);
- adev->dev.platform_data = pdata;
desc.type = CORESIGHT_DEV_TYPE_SOURCE; @@ -1515,17 +1531,19 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) desc.dev = dev; desc.groups = coresight_etmv4_groups; drvdata->csdev = coresight_register(&desc);
- if (IS_ERR(drvdata->csdev)) {
ret = PTR_ERR(drvdata->csdev);
goto err_arch_supported;
- }
- if (IS_ERR(drvdata->csdev))
return PTR_ERR(drvdata->csdev);
ret = etm_perf_symlink(drvdata->csdev, true); if (ret) { coresight_unregister(drvdata->csdev);
goto err_arch_supported;
}return ret;
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = drvdata;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- pm_runtime_put(&adev->dev); dev_info(&drvdata->csdev->dev, "CPU%d: ETM v%d.%d initialized\n", drvdata->cpu, drvdata->arch >> 4, drvdata->arch & 0xf);
@@ -1536,10 +1554,6 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) } return 0;
-err_arch_supported:
- etmdrvdata[drvdata->cpu] = NULL;
- return ret;
} static struct amba_cs_uci_id uci_id_etm4[] = { @@ -1551,6 +1565,22 @@ static struct amba_cs_uci_id uci_id_etm4[] = { } }; +static int __exit etm4_remove(struct amba_device *adev) +{
- struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = NULL;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- coresight_unregister(drvdata->csdev);
- return 0;
+}
static const struct amba_id etm4_ids[] = { CS_AMBA_ID(0x000bb95d), /* Cortex-A53 */ CS_AMBA_ID(0x000bb95e), /* Cortex-A57 */ @@ -1568,18 +1598,26 @@ static const struct amba_id etm4_ids[] = { {}, }; +MODULE_DEVICE_TABLE(amba, etm4_ids);
static struct amba_driver etm4x_driver = { .drv = { .name = "coresight-etm4x",
.suppress_bind_attrs = true, }, .probe = etm4_probe,.owner = THIS_MODULE,
- .remove = etm4_remove, .id_table = etm4_ids,
}; static int __init etm4x_init(void) { int ret;
- int cpu;
- for_each_possible_cpu(cpu)
mutex_init(&per_cpu(etmdrvdata_lock, cpu));
ret = etm4_pm_setup(); @@ -1595,4 +1633,20 @@ static int __init etm4x_init(void) return ret; } -device_initcall(etm4x_init);
+static void __exit etm4x_exit(void) +{
- int cpu;
- amba_driver_unregister(&etm4x_driver);
- etm4_pm_clear();
- for_each_possible_cpu(cpu)
mutex_destroy(&per_cpu(etmdrvdata_lock, cpu));
+} +module_init(etm4x_init); +module_exit(etm4x_exit);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace v4 driver");
... And this should be "Arm CoreSight Embedded Trace Macrocell 4.x driver".
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Fri, Aug 14, 2020 at 03:40:57AM +0800, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118 +++++++++++++----- 3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c =>
coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig
b/drivers/hwtracing/coresight/Kconfig
index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented
version
data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 diff --git a/drivers/hwtracing/coresight/Makefile
b/drivers/hwtracing/coresight/Makefile
index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) +=
coresight-funnel.o \
obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c
b/drivers/hwtracing/coresight/coresight-etm4x-core.c
similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted"); +static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config
*config)
static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
if (etmdrvdata[cpu]->boot_enable &&
!etmdrvdata[cpu]->sticky_enable)
coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock) @@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_dying_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (local_read(&etmdrvdata[cpu]->mode)) etm4_disable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} @@ -1360,24 +1373,30 @@ static int etm4_cpu_pm_notify(struct
notifier_block *nb, unsigned long cmd,
{ struct etmv4_drvdata *drvdata; unsigned int cpu = smp_processor_id();
- int ret = NOTIFY_OK;
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); if (!etmdrvdata[cpu])
return NOTIFY_OK;
goto out;
drvdata = etmdrvdata[cpu]; if (!drvdata->save_state)
return NOTIFY_OK;
goto out;
- if (WARN_ON_ONCE(drvdata->cpu != cpu))
return NOTIFY_BAD;
- if (WARN_ON_ONCE(drvdata->cpu != cpu)) {
ret = NOTIFY_BAD;
goto out;
- }
switch (cmd) { case CPU_PM_ENTER: /* save the state if self-hosted coresight is in use */ if (local_read(&drvdata->mode))
if (etm4_cpu_save(drvdata))
return NOTIFY_BAD;
if (etm4_cpu_save(drvdata)) {
ret = NOTIFY_BAD;
goto out;
break; case CPU_PM_EXIT: /* fallthrough */}
@@ -1386,10 +1405,12 @@ static int etm4_cpu_pm_notify(struct
notifier_block *nb, unsigned long cmd,
etm4_cpu_restore(drvdata); break;
default:
return NOTIFY_DONE;
}goto out;
- return NOTIFY_OK;
+out:
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- return ret;
} static struct notifier_block etm4_cpu_pm_nb = { @@ -1430,7 +1451,7 @@ static int __init etm4_pm_setup(void) return ret; } -static void __init etm4_pm_clear(void) +static void etm4_pm_clear(void) { cpu_pm_unregister_notifier(&etm4_cpu_pm_nb); cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING); @@ -1487,25 +1508,20 @@ static int etm4_probe(struct amba_device *adev,
const struct amba_id *id)
if (!desc.name) return -ENOMEM;
- etmdrvdata[drvdata->cpu] = drvdata;
- if (smp_call_function_single(drvdata->cpu, etm4_init_arch_data, drvdata, 1)) dev_err(dev, "ETM arch init failed\n");
- if (etm4_arch_supported(drvdata->arch) == false) {
ret = -EINVAL;
goto err_arch_supported;
- }
- if (etm4_arch_supported(drvdata->arch) == false)
return -EINVAL;
etm4_init_trace_id(drvdata); etm4_set_default(&drvdata->config); pdata = coresight_get_platform_data(dev);
- if (IS_ERR(pdata)) {
ret = PTR_ERR(pdata);
goto err_arch_supported;
- }
- if (IS_ERR(pdata))
return PTR_ERR(pdata);
- adev->dev.platform_data = pdata;
desc.type = CORESIGHT_DEV_TYPE_SOURCE; @@ -1515,17 +1531,19 @@ static int etm4_probe(struct amba_device *adev,
const struct amba_id *id)
desc.dev = dev; desc.groups = coresight_etmv4_groups; drvdata->csdev = coresight_register(&desc);
- if (IS_ERR(drvdata->csdev)) {
ret = PTR_ERR(drvdata->csdev);
goto err_arch_supported;
- }
- if (IS_ERR(drvdata->csdev))
return PTR_ERR(drvdata->csdev);
ret = etm_perf_symlink(drvdata->csdev, true); if (ret) { coresight_unregister(drvdata->csdev);
goto err_arch_supported;
}return ret;
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = drvdata;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- pm_runtime_put(&adev->dev); dev_info(&drvdata->csdev->dev, "CPU%d: ETM v%d.%d initialized\n", drvdata->cpu, drvdata->arch >> 4, drvdata->arch & 0xf);
@@ -1536,10 +1554,6 @@ static int etm4_probe(struct amba_device *adev,
const struct amba_id *id)
} return 0;
-err_arch_supported:
- etmdrvdata[drvdata->cpu] = NULL;
- return ret;
} static struct amba_cs_uci_id uci_id_etm4[] = { @@ -1551,6 +1565,22 @@ static struct amba_cs_uci_id uci_id_etm4[] = { } }; +static int __exit etm4_remove(struct amba_device *adev) +{
- struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = NULL;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- coresight_unregister(drvdata->csdev);
- return 0;
+}
static const struct amba_id etm4_ids[] = { CS_AMBA_ID(0x000bb95d), /* Cortex-A53 */ CS_AMBA_ID(0x000bb95e), /* Cortex-A57 */ @@ -1568,18 +1598,26 @@ static const struct amba_id etm4_ids[] = { {}, }; +MODULE_DEVICE_TABLE(amba, etm4_ids);
static struct amba_driver etm4x_driver = { .drv = { .name = "coresight-etm4x",
.suppress_bind_attrs = true, }, .probe = etm4_probe,.owner = THIS_MODULE,
- .remove = etm4_remove, .id_table = etm4_ids,
}; static int __init etm4x_init(void) { int ret;
- int cpu;
- for_each_possible_cpu(cpu)
mutex_init(&per_cpu(etmdrvdata_lock, cpu));
ret = etm4_pm_setup(); @@ -1595,4 +1633,20 @@ static int __init etm4x_init(void) return ret; } -device_initcall(etm4x_init);
+static void __exit etm4x_exit(void) +{
- int cpu;
- amba_driver_unregister(&etm4x_driver);
- etm4_pm_clear();
- for_each_possible_cpu(cpu)
mutex_destroy(&per_cpu(etmdrvdata_lock, cpu));
+} +module_init(etm4x_init); +module_exit(etm4x_exit);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace v4 driver");
... And this should be "Arm CoreSight Embedded Trace Macrocell 4.x driver".
This will be fixed in v9.
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project
linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118 +++++++++++++----- 3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c => coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented version data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x-core.c similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted"); +static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config *config) static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
if (etmdrvdata[cpu]->boot_enable && !etmdrvdata[cpu]->sticky_enable) coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock) @@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_dying_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (local_read(&etmdrvdata[cpu]->mode)) etm4_disable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} @@ -1360,24 +1373,30 @@ static int etm4_cpu_pm_notify(struct notifier_block *nb, unsigned long cmd, { struct etmv4_drvdata *drvdata; unsigned int cpu = smp_processor_id();
- int ret = NOTIFY_OK;
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); if (!etmdrvdata[cpu])
return NOTIFY_OK;
goto out;
drvdata = etmdrvdata[cpu]; if (!drvdata->save_state)
return NOTIFY_OK;
goto out;
- if (WARN_ON_ONCE(drvdata->cpu != cpu))
return NOTIFY_BAD;
- if (WARN_ON_ONCE(drvdata->cpu != cpu)) {
ret = NOTIFY_BAD;
goto out;
- }
switch (cmd) { case CPU_PM_ENTER: /* save the state if self-hosted coresight is in use */ if (local_read(&drvdata->mode))
if (etm4_cpu_save(drvdata))
return NOTIFY_BAD;
if (etm4_cpu_save(drvdata)) {
ret = NOTIFY_BAD;
goto out;
break; case CPU_PM_EXIT: /* fallthrough */}
@@ -1386,10 +1405,12 @@ static int etm4_cpu_pm_notify(struct notifier_block *nb, unsigned long cmd, etm4_cpu_restore(drvdata); break; default:
return NOTIFY_DONE;
}goto out;
- return NOTIFY_OK;
+out:
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- return ret;
} static struct notifier_block etm4_cpu_pm_nb = { @@ -1430,7 +1451,7 @@ static int __init etm4_pm_setup(void) return ret; } -static void __init etm4_pm_clear(void) +static void etm4_pm_clear(void) { cpu_pm_unregister_notifier(&etm4_cpu_pm_nb); cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING); @@ -1487,25 +1508,20 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) if (!desc.name) return -ENOMEM;
- etmdrvdata[drvdata->cpu] = drvdata;
- if (smp_call_function_single(drvdata->cpu, etm4_init_arch_data, drvdata, 1)) dev_err(dev, "ETM arch init failed\n");
- if (etm4_arch_supported(drvdata->arch) == false) {
ret = -EINVAL;
goto err_arch_supported;
- }
- if (etm4_arch_supported(drvdata->arch) == false)
return -EINVAL;
etm4_init_trace_id(drvdata); etm4_set_default(&drvdata->config); pdata = coresight_get_platform_data(dev);
- if (IS_ERR(pdata)) {
ret = PTR_ERR(pdata);
goto err_arch_supported;
- }
- if (IS_ERR(pdata))
return PTR_ERR(pdata);
- adev->dev.platform_data = pdata;
desc.type = CORESIGHT_DEV_TYPE_SOURCE; @@ -1515,17 +1531,19 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) desc.dev = dev; desc.groups = coresight_etmv4_groups; drvdata->csdev = coresight_register(&desc);
- if (IS_ERR(drvdata->csdev)) {
ret = PTR_ERR(drvdata->csdev);
goto err_arch_supported;
- }
- if (IS_ERR(drvdata->csdev))
return PTR_ERR(drvdata->csdev);
ret = etm_perf_symlink(drvdata->csdev, true); if (ret) { coresight_unregister(drvdata->csdev);
goto err_arch_supported;
}return ret;
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = drvdata;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- pm_runtime_put(&adev->dev); dev_info(&drvdata->csdev->dev, "CPU%d: ETM v%d.%d initialized\n", drvdata->cpu, drvdata->arch >> 4, drvdata->arch & 0xf);
@@ -1536,10 +1554,6 @@ static int etm4_probe(struct amba_device *adev, const struct amba_id *id) } return 0;
-err_arch_supported:
- etmdrvdata[drvdata->cpu] = NULL;
- return ret;
} static struct amba_cs_uci_id uci_id_etm4[] = { @@ -1551,6 +1565,22 @@ static struct amba_cs_uci_id uci_id_etm4[] = { } }; +static int __exit etm4_remove(struct amba_device *adev) +{
- struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = NULL;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- coresight_unregister(drvdata->csdev);
- return 0;
+}
Extra line
static const struct amba_id etm4_ids[] = { CS_AMBA_ID(0x000bb95d), /* Cortex-A53 */ CS_AMBA_ID(0x000bb95e), /* Cortex-A57 */ @@ -1568,18 +1598,26 @@ static const struct amba_id etm4_ids[] = { {}, }; +MODULE_DEVICE_TABLE(amba, etm4_ids);
static struct amba_driver etm4x_driver = { .drv = { .name = "coresight-etm4x",
.suppress_bind_attrs = true, }, .probe = etm4_probe,.owner = THIS_MODULE,
- .remove = etm4_remove, .id_table = etm4_ids,
}; static int __init etm4x_init(void) { int ret;
- int cpu;
- for_each_possible_cpu(cpu)
mutex_init(&per_cpu(etmdrvdata_lock, cpu));
ret = etm4_pm_setup(); @@ -1595,4 +1633,20 @@ static int __init etm4x_init(void) return ret; } -device_initcall(etm4x_init);
+static void __exit etm4x_exit(void) +{
- int cpu;
- amba_driver_unregister(&etm4x_driver);
- etm4_pm_clear();
- for_each_possible_cpu(cpu)
mutex_destroy(&per_cpu(etmdrvdata_lock, cpu));
+} +module_init(etm4x_init); +module_exit(etm4x_exit);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace v4 driver");
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Fri, Aug 14, 2020 at 03:45:06AM +0800, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:39PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etm4x as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-etm4x by the Makefile
- add an etm4_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
- protect etmdrvdata[] with mutex lock from racing between module unload and CPU hotplug
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- drivers/hwtracing/coresight/Makefile | 4 +- ...resight-etm4x.c => coresight-etm4x-core.c} | 118 +++++++++++++----- 3 files changed, 92 insertions(+), 35 deletions(-) rename drivers/hwtracing/coresight/{coresight-etm4x.c =>
coresight-etm4x-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig
b/drivers/hwtracing/coresight/Kconfig
index 8fd9fd139cf3..d6e107bbd30b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -78,7 +78,7 @@ config CORESIGHT_SOURCE_ETM3X module will be called coresight-etm3x. config CORESIGHT_SOURCE_ETM4X
- bool "CoreSight Embedded Trace Macrocell 4.x driver"
- tristate "CoreSight Embedded Trace Macrocell 4.x driver" depends on ARM64 select CORESIGHT_LINKS_AND_SINKS select PID_IN_CONTEXTIDR
@@ -88,6 +88,9 @@ config CORESIGHT_SOURCE_ETM4X for instruction level tracing. Depending on the implemented
version
data tracing may also be available.
To compile this driver as a module, choose M here: the
module will be called coresight-etm4x.
config CORESIGHT_STM tristate "CoreSight System Trace Macrocell driver" depends on (ARM && !(CPU_32v3 || CPU_32v4 || CPU_32v4T)) || ARM64 diff --git a/drivers/hwtracing/coresight/Makefile
b/drivers/hwtracing/coresight/Makefile
index d619cfd0abd8..271dc255454f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -14,8 +14,8 @@ obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) +=
coresight-funnel.o \
obj-$(CONFIG_CORESIGHT_SOURCE_ETM3X) += coresight-etm3x.o coresight-etm3x-y := coresight-etm3x-core.o coresight-etm-cp14.o \ coresight-etm3x-sysfs.o -obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o \
coresight-etm4x-sysfs.o
+obj-$(CONFIG_CORESIGHT_SOURCE_ETM4X) += coresight-etm4x.o +coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c
b/drivers/hwtracing/coresight/coresight-etm4x-core.c
similarity index 95% rename from drivers/hwtracing/coresight/coresight-etm4x.c rename to drivers/hwtracing/coresight/coresight-etm4x-core.c index fddfd93b9a7b..ae9847e194a3 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -48,6 +48,7 @@ module_param(pm_save_enable, int, 0444); MODULE_PARM_DESC(pm_save_enable, "Save/restore state on power down: 1 = never, 2 = self-hosted"); +static DEFINE_PER_CPU(struct mutex, etmdrvdata_lock); static struct etmv4_drvdata *etmdrvdata[NR_CPUS]; static void etm4_set_default_config(struct etmv4_config *config); static int etm4_set_event_filters(struct etmv4_drvdata *drvdata, @@ -1089,18 +1090,25 @@ void etm4_config_trace_mode(struct etmv4_config
*config)
static int etm4_online_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
if (etmdrvdata[cpu]->boot_enable &&
!etmdrvdata[cpu]->sticky_enable)
coresight_enable(etmdrvdata[cpu]->csdev);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_starting_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (!etmdrvdata[cpu]->os_unlock) @@ -1109,18 +1117,23 @@ static int etm4_starting_cpu(unsigned int cpu) if (local_read(&etmdrvdata[cpu]->mode)) etm4_enable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} static int etm4_dying_cpu(unsigned int cpu) {
- if (!etmdrvdata[cpu])
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu));
- if (!etmdrvdata[cpu]) {
return 0;mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- }
spin_lock(&etmdrvdata[cpu]->spinlock); if (local_read(&etmdrvdata[cpu]->mode)) etm4_disable_hw(etmdrvdata[cpu]); spin_unlock(&etmdrvdata[cpu]->spinlock);
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu)); return 0;
} @@ -1360,24 +1373,30 @@ static int etm4_cpu_pm_notify(struct
notifier_block *nb, unsigned long cmd,
{ struct etmv4_drvdata *drvdata; unsigned int cpu = smp_processor_id();
- int ret = NOTIFY_OK;
- mutex_lock(&per_cpu(etmdrvdata_lock, cpu)); if (!etmdrvdata[cpu])
return NOTIFY_OK;
goto out;
drvdata = etmdrvdata[cpu]; if (!drvdata->save_state)
return NOTIFY_OK;
goto out;
- if (WARN_ON_ONCE(drvdata->cpu != cpu))
return NOTIFY_BAD;
- if (WARN_ON_ONCE(drvdata->cpu != cpu)) {
ret = NOTIFY_BAD;
goto out;
- }
switch (cmd) { case CPU_PM_ENTER: /* save the state if self-hosted coresight is in use */ if (local_read(&drvdata->mode))
if (etm4_cpu_save(drvdata))
return NOTIFY_BAD;
if (etm4_cpu_save(drvdata)) {
ret = NOTIFY_BAD;
goto out;
break; case CPU_PM_EXIT: /* fallthrough */}
@@ -1386,10 +1405,12 @@ static int etm4_cpu_pm_notify(struct
notifier_block *nb, unsigned long cmd,
etm4_cpu_restore(drvdata); break;
default:
return NOTIFY_DONE;
}goto out;
- return NOTIFY_OK;
+out:
- mutex_unlock(&per_cpu(etmdrvdata_lock, cpu));
- return ret;
} static struct notifier_block etm4_cpu_pm_nb = { @@ -1430,7 +1451,7 @@ static int __init etm4_pm_setup(void) return ret; } -static void __init etm4_pm_clear(void) +static void etm4_pm_clear(void) { cpu_pm_unregister_notifier(&etm4_cpu_pm_nb); cpuhp_remove_state_nocalls(CPUHP_AP_ARM_CORESIGHT_STARTING); @@ -1487,25 +1508,20 @@ static int etm4_probe(struct amba_device *adev,
const struct amba_id *id)
if (!desc.name) return -ENOMEM;
- etmdrvdata[drvdata->cpu] = drvdata;
- if (smp_call_function_single(drvdata->cpu, etm4_init_arch_data, drvdata, 1)) dev_err(dev, "ETM arch init failed\n");
- if (etm4_arch_supported(drvdata->arch) == false) {
ret = -EINVAL;
goto err_arch_supported;
- }
- if (etm4_arch_supported(drvdata->arch) == false)
return -EINVAL;
etm4_init_trace_id(drvdata); etm4_set_default(&drvdata->config); pdata = coresight_get_platform_data(dev);
- if (IS_ERR(pdata)) {
ret = PTR_ERR(pdata);
goto err_arch_supported;
- }
- if (IS_ERR(pdata))
return PTR_ERR(pdata);
- adev->dev.platform_data = pdata;
desc.type = CORESIGHT_DEV_TYPE_SOURCE; @@ -1515,17 +1531,19 @@ static int etm4_probe(struct amba_device *adev,
const struct amba_id *id)
desc.dev = dev; desc.groups = coresight_etmv4_groups; drvdata->csdev = coresight_register(&desc);
- if (IS_ERR(drvdata->csdev)) {
ret = PTR_ERR(drvdata->csdev);
goto err_arch_supported;
- }
- if (IS_ERR(drvdata->csdev))
return PTR_ERR(drvdata->csdev);
ret = etm_perf_symlink(drvdata->csdev, true); if (ret) { coresight_unregister(drvdata->csdev);
goto err_arch_supported;
}return ret;
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = drvdata;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- pm_runtime_put(&adev->dev); dev_info(&drvdata->csdev->dev, "CPU%d: ETM v%d.%d initialized\n", drvdata->cpu, drvdata->arch >> 4, drvdata->arch & 0xf);
@@ -1536,10 +1554,6 @@ static int etm4_probe(struct amba_device *adev,
const struct amba_id *id)
} return 0;
-err_arch_supported:
- etmdrvdata[drvdata->cpu] = NULL;
- return ret;
} static struct amba_cs_uci_id uci_id_etm4[] = { @@ -1551,6 +1565,22 @@ static struct amba_cs_uci_id uci_id_etm4[] = { } }; +static int __exit etm4_remove(struct amba_device *adev) +{
- struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- etm_perf_symlink(drvdata->csdev, false);
- mutex_lock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- etmdrvdata[drvdata->cpu] = NULL;
- mutex_unlock(&per_cpu(etmdrvdata_lock, drvdata->cpu));
- coresight_unregister(drvdata->csdev);
- return 0;
+}
Extra line
Will fix in next revision.
