From: Geert Uytterhoeven <geert+renesas(a)glider.be>
[ Upstream commit 202dc3cc10b4d37e5251431acf8d5040a8876c7d ]
On SCIFA and SCIFB serial ports with DMA support (i.e. some ports on
R-Car Gen2 and RZ/G1 SoCs), receive DMA operations are submitted before
the DMA channel pointer is initialized. Hence this fails, and the
driver tries to fall back to PIO. However, at this early phase in the
initialization sequence, fallback to PIO does not work, leading to a
serial port that cannot receive any data.
Fix this by calling sci_submit_rx() after initialization of the DMA
channel pointer.
Reported-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh(a)renesas.com>
Fixes: 2c4ee23530ffc022 ("serial: sh-sci: Postpone DMA release when falling back to PIO")
Signed-off-by: Geert Uytterhoeven <geert+renesas(a)glider.be>
Reviewed-by: Wolfram Sang <wsa+renesas(a)sang-engineering.com>
Tested-by: Wolfram Sang <wsa+renesas(a)sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal(a)kernel.org>
---
drivers/tty/serial/sh-sci.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
index ab3f6e91853d..e19bfbba8a01 100644
--- a/drivers/tty/serial/sh-sci.c
+++ b/drivers/tty/serial/sh-sci.c
@@ -1614,10 +1614,10 @@ static void sci_request_dma(struct uart_port *port)
hrtimer_init(&s->rx_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
s->rx_timer.function = rx_timer_fn;
+ s->chan_rx_saved = s->chan_rx = chan;
+
if (port->type == PORT_SCIFA || port->type == PORT_SCIFB)
sci_submit_rx(s);
-
- s->chan_rx_saved = s->chan_rx = chan;
}
}
--
2.17.1
The leak was found when opening/closing a serial port a great number of
time, increasing kmalloc-32 in slabinfo.
Each time the port was opened, dma_request_slave_channel() was called.
Then, in at_dma_xlate(), atslave was allocated with devm_kzalloc() and
never freed. (Well, it was free at module unload, but that's not what we
want).
So, here, kzalloc is more suited for the job since it has to be freed in
atc_free_chan_resources().
Cc: stable(a)vger.kernel.org
Fixes: bbe89c8e3d59 ("at_hdmac: move to generic DMA binding")
Reported-by: Mario Forner <m.forner(a)be4energy.com>
Suggested-by: Alexandre Belloni <alexandre.belloni(a)bootlin.com>
Acked-by: Alexandre Belloni <alexandre.belloni(a)bootlin.com>
Signed-off-by: Richard Genoud <richard.genoud(a)gmail.com>
---
drivers/dma/at_hdmac.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/drivers/dma/at_hdmac.c b/drivers/dma/at_hdmac.c
index 7cbac6e8c113..1b7f0ca0d5cd 100644
--- a/drivers/dma/at_hdmac.c
+++ b/drivers/dma/at_hdmac.c
@@ -1641,6 +1641,12 @@ static void atc_free_chan_resources(struct dma_chan *chan)
atchan->descs_allocated = 0;
atchan->status = 0;
+ /*
+ * Free atslave allocated in at_dma_xlate()
+ */
+ kfree(chan->private);
+ chan->private = NULL;
+
dev_vdbg(chan2dev(chan), "free_chan_resources: done\n");
}
@@ -1675,7 +1681,7 @@ static struct dma_chan *at_dma_xlate(struct of_phandle_args *dma_spec,
dma_cap_zero(mask);
dma_cap_set(DMA_SLAVE, mask);
- atslave = devm_kzalloc(&dmac_pdev->dev, sizeof(*atslave), GFP_KERNEL);
+ atslave = kzalloc(sizeof(*atslave), GFP_KERNEL);
if (!atslave)
return NULL;
Sending the exact same hotplug event is not great uapi. Luckily the
only already merged implementation of leases (in the -modesetting
driver) doesn't care about what kind of uevent it gets, and
unconditionally processes both hotplug and lease changes. So we can
still adjust the uapi here.
But e.g. weston tries to filter stuff, and I guess others might want
to do that too. Try to make that possible. Cc: stable since it's uapi
adjustement that we want to roll out everywhere.