static const struct amba_id etm4_ids[] = { CS_AMBA_ID(0x000bb95d), /* Cortex-A53 */ CS_AMBA_ID(0x000bb95e), /* Cortex-A57 */ @@ -1568,18 +1598,26 @@ static const struct amba_id etm4_ids[] = { {}, }; +MODULE_DEVICE_TABLE(amba, etm4_ids);
static struct amba_driver etm4x_driver = { .drv = { .name = "coresight-etm4x",
.suppress_bind_attrs = true, }, .probe = etm4_probe,.owner = THIS_MODULE,
- .remove = etm4_remove, .id_table = etm4_ids,
}; static int __init etm4x_init(void) { int ret;
- int cpu;
- for_each_possible_cpu(cpu)
mutex_init(&per_cpu(etmdrvdata_lock, cpu));
ret = etm4_pm_setup(); @@ -1595,4 +1633,20 @@ static int __init etm4x_init(void) return ret; } -device_initcall(etm4x_init);
+static void __exit etm4x_exit(void) +{
- int cpu;
- amba_driver_unregister(&etm4x_driver);
- etm4_pm_clear();
- for_each_possible_cpu(cpu)
mutex_destroy(&per_cpu(etmdrvdata_lock, cpu));
+} +module_init(etm4x_init); +module_exit(etm4x_exit);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Program Flow Trace v4 driver");
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-etb10 as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m - add an etb_remove function, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/Kconfig | 5 +++- drivers/hwtracing/coresight/coresight-etb10.c | 25 ++++++++++++++++++- 2 files changed, 28 insertions(+), 2 deletions(-)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index d6e107bbd30b..996d84a1edb8 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -57,13 +57,16 @@ config CORESIGHT_SINK_TPIU the on-board coresight memory can handle.
config CORESIGHT_SINK_ETBV10 - bool "Coresight ETBv1.0 driver" + tristate "Coresight ETBv1.0 driver" depends on CORESIGHT_LINKS_AND_SINKS help This enables support for the Embedded Trace Buffer version 1.0 driver that complies with the generic implementation of the component without special enhancement or added features.
+ To compile this driver as a module, choose M here: the + module will be called coresight-etb10. + config CORESIGHT_SOURCE_ETM3X tristate "CoreSight Embedded Trace Macrocell 3.x driver" depends on !ARM64 diff --git a/drivers/hwtracing/coresight/coresight-etb10.c b/drivers/hwtracing/coresight/coresight-etb10.c index 04ee9cda988d..29ef6c7f7be5 100644 --- a/drivers/hwtracing/coresight/coresight-etb10.c +++ b/drivers/hwtracing/coresight/coresight-etb10.c @@ -801,6 +801,21 @@ static int etb_probe(struct amba_device *adev, const struct amba_id *id) return ret; }
+static int __exit etb_remove(struct amba_device *adev) +{ + struct etb_drvdata *drvdata = dev_get_drvdata(&adev->dev); + + /* + * Since misc_open() holds a refcount on the f_ops, which is + * etb fops in this case, device is there until last file + * handler to this device is closed. + */ + misc_deregister(&drvdata->miscdev); + coresight_unregister(drvdata->csdev); + + return 0; +} + #ifdef CONFIG_PM static int etb_runtime_suspend(struct device *dev) { @@ -835,6 +850,8 @@ static const struct amba_id etb_ids[] = { { 0, 0}, };
+MODULE_DEVICE_TABLE(amba, etb_ids); + static struct amba_driver etb_driver = { .drv = { .name = "coresight-etb10", @@ -844,6 +861,12 @@ static struct amba_driver etb_driver = {
}, .probe = etb_probe, + .remove = etb_remove, .id_table = etb_ids, }; -builtin_amba_driver(etb_driver); +module_amba_driver(etb_driver); + +MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Embedded Trace Buffer driver"); +MODULE_LICENSE("GPL v2");
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-tpiu as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m - add a tpiu_remove function, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/Kconfig | 5 ++++- drivers/hwtracing/coresight/coresight-tpiu.c | 19 ++++++++++++++++++- 2 files changed, 22 insertions(+), 2 deletions(-)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 996d84a1edb8..8fd9887fb03b 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -46,7 +46,7 @@ config CORESIGHT_CATU mode where the address is not translated.
config CORESIGHT_SINK_TPIU - bool "Coresight generic TPIU driver" + tristate "Coresight generic TPIU driver" depends on CORESIGHT_LINKS_AND_SINKS help This enables support for the Trace Port Interface Unit driver, @@ -56,6 +56,9 @@ config CORESIGHT_SINK_TPIU connected to an external host for use case capturing more traces than the on-board coresight memory can handle.
+ To compile this driver as a module, choose M here: the + module will be called coresight-tpiu. + config CORESIGHT_SINK_ETBV10 tristate "Coresight ETBv1.0 driver" depends on CORESIGHT_LINKS_AND_SINKS diff --git a/drivers/hwtracing/coresight/coresight-tpiu.c b/drivers/hwtracing/coresight/coresight-tpiu.c index f8583e4032a6..e4ddd9801535 100644 --- a/drivers/hwtracing/coresight/coresight-tpiu.c +++ b/drivers/hwtracing/coresight/coresight-tpiu.c @@ -173,6 +173,15 @@ static int tpiu_probe(struct amba_device *adev, const struct amba_id *id) return PTR_ERR(drvdata->csdev); }
+static int __exit tpiu_remove(struct amba_device *adev) +{ + struct tpiu_drvdata *drvdata = dev_get_drvdata(&adev->dev); + + coresight_unregister(drvdata->csdev); + + return 0; +} + #ifdef CONFIG_PM static int tpiu_runtime_suspend(struct device *dev) { @@ -216,6 +225,8 @@ static const struct amba_id tpiu_ids[] = { { 0, 0}, };
+MODULE_DEVICE_TABLE(amba, tpiu_ids); + static struct amba_driver tpiu_driver = { .drv = { .name = "coresight-tpiu", @@ -224,6 +235,12 @@ static struct amba_driver tpiu_driver = { .suppress_bind_attrs = true, }, .probe = tpiu_probe, + .remove = tpiu_remove, .id_table = tpiu_ids, }; -builtin_amba_driver(tpiu_driver); +module_amba_driver(tpiu_driver); + +MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight TPIU (Trace Port Interface Unit) driver"); +MODULE_LICENSE("GPL v2");
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-tmc as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m - append -core to source file name to allow module to be called coresight-tmc by the Makefile - add an tmc_remove function, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/Kconfig | 6 ++++- drivers/hwtracing/coresight/Makefile | 6 ++--- .../{coresight-tmc.c => coresight-tmc-core.c} | 24 ++++++++++++++++++- 3 files changed, 31 insertions(+), 5 deletions(-) rename drivers/hwtracing/coresight/{coresight-tmc.c => coresight-tmc-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 8fd9887fb03b..fc48ae086746 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -25,7 +25,8 @@ config CORESIGHT_LINKS_AND_SINKS entity at run time to form a complete trace path.
config CORESIGHT_LINK_AND_SINK_TMC - bool "Coresight generic TMC driver" + tristate "Coresight generic TMC driver" + depends on CORESIGHT_LINKS_AND_SINKS help This enables support for the Trace Memory Controller driver. @@ -34,6 +35,9 @@ config CORESIGHT_LINK_AND_SINK_TMC complies with the generic implementation of the component without special enhancement or added features.
+ To compile this driver as a module, choose M here: the + module will be called coresight-tmc. + config CORESIGHT_CATU bool "Coresight Address Translation Unit (CATU) driver" depends on CORESIGHT_LINK_AND_SINK_TMC diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 271dc255454f..f2a568b969c4 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -4,9 +4,9 @@ # obj-$(CONFIG_CORESIGHT) += coresight.o coresight-etm-perf.o \ coresight-platform.o coresight-sysfs.o -obj-$(CONFIG_CORESIGHT_LINK_AND_SINK_TMC) += coresight-tmc.o \ - coresight-tmc-etf.o \ - coresight-tmc-etr.o +obj-$(CONFIG_CORESIGHT_LINK_AND_SINK_TMC) += coresight-tmc.o +coresight-tmc-y := coresight-tmc-core.o coresight-tmc-etf.o \ + coresight-tmc-etr.o obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ diff --git a/drivers/hwtracing/coresight/coresight-tmc.c b/drivers/hwtracing/coresight/coresight-tmc-core.c similarity index 95% rename from drivers/hwtracing/coresight/coresight-tmc.c rename to drivers/hwtracing/coresight/coresight-tmc-core.c index 7040d583bed9..f46852489adf 100644 --- a/drivers/hwtracing/coresight/coresight-tmc.c +++ b/drivers/hwtracing/coresight/coresight-tmc-core.c @@ -561,6 +561,21 @@ static void tmc_shutdown(struct amba_device *adev) spin_unlock_irqrestore(&drvdata->spinlock, flags); }
+static int __exit tmc_remove(struct amba_device *adev) +{ + struct tmc_drvdata *drvdata = dev_get_drvdata(&adev->dev); + + /* + * Since misc_open() holds a refcount on the f_ops, which is + * etb fops in this case, device is there until last file + * handler to this device is closed. + */ + misc_deregister(&drvdata->miscdev); + coresight_unregister(drvdata->csdev); + + return 0; +} + static const struct amba_id tmc_ids[] = { CS_AMBA_ID(0x000bb961), /* Coresight SoC 600 TMC-ETR/ETS */ @@ -572,6 +587,8 @@ static const struct amba_id tmc_ids[] = { { 0, 0}, };
+MODULE_DEVICE_TABLE(amba, tmc_ids); + static struct amba_driver tmc_driver = { .drv = { .name = "coresight-tmc", @@ -580,6 +597,11 @@ static struct amba_driver tmc_driver = { }, .probe = tmc_probe, .shutdown = tmc_shutdown, + .remove = tmc_remove, .id_table = tmc_ids, }; -builtin_amba_driver(tmc_driver); +module_amba_driver(tmc_driver); + +MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_DESCRIPTION("Arm CoreSight Trace Memory Controller driver"); +MODULE_LICENSE("GPL v2");
On Fri, Aug 07, 2020 at 07:11:42PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-tmc as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-tmc by the Makefile
- add an tmc_remove function, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
For patches 11, 12 and 13:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
drivers/hwtracing/coresight/Kconfig | 6 ++++- drivers/hwtracing/coresight/Makefile | 6 ++--- .../{coresight-tmc.c => coresight-tmc-core.c} | 24 ++++++++++++++++++- 3 files changed, 31 insertions(+), 5 deletions(-) rename drivers/hwtracing/coresight/{coresight-tmc.c => coresight-tmc-core.c} (95%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index 8fd9887fb03b..fc48ae086746 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -25,7 +25,8 @@ config CORESIGHT_LINKS_AND_SINKS entity at run time to form a complete trace path. config CORESIGHT_LINK_AND_SINK_TMC
- bool "Coresight generic TMC driver"
- tristate "Coresight generic TMC driver"
- depends on CORESIGHT_LINKS_AND_SINKS help This enables support for the Trace Memory Controller driver.
@@ -34,6 +35,9 @@ config CORESIGHT_LINK_AND_SINK_TMC complies with the generic implementation of the component without special enhancement or added features.
To compile this driver as a module, choose M here: the
module will be called coresight-tmc.
config CORESIGHT_CATU bool "Coresight Address Translation Unit (CATU) driver" depends on CORESIGHT_LINK_AND_SINK_TMC diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 271dc255454f..f2a568b969c4 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -4,9 +4,9 @@ # obj-$(CONFIG_CORESIGHT) += coresight.o coresight-etm-perf.o \ coresight-platform.o coresight-sysfs.o -obj-$(CONFIG_CORESIGHT_LINK_AND_SINK_TMC) += coresight-tmc.o \
coresight-tmc-etf.o \
coresight-tmc-etr.o
+obj-$(CONFIG_CORESIGHT_LINK_AND_SINK_TMC) += coresight-tmc.o +coresight-tmc-y := coresight-tmc-core.o coresight-tmc-etf.o \
coresight-tmc-etr.o
obj-$(CONFIG_CORESIGHT_SINK_TPIU) += coresight-tpiu.o obj-$(CONFIG_CORESIGHT_SINK_ETBV10) += coresight-etb10.o obj-$(CONFIG_CORESIGHT_LINKS_AND_SINKS) += coresight-funnel.o \ diff --git a/drivers/hwtracing/coresight/coresight-tmc.c b/drivers/hwtracing/coresight/coresight-tmc-core.c similarity index 95% rename from drivers/hwtracing/coresight/coresight-tmc.c rename to drivers/hwtracing/coresight/coresight-tmc-core.c index 7040d583bed9..f46852489adf 100644 --- a/drivers/hwtracing/coresight/coresight-tmc.c +++ b/drivers/hwtracing/coresight/coresight-tmc-core.c @@ -561,6 +561,21 @@ static void tmc_shutdown(struct amba_device *adev) spin_unlock_irqrestore(&drvdata->spinlock, flags); } +static int __exit tmc_remove(struct amba_device *adev) +{
- struct tmc_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- /*
* Since misc_open() holds a refcount on the f_ops, which is
* etb fops in this case, device is there until last file
* handler to this device is closed.
*/
- misc_deregister(&drvdata->miscdev);
- coresight_unregister(drvdata->csdev);
- return 0;
+}
static const struct amba_id tmc_ids[] = { CS_AMBA_ID(0x000bb961), /* Coresight SoC 600 TMC-ETR/ETS */ @@ -572,6 +587,8 @@ static const struct amba_id tmc_ids[] = { { 0, 0}, }; +MODULE_DEVICE_TABLE(amba, tmc_ids);
static struct amba_driver tmc_driver = { .drv = { .name = "coresight-tmc", @@ -580,6 +597,11 @@ static struct amba_driver tmc_driver = { }, .probe = tmc_probe, .shutdown = tmc_shutdown,
- .remove = tmc_remove, .id_table = tmc_ids,
}; -builtin_amba_driver(tmc_driver); +module_amba_driver(tmc_driver);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_DESCRIPTION("Arm CoreSight Trace Memory Controller driver");
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-funnel as module, for ease of development.
- combine static and dynamic funnel init into single module_init/exit call - add funnel_remove functions, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Co-developed-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org --- .../hwtracing/coresight/coresight-funnel.c | 62 ++++++++++++++++++- 1 file changed, 60 insertions(+), 2 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-funnel.c b/drivers/hwtracing/coresight/coresight-funnel.c index 900690a9f7f0..062694ef9879 100644 --- a/drivers/hwtracing/coresight/coresight-funnel.c +++ b/drivers/hwtracing/coresight/coresight-funnel.c @@ -274,6 +274,15 @@ static int funnel_probe(struct device *dev, struct resource *res) return ret; }
+static int __exit funnel_remove(struct device *dev) +{ + struct funnel_drvdata *drvdata = dev_get_drvdata(dev); + + coresight_unregister(drvdata->csdev); + + return 0; +} + #ifdef CONFIG_PM static int funnel_runtime_suspend(struct device *dev) { @@ -319,20 +328,31 @@ static int static_funnel_probe(struct platform_device *pdev) return ret; }
+static int __exit static_funnel_remove(struct platform_device *pdev) +{ + funnel_remove(&pdev->dev); + pm_runtime_disable(&pdev->dev); + return 0; +} + static const struct of_device_id static_funnel_match[] = { {.compatible = "arm,coresight-static-funnel"}, {} };
+MODULE_DEVICE_TABLE(of, static_funnel_match); + #ifdef CONFIG_ACPI static const struct acpi_device_id static_funnel_ids[] = { {"ARMHC9FE", 0}, {}, }; +MODULE_DEVICE_TABLE(acpi, static_funnel_ids); #endif
static struct platform_driver static_funnel_driver = { .probe = static_funnel_probe, + .probe = static_funnel_remove, .driver = { .name = "coresight-static-funnel", .of_match_table = static_funnel_match, @@ -341,7 +361,6 @@ static struct platform_driver static_funnel_driver = { .suppress_bind_attrs = true, }, }; -builtin_platform_driver(static_funnel_driver);
static int dynamic_funnel_probe(struct amba_device *adev, const struct amba_id *id) @@ -349,6 +368,11 @@ static int dynamic_funnel_probe(struct amba_device *adev, return funnel_probe(&adev->dev, &adev->res); }
+static int __exit dynamic_funnel_remove(struct amba_device *adev) +{ + return funnel_remove(&adev->dev); +} + static const struct amba_id dynamic_funnel_ids[] = { { .id = 0x000bb908, @@ -362,6 +386,8 @@ static const struct amba_id dynamic_funnel_ids[] = { { 0, 0}, };
+MODULE_DEVICE_TABLE(amba, dynamic_funnel_ids); + static struct amba_driver dynamic_funnel_driver = { .drv = { .name = "coresight-dynamic-funnel", @@ -370,6 +396,38 @@ static struct amba_driver dynamic_funnel_driver = { .suppress_bind_attrs = true, }, .probe = dynamic_funnel_probe, + .remove = dynamic_funnel_remove, .id_table = dynamic_funnel_ids, }; -builtin_amba_driver(dynamic_funnel_driver); + +static int __init funnel_init(void) +{ + int ret; + + ret = platform_driver_register(&static_funnel_driver); + if (ret) { + pr_info("Error registering platform driver\n"); + return ret; + } + + ret = amba_driver_register(&dynamic_funnel_driver); + if (ret) { + pr_info("Error registering amba driver\n"); + platform_driver_unregister(&static_funnel_driver); + } + + return ret; +} + +static void __exit funnel_exit(void) +{ + platform_driver_unregister(&static_funnel_driver); + amba_driver_unregister(&dynamic_funnel_driver); +} + +module_init(funnel_init); +module_exit(funnel_exit); + +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Funnel Driver"); +MODULE_LICENSE("GPL v2");
On Fri, Aug 07, 2020 at 07:11:43PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-funnel as module, for ease of development.
- combine static and dynamic funnel init into single module_init/exit call
- add funnel_remove functions, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Co-developed-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
.../hwtracing/coresight/coresight-funnel.c | 62 ++++++++++++++++++- 1 file changed, 60 insertions(+), 2 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-funnel.c b/drivers/hwtracing/coresight/coresight-funnel.c index 900690a9f7f0..062694ef9879 100644 --- a/drivers/hwtracing/coresight/coresight-funnel.c +++ b/drivers/hwtracing/coresight/coresight-funnel.c @@ -274,6 +274,15 @@ static int funnel_probe(struct device *dev, struct resource *res) return ret; } +static int __exit funnel_remove(struct device *dev) +{
- struct funnel_drvdata *drvdata = dev_get_drvdata(dev);
- coresight_unregister(drvdata->csdev);
- return 0;
+}
#ifdef CONFIG_PM static int funnel_runtime_suspend(struct device *dev) { @@ -319,20 +328,31 @@ static int static_funnel_probe(struct platform_device *pdev) return ret; } +static int __exit static_funnel_remove(struct platform_device *pdev) +{
- funnel_remove(&pdev->dev);
- pm_runtime_disable(&pdev->dev);
- return 0;
+}
static const struct of_device_id static_funnel_match[] = { {.compatible = "arm,coresight-static-funnel"}, {} }; +MODULE_DEVICE_TABLE(of, static_funnel_match);
#ifdef CONFIG_ACPI static const struct acpi_device_id static_funnel_ids[] = { {"ARMHC9FE", 0}, {}, };
Add extra line to be consistent with the rest of your patchset.
With the that:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
+MODULE_DEVICE_TABLE(acpi, static_funnel_ids); #endif static struct platform_driver static_funnel_driver = { .probe = static_funnel_probe,
- .probe = static_funnel_remove, .driver = { .name = "coresight-static-funnel", .of_match_table = static_funnel_match,
@@ -341,7 +361,6 @@ static struct platform_driver static_funnel_driver = { .suppress_bind_attrs = true, }, }; -builtin_platform_driver(static_funnel_driver); static int dynamic_funnel_probe(struct amba_device *adev, const struct amba_id *id) @@ -349,6 +368,11 @@ static int dynamic_funnel_probe(struct amba_device *adev, return funnel_probe(&adev->dev, &adev->res); } +static int __exit dynamic_funnel_remove(struct amba_device *adev) +{
- return funnel_remove(&adev->dev);
+}
static const struct amba_id dynamic_funnel_ids[] = { { .id = 0x000bb908, @@ -362,6 +386,8 @@ static const struct amba_id dynamic_funnel_ids[] = { { 0, 0}, }; +MODULE_DEVICE_TABLE(amba, dynamic_funnel_ids);
static struct amba_driver dynamic_funnel_driver = { .drv = { .name = "coresight-dynamic-funnel", @@ -370,6 +396,38 @@ static struct amba_driver dynamic_funnel_driver = { .suppress_bind_attrs = true, }, .probe = dynamic_funnel_probe,
- .remove = dynamic_funnel_remove, .id_table = dynamic_funnel_ids,
}; -builtin_amba_driver(dynamic_funnel_driver);
+static int __init funnel_init(void) +{
- int ret;
- ret = platform_driver_register(&static_funnel_driver);
- if (ret) {
pr_info("Error registering platform driver\n");
return ret;
- }
- ret = amba_driver_register(&dynamic_funnel_driver);
- if (ret) {
pr_info("Error registering amba driver\n");
platform_driver_unregister(&static_funnel_driver);
- }
- return ret;
+}
+static void __exit funnel_exit(void) +{
- platform_driver_unregister(&static_funnel_driver);
- amba_driver_unregister(&dynamic_funnel_driver);
+}
+module_init(funnel_init); +module_exit(funnel_exit);
+MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Funnel Driver");
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-replicator as modules, for ease of development.
- Kconfig becomes a tristate, to allow =m - combine static and dynamic replicator init into single module_init/exit call - add replicator_remove functions, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Co-developed-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org --- drivers/hwtracing/coresight/Kconfig | 5 +- .../coresight/coresight-replicator.c | 63 ++++++++++++++++++- 2 files changed, 65 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index fc48ae086746..f31778dd0b5d 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -17,13 +17,16 @@ menuconfig CORESIGHT
if CORESIGHT config CORESIGHT_LINKS_AND_SINKS - bool "CoreSight Link and Sink drivers" + tristate "CoreSight Link and Sink drivers" help This enables support for CoreSight link and sink drivers that are responsible for transporting and collecting the trace data respectively. Link and sinks are dynamically aggregated with a trace entity at run time to form a complete trace path.
+ To compile these drivers as modules, choose M here: the + modules will be called coresight-funnel and coresight-replicator. + config CORESIGHT_LINK_AND_SINK_TMC tristate "Coresight generic TMC driver"
diff --git a/drivers/hwtracing/coresight/coresight-replicator.c b/drivers/hwtracing/coresight/coresight-replicator.c index 78acf29c49ca..05db434c79d0 100644 --- a/drivers/hwtracing/coresight/coresight-replicator.c +++ b/drivers/hwtracing/coresight/coresight-replicator.c @@ -291,6 +291,14 @@ static int replicator_probe(struct device *dev, struct resource *res) return ret; }
+static int __exit replicator_remove(struct device *dev) +{ + struct replicator_drvdata *drvdata = dev_get_drvdata(dev); + + coresight_unregister(drvdata->csdev); + return 0; +} + static int static_replicator_probe(struct platform_device *pdev) { int ret; @@ -310,6 +318,13 @@ static int static_replicator_probe(struct platform_device *pdev) return ret; }
+static int __exit static_replicator_remove(struct platform_device *pdev) +{ + replicator_remove(&pdev->dev); + pm_runtime_disable(&pdev->dev); + return 0; +} + #ifdef CONFIG_PM static int replicator_runtime_suspend(struct device *dev) { @@ -342,25 +357,28 @@ static const struct of_device_id static_replicator_match[] = { {.compatible = "arm,coresight-static-replicator"}, {} }; +MODULE_DEVICE_TABLE(of, static_replicator_match);
#ifdef CONFIG_ACPI static const struct acpi_device_id static_replicator_acpi_ids[] = { {"ARMHC985", 0}, /* ARM CoreSight Static Replicator */ {} }; +MODULE_DEVICE_TABLE(acpi, static_replicator_acpi_ids); #endif
static struct platform_driver static_replicator_driver = { .probe = static_replicator_probe, + .remove = static_replicator_remove, .driver = { .name = "coresight-static-replicator", + .owner = THIS_MODULE, .of_match_table = of_match_ptr(static_replicator_match), .acpi_match_table = ACPI_PTR(static_replicator_acpi_ids), .pm = &replicator_dev_pm_ops, .suppress_bind_attrs = true, }, }; -builtin_platform_driver(static_replicator_driver);
static int dynamic_replicator_probe(struct amba_device *adev, const struct amba_id *id) @@ -368,19 +386,60 @@ static int dynamic_replicator_probe(struct amba_device *adev, return replicator_probe(&adev->dev, &adev->res); }
+static int __exit dynamic_replicator_remove(struct amba_device *adev) +{ + return replicator_remove(&adev->dev); +} + static const struct amba_id dynamic_replicator_ids[] = { CS_AMBA_ID(0x000bb909), CS_AMBA_ID(0x000bb9ec), /* Coresight SoC-600 */ {}, };
+MODULE_DEVICE_TABLE(amba, dynamic_replicator_ids); + static struct amba_driver dynamic_replicator_driver = { .drv = { .name = "coresight-dynamic-replicator", .pm = &replicator_dev_pm_ops, + .owner = THIS_MODULE, .suppress_bind_attrs = true, }, .probe = dynamic_replicator_probe, + .remove = dynamic_replicator_remove, .id_table = dynamic_replicator_ids, }; -builtin_amba_driver(dynamic_replicator_driver); + +static int __init replicator_init(void) +{ + int ret; + + ret = platform_driver_register(&static_replicator_driver); + if (ret) { + pr_info("Error registering platform driver\n"); + return ret; + } + + ret = amba_driver_register(&dynamic_replicator_driver); + if (ret) { + pr_info("Error registering amba driver\n"); + platform_driver_unregister(&static_replicator_driver); + } + + return ret; +} + +static void __exit replicator_exit(void) +{ + platform_driver_unregister(&static_replicator_driver); + amba_driver_unregister(&dynamic_replicator_driver); +} + +module_init(replicator_init); +module_exit(replicator_exit); + +MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org"); +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Replicator Driver"); +MODULE_LICENSE("GPL v2");
On Fri, Aug 07, 2020 at 07:11:44PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-replicator as modules, for ease of development.
- Kconfig becomes a tristate, to allow =m
- combine static and dynamic replicator init into single module_init/exit call
- add replicator_remove functions, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Co-developed-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- .../coresight/coresight-replicator.c | 63 ++++++++++++++++++- 2 files changed, 65 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index fc48ae086746..f31778dd0b5d 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -17,13 +17,16 @@ menuconfig CORESIGHT if CORESIGHT config CORESIGHT_LINKS_AND_SINKS
- bool "CoreSight Link and Sink drivers"
- tristate "CoreSight Link and Sink drivers" help This enables support for CoreSight link and sink drivers that are responsible for transporting and collecting the trace data respectively. Link and sinks are dynamically aggregated with a trace entity at run time to form a complete trace path.
With the passage of time and the refinement of the supported components, the CONFIG_ and description are no longer valid. I'll leave it up to you to decide if you want to fix it as part of this set, the choice is entirely up to you.
To compile these drivers as modules, choose M here: the
modules will be called coresight-funnel and coresight-replicator.
config CORESIGHT_LINK_AND_SINK_TMC tristate "Coresight generic TMC driver" diff --git a/drivers/hwtracing/coresight/coresight-replicator.c b/drivers/hwtracing/coresight/coresight-replicator.c index 78acf29c49ca..05db434c79d0 100644 --- a/drivers/hwtracing/coresight/coresight-replicator.c +++ b/drivers/hwtracing/coresight/coresight-replicator.c @@ -291,6 +291,14 @@ static int replicator_probe(struct device *dev, struct resource *res) return ret; } +static int __exit replicator_remove(struct device *dev) +{
- struct replicator_drvdata *drvdata = dev_get_drvdata(dev);
- coresight_unregister(drvdata->csdev);
- return 0;
+}
static int static_replicator_probe(struct platform_device *pdev) { int ret; @@ -310,6 +318,13 @@ static int static_replicator_probe(struct platform_device *pdev) return ret; } +static int __exit static_replicator_remove(struct platform_device *pdev) +{
- replicator_remove(&pdev->dev);
- pm_runtime_disable(&pdev->dev);
- return 0;
+}
#ifdef CONFIG_PM static int replicator_runtime_suspend(struct device *dev) { @@ -342,25 +357,28 @@ static const struct of_device_id static_replicator_match[] = { {.compatible = "arm,coresight-static-replicator"}, {} }; +MODULE_DEVICE_TABLE(of, static_replicator_match);
Pick a way to do this, a newline or not, but stick with it for _all_ the drivers in this set.
#ifdef CONFIG_ACPI static const struct acpi_device_id static_replicator_acpi_ids[] = { {"ARMHC985", 0}, /* ARM CoreSight Static Replicator */ {} }; +MODULE_DEVICE_TABLE(acpi, static_replicator_acpi_ids);
Same as above
#endif static struct platform_driver static_replicator_driver = { .probe = static_replicator_probe,
- .remove = static_replicator_remove, .driver = { .name = "coresight-static-replicator",
.of_match_table = of_match_ptr(static_replicator_match), .acpi_match_table = ACPI_PTR(static_replicator_acpi_ids), .pm = &replicator_dev_pm_ops, .suppress_bind_attrs = true, },.owner = THIS_MODULE,
}; -builtin_platform_driver(static_replicator_driver); static int dynamic_replicator_probe(struct amba_device *adev, const struct amba_id *id) @@ -368,19 +386,60 @@ static int dynamic_replicator_probe(struct amba_device *adev, return replicator_probe(&adev->dev, &adev->res); } +static int __exit dynamic_replicator_remove(struct amba_device *adev) +{
- return replicator_remove(&adev->dev);
+}
static const struct amba_id dynamic_replicator_ids[] = { CS_AMBA_ID(0x000bb909), CS_AMBA_ID(0x000bb9ec), /* Coresight SoC-600 */ {}, }; +MODULE_DEVICE_TABLE(amba, dynamic_replicator_ids);
static struct amba_driver dynamic_replicator_driver = { .drv = { .name = "coresight-dynamic-replicator", .pm = &replicator_dev_pm_ops,
.suppress_bind_attrs = true, }, .probe = dynamic_replicator_probe,.owner = THIS_MODULE,
- .remove = dynamic_replicator_remove, .id_table = dynamic_replicator_ids,
}; -builtin_amba_driver(dynamic_replicator_driver);
+static int __init replicator_init(void) +{
- int ret;
- ret = platform_driver_register(&static_replicator_driver);
- if (ret) {
pr_info("Error registering platform driver\n");
return ret;
- }
- ret = amba_driver_register(&dynamic_replicator_driver);
- if (ret) {
pr_info("Error registering amba driver\n");
platform_driver_unregister(&static_replicator_driver);
- }
- return ret;
+}
+static void __exit replicator_exit(void) +{
- platform_driver_unregister(&static_replicator_driver);
- amba_driver_unregister(&dynamic_replicator_driver);
+}
Same here, choose to add a newline or not but stay consistent for all the drives.
+module_init(replicator_init); +module_exit(replicator_exit);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org");
Pratik should also be mentioned in the funnel driver.
+MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Replicator Driver");
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Fri, Aug 14, 2020 at 04:35:11AM +0800, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:44PM +0800, Tingwei Zhang wrote:
From: Kim Phillips kim.phillips@arm.com
Allow to build coresight-replicator as modules, for ease of development.
- Kconfig becomes a tristate, to allow =m
- combine static and dynamic replicator init into single module_init/exit call
- add replicator_remove functions, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Co-developed-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 +- .../coresight/coresight-replicator.c | 63 ++++++++++++++++++- 2 files changed, 65 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/Kconfig
b/drivers/hwtracing/coresight/Kconfig
index fc48ae086746..f31778dd0b5d 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -17,13 +17,16 @@ menuconfig CORESIGHT if CORESIGHT config CORESIGHT_LINKS_AND_SINKS
- bool "CoreSight Link and Sink drivers"
- tristate "CoreSight Link and Sink drivers" help This enables support for CoreSight link and sink drivers that
are
responsible for transporting and collecting the trace data respectively. Link and sinks are dynamically aggregated with a
trace
entity at run time to form a complete trace path.
With the passage of time and the refinement of the supported components, the CONFIG_ and description are no longer valid. I'll leave it up to you to decide if you want to fix it as part of this set, the choice is entirely up to you.
I'll leave that to another patch and focus this patch set to make driver as module.
To compile these drivers as modules, choose M here: the
modules will be called coresight-funnel and
coresight-replicator.
config CORESIGHT_LINK_AND_SINK_TMC tristate "Coresight generic TMC driver" diff --git a/drivers/hwtracing/coresight/coresight-replicator.c
b/drivers/hwtracing/coresight/coresight-replicator.c
index 78acf29c49ca..05db434c79d0 100644 --- a/drivers/hwtracing/coresight/coresight-replicator.c +++ b/drivers/hwtracing/coresight/coresight-replicator.c @@ -291,6 +291,14 @@ static int replicator_probe(struct device *dev,
struct resource *res)
return ret; } +static int __exit replicator_remove(struct device *dev) +{
- struct replicator_drvdata *drvdata = dev_get_drvdata(dev);
- coresight_unregister(drvdata->csdev);
- return 0;
+}
static int static_replicator_probe(struct platform_device *pdev) { int ret; @@ -310,6 +318,13 @@ static int static_replicator_probe(struct
platform_device *pdev)
return ret; } +static int __exit static_replicator_remove(struct platform_device
*pdev)
+{
- replicator_remove(&pdev->dev);
- pm_runtime_disable(&pdev->dev);
- return 0;
+}
#ifdef CONFIG_PM static int replicator_runtime_suspend(struct device *dev) { @@ -342,25 +357,28 @@ static const struct of_device_id
static_replicator_match[] = {
{.compatible = "arm,coresight-static-replicator"}, {} }; +MODULE_DEVICE_TABLE(of, static_replicator_match);
Pick a way to do this, a newline or not, but stick with it for _all_ the drivers in this set.
I will add a new line and check all the coresight drivers.
#ifdef CONFIG_ACPI static const struct acpi_device_id static_replicator_acpi_ids[] = { {"ARMHC985", 0}, /* ARM CoreSight Static Replicator */ {} }; +MODULE_DEVICE_TABLE(acpi, static_replicator_acpi_ids);
Same as above
I will add a new line.
#endif static struct platform_driver static_replicator_driver = { .probe = static_replicator_probe,
- .remove = static_replicator_remove, .driver = { .name = "coresight-static-replicator",
.of_match_table = of_match_ptr(static_replicator_match), .acpi_match_table = ACPI_PTR(static_replicator_acpi_ids), .pm = &replicator_dev_pm_ops, .suppress_bind_attrs = true, },.owner = THIS_MODULE,
}; -builtin_platform_driver(static_replicator_driver); static int dynamic_replicator_probe(struct amba_device *adev, const struct amba_id *id) @@ -368,19 +386,60 @@ static int dynamic_replicator_probe(struct
amba_device *adev,
return replicator_probe(&adev->dev, &adev->res); } +static int __exit dynamic_replicator_remove(struct amba_device *adev) +{
- return replicator_remove(&adev->dev);
+}
static const struct amba_id dynamic_replicator_ids[] = { CS_AMBA_ID(0x000bb909), CS_AMBA_ID(0x000bb9ec), /* Coresight SoC-600 */ {}, }; +MODULE_DEVICE_TABLE(amba, dynamic_replicator_ids);
static struct amba_driver dynamic_replicator_driver = { .drv = { .name = "coresight-dynamic-replicator", .pm = &replicator_dev_pm_ops,
.suppress_bind_attrs = true, }, .probe = dynamic_replicator_probe,.owner = THIS_MODULE,
- .remove = dynamic_replicator_remove, .id_table = dynamic_replicator_ids,
}; -builtin_amba_driver(dynamic_replicator_driver);
+static int __init replicator_init(void) +{
- int ret;
- ret = platform_driver_register(&static_replicator_driver);
- if (ret) {
pr_info("Error registering platform driver\n");
return ret;
- }
- ret = amba_driver_register(&dynamic_replicator_driver);
- if (ret) {
pr_info("Error registering amba driver\n");
platform_driver_unregister(&static_replicator_driver);
- }
- return ret;
+}
+static void __exit replicator_exit(void) +{
- platform_driver_unregister(&static_replicator_driver);
- amba_driver_unregister(&dynamic_replicator_driver);
+}
Same here, choose to add a newline or not but stay consistent for all the drives.