Cc: Keith Packard <keithp(a)keithp.com>
Cc: Dave Airlie <airlied(a)redhat.com>
Cc: stable(a)vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter(a)intel.com>
---
drivers/gpu/drm/drm_internal.h | 2 ++
drivers/gpu/drm/drm_lease.c | 2 +-
drivers/gpu/drm/drm_sysfs.c | 10 ++++++++++
3 files changed, 13 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/drm_internal.h b/drivers/gpu/drm/drm_internal.h
index c7a7d7ce5d1c..d9caf205e0b3 100644
--- a/drivers/gpu/drm/drm_internal.h
+++ b/drivers/gpu/drm/drm_internal.h
@@ -99,6 +99,8 @@ struct device *drm_sysfs_minor_alloc(struct drm_minor *minor);
int drm_sysfs_connector_add(struct drm_connector *connector);
void drm_sysfs_connector_remove(struct drm_connector *connector);
+void drm_sysfs_lease_event(struct drm_device *dev);
+
/* drm_gem.c */
int drm_gem_init(struct drm_device *dev);
void drm_gem_destroy(struct drm_device *dev);
diff --git a/drivers/gpu/drm/drm_lease.c b/drivers/gpu/drm/drm_lease.c
index 3650d3c46718..99cba8ea5d82 100644
--- a/drivers/gpu/drm/drm_lease.c
+++ b/drivers/gpu/drm/drm_lease.c
@@ -292,7 +292,7 @@ void drm_lease_destroy(struct drm_master *master)
if (master->lessor) {
/* Tell the master to check the lessee list */
- drm_sysfs_hotplug_event(dev);
+ drm_sysfs_lease_event(dev);
drm_master_put(&master->lessor);
}
diff --git a/drivers/gpu/drm/drm_sysfs.c b/drivers/gpu/drm/drm_sysfs.c
index b3c1daad1169..ecb7b33002bb 100644
--- a/drivers/gpu/drm/drm_sysfs.c
+++ b/drivers/gpu/drm/drm_sysfs.c
@@ -301,6 +301,16 @@ void drm_sysfs_connector_remove(struct drm_connector *connector)
connector->kdev = NULL;
}
+void drm_sysfs_lease_event(struct drm_device *dev)
+{
+ char *event_string = "LEASE=1";
+ char *envp[] = { event_string, NULL };
+
+ DRM_DEBUG("generating lease event\n");
+
+ kobject_uevent_env(&dev->primary->kdev->kobj, KOBJ_CHANGE, envp);
+}
+
/**
* drm_sysfs_hotplug_event - generate a DRM uevent
* @dev: DRM device
--
2.19.1
From: Zenghui Yu <yuzenghui(a)huawei.com>
After enabling KVM event tracing, almost all of trace_kvm_exit()'s
printk shows
"kvm_exit: IRQ: ..."
even if the actual exception_type is NOT IRQ. More specifically,
trace_kvm_exit() is defined in virt/kvm/arm/trace.h by TRACE_EVENT.
This slight problem may have existed after commit e6753f23d961
("tracepoint: Make rcuidle tracepoint callers use SRCU"). There are
two variables in trace_kvm_exit() and __DO_TRACE() which have the
same name, *idx*. Thus the actual value of *idx* will be overwritten
when tracing. Fix it by adding a simple prefix.
Cc: Joel Fernandes <joel(a)joelfernandes.org>
Cc: Wang Haibin <wanghaibin.wang(a)huawei.com>
Cc: linux-trace-devel(a)vger.kernel.org
Cc: stable(a)vger.kernel.org
Fixes: e6753f23d961 ("tracepoint: Make rcuidle tracepoint callers use SRCU")
Reviewed-by: Joel Fernandes (Google) <joel(a)joelfernandes.org>
Signed-off-by: Zenghui Yu <yuzenghui(a)huawei.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt(a)goodmis.org>
---
include/linux/tracepoint.h | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/include/linux/tracepoint.h b/include/linux/tracepoint.h
index 538ba1a58f5b..e9de8ad0bad7 100644
--- a/include/linux/tracepoint.h
+++ b/include/linux/tracepoint.h
@@ -166,7 +166,7 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p)
struct tracepoint_func *it_func_ptr; \
void *it_func; \
void *__data; \
- int __maybe_unused idx = 0; \
+ int __maybe_unused __idx = 0; \
\
if (!(cond)) \
return; \
@@ -182,7 +182,7 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p)
* doesn't work from the idle path. \
*/ \
if (rcuidle) { \
- idx = srcu_read_lock_notrace(&tracepoint_srcu); \
+ __idx = srcu_read_lock_notrace(&tracepoint_srcu);\
rcu_irq_enter_irqson(); \
} \
\
@@ -198,7 +198,7 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p)
\
if (rcuidle) { \
rcu_irq_exit_irqson(); \
- srcu_read_unlock_notrace(&tracepoint_srcu, idx);\
+ srcu_read_unlock_notrace(&tracepoint_srcu, __idx);\
} \
\
preempt_enable_notrace(); \
--
2.19.1