Sure.
+module_init(replicator_init); +module_exit(replicator_exit);
+MODULE_AUTHOR("Pratik Patel pratikp@codeaurora.org");
Pratik should also be mentioned in the funnel driver.
I'll add Pratik in v9.
+MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight Replicator Driver");
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project
Add static cti_assoc_ops to coresight core driver. Let cti driver register the add_assoc and remove_assoc call back. Avoid coresight core driver to depend on cti driver.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org --- drivers/hwtracing/coresight/coresight-cti.c | 32 ++++++++++++++++++-- drivers/hwtracing/coresight/coresight-priv.h | 14 ++++----- drivers/hwtracing/coresight/coresight.c | 21 +++++++++++-- 3 files changed, 53 insertions(+), 14 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c index 3ccc703dc940..1f470c47ba70 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -589,7 +589,6 @@ void cti_add_assoc_to_csdev(struct coresight_device *csdev) cti_add_done: mutex_unlock(&ect_mutex); } -EXPORT_SYMBOL_GPL(cti_add_assoc_to_csdev);
/* * Removing the associated devices is easier. @@ -616,7 +615,15 @@ void cti_remove_assoc_from_csdev(struct coresight_device *csdev) } mutex_unlock(&ect_mutex); } -EXPORT_SYMBOL_GPL(cti_remove_assoc_from_csdev); + +/* + * Operations to add and remove associated CTI. + * Register to coresight core driver as call back function. + */ +static struct cti_assoc_op cti_assoc_ops = { + .add = cti_add_assoc_to_csdev, + .remove = cti_remove_assoc_from_csdev +};
/* * Update the cross references where the associated device was found @@ -972,4 +979,23 @@ static struct amba_driver cti_driver = { .probe = cti_probe, .id_table = cti_ids, }; -builtin_amba_driver(cti_driver); + +static int __init cti_init(void) +{ + int ret; + + ret = amba_driver_register(&cti_driver); + if (ret) + pr_info("Error registering cti driver\n"); + coresight_set_cti_ops(&cti_assoc_ops); + return ret; +} + +static void __exit cti_exit(void) +{ + coresight_remove_cti_ops(); + amba_driver_unregister(&cti_driver); +} + +module_init(cti_init); +module_exit(cti_exit); diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index dcb8aeb6af62..6cde6cf42554 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -173,15 +173,13 @@ static inline int etm_readl_cp14(u32 off, unsigned int *val) { return 0; } static inline int etm_writel_cp14(u32 off, u32 val) { return 0; } #endif
-#ifdef CONFIG_CORESIGHT_CTI -extern void cti_add_assoc_to_csdev(struct coresight_device *csdev); -extern void cti_remove_assoc_from_csdev(struct coresight_device *csdev); +struct cti_assoc_op { + void (*add)(struct coresight_device *csdev); + void (*remove)(struct coresight_device *csdev); +};
-#else -static inline void cti_add_assoc_to_csdev(struct coresight_device *csdev) {} -static inline void -cti_remove_assoc_from_csdev(struct coresight_device *csdev) {} -#endif +extern void coresight_set_cti_ops(const struct cti_assoc_op *cti_op); +extern void coresight_remove_cti_ops(void);
/* * Macros and inline functions to handle CoreSight UCI data and driver diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 0b0e31577b9b..31b9ec8d3b9c 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -57,6 +57,20 @@ const u32 coresight_barrier_pkt[4] = { 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff}; EXPORT_SYMBOL_GPL(coresight_barrier_pkt);
+static const struct cti_assoc_op *cti_assoc_ops; + +void coresight_set_cti_ops(const struct cti_assoc_op *cti_op) +{ + cti_assoc_ops = cti_op; +} +EXPORT_SYMBOL_GPL(coresight_set_cti_ops); + +void coresight_remove_cti_ops(void) +{ + cti_assoc_ops = NULL; +} +EXPORT_SYMBOL_GPL(coresight_remove_cti_ops); + static int coresight_id_match(struct device *dev, void *data) { int trace_id, i_trace_id; @@ -1242,7 +1256,8 @@ static void coresight_device_release(struct device *dev) { struct coresight_device *csdev = to_coresight_device(dev);
- cti_remove_assoc_from_csdev(csdev); + if (cti_assoc_ops && cti_assoc_ops->remove) + cti_assoc_ops->remove(csdev); fwnode_handle_put(csdev->dev.fwnode); kfree(csdev->refcnt); kfree(csdev); @@ -1553,8 +1568,8 @@ struct coresight_device *coresight_register(struct coresight_desc *desc) ret = coresight_fixup_device_conns(csdev); if (!ret) ret = coresight_fixup_orphan_conns(csdev); - if (!ret) - cti_add_assoc_to_csdev(csdev); + if (!ret && cti_assoc_ops && cti_assoc_ops->add) + cti_assoc_ops->add(csdev);
mutex_unlock(&coresight_mutex); if (ret) {
On Fri, Aug 07, 2020 at 07:11:45PM +0800, Tingwei Zhang wrote:
Add static cti_assoc_ops to coresight core driver. Let cti driver register the add_assoc and remove_assoc call back. Avoid coresight core driver to depend on cti driver.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/coresight-cti.c | 32 ++++++++++++++++++-- drivers/hwtracing/coresight/coresight-priv.h | 14 ++++----- drivers/hwtracing/coresight/coresight.c | 21 +++++++++++-- 3 files changed, 53 insertions(+), 14 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c index 3ccc703dc940..1f470c47ba70 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -589,7 +589,6 @@ void cti_add_assoc_to_csdev(struct coresight_device *csdev) cti_add_done: mutex_unlock(&ect_mutex); } -EXPORT_SYMBOL_GPL(cti_add_assoc_to_csdev); /*
- Removing the associated devices is easier.
@@ -616,7 +615,15 @@ void cti_remove_assoc_from_csdev(struct coresight_device *csdev) } mutex_unlock(&ect_mutex); } -EXPORT_SYMBOL_GPL(cti_remove_assoc_from_csdev);
Mike and Tingwei,
It seems to me that functions cti_add_assoc_to_csdev() and cti_remote_assoc_from_csdev() don't have to be exported. They could be called in cti_probe() and cti_device_release() with the ect_mutex held, the same way cti_update_conn_xrefs() and cti_remove_conn_xrefs() are.
If my assessment is correct this patch would be simpler.
Thanks, Mathieu
+/*
- Operations to add and remove associated CTI.
- Register to coresight core driver as call back function.
- */
+static struct cti_assoc_op cti_assoc_ops = {
- .add = cti_add_assoc_to_csdev,
- .remove = cti_remove_assoc_from_csdev
+}; /*
- Update the cross references where the associated device was found
@@ -972,4 +979,23 @@ static struct amba_driver cti_driver = { .probe = cti_probe, .id_table = cti_ids, }; -builtin_amba_driver(cti_driver);
+static int __init cti_init(void) +{
- int ret;
- ret = amba_driver_register(&cti_driver);
- if (ret)
pr_info("Error registering cti driver\n");
- coresight_set_cti_ops(&cti_assoc_ops);
- return ret;
+}
+static void __exit cti_exit(void) +{
- coresight_remove_cti_ops();
- amba_driver_unregister(&cti_driver);
+}
+module_init(cti_init); +module_exit(cti_exit); diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index dcb8aeb6af62..6cde6cf42554 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -173,15 +173,13 @@ static inline int etm_readl_cp14(u32 off, unsigned int *val) { return 0; } static inline int etm_writel_cp14(u32 off, u32 val) { return 0; } #endif -#ifdef CONFIG_CORESIGHT_CTI -extern void cti_add_assoc_to_csdev(struct coresight_device *csdev); -extern void cti_remove_assoc_from_csdev(struct coresight_device *csdev); +struct cti_assoc_op {
- void (*add)(struct coresight_device *csdev);
- void (*remove)(struct coresight_device *csdev);
+}; -#else -static inline void cti_add_assoc_to_csdev(struct coresight_device *csdev) {} -static inline void -cti_remove_assoc_from_csdev(struct coresight_device *csdev) {} -#endif +extern void coresight_set_cti_ops(const struct cti_assoc_op *cti_op); +extern void coresight_remove_cti_ops(void); /*
- Macros and inline functions to handle CoreSight UCI data and driver
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 0b0e31577b9b..31b9ec8d3b9c 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -57,6 +57,20 @@ const u32 coresight_barrier_pkt[4] = { 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff}; EXPORT_SYMBOL_GPL(coresight_barrier_pkt); +static const struct cti_assoc_op *cti_assoc_ops;
+void coresight_set_cti_ops(const struct cti_assoc_op *cti_op) +{
- cti_assoc_ops = cti_op;
+} +EXPORT_SYMBOL_GPL(coresight_set_cti_ops);
+void coresight_remove_cti_ops(void) +{
- cti_assoc_ops = NULL;
+} +EXPORT_SYMBOL_GPL(coresight_remove_cti_ops);
static int coresight_id_match(struct device *dev, void *data) { int trace_id, i_trace_id; @@ -1242,7 +1256,8 @@ static void coresight_device_release(struct device *dev) { struct coresight_device *csdev = to_coresight_device(dev);
- cti_remove_assoc_from_csdev(csdev);
- if (cti_assoc_ops && cti_assoc_ops->remove)
fwnode_handle_put(csdev->dev.fwnode); kfree(csdev->refcnt); kfree(csdev);cti_assoc_ops->remove(csdev);
@@ -1553,8 +1568,8 @@ struct coresight_device *coresight_register(struct coresight_desc *desc) ret = coresight_fixup_device_conns(csdev); if (!ret) ret = coresight_fixup_orphan_conns(csdev);
- if (!ret)
cti_add_assoc_to_csdev(csdev);
- if (!ret && cti_assoc_ops && cti_assoc_ops->add)
cti_assoc_ops->add(csdev);
mutex_unlock(&coresight_mutex); if (ret) { -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Hi Mathieu,
On Mon, 17 Aug 2020 at 17:02, Mathieu Poirier mathieu.poirier@linaro.org wrote:
On Fri, Aug 07, 2020 at 07:11:45PM +0800, Tingwei Zhang wrote:
Add static cti_assoc_ops to coresight core driver. Let cti driver register the add_assoc and remove_assoc call back. Avoid coresight core driver to depend on cti driver.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/coresight-cti.c | 32 ++++++++++++++++++-- drivers/hwtracing/coresight/coresight-priv.h | 14 ++++----- drivers/hwtracing/coresight/coresight.c | 21 +++++++++++-- 3 files changed, 53 insertions(+), 14 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c index 3ccc703dc940..1f470c47ba70 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -589,7 +589,6 @@ void cti_add_assoc_to_csdev(struct coresight_device *csdev) cti_add_done: mutex_unlock(&ect_mutex); } -EXPORT_SYMBOL_GPL(cti_add_assoc_to_csdev);
/*
- Removing the associated devices is easier.
@@ -616,7 +615,15 @@ void cti_remove_assoc_from_csdev(struct coresight_device *csdev) } mutex_unlock(&ect_mutex); } -EXPORT_SYMBOL_GPL(cti_remove_assoc_from_csdev);
Mike and Tingwei,
It seems to me that functions cti_add_assoc_to_csdev() and cti_remote_assoc_from_csdev() don't have to be exported. They could be called in cti_probe() and cti_device_release() with the ect_mutex held, the same way cti_update_conn_xrefs() and cti_remove_conn_xrefs() are.
The different functions handle two different initialisation & two different teardown ordering scenarios a) csdev - e.g. ETM is discovered & initialised before the associated CTI. Therefore the CTI has to establish interconnections on creation - calls cti_update_conn_xrefs() . b) CTI is discovered before csdev - therefore the csdev calls cti_add_assoc_to_csdev(). c) csdev is torn down before CTI - therefore csdev calls cti_remove_assoc_from_csdev() d) CTI is torn down before csdev - therefore calls cti_remove_conn_xrefs()
We can't call cti_add_assoc_to_csdev() in cti probe as the csdev does not exist yet. Bear in mind that one CTI can actually be associated with multiple different csdev devices - especially in the none CPU bound infrastructure such as funnels and sinks. This has always been the case since the first CTI set - but is even more important now the modules can be removed / added as required.
Regards
Mike
If my assessment is correct this patch would be simpler.
Thanks, Mathieu
+/*
- Operations to add and remove associated CTI.
- Register to coresight core driver as call back function.
- */
+static struct cti_assoc_op cti_assoc_ops = {
.add = cti_add_assoc_to_csdev,
.remove = cti_remove_assoc_from_csdev
+};
/*
- Update the cross references where the associated device was found
@@ -972,4 +979,23 @@ static struct amba_driver cti_driver = { .probe = cti_probe, .id_table = cti_ids, }; -builtin_amba_driver(cti_driver);
+static int __init cti_init(void) +{
int ret;
ret = amba_driver_register(&cti_driver);
if (ret)
pr_info("Error registering cti driver\n");
coresight_set_cti_ops(&cti_assoc_ops);
return ret;
+}
+static void __exit cti_exit(void) +{
coresight_remove_cti_ops();
amba_driver_unregister(&cti_driver);
+}
+module_init(cti_init); +module_exit(cti_exit); diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index dcb8aeb6af62..6cde6cf42554 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -173,15 +173,13 @@ static inline int etm_readl_cp14(u32 off, unsigned int *val) { return 0; } static inline int etm_writel_cp14(u32 off, u32 val) { return 0; } #endif
-#ifdef CONFIG_CORESIGHT_CTI -extern void cti_add_assoc_to_csdev(struct coresight_device *csdev); -extern void cti_remove_assoc_from_csdev(struct coresight_device *csdev); +struct cti_assoc_op {
void (*add)(struct coresight_device *csdev);
void (*remove)(struct coresight_device *csdev);
+};
-#else -static inline void cti_add_assoc_to_csdev(struct coresight_device *csdev) {} -static inline void -cti_remove_assoc_from_csdev(struct coresight_device *csdev) {} -#endif +extern void coresight_set_cti_ops(const struct cti_assoc_op *cti_op); +extern void coresight_remove_cti_ops(void);
/*
- Macros and inline functions to handle CoreSight UCI data and driver
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 0b0e31577b9b..31b9ec8d3b9c 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -57,6 +57,20 @@ const u32 coresight_barrier_pkt[4] = { 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff}; EXPORT_SYMBOL_GPL(coresight_barrier_pkt);
+static const struct cti_assoc_op *cti_assoc_ops;
+void coresight_set_cti_ops(const struct cti_assoc_op *cti_op) +{
cti_assoc_ops = cti_op;
+} +EXPORT_SYMBOL_GPL(coresight_set_cti_ops);
+void coresight_remove_cti_ops(void) +{
cti_assoc_ops = NULL;
+} +EXPORT_SYMBOL_GPL(coresight_remove_cti_ops);
static int coresight_id_match(struct device *dev, void *data) { int trace_id, i_trace_id; @@ -1242,7 +1256,8 @@ static void coresight_device_release(struct device *dev) { struct coresight_device *csdev = to_coresight_device(dev);
cti_remove_assoc_from_csdev(csdev);
if (cti_assoc_ops && cti_assoc_ops->remove)
cti_assoc_ops->remove(csdev); fwnode_handle_put(csdev->dev.fwnode); kfree(csdev->refcnt); kfree(csdev);
@@ -1553,8 +1568,8 @@ struct coresight_device *coresight_register(struct coresight_desc *desc) ret = coresight_fixup_device_conns(csdev); if (!ret) ret = coresight_fixup_orphan_conns(csdev);
if (!ret)
cti_add_assoc_to_csdev(csdev);
if (!ret && cti_assoc_ops && cti_assoc_ops->add)
cti_assoc_ops->add(csdev); mutex_unlock(&coresight_mutex); if (ret) {
-- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Good day,
On Mon, Aug 17, 2020 at 07:49:16PM +0100, Mike Leach wrote:
Hi Mathieu,
On Mon, 17 Aug 2020 at 17:02, Mathieu Poirier mathieu.poirier@linaro.org wrote:
On Fri, Aug 07, 2020 at 07:11:45PM +0800, Tingwei Zhang wrote:
Add static cti_assoc_ops to coresight core driver. Let cti driver register the add_assoc and remove_assoc call back. Avoid coresight core driver to depend on cti driver.
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/coresight-cti.c | 32 ++++++++++++++++++-- drivers/hwtracing/coresight/coresight-priv.h | 14 ++++----- drivers/hwtracing/coresight/coresight.c | 21 +++++++++++-- 3 files changed, 53 insertions(+), 14 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c index 3ccc703dc940..1f470c47ba70 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -589,7 +589,6 @@ void cti_add_assoc_to_csdev(struct coresight_device *csdev) cti_add_done: mutex_unlock(&ect_mutex); } -EXPORT_SYMBOL_GPL(cti_add_assoc_to_csdev);
/*
- Removing the associated devices is easier.
@@ -616,7 +615,15 @@ void cti_remove_assoc_from_csdev(struct coresight_device *csdev) } mutex_unlock(&ect_mutex); } -EXPORT_SYMBOL_GPL(cti_remove_assoc_from_csdev);
Mike and Tingwei,
It seems to me that functions cti_add_assoc_to_csdev() and cti_remote_assoc_from_csdev() don't have to be exported. They could be called in cti_probe() and cti_device_release() with the ect_mutex held, the same way cti_update_conn_xrefs() and cti_remove_conn_xrefs() are.
The different functions handle two different initialisation & two different teardown ordering scenarios a) csdev - e.g. ETM is discovered & initialised before the associated CTI. Therefore the CTI has to establish interconnections on creation - calls cti_update_conn_xrefs() . b) CTI is discovered before csdev - therefore the csdev calls cti_add_assoc_to_csdev(). c) csdev is torn down before CTI - therefore csdev calls cti_remove_assoc_from_csdev() d) CTI is torn down before csdev - therefore calls cti_remove_conn_xrefs()
We can't call cti_add_assoc_to_csdev() in cti probe as the csdev does not exist yet. Bear in mind that one CTI can actually be associated with multiple different csdev devices - especially in the none CPU bound infrastructure such as funnels and sinks.
Yes, I see your point now. The problem needs to be viewed from other devices' (than CTI) perspective. And that part is done in coresight_register/unregister().
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
This has always been the case since the first CTI set - but is even more important now the modules can be removed / added as required.
Regards
Mike
If my assessment is correct this patch would be simpler.
Thanks, Mathieu
+/*
- Operations to add and remove associated CTI.
- Register to coresight core driver as call back function.
- */
+static struct cti_assoc_op cti_assoc_ops = {
.add = cti_add_assoc_to_csdev,
.remove = cti_remove_assoc_from_csdev
+};
/*
- Update the cross references where the associated device was found
@@ -972,4 +979,23 @@ static struct amba_driver cti_driver = { .probe = cti_probe, .id_table = cti_ids, }; -builtin_amba_driver(cti_driver);
+static int __init cti_init(void) +{
int ret;
ret = amba_driver_register(&cti_driver);
if (ret)
pr_info("Error registering cti driver\n");
coresight_set_cti_ops(&cti_assoc_ops);
return ret;
+}
+static void __exit cti_exit(void) +{
coresight_remove_cti_ops();
amba_driver_unregister(&cti_driver);
+}
+module_init(cti_init); +module_exit(cti_exit); diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index dcb8aeb6af62..6cde6cf42554 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -173,15 +173,13 @@ static inline int etm_readl_cp14(u32 off, unsigned int *val) { return 0; } static inline int etm_writel_cp14(u32 off, u32 val) { return 0; } #endif
-#ifdef CONFIG_CORESIGHT_CTI -extern void cti_add_assoc_to_csdev(struct coresight_device *csdev); -extern void cti_remove_assoc_from_csdev(struct coresight_device *csdev); +struct cti_assoc_op {
void (*add)(struct coresight_device *csdev);
void (*remove)(struct coresight_device *csdev);
+};
-#else -static inline void cti_add_assoc_to_csdev(struct coresight_device *csdev) {} -static inline void -cti_remove_assoc_from_csdev(struct coresight_device *csdev) {} -#endif +extern void coresight_set_cti_ops(const struct cti_assoc_op *cti_op); +extern void coresight_remove_cti_ops(void);
/*
- Macros and inline functions to handle CoreSight UCI data and driver
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 0b0e31577b9b..31b9ec8d3b9c 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -57,6 +57,20 @@ const u32 coresight_barrier_pkt[4] = { 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff}; EXPORT_SYMBOL_GPL(coresight_barrier_pkt);
+static const struct cti_assoc_op *cti_assoc_ops;
+void coresight_set_cti_ops(const struct cti_assoc_op *cti_op) +{
cti_assoc_ops = cti_op;
+} +EXPORT_SYMBOL_GPL(coresight_set_cti_ops);
+void coresight_remove_cti_ops(void) +{
cti_assoc_ops = NULL;
+} +EXPORT_SYMBOL_GPL(coresight_remove_cti_ops);
static int coresight_id_match(struct device *dev, void *data) { int trace_id, i_trace_id; @@ -1242,7 +1256,8 @@ static void coresight_device_release(struct device *dev) { struct coresight_device *csdev = to_coresight_device(dev);
cti_remove_assoc_from_csdev(csdev);
if (cti_assoc_ops && cti_assoc_ops->remove)
cti_assoc_ops->remove(csdev); fwnode_handle_put(csdev->dev.fwnode); kfree(csdev->refcnt); kfree(csdev);
@@ -1553,8 +1568,8 @@ struct coresight_device *coresight_register(struct coresight_desc *desc) ret = coresight_fixup_device_conns(csdev); if (!ret) ret = coresight_fixup_orphan_conns(csdev);
if (!ret)
cti_add_assoc_to_csdev(csdev);
if (!ret && cti_assoc_ops && cti_assoc_ops->add)
cti_assoc_ops->add(csdev); mutex_unlock(&coresight_mutex); if (ret) {
-- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
-- Mike Leach Principal Engineer, ARM Ltd. Manchester Design Centre. UK
From: Mike Leach mike.leach@linaro.org
CTI code to remove sysfs link to other devices on shutdown, incorrectly tries to remove a single ended link when these are all double ended. This implementation leaves elements in the link info structure undefined which results in a crash in recent tests for driver module unload.
This patch corrects the link removal code.
Fixes: 73274abb6557 (coresight: cti: Add in sysfs links to other coresight devices) Reported-by: Tingwei Zhang tingwei@codeaurora.org Signed-off-by: Mike Leach mike.leach@linaro.org --- drivers/hwtracing/coresight/coresight-cti.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c index 1f470c47ba70..1c34db7d1b1b 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -511,12 +511,15 @@ static bool cti_add_sysfs_link(struct cti_drvdata *drvdata, return !link_err; }
-static void cti_remove_sysfs_link(struct cti_trig_con *tc) +static void cti_remove_sysfs_link(struct cti_drvdata *drvdata, + struct cti_trig_con *tc) { struct coresight_sysfs_link link_info;
+ link_info.orig = drvdata->csdev; link_info.orig_name = tc->con_dev_name; link_info.target = tc->con_dev; + link_info.target_name = dev_name(&drvdata->csdev->dev); coresight_remove_sysfs_link(&link_info); }
@@ -606,7 +609,7 @@ void cti_remove_assoc_from_csdev(struct coresight_device *csdev) ctidev = &ctidrv->ctidev; list_for_each_entry(tc, &ctidev->trig_cons, node) { if (tc->con_dev == csdev->ect_dev) { - cti_remove_sysfs_link(tc); + cti_remove_sysfs_link(ctidrv, tc); tc->con_dev = NULL; break; } @@ -658,7 +661,7 @@ static void cti_remove_conn_xrefs(struct cti_drvdata *drvdata) if (tc->con_dev) { coresight_set_assoc_ectdev_mutex(tc->con_dev, NULL); - cti_remove_sysfs_link(tc); + cti_remove_sysfs_link(drvdata, tc); tc->con_dev = NULL; } }
On Fri, Aug 07, 2020 at 07:11:46PM +0800, Tingwei Zhang wrote:
From: Mike Leach mike.leach@linaro.org
CTI code to remove sysfs link to other devices on shutdown, incorrectly tries to remove a single ended link when these are all double ended. This implementation leaves elements in the link info structure undefined which results in a crash in recent tests for driver module unload.
This patch corrects the link removal code.
Fixes: 73274abb6557 (coresight: cti: Add in sysfs links to other coresight devices)
This should be:
Fixes: 73274abb6557 ("coresight: cti: Add in sysfs links to other coresight devices")
Reported-by: Tingwei Zhang tingwei@codeaurora.org Signed-off-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/coresight-cti.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c index 1f470c47ba70..1c34db7d1b1b 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -511,12 +511,15 @@ static bool cti_add_sysfs_link(struct cti_drvdata *drvdata, return !link_err; } -static void cti_remove_sysfs_link(struct cti_trig_con *tc) +static void cti_remove_sysfs_link(struct cti_drvdata *drvdata,
struct cti_trig_con *tc)
{ struct coresight_sysfs_link link_info;
- link_info.orig = drvdata->csdev; link_info.orig_name = tc->con_dev_name; link_info.target = tc->con_dev;
- link_info.target_name = dev_name(&drvdata->csdev->dev); coresight_remove_sysfs_link(&link_info);
} @@ -606,7 +609,7 @@ void cti_remove_assoc_from_csdev(struct coresight_device *csdev) ctidev = &ctidrv->ctidev; list_for_each_entry(tc, &ctidev->trig_cons, node) { if (tc->con_dev == csdev->ect_dev) {
cti_remove_sysfs_link(tc);
cti_remove_sysfs_link(ctidrv, tc); tc->con_dev = NULL; break; }
@@ -658,7 +661,7 @@ static void cti_remove_conn_xrefs(struct cti_drvdata *drvdata) if (tc->con_dev) { coresight_set_assoc_ectdev_mutex(tc->con_dev, NULL);
cti_remove_sysfs_link(tc);
cti_remove_sysfs_link(drvdata, tc);
With the above:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
tc->con_dev = NULL; }
}
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Tue, Aug 18, 2020 at 12:15:31AM +0800, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:46PM +0800, Tingwei Zhang wrote:
From: Mike Leach mike.leach@linaro.org
CTI code to remove sysfs link to other devices on shutdown, incorrectly tries to remove a single ended link when these are all double ended.
This
implementation leaves elements in the link info structure undefined
which
results in a crash in recent tests for driver module unload.
This patch corrects the link removal code.
Fixes: 73274abb6557 (coresight: cti: Add in sysfs links to other
coresight devices)
This should be:
Fixes: 73274abb6557 ("coresight: cti: Add in sysfs links to other coresight devices")
Sure. I'll put this into v9.
Thanks, Tingwei
Reported-by: Tingwei Zhang tingwei@codeaurora.org Signed-off-by: Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/coresight-cti.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c
b/drivers/hwtracing/coresight/coresight-cti.c
index 1f470c47ba70..1c34db7d1b1b 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -511,12 +511,15 @@ static bool cti_add_sysfs_link(struct cti_drvdata
*drvdata,
return !link_err; } -static void cti_remove_sysfs_link(struct cti_trig_con *tc) +static void cti_remove_sysfs_link(struct cti_drvdata *drvdata,
struct cti_trig_con *tc)
{ struct coresight_sysfs_link link_info;
- link_info.orig = drvdata->csdev; link_info.orig_name = tc->con_dev_name; link_info.target = tc->con_dev;
- link_info.target_name = dev_name(&drvdata->csdev->dev); coresight_remove_sysfs_link(&link_info);
} @@ -606,7 +609,7 @@ void cti_remove_assoc_from_csdev(struct
coresight_device *csdev)
ctidev = &ctidrv->ctidev; list_for_each_entry(tc, &ctidev->trig_cons, node) { if (tc->con_dev == csdev->ect_dev) {
cti_remove_sysfs_link(tc);
cti_remove_sysfs_link(ctidrv, tc); tc->con_dev = NULL; break; }
@@ -658,7 +661,7 @@ static void cti_remove_conn_xrefs(struct cti_drvdata
*drvdata)
if (tc->con_dev) { coresight_set_assoc_ectdev_mutex(tc->con_dev, NULL);
cti_remove_sysfs_link(tc);
cti_remove_sysfs_link(drvdata, tc);
With the above:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
tc->con_dev = NULL; }
}
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project
From: Mike Leach mike.leach@linaro.org
During module unload, a coresight driver module will call back into the CTI driver to remove any links between the two devices.
The current code has 2 issues:- 1) in the CTI driver the matching code is matching to the wrong device so misses all the links. 2) The callback is called too late in the unload process resulting in a crash.
This fixes both the issues.
Fixes: 177af8285b59 (coresight: cti: Enable CTI associated with devices) Reported-by: Tingwei Zhang tingwei@codeaurora.org Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Acked-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight-cti.c | 2 +- drivers/hwtracing/coresight/coresight.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c index 1c34db7d1b1b..6bb06c5715db 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -608,7 +608,7 @@ void cti_remove_assoc_from_csdev(struct coresight_device *csdev) ctidrv = csdev_to_cti_drvdata(csdev->ect_dev); ctidev = &ctidrv->ctidev; list_for_each_entry(tc, &ctidev->trig_cons, node) { - if (tc->con_dev == csdev->ect_dev) { + if (tc->con_dev == csdev) { cti_remove_sysfs_link(ctidrv, tc); tc->con_dev = NULL; break; diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 31b9ec8d3b9c..d066411aa794 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -1256,8 +1256,6 @@ static void coresight_device_release(struct device *dev) { struct coresight_device *csdev = to_coresight_device(dev);
- if (cti_assoc_ops && cti_assoc_ops->remove) - cti_assoc_ops->remove(csdev); fwnode_handle_put(csdev->dev.fwnode); kfree(csdev->refcnt); kfree(csdev); @@ -1592,6 +1590,8 @@ void coresight_unregister(struct coresight_device *csdev) { etm_perf_del_symlink_sink(csdev); /* Remove references of that device in the topology */ + if (cti_assoc_ops && cti_assoc_ops->remove) + cti_assoc_ops->remove(csdev); coresight_remove_conns(csdev); coresight_clear_default_sink(csdev); coresight_release_platform_data(csdev, csdev->pdata);
On Fri, Aug 07, 2020 at 07:11:47PM +0800, Tingwei Zhang wrote:
From: Mike Leach mike.leach@linaro.org
During module unload, a coresight driver module will call back into the CTI driver to remove any links between the two devices.
The current code has 2 issues:-
- in the CTI driver the matching code is matching to the wrong device
so misses all the links. 2) The callback is called too late in the unload process resulting in a crash.
This fixes both the issues.
This patch may no longer be required if my recommendation from patch 16 is followed.
Fixes: 177af8285b59 (coresight: cti: Enable CTI associated with devices)
Same comment as in my previous patch, we need a set of double quotes for the title.
Reported-by: Tingwei Zhang tingwei@codeaurora.org Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Acked-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/coresight-cti.c | 2 +- drivers/hwtracing/coresight/coresight.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c index 1c34db7d1b1b..6bb06c5715db 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti.c @@ -608,7 +608,7 @@ void cti_remove_assoc_from_csdev(struct coresight_device *csdev) ctidrv = csdev_to_cti_drvdata(csdev->ect_dev); ctidev = &ctidrv->ctidev; list_for_each_entry(tc, &ctidev->trig_cons, node) {
if (tc->con_dev == csdev->ect_dev) {
if (tc->con_dev == csdev) { cti_remove_sysfs_link(ctidrv, tc); tc->con_dev = NULL; break;
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 31b9ec8d3b9c..d066411aa794 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -1256,8 +1256,6 @@ static void coresight_device_release(struct device *dev) { struct coresight_device *csdev = to_coresight_device(dev);
- if (cti_assoc_ops && cti_assoc_ops->remove)
fwnode_handle_put(csdev->dev.fwnode); kfree(csdev->refcnt); kfree(csdev);cti_assoc_ops->remove(csdev);
@@ -1592,6 +1590,8 @@ void coresight_unregister(struct coresight_device *csdev) { etm_perf_del_symlink_sink(csdev); /* Remove references of that device in the topology */
- if (cti_assoc_ops && cti_assoc_ops->remove)
coresight_remove_conns(csdev); coresight_clear_default_sink(csdev); coresight_release_platform_data(csdev, csdev->pdata);cti_assoc_ops->remove(csdev);
-- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
If associated ect device is not enabled at first place, disable routine should not be called. Add ect_enabled flag to check whether ect device is enabled. Fix the issue in below case. Ect device is not available when associated coresight device enabled and the association is established after coresight device is enabled.
Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org --- drivers/hwtracing/coresight/coresight.c | 11 ++++++++--- include/linux/coresight.h | 1 + 2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index d066411aa794..27ad8317e3cf 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -245,13 +245,18 @@ coresight_control_assoc_ectdev(struct coresight_device *csdev, bool enable)
if (!ect_csdev) return 0; + if ((!ect_ops(ect_csdev)->enable) || (!ect_ops(ect_csdev)->disable)) + return 0;
if (enable) { - if (ect_ops(ect_csdev)->enable) - ect_ret = ect_ops(ect_csdev)->enable(ect_csdev); + ect_ret = ect_ops(ect_csdev)->enable(ect_csdev); + if (!ect_ret) + csdev->ect_enabled = true; } else { - if (ect_ops(ect_csdev)->disable) + if (csdev->ect_enabled) { ect_ret = ect_ops(ect_csdev)->disable(ect_csdev); + csdev->ect_enabled = false; + } }
/* output warning if ECT enable is preventing trace operation */ diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 3bb738f9a326..7d3c87e5b97c 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -208,6 +208,7 @@ struct coresight_device { /* sysfs links between components */ int nr_links; bool has_conns_grp; + bool ect_enabled; /* true only if associated ect device is enabled */ };
/*
On Fri, Aug 07, 2020 at 07:11:48PM +0800, Tingwei Zhang wrote:
If associated ect device is not enabled at first place, disable routine should not be called. Add ect_enabled flag to check whether ect device is enabled. Fix the issue in below case. Ect device is not available when associated coresight device enabled and the association is established after coresight device is enabled.
Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org
drivers/hwtracing/coresight/coresight.c | 11 ++++++++--- include/linux/coresight.h | 1 + 2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index d066411aa794..27ad8317e3cf 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -245,13 +245,18 @@ coresight_control_assoc_ectdev(struct coresight_device *csdev, bool enable) if (!ect_csdev) return 0;
- if ((!ect_ops(ect_csdev)->enable) || (!ect_ops(ect_csdev)->disable))
return 0;
if (enable) {
if (ect_ops(ect_csdev)->enable)
ect_ret = ect_ops(ect_csdev)->enable(ect_csdev);
ect_ret = ect_ops(ect_csdev)->enable(ect_csdev);
if (!ect_ret)
} else {csdev->ect_enabled = true;
if (ect_ops(ect_csdev)->disable)
if (csdev->ect_enabled) { ect_ret = ect_ops(ect_csdev)->disable(ect_csdev);
csdev->ect_enabled = false;
}}
/* output warning if ECT enable is preventing trace operation */ diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 3bb738f9a326..7d3c87e5b97c 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -208,6 +208,7 @@ struct coresight_device { /* sysfs links between components */ int nr_links; bool has_conns_grp;
- bool ect_enabled; /* true only if associated ect device is enabled */
We have cti_config::enable_req_count and cti_config::hw_enabled, both used in cti_enable_hw() and cti_disable_hw(). I would have thought they'd be sufficient to address the counting problems. If they are not I would much rather see a solution confined to the cti driver than in the core itself.
Thanks, Mathieu
}; /* -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Hi Mathieu,
On Mon, 17 Aug 2020 at 17:38, Mathieu Poirier mathieu.poirier@linaro.org wrote:
On Fri, Aug 07, 2020 at 07:11:48PM +0800, Tingwei Zhang wrote:
If associated ect device is not enabled at first place, disable routine should not be called. Add ect_enabled flag to check whether ect device is enabled. Fix the issue in below case. Ect device is not available when associated coresight device enabled and the association is established after coresight device is enabled.
Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org
drivers/hwtracing/coresight/coresight.c | 11 ++++++++--- include/linux/coresight.h | 1 + 2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index d066411aa794..27ad8317e3cf 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -245,13 +245,18 @@ coresight_control_assoc_ectdev(struct coresight_device *csdev, bool enable)
if (!ect_csdev) return 0;
if ((!ect_ops(ect_csdev)->enable) || (!ect_ops(ect_csdev)->disable))
return 0; if (enable) {
if (ect_ops(ect_csdev)->enable)
ect_ret = ect_ops(ect_csdev)->enable(ect_csdev);
ect_ret = ect_ops(ect_csdev)->enable(ect_csdev);
if (!ect_ret)
csdev->ect_enabled = true; } else {
if (ect_ops(ect_csdev)->disable)
if (csdev->ect_enabled) { ect_ret = ect_ops(ect_csdev)->disable(ect_csdev);
csdev->ect_enabled = false;
} } /* output warning if ECT enable is preventing trace operation */
diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 3bb738f9a326..7d3c87e5b97c 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -208,6 +208,7 @@ struct coresight_device { /* sysfs links between components */ int nr_links; bool has_conns_grp;
bool ect_enabled; /* true only if associated ect device is enabled */
We have cti_config::enable_req_count and cti_config::hw_enabled, both used in cti_enable_hw() and cti_disable_hw(). I would have thought they'd be sufficient to address the counting problems. If they are not I would much rather see a solution confined to the cti driver than in the core itself.
This is related to the fact that under sysfs it is possible under sysfs to enable an etm e.g. etm1 without the cti module present, then insert the CTI module, then enable another ETM e.g etm2. This is an issue that is caused by the possibility of module load and unload, and though inadvisable from a system usage point of view - Tingwei correctly points out that it could happen.
At the point that the first ETM is enabled, the associated ect pointer would be NULL, and thus no attempt to enable the ect/CTI is made. The CTI module on load will set the ect pointers on all registered csdev devices, including ones that are already enabled - etm1. So when we come to disable etm1, it will try to disable a CTI that it did not enable - a fact that cannot be counted in the CTI driver as it was not there when the etm was enabled. So we have a flag in csdev to record if this csdev did in fact enable the associated device, so it is clear to disable it on shutdown.
Regards
Mike
Thanks, Mathieu
};
/*
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Mon, Aug 17, 2020 at 08:04:06PM +0100, Mike Leach wrote:
Hi Mathieu,
On Mon, 17 Aug 2020 at 17:38, Mathieu Poirier mathieu.poirier@linaro.org wrote:
On Fri, Aug 07, 2020 at 07:11:48PM +0800, Tingwei Zhang wrote:
If associated ect device is not enabled at first place, disable routine should not be called. Add ect_enabled flag to check whether ect device is enabled. Fix the issue in below case. Ect device is not available when associated coresight device enabled and the association is established after coresight device is enabled.
Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org
drivers/hwtracing/coresight/coresight.c | 11 ++++++++--- include/linux/coresight.h | 1 + 2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index d066411aa794..27ad8317e3cf 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -245,13 +245,18 @@ coresight_control_assoc_ectdev(struct coresight_device *csdev, bool enable)
if (!ect_csdev) return 0;
if ((!ect_ops(ect_csdev)->enable) || (!ect_ops(ect_csdev)->disable))
return 0; if (enable) {
if (ect_ops(ect_csdev)->enable)
ect_ret = ect_ops(ect_csdev)->enable(ect_csdev);
ect_ret = ect_ops(ect_csdev)->enable(ect_csdev);
if (!ect_ret)
csdev->ect_enabled = true; } else {
if (ect_ops(ect_csdev)->disable)
if (csdev->ect_enabled) { ect_ret = ect_ops(ect_csdev)->disable(ect_csdev);
csdev->ect_enabled = false;
} } /* output warning if ECT enable is preventing trace operation */
diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 3bb738f9a326..7d3c87e5b97c 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -208,6 +208,7 @@ struct coresight_device { /* sysfs links between components */ int nr_links; bool has_conns_grp;
bool ect_enabled; /* true only if associated ect device is enabled */
We have cti_config::enable_req_count and cti_config::hw_enabled, both used in cti_enable_hw() and cti_disable_hw(). I would have thought they'd be sufficient to address the counting problems. If they are not I would much rather see a solution confined to the cti driver than in the core itself.
This is related to the fact that under sysfs it is possible under sysfs to enable an etm e.g. etm1 without the cti module present, then insert the CTI module, then enable another ETM e.g etm2. This is an issue that is caused by the possibility of module load and unload, and though inadvisable from a system usage point of view - Tingwei correctly points out that it could happen.
At the point that the first ETM is enabled, the associated ect pointer would be NULL, and thus no attempt to enable the ect/CTI is made. The CTI module on load will set the ect pointers on all registered csdev devices, including ones that are already enabled - etm1. So when we come to disable etm1, it will try to disable a CTI that it did not enable - a fact that cannot be counted in the CTI driver as it was not there when the etm was enabled. So we have a flag in csdev to record if this csdev did in fact enable the associated device, so it is clear to disable it on shutdown.
In the above example the csdev structures associated with ETM1 and ETM2 will have different csdev->ect_dev pointers, one for each CTI it is assocated with. And I would think the reference count for CTI1 would not have been incremented since it was never enabled.
What am I missing?
Regards
Mike
Thanks, Mathieu
};
/*
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
-- Mike Leach Principal Engineer, ARM Ltd. Manchester Design Centre. UK
On Wed, Aug 19, 2020 at 01:39:09AM +0800, Mathieu Poirier wrote:
On Mon, Aug 17, 2020 at 08:04:06PM +0100, Mike Leach wrote:
Hi Mathieu,
On Mon, 17 Aug 2020 at 17:38, Mathieu Poirier mathieu.poirier@linaro.org wrote:
On Fri, Aug 07, 2020 at 07:11:48PM +0800, Tingwei Zhang wrote:
If associated ect device is not enabled at first place, disable routine should not be called. Add ect_enabled flag to check whether ect device is enabled. Fix the issue in below case. Ect device is not available when associated coresight device enabled and the association is established after coresight device is enabled.
Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org
drivers/hwtracing/coresight/coresight.c | 11 ++++++++--- include/linux/coresight.h | 1 + 2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c
b/drivers/hwtracing/coresight/coresight.c
index d066411aa794..27ad8317e3cf 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -245,13 +245,18 @@ coresight_control_assoc_ectdev(struct
coresight_device *csdev, bool enable)
if (!ect_csdev) return 0;
if ((!ect_ops(ect_csdev)->enable) ||
(!ect_ops(ect_csdev)->disable))
return 0; if (enable) {
if (ect_ops(ect_csdev)->enable)
ect_ret =
ect_ops(ect_csdev)->enable(ect_csdev);
ect_ret = ect_ops(ect_csdev)->enable(ect_csdev);
if (!ect_ret)
csdev->ect_enabled = true; } else {
if (ect_ops(ect_csdev)->disable)
if (csdev->ect_enabled) { ect_ret =
ect_ops(ect_csdev)->disable(ect_csdev);
csdev->ect_enabled = false;
} } /* output warning if ECT enable is preventing trace operation
*/
diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 3bb738f9a326..7d3c87e5b97c 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -208,6 +208,7 @@ struct coresight_device { /* sysfs links between components */ int nr_links; bool has_conns_grp;
bool ect_enabled; /* true only if associated ect device is
enabled */
We have cti_config::enable_req_count and cti_config::hw_enabled, both
used in
cti_enable_hw() and cti_disable_hw(). I would have thought they'd be
sufficient
to address the counting problems. If they are not I would much rather
see a
solution confined to the cti driver than in the core itself.
This is related to the fact that under sysfs it is possible under sysfs to enable an etm e.g. etm1 without the cti module present, then insert the CTI module, then enable another ETM e.g etm2. This is an issue that is caused by the possibility of module load and unload, and though inadvisable from a system usage point of view - Tingwei correctly points out that it could happen.
At the point that the first ETM is enabled, the associated ect pointer would be NULL, and thus no attempt to enable the ect/CTI is made. The CTI module on load will set the ect pointers on all registered csdev devices, including ones that are already enabled - etm1. So when we come to disable etm1, it will try to disable a CTI that it did not enable - a fact that cannot be counted in the CTI driver as it was not there when the etm was enabled. So we have a flag in csdev to record if this csdev did in fact enable the associated device, so it is clear to disable it on shutdown.
In the above example the csdev structures associated with ETM1 and ETM2 will have different csdev->ect_dev pointers, one for each CTI it is assocated with. And I would think the reference count for CTI1 would not have been incremented since it was never enabled.
What am I missing?
The issue here is we could have two or more coresght devices associated to single CTI device, especially for sys cti case. For example, we have coresight devcie A and B. They are assocaited to CTI device C. CTI module is not loaded at first. Device A is enabled. CTI C is not enabled. Then CTI module is loaded. Device B is enabled. CTI C is enabled with enable_req_count = 1. Device A is disabled. It decrease enable_req_count to 0, so CTI C is disabled. It's kinda weird to me that CTI C is enabled with device B and disabled by device A.
The point is we need some flag in coresight device structure to save the status of associated CTI enablement when CTI module is not loaded yet.
Thanks, Tingwei
Regards
Mike
Thanks, Mathieu
};
/*
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project
-- Mike Leach Principal Engineer, ARM Ltd. Manchester Design Centre. UK
linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
On Wed, Aug 19, 2020 at 09:55:33AM +0800, Tingwei Zhang wrote:
On Wed, Aug 19, 2020 at 01:39:09AM +0800, Mathieu Poirier wrote:
On Mon, Aug 17, 2020 at 08:04:06PM +0100, Mike Leach wrote:
Hi Mathieu,
On Mon, 17 Aug 2020 at 17:38, Mathieu Poirier mathieu.poirier@linaro.org wrote:
On Fri, Aug 07, 2020 at 07:11:48PM +0800, Tingwei Zhang wrote:
If associated ect device is not enabled at first place, disable routine should not be called. Add ect_enabled flag to check whether ect device is enabled. Fix the issue in below case. Ect device is not available when associated coresight device enabled and the association is established after coresight device is enabled.
Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org
drivers/hwtracing/coresight/coresight.c | 11 ++++++++--- include/linux/coresight.h | 1 + 2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c
b/drivers/hwtracing/coresight/coresight.c
index d066411aa794..27ad8317e3cf 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -245,13 +245,18 @@ coresight_control_assoc_ectdev(struct
coresight_device *csdev, bool enable)
if (!ect_csdev) return 0;
if ((!ect_ops(ect_csdev)->enable) ||
(!ect_ops(ect_csdev)->disable))
return 0; if (enable) {
if (ect_ops(ect_csdev)->enable)
ect_ret =
ect_ops(ect_csdev)->enable(ect_csdev);
ect_ret = ect_ops(ect_csdev)->enable(ect_csdev);
if (!ect_ret)
csdev->ect_enabled = true; } else {
if (ect_ops(ect_csdev)->disable)
if (csdev->ect_enabled) { ect_ret =
ect_ops(ect_csdev)->disable(ect_csdev);
csdev->ect_enabled = false;
} } /* output warning if ECT enable is preventing trace operation
*/
diff --git a/include/linux/coresight.h b/include/linux/coresight.h index 3bb738f9a326..7d3c87e5b97c 100644 --- a/include/linux/coresight.h +++ b/include/linux/coresight.h @@ -208,6 +208,7 @@ struct coresight_device { /* sysfs links between components */ int nr_links; bool has_conns_grp;
bool ect_enabled; /* true only if associated ect device is
enabled */
We have cti_config::enable_req_count and cti_config::hw_enabled, both
used in
cti_enable_hw() and cti_disable_hw(). I would have thought they'd be
sufficient
to address the counting problems. If they are not I would much rather
see a
solution confined to the cti driver than in the core itself.
This is related to the fact that under sysfs it is possible under sysfs to enable an etm e.g. etm1 without the cti module present, then insert the CTI module, then enable another ETM e.g etm2. This is an issue that is caused by the possibility of module load and unload, and though inadvisable from a system usage point of view - Tingwei correctly points out that it could happen.
At the point that the first ETM is enabled, the associated ect pointer would be NULL, and thus no attempt to enable the ect/CTI is made. The CTI module on load will set the ect pointers on all registered csdev devices, including ones that are already enabled - etm1. So when we come to disable etm1, it will try to disable a CTI that it did not enable - a fact that cannot be counted in the CTI driver as it was not there when the etm was enabled. So we have a flag in csdev to record if this csdev did in fact enable the associated device, so it is clear to disable it on shutdown.
In the above example the csdev structures associated with ETM1 and ETM2 will have different csdev->ect_dev pointers, one for each CTI it is assocated with. And I would think the reference count for CTI1 would not have been incremented since it was never enabled.
What am I missing?
The issue here is we could have two or more coresght devices associated to single CTI device, especially for sys cti case. For example, we have coresight devcie A and B. They are assocaited to CTI device C. CTI module is not loaded at first. Device A is enabled. CTI C is not enabled. Then CTI module is loaded. Device B is enabled. CTI C is enabled with enable_req_count = 1. Device A is disabled. It decrease enable_req_count to 0, so CTI C is disabled. It's kinda weird to me that CTI C is enabled with device B and disabled by device A.
Thanks for taking the time to write this out, I now have a better understanding of the problem. On the flip side I hope to find a better solution, something I will have to think about. No need to wait on me though, resubmit a new revision when ready and I'll see then.
The point is we need some flag in coresight device structure to save the status of associated CTI enablement when CTI module is not loaded yet.
Thanks, Tingwei
Regards
Mike
Thanks, Mathieu
};
/*
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project
-- Mike Leach Principal Engineer, ARM Ltd. Manchester Design Centre. UK
linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
CTI device is enabled when associated coresight device is enabled. Increase the module and device reference count for CTI device when it's enabled. This can prevent CTI device be removed or module be unloaded when CTI device is enabled by an active trace session.
Signed-off-by: Mike Leach mike.leach@linaro.org Signed-off-by: Tingwei Zhang tingwei@codeaurora.org --- drivers/hwtracing/coresight/coresight.c | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 27ad8317e3cf..8dcd73ea33b4 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -242,19 +242,30 @@ coresight_control_assoc_ectdev(struct coresight_device *csdev, bool enable) { int ect_ret = 0; struct coresight_device *ect_csdev = csdev->ect_dev; + struct module *mod;
if (!ect_csdev) return 0; if ((!ect_ops(ect_csdev)->enable) || (!ect_ops(ect_csdev)->disable)) return 0;
+ mod = ect_csdev->dev.parent->driver->owner; if (enable) { - ect_ret = ect_ops(ect_csdev)->enable(ect_csdev); - if (!ect_ret) - csdev->ect_enabled = true; + if (try_module_get(mod)) { + ect_ret = ect_ops(ect_csdev)->enable(ect_csdev); + if (ect_ret) { + module_put(mod); + } else { + get_device(ect_csdev->dev.parent); + csdev->ect_enabled = true; + } + } else + ect_ret = -ENODEV; } else { if (csdev->ect_enabled) { ect_ret = ect_ops(ect_csdev)->disable(ect_csdev); + put_device(ect_csdev->dev.parent); + module_put(mod); csdev->ect_enabled = false; } }
Allow to build coresight-cti as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m - append -core to source file name to allow module to be called coresight-cti by the Makefile - add an cti_remove function, for module unload - move cti_remove_conn_xrefs to cti_remove - add a MODULE_DEVICE_TABLE for autoloading on boot
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org --- drivers/hwtracing/coresight/Kconfig | 5 ++++- drivers/hwtracing/coresight/Makefile | 4 ++-- .../{coresight-cti.c => coresight-cti-core.c} | 19 ++++++++++++++++++- .../hwtracing/coresight/coresight-platform.c | 1 + drivers/hwtracing/coresight/coresight.c | 1 + 5 files changed, 26 insertions(+), 4 deletions(-) rename drivers/hwtracing/coresight/{coresight-cti.c => coresight-cti-core.c} (98%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index f31778dd0b5d..b04aae2ceecc 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -136,7 +136,7 @@ config CORESIGHT_CPU_DEBUG module will be called coresight-cpu-debug.
config CORESIGHT_CTI - bool "CoreSight Cross Trigger Interface (CTI) driver" + tristate "CoreSight Cross Trigger Interface (CTI) driver" depends on ARM || ARM64 help This driver provides support for CoreSight CTI and CTM components. @@ -147,6 +147,9 @@ config CORESIGHT_CTI halt compared to disabling sources and sinks normally in driver software.
+ To compile this driver as a module, choose M here: the + module will be called coresight-cti. + config CORESIGHT_CTI_INTEGRATION_REGS bool "Access CTI CoreSight Integration Registers" depends on CORESIGHT_CTI diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index f2a568b969c4..0359d5a1588f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -19,6 +19,6 @@ coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o -obj-$(CONFIG_CORESIGHT_CTI) += coresight-cti.o \ - coresight-cti-platform.o \ +obj-$(CONFIG_CORESIGHT_CTI) += coresight-cti.o +coresight-cti-y := coresight-cti-core.o coresight-cti-platform.o \ coresight-cti-sysfs.o diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti-core.c similarity index 98% rename from drivers/hwtracing/coresight/coresight-cti.c rename to drivers/hwtracing/coresight/coresight-cti-core.c index 6bb06c5715db..ca31da6cf1c0 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti-core.c @@ -838,7 +838,6 @@ static void cti_device_release(struct device *dev) struct cti_drvdata *ect_item, *ect_tmp;
mutex_lock(&ect_mutex); - cti_remove_conn_xrefs(drvdata); cti_pm_release(drvdata);
/* remove from the list */ @@ -853,6 +852,18 @@ static void cti_device_release(struct device *dev) if (drvdata->csdev_release) drvdata->csdev_release(dev); } +static int __exit cti_remove(struct amba_device *adev) +{ + struct cti_drvdata *drvdata = dev_get_drvdata(&adev->dev); + + mutex_lock(&ect_mutex); + cti_remove_conn_xrefs(drvdata); + mutex_unlock(&ect_mutex); + + coresight_unregister(drvdata->csdev); + + return 0; +}
static int cti_probe(struct amba_device *adev, const struct amba_id *id) { @@ -972,6 +983,7 @@ static const struct amba_id cti_ids[] = { CS_AMBA_UCI_ID(0x000bb9ed, uci_id_cti), /* Coresight CTI (SoC 600) */ { 0, 0}, }; +MODULE_DEVICE_TABLE(amba, cti_ids);
static struct amba_driver cti_driver = { .drv = { @@ -980,6 +992,7 @@ static struct amba_driver cti_driver = { .suppress_bind_attrs = true, }, .probe = cti_probe, + .remove = cti_remove, .id_table = cti_ids, };
@@ -1002,3 +1015,7 @@ static void __exit cti_exit(void)
module_init(cti_init); module_exit(cti_exit); + +MODULE_AUTHOR("Mike Leach mike.leach@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight CTI Driver"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/hwtracing/coresight/coresight-platform.c b/drivers/hwtracing/coresight/coresight-platform.c index bfd44231d7ad..635d55c1d580 100644 --- a/drivers/hwtracing/coresight/coresight-platform.c +++ b/drivers/hwtracing/coresight/coresight-platform.c @@ -75,6 +75,7 @@ coresight_find_csdev_by_fwnode(struct fwnode_handle *r_fwnode) } return csdev; } +EXPORT_SYMBOL_GPL(coresight_find_csdev_by_fwnode);
#ifdef CONFIG_OF static inline bool of_coresight_legacy_ep_is_input(struct device_node *ep) diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 8dcd73ea33b4..f9566d667261 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -289,6 +289,7 @@ void coresight_set_assoc_ectdev_mutex(struct coresight_device *csdev, csdev->ect_dev = ect_csdev; mutex_unlock(&coresight_mutex); } +EXPORT_SYMBOL_GPL(coresight_set_assoc_ectdev_mutex);
static int coresight_enable_sink(struct coresight_device *csdev, u32 mode, void *data)
On Fri, Aug 07, 2020 at 07:11:50PM +0800, Tingwei Zhang wrote:
Allow to build coresight-cti as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-cti by the Makefile
- add an cti_remove function, for module unload
- move cti_remove_conn_xrefs to cti_remove
- add a MODULE_DEVICE_TABLE for autoloading on boot
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
Mike,
Please add a "Reviewed-by" tag once you have reviewed this patch.
drivers/hwtracing/coresight/Kconfig | 5 ++++- drivers/hwtracing/coresight/Makefile | 4 ++-- .../{coresight-cti.c => coresight-cti-core.c} | 19 ++++++++++++++++++- .../hwtracing/coresight/coresight-platform.c | 1 + drivers/hwtracing/coresight/coresight.c | 1 + 5 files changed, 26 insertions(+), 4 deletions(-) rename drivers/hwtracing/coresight/{coresight-cti.c => coresight-cti-core.c} (98%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index f31778dd0b5d..b04aae2ceecc 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -136,7 +136,7 @@ config CORESIGHT_CPU_DEBUG module will be called coresight-cpu-debug. config CORESIGHT_CTI
- bool "CoreSight Cross Trigger Interface (CTI) driver"
- tristate "CoreSight Cross Trigger Interface (CTI) driver" depends on ARM || ARM64 help This driver provides support for CoreSight CTI and CTM components.
@@ -147,6 +147,9 @@ config CORESIGHT_CTI halt compared to disabling sources and sinks normally in driver software.
To compile this driver as a module, choose M here: the
module will be called coresight-cti.
config CORESIGHT_CTI_INTEGRATION_REGS bool "Access CTI CoreSight Integration Registers" depends on CORESIGHT_CTI diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index f2a568b969c4..0359d5a1588f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -19,6 +19,6 @@ coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o -obj-$(CONFIG_CORESIGHT_CTI) += coresight-cti.o \
coresight-cti-platform.o \
+obj-$(CONFIG_CORESIGHT_CTI) += coresight-cti.o +coresight-cti-y := coresight-cti-core.o coresight-cti-platform.o \ coresight-cti-sysfs.o
Please align "coresight-cti-sysfs.o" with "coresight-cti-core.o". The same should be done for the etm3x driver.
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti-core.c similarity index 98% rename from drivers/hwtracing/coresight/coresight-cti.c rename to drivers/hwtracing/coresight/coresight-cti-core.c index 6bb06c5715db..ca31da6cf1c0 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti-core.c @@ -838,7 +838,6 @@ static void cti_device_release(struct device *dev) struct cti_drvdata *ect_item, *ect_tmp; mutex_lock(&ect_mutex);
- cti_remove_conn_xrefs(drvdata); cti_pm_release(drvdata);
/* remove from the list */ @@ -853,6 +852,18 @@ static void cti_device_release(struct device *dev) if (drvdata->csdev_release) drvdata->csdev_release(dev); } +static int __exit cti_remove(struct amba_device *adev) +{
- struct cti_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- mutex_lock(&ect_mutex);
- cti_remove_conn_xrefs(drvdata);
- mutex_unlock(&ect_mutex);
The changelog point out this change but now _why_ it is done - please revise.
- coresight_unregister(drvdata->csdev);
- return 0;
+} static int cti_probe(struct amba_device *adev, const struct amba_id *id) { @@ -972,6 +983,7 @@ static const struct amba_id cti_ids[] = { CS_AMBA_UCI_ID(0x000bb9ed, uci_id_cti), /* Coresight CTI (SoC 600) */ { 0, 0}, }; +MODULE_DEVICE_TABLE(amba, cti_ids); static struct amba_driver cti_driver = { .drv = { @@ -980,6 +992,7 @@ static struct amba_driver cti_driver = { .suppress_bind_attrs = true, }, .probe = cti_probe,
- .remove = cti_remove, .id_table = cti_ids,
}; @@ -1002,3 +1015,7 @@ static void __exit cti_exit(void) module_init(cti_init); module_exit(cti_exit);
+MODULE_AUTHOR("Mike Leach mike.leach@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight CTI Driver"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/hwtracing/coresight/coresight-platform.c b/drivers/hwtracing/coresight/coresight-platform.c index bfd44231d7ad..635d55c1d580 100644 --- a/drivers/hwtracing/coresight/coresight-platform.c +++ b/drivers/hwtracing/coresight/coresight-platform.c @@ -75,6 +75,7 @@ coresight_find_csdev_by_fwnode(struct fwnode_handle *r_fwnode) } return csdev; } +EXPORT_SYMBOL_GPL(coresight_find_csdev_by_fwnode); #ifdef CONFIG_OF static inline bool of_coresight_legacy_ep_is_input(struct device_node *ep) diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 8dcd73ea33b4..f9566d667261 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -289,6 +289,7 @@ void coresight_set_assoc_ectdev_mutex(struct coresight_device *csdev, csdev->ect_dev = ect_csdev; mutex_unlock(&coresight_mutex); } +EXPORT_SYMBOL_GPL(coresight_set_assoc_ectdev_mutex); static int coresight_enable_sink(struct coresight_device *csdev, u32 mode, void *data) -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Hi,
On Mon, 17 Aug 2020 at 18:13, Mathieu Poirier mathieu.poirier@linaro.org wrote:
On Fri, Aug 07, 2020 at 07:11:50PM +0800, Tingwei Zhang wrote:
Allow to build coresight-cti as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-cti by the Makefile
- add an cti_remove function, for module unload
- move cti_remove_conn_xrefs to cti_remove
- add a MODULE_DEVICE_TABLE for autoloading on boot
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
Mike,
Please add a "Reviewed-by" tag once you have reviewed this patch.
Sorry - not sure of the protocol - assumed that my tested-by implied reviewed by.
Anyway...
Reviewed-by Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 ++++- drivers/hwtracing/coresight/Makefile | 4 ++-- .../{coresight-cti.c => coresight-cti-core.c} | 19 ++++++++++++++++++- .../hwtracing/coresight/coresight-platform.c | 1 + drivers/hwtracing/coresight/coresight.c | 1 + 5 files changed, 26 insertions(+), 4 deletions(-) rename drivers/hwtracing/coresight/{coresight-cti.c => coresight-cti-core.c} (98%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index f31778dd0b5d..b04aae2ceecc 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -136,7 +136,7 @@ config CORESIGHT_CPU_DEBUG module will be called coresight-cpu-debug.
config CORESIGHT_CTI
bool "CoreSight Cross Trigger Interface (CTI) driver"
tristate "CoreSight Cross Trigger Interface (CTI) driver" depends on ARM || ARM64 help This driver provides support for CoreSight CTI and CTM components.
@@ -147,6 +147,9 @@ config CORESIGHT_CTI halt compared to disabling sources and sinks normally in driver software.
To compile this driver as a module, choose M here: the
module will be called coresight-cti.
config CORESIGHT_CTI_INTEGRATION_REGS bool "Access CTI CoreSight Integration Registers" depends on CORESIGHT_CTI diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index f2a568b969c4..0359d5a1588f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -19,6 +19,6 @@ coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o -obj-$(CONFIG_CORESIGHT_CTI) += coresight-cti.o \
coresight-cti-platform.o \
+obj-$(CONFIG_CORESIGHT_CTI) += coresight-cti.o +coresight-cti-y := coresight-cti-core.o coresight-cti-platform.o \ coresight-cti-sysfs.o
Please align "coresight-cti-sysfs.o" with "coresight-cti-core.o". The same should be done for the etm3x driver.
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti-core.c similarity index 98% rename from drivers/hwtracing/coresight/coresight-cti.c rename to drivers/hwtracing/coresight/coresight-cti-core.c index 6bb06c5715db..ca31da6cf1c0 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti-core.c @@ -838,7 +838,6 @@ static void cti_device_release(struct device *dev) struct cti_drvdata *ect_item, *ect_tmp;
mutex_lock(&ect_mutex);
cti_remove_conn_xrefs(drvdata); cti_pm_release(drvdata); /* remove from the list */
@@ -853,6 +852,18 @@ static void cti_device_release(struct device *dev) if (drvdata->csdev_release) drvdata->csdev_release(dev); } +static int __exit cti_remove(struct amba_device *adev) +{
struct cti_drvdata *drvdata = dev_get_drvdata(&adev->dev);
mutex_lock(&ect_mutex);
cti_remove_conn_xrefs(drvdata);
mutex_unlock(&ect_mutex);
The changelog point out this change but now _why_ it is done - please revise.
By the time the system calls device_release, all the sysfs links have gone so trying to remove them results in errors / crash. Not obvious when the drivers were built in as this would be called while the system was powering down, but very much noticeable when removing the CTI moduie from a running system.
Regards
Mike
coresight_unregister(drvdata->csdev);
return 0;
+}
static int cti_probe(struct amba_device *adev, const struct amba_id *id) { @@ -972,6 +983,7 @@ static const struct amba_id cti_ids[] = { CS_AMBA_UCI_ID(0x000bb9ed, uci_id_cti), /* Coresight CTI (SoC 600) */ { 0, 0}, }; +MODULE_DEVICE_TABLE(amba, cti_ids);
static struct amba_driver cti_driver = { .drv = { @@ -980,6 +992,7 @@ static struct amba_driver cti_driver = { .suppress_bind_attrs = true, }, .probe = cti_probe,
.remove = cti_remove, .id_table = cti_ids,
};
@@ -1002,3 +1015,7 @@ static void __exit cti_exit(void)
module_init(cti_init); module_exit(cti_exit);
+MODULE_AUTHOR("Mike Leach mike.leach@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight CTI Driver"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/hwtracing/coresight/coresight-platform.c b/drivers/hwtracing/coresight/coresight-platform.c index bfd44231d7ad..635d55c1d580 100644 --- a/drivers/hwtracing/coresight/coresight-platform.c +++ b/drivers/hwtracing/coresight/coresight-platform.c @@ -75,6 +75,7 @@ coresight_find_csdev_by_fwnode(struct fwnode_handle *r_fwnode) } return csdev; } +EXPORT_SYMBOL_GPL(coresight_find_csdev_by_fwnode);
#ifdef CONFIG_OF static inline bool of_coresight_legacy_ep_is_input(struct device_node *ep) diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 8dcd73ea33b4..f9566d667261 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -289,6 +289,7 @@ void coresight_set_assoc_ectdev_mutex(struct coresight_device *csdev, csdev->ect_dev = ect_csdev; mutex_unlock(&coresight_mutex); } +EXPORT_SYMBOL_GPL(coresight_set_assoc_ectdev_mutex);
static int coresight_enable_sink(struct coresight_device *csdev, u32 mode, void *data) -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Mon, Aug 17, 2020 at 08:14:29PM +0100, Mike Leach wrote:
Hi,
On Mon, 17 Aug 2020 at 18:13, Mathieu Poirier mathieu.poirier@linaro.org wrote:
On Fri, Aug 07, 2020 at 07:11:50PM +0800, Tingwei Zhang wrote:
Allow to build coresight-cti as a module, for ease of development.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight-cti by the Makefile
- add an cti_remove function, for module unload
- move cti_remove_conn_xrefs to cti_remove
- add a MODULE_DEVICE_TABLE for autoloading on boot
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org
Mike,
Please add a "Reviewed-by" tag once you have reviewed this patch.
Sorry - not sure of the protocol - assumed that my tested-by implied reviewed by.
One does not imply the other - in fact many people keep to testing patches only.
Anyway...
Reviewed-by Mike Leach mike.leach@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 ++++- drivers/hwtracing/coresight/Makefile | 4 ++-- .../{coresight-cti.c => coresight-cti-core.c} | 19 ++++++++++++++++++- .../hwtracing/coresight/coresight-platform.c | 1 + drivers/hwtracing/coresight/coresight.c | 1 + 5 files changed, 26 insertions(+), 4 deletions(-) rename drivers/hwtracing/coresight/{coresight-cti.c => coresight-cti-core.c} (98%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index f31778dd0b5d..b04aae2ceecc 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -136,7 +136,7 @@ config CORESIGHT_CPU_DEBUG module will be called coresight-cpu-debug.
config CORESIGHT_CTI
bool "CoreSight Cross Trigger Interface (CTI) driver"
tristate "CoreSight Cross Trigger Interface (CTI) driver" depends on ARM || ARM64 help This driver provides support for CoreSight CTI and CTM components.
@@ -147,6 +147,9 @@ config CORESIGHT_CTI halt compared to disabling sources and sinks normally in driver software.
To compile this driver as a module, choose M here: the
module will be called coresight-cti.
config CORESIGHT_CTI_INTEGRATION_REGS bool "Access CTI CoreSight Integration Registers" depends on CORESIGHT_CTI diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index f2a568b969c4..0359d5a1588f 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -19,6 +19,6 @@ coresight-etm4x-y := coresight-etm4x-core.o coresight-etm4x-sysfs.o obj-$(CONFIG_CORESIGHT_STM) += coresight-stm.o obj-$(CONFIG_CORESIGHT_CPU_DEBUG) += coresight-cpu-debug.o obj-$(CONFIG_CORESIGHT_CATU) += coresight-catu.o -obj-$(CONFIG_CORESIGHT_CTI) += coresight-cti.o \
coresight-cti-platform.o \
+obj-$(CONFIG_CORESIGHT_CTI) += coresight-cti.o +coresight-cti-y := coresight-cti-core.o coresight-cti-platform.o \ coresight-cti-sysfs.o
Please align "coresight-cti-sysfs.o" with "coresight-cti-core.o". The same should be done for the etm3x driver.
diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti-core.c similarity index 98% rename from drivers/hwtracing/coresight/coresight-cti.c rename to drivers/hwtracing/coresight/coresight-cti-core.c index 6bb06c5715db..ca31da6cf1c0 100644 --- a/drivers/hwtracing/coresight/coresight-cti.c +++ b/drivers/hwtracing/coresight/coresight-cti-core.c @@ -838,7 +838,6 @@ static void cti_device_release(struct device *dev) struct cti_drvdata *ect_item, *ect_tmp;
mutex_lock(&ect_mutex);
cti_remove_conn_xrefs(drvdata); cti_pm_release(drvdata); /* remove from the list */
@@ -853,6 +852,18 @@ static void cti_device_release(struct device *dev) if (drvdata->csdev_release) drvdata->csdev_release(dev); } +static int __exit cti_remove(struct amba_device *adev) +{
struct cti_drvdata *drvdata = dev_get_drvdata(&adev->dev);
mutex_lock(&ect_mutex);
cti_remove_conn_xrefs(drvdata);
mutex_unlock(&ect_mutex);
The changelog point out this change but now _why_ it is done - please revise.
By the time the system calls device_release, all the sysfs links have gone so trying to remove them results in errors / crash. Not obvious when the drivers were built in as this would be called while the system was powering down, but very much noticeable when removing the CTI moduie from a running system.
Regards
Mike
coresight_unregister(drvdata->csdev);
return 0;
+}
static int cti_probe(struct amba_device *adev, const struct amba_id *id) { @@ -972,6 +983,7 @@ static const struct amba_id cti_ids[] = { CS_AMBA_UCI_ID(0x000bb9ed, uci_id_cti), /* Coresight CTI (SoC 600) */ { 0, 0}, }; +MODULE_DEVICE_TABLE(amba, cti_ids);
static struct amba_driver cti_driver = { .drv = { @@ -980,6 +992,7 @@ static struct amba_driver cti_driver = { .suppress_bind_attrs = true, }, .probe = cti_probe,
.remove = cti_remove, .id_table = cti_ids,
};
@@ -1002,3 +1015,7 @@ static void __exit cti_exit(void)
module_init(cti_init); module_exit(cti_exit);
+MODULE_AUTHOR("Mike Leach mike.leach@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight CTI Driver"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/hwtracing/coresight/coresight-platform.c b/drivers/hwtracing/coresight/coresight-platform.c index bfd44231d7ad..635d55c1d580 100644 --- a/drivers/hwtracing/coresight/coresight-platform.c +++ b/drivers/hwtracing/coresight/coresight-platform.c @@ -75,6 +75,7 @@ coresight_find_csdev_by_fwnode(struct fwnode_handle *r_fwnode) } return csdev; } +EXPORT_SYMBOL_GPL(coresight_find_csdev_by_fwnode);
#ifdef CONFIG_OF static inline bool of_coresight_legacy_ep_is_input(struct device_node *ep) diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index 8dcd73ea33b4..f9566d667261 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -289,6 +289,7 @@ void coresight_set_assoc_ectdev_mutex(struct coresight_device *csdev, csdev->ect_dev = ect_csdev; mutex_unlock(&coresight_mutex); } +EXPORT_SYMBOL_GPL(coresight_set_assoc_ectdev_mutex);
static int coresight_enable_sink(struct coresight_device *csdev, u32 mode, void *data) -- The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
-- Mike Leach Principal Engineer, ARM Ltd. Manchester Design Centre. UK
From: Mian Yousaf Kaukab ykaukab@suse.de
Make etr_catu_buf_ops static. Instead of directly accessing it in etr_buf_ops[], add a function to let catu driver register the ops at runtime. Break circular dependency between tmc-etr and catu drivers.
Signed-off-by: Mian Yousaf Kaukab ykaukab@suse.de Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by : Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/coresight-catu.c | 22 +++++++++++++++++-- drivers/hwtracing/coresight/coresight-catu.h | 2 -- .../hwtracing/coresight/coresight-tmc-etr.c | 15 +++++++++++-- drivers/hwtracing/coresight/coresight-tmc.h | 3 +++ 4 files changed, 36 insertions(+), 6 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-catu.c b/drivers/hwtracing/coresight/coresight-catu.c index 1801804a7762..47696a7d24a7 100644 --- a/drivers/hwtracing/coresight/coresight-catu.c +++ b/drivers/hwtracing/coresight/coresight-catu.c @@ -358,7 +358,7 @@ static int catu_alloc_etr_buf(struct tmc_drvdata *tmc_drvdata, return 0; }
-const struct etr_buf_operations etr_catu_buf_ops = { +static const struct etr_buf_operations etr_catu_buf_ops = { .alloc = catu_alloc_etr_buf, .free = catu_free_etr_buf, .sync = catu_sync_etr_buf, @@ -582,4 +582,22 @@ static struct amba_driver catu_driver = { .id_table = catu_ids, };
-builtin_amba_driver(catu_driver); +static int __init catu_init(void) +{ + int ret; + + ret = amba_driver_register(&catu_driver); + if (ret) + pr_info("Error registering catu driver\n"); + tmc_etr_set_catu_ops(&etr_catu_buf_ops); + return ret; +} + +static void __exit catu_exit(void) +{ + tmc_etr_remove_catu_ops(); + amba_driver_unregister(&catu_driver); +} + +module_init(catu_init); +module_exit(catu_exit); diff --git a/drivers/hwtracing/coresight/coresight-catu.h b/drivers/hwtracing/coresight/coresight-catu.h index 80ceee3c739c..6160c2d75a56 100644 --- a/drivers/hwtracing/coresight/coresight-catu.h +++ b/drivers/hwtracing/coresight/coresight-catu.h @@ -108,6 +108,4 @@ static inline bool coresight_is_catu_device(struct coresight_device *csdev) return true; }
-extern const struct etr_buf_operations etr_catu_buf_ops; - #endif diff --git a/drivers/hwtracing/coresight/coresight-tmc-etr.c b/drivers/hwtracing/coresight/coresight-tmc-etr.c index ad991a37e2d2..714f9e867e5f 100644 --- a/drivers/hwtracing/coresight/coresight-tmc-etr.c +++ b/drivers/hwtracing/coresight/coresight-tmc-etr.c @@ -794,10 +794,21 @@ static inline void tmc_etr_disable_catu(struct tmc_drvdata *drvdata) static const struct etr_buf_operations *etr_buf_ops[] = { [ETR_MODE_FLAT] = &etr_flat_buf_ops, [ETR_MODE_ETR_SG] = &etr_sg_buf_ops, - [ETR_MODE_CATU] = IS_ENABLED(CONFIG_CORESIGHT_CATU) - ? &etr_catu_buf_ops : NULL, + [ETR_MODE_CATU] = NULL, };
+void tmc_etr_set_catu_ops(const struct etr_buf_operations *catu) +{ + etr_buf_ops[ETR_MODE_CATU] = catu; +} +EXPORT_SYMBOL_GPL(tmc_etr_set_catu_ops); + +void tmc_etr_remove_catu_ops(void) +{ + etr_buf_ops[ETR_MODE_CATU] = NULL; +} +EXPORT_SYMBOL_GPL(tmc_etr_remove_catu_ops); + static inline int tmc_etr_mode_alloc_buf(int mode, struct tmc_drvdata *drvdata, struct etr_buf *etr_buf, int node, diff --git a/drivers/hwtracing/coresight/coresight-tmc.h b/drivers/hwtracing/coresight/coresight-tmc.h index 6e8d2dc33d17..b91ec7dde7bc 100644 --- a/drivers/hwtracing/coresight/coresight-tmc.h +++ b/drivers/hwtracing/coresight/coresight-tmc.h @@ -326,4 +326,7 @@ tmc_sg_table_buf_size(struct tmc_sg_table *sg_table)
struct coresight_device *tmc_etr_get_catu_device(struct tmc_drvdata *drvdata);
+void tmc_etr_set_catu_ops(const struct etr_buf_operations *catu); +void tmc_etr_remove_catu_ops(void); + #endif
Allow to build coresight-catu as modules, for ease of development. - Kconfig becomes a tristate, to allow =m - add catu_remove functions, for module unload - add a MODULE_DEVICE_TABLE for autoloading on boot
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/Kconfig | 5 ++++- drivers/hwtracing/coresight/coresight-catu.c | 15 +++++++++++++++ 2 files changed, 19 insertions(+), 1 deletion(-)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index b04aae2ceecc..dfe407cde262 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -42,7 +42,7 @@ config CORESIGHT_LINK_AND_SINK_TMC module will be called coresight-tmc.
config CORESIGHT_CATU - bool "Coresight Address Translation Unit (CATU) driver" + tristate "Coresight Address Translation Unit (CATU) driver" depends on CORESIGHT_LINK_AND_SINK_TMC help Enable support for the Coresight Address Translation Unit (CATU). @@ -52,6 +52,9 @@ config CORESIGHT_CATU by looking up the provided table. CATU can also be used in pass-through mode where the address is not translated.
+ To compile this driver as a module, choose M here: the + module will be called coresight-catu. + config CORESIGHT_SINK_TPIU tristate "Coresight generic TPIU driver" depends on CORESIGHT_LINKS_AND_SINKS diff --git a/drivers/hwtracing/coresight/coresight-catu.c b/drivers/hwtracing/coresight/coresight-catu.c index 47696a7d24a7..99430f6cf5a5 100644 --- a/drivers/hwtracing/coresight/coresight-catu.c +++ b/drivers/hwtracing/coresight/coresight-catu.c @@ -567,11 +567,21 @@ static int catu_probe(struct amba_device *adev, const struct amba_id *id) return ret; }
+static int __exit catu_remove(struct amba_device *adev) +{ + struct catu_drvdata *drvdata = dev_get_drvdata(&adev->dev); + + coresight_unregister(drvdata->csdev); + return 0; +} + static struct amba_id catu_ids[] = { CS_AMBA_ID(0x000bb9ee), {}, };
+MODULE_DEVICE_TABLE(amba, catu_ids); + static struct amba_driver catu_driver = { .drv = { .name = "coresight-catu", @@ -579,6 +589,7 @@ static struct amba_driver catu_driver = { .suppress_bind_attrs = true, }, .probe = catu_probe, + .remove = catu_remove, .id_table = catu_ids, };
@@ -601,3 +612,7 @@ static void __exit catu_exit(void)
module_init(catu_init); module_exit(catu_exit); + +MODULE_AUTHOR("Suzuki K Poulose suzuki.poulose@arm.com"); +MODULE_DESCRIPTION("Arm CoreSight Address Translation Unit (CATU) Driver"); +MODULE_LICENSE("GPL v2");
On Fri, Aug 07, 2020 at 07:11:52PM +0800, Tingwei Zhang wrote:
Allow to build coresight-catu as modules, for ease of development.
- Kconfig becomes a tristate, to allow =m
- add catu_remove functions, for module unload
- add a MODULE_DEVICE_TABLE for autoloading on boot
Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Reviewed-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
For patches 22 and 23:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
drivers/hwtracing/coresight/Kconfig | 5 ++++- drivers/hwtracing/coresight/coresight-catu.c | 15 +++++++++++++++ 2 files changed, 19 insertions(+), 1 deletion(-)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index b04aae2ceecc..dfe407cde262 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -42,7 +42,7 @@ config CORESIGHT_LINK_AND_SINK_TMC module will be called coresight-tmc. config CORESIGHT_CATU
- bool "Coresight Address Translation Unit (CATU) driver"
- tristate "Coresight Address Translation Unit (CATU) driver" depends on CORESIGHT_LINK_AND_SINK_TMC help Enable support for the Coresight Address Translation Unit (CATU).
@@ -52,6 +52,9 @@ config CORESIGHT_CATU by looking up the provided table. CATU can also be used in pass-through mode where the address is not translated.
To compile this driver as a module, choose M here: the
module will be called coresight-catu.
config CORESIGHT_SINK_TPIU tristate "Coresight generic TPIU driver" depends on CORESIGHT_LINKS_AND_SINKS diff --git a/drivers/hwtracing/coresight/coresight-catu.c b/drivers/hwtracing/coresight/coresight-catu.c index 47696a7d24a7..99430f6cf5a5 100644 --- a/drivers/hwtracing/coresight/coresight-catu.c +++ b/drivers/hwtracing/coresight/coresight-catu.c @@ -567,11 +567,21 @@ static int catu_probe(struct amba_device *adev, const struct amba_id *id) return ret; } +static int __exit catu_remove(struct amba_device *adev) +{
- struct catu_drvdata *drvdata = dev_get_drvdata(&adev->dev);
- coresight_unregister(drvdata->csdev);
- return 0;
+}
static struct amba_id catu_ids[] = { CS_AMBA_ID(0x000bb9ee), {}, }; +MODULE_DEVICE_TABLE(amba, catu_ids);
static struct amba_driver catu_driver = { .drv = { .name = "coresight-catu", @@ -579,6 +589,7 @@ static struct amba_driver catu_driver = { .suppress_bind_attrs = true, }, .probe = catu_probe,
- .remove = catu_remove, .id_table = catu_ids,
}; @@ -601,3 +612,7 @@ static void __exit catu_exit(void) module_init(catu_init); module_exit(catu_exit);
+MODULE_AUTHOR("Suzuki K Poulose suzuki.poulose@arm.com"); +MODULE_DESCRIPTION("Arm CoreSight Address Translation Unit (CATU) Driver");
+MODULE_LICENSE("GPL v2");
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Enhance coresight developer's efficiency to debug coresight drivers. - Kconfig becomes a tristate, to allow =m - append -core to source file name to allow module to be called coresight by the Makefile - modules can have only one init/exit, so we add the etm_perf register/unregister function calls to the core init/exit functions. - add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com --- drivers/hwtracing/coresight/Kconfig | 5 ++- drivers/hwtracing/coresight/Makefile | 5 ++- .../{coresight.c => coresight-core.c} | 42 ++++++++++++++----- .../hwtracing/coresight/coresight-etm-perf.c | 8 +++- .../hwtracing/coresight/coresight-etm-perf.h | 3 ++ 5 files changed, 48 insertions(+), 15 deletions(-) rename drivers/hwtracing/coresight/{coresight.c => coresight-core.c} (98%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index dfe407cde262..c1198245461d 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -3,7 +3,7 @@ # Coresight configuration # menuconfig CORESIGHT - bool "CoreSight Tracing Support" + tristate "CoreSight Tracing Support" depends on ARM || ARM64 depends on OF || ACPI select ARM_AMBA @@ -15,6 +15,9 @@ menuconfig CORESIGHT specification and configure the right series of components when a trace source gets enabled.
+ To compile this driver as a module, choose M here: the + module will be called coresight. + if CORESIGHT config CORESIGHT_LINKS_AND_SINKS tristate "CoreSight Link and Sink drivers" diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 0359d5a1588f..1b35b55bd420 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -2,8 +2,9 @@ # # Makefile for CoreSight drivers. # -obj-$(CONFIG_CORESIGHT) += coresight.o coresight-etm-perf.o \ - coresight-platform.o coresight-sysfs.o +obj-$(CONFIG_CORESIGHT) += coresight.o +coresight-y := coresight-core.o coresight-etm-perf.o coresight-platform.o \ + coresight-sysfs.o obj-$(CONFIG_CORESIGHT_LINK_AND_SINK_TMC) += coresight-tmc.o coresight-tmc-y := coresight-tmc-core.o coresight-tmc-etf.o \ coresight-tmc-etr.o diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight-core.c similarity index 98% rename from drivers/hwtracing/coresight/coresight.c rename to drivers/hwtracing/coresight/coresight-core.c index f9566d667261..045416ac962f 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight-core.c @@ -1462,16 +1462,6 @@ int coresight_timeout(void __iomem *addr, u32 offset, int position, int value) } EXPORT_SYMBOL_GPL(coresight_timeout);
-struct bus_type coresight_bustype = { - .name = "coresight", -}; - -static int __init coresight_init(void) -{ - return bus_register(&coresight_bustype); -} -postcore_initcall(coresight_init); - /* * coresight_release_platform_data: Release references to the devices connected * to the output port of this device. @@ -1680,3 +1670,35 @@ char *coresight_alloc_device_name(struct coresight_dev_list *dict, return name; } EXPORT_SYMBOL_GPL(coresight_alloc_device_name); + +struct bus_type coresight_bustype = { + .name = "coresight", +}; + +static int __init coresight_init(void) +{ + int ret; + + ret = bus_register(&coresight_bustype); + if (ret) + return ret; + + ret = etm_perf_init(); + if (ret) + bus_unregister(&coresight_bustype); + + return ret; +} + +static void __exit coresight_exit(void) +{ + etm_perf_exit(); + bus_unregister(&coresight_bustype); +} + +module_init(coresight_init); +module_exit(coresight_exit); + +MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org"); +MODULE_DESCRIPTION("Arm CoreSight tracer driver"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index 3728c44e5763..668b3ff11576 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -591,7 +591,7 @@ void etm_perf_del_symlink_sink(struct coresight_device *csdev) csdev->ea = NULL; }
-static int __init etm_perf_init(void) +int __init etm_perf_init(void) { int ret;
@@ -618,4 +618,8 @@ static int __init etm_perf_init(void)
return ret; } -device_initcall(etm_perf_init); + +void __exit etm_perf_exit(void) +{ + perf_pmu_unregister(&etm_pmu); +} diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.h b/drivers/hwtracing/coresight/coresight-etm-perf.h index 05f89723e282..3e4f2ad5e193 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.h +++ b/drivers/hwtracing/coresight/coresight-etm-perf.h @@ -82,4 +82,7 @@ static inline void *etm_perf_sink_config(struct perf_output_handle *handle)
#endif /* CONFIG_CORESIGHT */
+int __init etm_perf_init(void); +void __exit etm_perf_exit(void); + #endif
On Fri, Aug 07, 2020 at 07:11:53PM +0800, Tingwei Zhang wrote:
Enhance coresight developer's efficiency to debug coresight drivers.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight by the Makefile
- modules can have only one init/exit, so we add the etm_perf register/unregister function calls to the core init/exit functions.
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/Kconfig | 5 ++- drivers/hwtracing/coresight/Makefile | 5 ++- .../{coresight.c => coresight-core.c} | 42 ++++++++++++++----- .../hwtracing/coresight/coresight-etm-perf.c | 8 +++- .../hwtracing/coresight/coresight-etm-perf.h | 3 ++ 5 files changed, 48 insertions(+), 15 deletions(-) rename drivers/hwtracing/coresight/{coresight.c => coresight-core.c} (98%)
diff --git a/drivers/hwtracing/coresight/Kconfig b/drivers/hwtracing/coresight/Kconfig index dfe407cde262..c1198245461d 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -3,7 +3,7 @@ # Coresight configuration # menuconfig CORESIGHT
- bool "CoreSight Tracing Support"
- tristate "CoreSight Tracing Support" depends on ARM || ARM64 depends on OF || ACPI select ARM_AMBA
@@ -15,6 +15,9 @@ menuconfig CORESIGHT specification and configure the right series of components when a trace source gets enabled.
To compile this driver as a module, choose M here: the
module will be called coresight.
if CORESIGHT config CORESIGHT_LINKS_AND_SINKS tristate "CoreSight Link and Sink drivers" diff --git a/drivers/hwtracing/coresight/Makefile b/drivers/hwtracing/coresight/Makefile index 0359d5a1588f..1b35b55bd420 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -2,8 +2,9 @@ # # Makefile for CoreSight drivers. # -obj-$(CONFIG_CORESIGHT) += coresight.o coresight-etm-perf.o \
coresight-platform.o coresight-sysfs.o
+obj-$(CONFIG_CORESIGHT) += coresight.o +coresight-y := coresight-core.o coresight-etm-perf.o coresight-platform.o \
coresight-sysfs.o
obj-$(CONFIG_CORESIGHT_LINK_AND_SINK_TMC) += coresight-tmc.o coresight-tmc-y := coresight-tmc-core.o coresight-tmc-etf.o \ coresight-tmc-etr.o diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight-core.c similarity index 98% rename from drivers/hwtracing/coresight/coresight.c rename to drivers/hwtracing/coresight/coresight-core.c index f9566d667261..045416ac962f 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight-core.c @@ -1462,16 +1462,6 @@ int coresight_timeout(void __iomem *addr, u32 offset, int position, int value) } EXPORT_SYMBOL_GPL(coresight_timeout); -struct bus_type coresight_bustype = {
- .name = "coresight",
-};
-static int __init coresight_init(void) -{
- return bus_register(&coresight_bustype);
-} -postcore_initcall(coresight_init);
/*
- coresight_release_platform_data: Release references to the devices connected
- to the output port of this device.
@@ -1680,3 +1670,35 @@ char *coresight_alloc_device_name(struct coresight_dev_list *dict, return name; } EXPORT_SYMBOL_GPL(coresight_alloc_device_name);
+struct bus_type coresight_bustype = {
- .name = "coresight",
+};
+static int __init coresight_init(void) +{
- int ret;
- ret = bus_register(&coresight_bustype);
- if (ret)
return ret;
- ret = etm_perf_init();
- if (ret)
bus_unregister(&coresight_bustype);
- return ret;
+}
+static void __exit coresight_exit(void) +{
- etm_perf_exit();
- bus_unregister(&coresight_bustype);
+}
+module_init(coresight_init); +module_exit(coresight_exit);
+MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org");
Pratik deserves credit too.
With that:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
I am done reviewing this set.
Thank you for the patience, Mathieu
+MODULE_DESCRIPTION("Arm CoreSight tracer driver"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index 3728c44e5763..668b3ff11576 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -591,7 +591,7 @@ void etm_perf_del_symlink_sink(struct coresight_device *csdev) csdev->ea = NULL; } -static int __init etm_perf_init(void) +int __init etm_perf_init(void) { int ret; @@ -618,4 +618,8 @@ static int __init etm_perf_init(void) return ret; } -device_initcall(etm_perf_init);
+void __exit etm_perf_exit(void) +{
- perf_pmu_unregister(&etm_pmu);
+} diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.h b/drivers/hwtracing/coresight/coresight-etm-perf.h index 05f89723e282..3e4f2ad5e193 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.h +++ b/drivers/hwtracing/coresight/coresight-etm-perf.h @@ -82,4 +82,7 @@ static inline void *etm_perf_sink_config(struct perf_output_handle *handle) #endif /* CONFIG_CORESIGHT */ +int __init etm_perf_init(void); +void __exit etm_perf_exit(void);
#endif
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
On Tue, Aug 18, 2020 at 01:47:22AM +0800, Mathieu Poirier wrote:
On Fri, Aug 07, 2020 at 07:11:53PM +0800, Tingwei Zhang wrote:
Enhance coresight developer's efficiency to debug coresight drivers.
- Kconfig becomes a tristate, to allow =m
- append -core to source file name to allow module to be called coresight by the Makefile
- modules can have only one init/exit, so we add the etm_perf register/unregister function calls to the core init/exit functions.
- add a MODULE_DEVICE_TABLE for autoloading on boot
Cc: Mathieu Poirier mathieu.poirier@linaro.org Cc: Leo Yan leo.yan@linaro.org Cc: Alexander Shishkin alexander.shishkin@linux.intel.com Cc: Randy Dunlap rdunlap@infradead.org Cc: Suzuki K Poulose Suzuki.Poulose@arm.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: Russell King linux@armlinux.org.uk Signed-off-by: Kim Phillips kim.phillips@arm.com Signed-off-by: Tingwei Zhang tingwei@codeaurora.org Tested-by: Mike Leach mike.leach@linaro.org Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
drivers/hwtracing/coresight/Kconfig | 5 ++- drivers/hwtracing/coresight/Makefile | 5 ++- .../{coresight.c => coresight-core.c} | 42 ++++++++++++++----- .../hwtracing/coresight/coresight-etm-perf.c | 8 +++- .../hwtracing/coresight/coresight-etm-perf.h | 3 ++ 5 files changed, 48 insertions(+), 15 deletions(-) rename drivers/hwtracing/coresight/{coresight.c => coresight-core.c}
(98%)
diff --git a/drivers/hwtracing/coresight/Kconfig
b/drivers/hwtracing/coresight/Kconfig
index dfe407cde262..c1198245461d 100644 --- a/drivers/hwtracing/coresight/Kconfig +++ b/drivers/hwtracing/coresight/Kconfig @@ -3,7 +3,7 @@ # Coresight configuration # menuconfig CORESIGHT
- bool "CoreSight Tracing Support"
- tristate "CoreSight Tracing Support" depends on ARM || ARM64 depends on OF || ACPI select ARM_AMBA
@@ -15,6 +15,9 @@ menuconfig CORESIGHT specification and configure the right series of components when
a
trace source gets enabled.
To compile this driver as a module, choose M here: the
module will be called coresight.
if CORESIGHT config CORESIGHT_LINKS_AND_SINKS tristate "CoreSight Link and Sink drivers" diff --git a/drivers/hwtracing/coresight/Makefile
b/drivers/hwtracing/coresight/Makefile
index 0359d5a1588f..1b35b55bd420 100644 --- a/drivers/hwtracing/coresight/Makefile +++ b/drivers/hwtracing/coresight/Makefile @@ -2,8 +2,9 @@ # # Makefile for CoreSight drivers. # -obj-$(CONFIG_CORESIGHT) += coresight.o coresight-etm-perf.o \
coresight-platform.o coresight-sysfs.o
+obj-$(CONFIG_CORESIGHT) += coresight.o +coresight-y := coresight-core.o coresight-etm-perf.o
coresight-platform.o \
coresight-sysfs.o
obj-$(CONFIG_CORESIGHT_LINK_AND_SINK_TMC) += coresight-tmc.o coresight-tmc-y := coresight-tmc-core.o coresight-tmc-etf.o \ coresight-tmc-etr.o diff --git a/drivers/hwtracing/coresight/coresight.c
b/drivers/hwtracing/coresight/coresight-core.c
similarity index 98% rename from drivers/hwtracing/coresight/coresight.c rename to drivers/hwtracing/coresight/coresight-core.c index f9566d667261..045416ac962f 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight-core.c @@ -1462,16 +1462,6 @@ int coresight_timeout(void __iomem *addr, u32
offset, int position, int value)
} EXPORT_SYMBOL_GPL(coresight_timeout); -struct bus_type coresight_bustype = {
- .name = "coresight",
-};
-static int __init coresight_init(void) -{
- return bus_register(&coresight_bustype);
-} -postcore_initcall(coresight_init);
/*
- coresight_release_platform_data: Release references to the devices
connected
- to the output port of this device.
@@ -1680,3 +1670,35 @@ char *coresight_alloc_device_name(struct
coresight_dev_list *dict,
return name; } EXPORT_SYMBOL_GPL(coresight_alloc_device_name);
+struct bus_type coresight_bustype = {
- .name = "coresight",
+};
+static int __init coresight_init(void) +{
- int ret;
- ret = bus_register(&coresight_bustype);
- if (ret)
return ret;
- ret = etm_perf_init();
- if (ret)
bus_unregister(&coresight_bustype);
- return ret;
+}
+static void __exit coresight_exit(void) +{
- etm_perf_exit();
- bus_unregister(&coresight_bustype);
+}
+module_init(coresight_init); +module_exit(coresight_exit);
+MODULE_AUTHOR("Mathieu Poirier mathieu.poirier@linaro.org");
Pratik deserves credit too.
With that:
Reviewed-by: Mathieu Poirier mathieu.poirier@linaro.org
I am done reviewing this set.
Thank you for the patience, Mathieu
Thanks for the review of this big patchset, Mathieu. Sorry for some silly mistakes.
Thanks, Tingwei
+MODULE_DESCRIPTION("Arm CoreSight tracer driver"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c
b/drivers/hwtracing/coresight/coresight-etm-perf.c
index 3728c44e5763..668b3ff11576 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -591,7 +591,7 @@ void etm_perf_del_symlink_sink(struct
coresight_device *csdev)
csdev->ea = NULL; } -static int __init etm_perf_init(void) +int __init etm_perf_init(void) { int ret; @@ -618,4 +618,8 @@ static int __init etm_perf_init(void) return ret; } -device_initcall(etm_perf_init);
+void __exit etm_perf_exit(void) +{
- perf_pmu_unregister(&etm_pmu);
+} diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.h
b/drivers/hwtracing/coresight/coresight-etm-perf.h
index 05f89723e282..3e4f2ad5e193 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.h +++ b/drivers/hwtracing/coresight/coresight-etm-perf.h @@ -82,4 +82,7 @@ static inline void *etm_perf_sink_config(struct
perf_output_handle *handle)
#endif /* CONFIG_CORESIGHT */ +int __init etm_perf_init(void); +void __exit etm_perf_exit(void);
#endif
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora
Forum,
a Linux Foundation Collaborative Project