v4: - Update patch 1 to move apply_wqattrs_lock() and apply_wqattrs_unlock() down into CONFIG_SYSFS block to avoid compilation warnings.
v3: - Break out a separate patch to make workqueue_set_unbound_cpumask() static and move it down to the CONFIG_SYSFS section. - Remove the "__DEBUG__." prefix and the CFTYPE_DEBUG flag from the new root only cpuset.cpus.isolated control files and update the test accordingly.
v2: - Add 2 read-only workqueue sysfs files to expose the user requested cpumask as well as the isolated CPUs to be excluded from wq_unbound_cpumask. - Ensure that caller of the new workqueue_unbound_exclude_cpumask() hold cpus_read_lock. - Update the cpuset code to make sure the cpus_read_lock is held whenever workqueue_unbound_exclude_cpumask() may be called.
Isolated cpuset partition can currently be created to contain an exclusive set of CPUs not used in other cgroups and with load balancing disabled to reduce interference from the scheduler.
The main purpose of this isolated partition type is to dynamically emulate what can be done via the "isolcpus" boot command line option, specifically the default domain flag. One effect of the "isolcpus" option is to remove the isolated CPUs from the cpumasks of unbound workqueues since running work functions in an isolated CPU can be a major source of interference. Changing the unbound workqueue cpumasks can be done at run time by writing an appropriate cpumask without the isolated CPUs to /sys/devices/virtual/workqueue/cpumask. So one can set up an isolated cpuset partition and then write to the cpumask sysfs file to achieve similar level of CPU isolation. However, this manual process can be error prone.
This patch series implements automatic exclusion of isolated CPUs from unbound workqueue cpumasks when an isolated cpuset partition is created and then adds those CPUs back when the isolated partition is destroyed.
There are also other places in the kernel that look at the HK_FLAG_DOMAIN cpumask or other HK_FLAG_* cpumasks and exclude the isolated CPUs from certain actions to further reduce interference. CPUs in an isolated cpuset partition will not be able to avoid those interferences yet. That may change in the future as the need arises.
Waiman Long (5): workqueue: Make workqueue_set_unbound_cpumask() static workqueue: Add workqueue_unbound_exclude_cpumask() to exclude CPUs from wq_unbound_cpumask selftests/cgroup: Minor code cleanup and reorganization of test_cpuset_prs.sh cgroup/cpuset: Keep track of CPUs in isolated partitions cgroup/cpuset: Take isolated CPUs out of workqueue unbound cpumask
Documentation/admin-guide/cgroup-v2.rst | 10 +- include/linux/workqueue.h | 2 +- kernel/cgroup/cpuset.c | 286 +++++++++++++----- kernel/workqueue.c | 165 +++++++--- .../selftests/cgroup/test_cpuset_prs.sh | 216 ++++++++----- 5 files changed, 475 insertions(+), 204 deletions(-)
The workqueue_set_unbound_cpumask() function is only used inside workqueue.c with CONFIG_SYSFS. Make it static and move it down to within the ifdef CONFIG_SYSFS block. Also move apply_wqattrs_lock() and apply_wqattrs_unlock() down as all their callers are within the CONFIG_SYSFS block.
Also mark workqueue_apply_unbound_cpumask() as __maybe_unused for now as it will be used by another function introduced in a later commit. There is no functional change.
Signed-off-by: Waiman Long longman@redhat.com --- include/linux/workqueue.h | 1 - kernel/workqueue.c | 102 +++++++++++++++++++------------------- 2 files changed, 51 insertions(+), 52 deletions(-)
diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h index 24b1e5070f4d..cf49b467bd57 100644 --- a/include/linux/workqueue.h +++ b/include/linux/workqueue.h @@ -491,7 +491,6 @@ struct workqueue_attrs *alloc_workqueue_attrs(void); void free_workqueue_attrs(struct workqueue_attrs *attrs); int apply_workqueue_attrs(struct workqueue_struct *wq, const struct workqueue_attrs *attrs); -int workqueue_set_unbound_cpumask(cpumask_var_t cpumask);
extern bool queue_work_on(int cpu, struct workqueue_struct *wq, struct work_struct *work); diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 6e578f576a6f..145159f993e9 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -4411,19 +4411,6 @@ static void apply_wqattrs_commit(struct apply_wqattrs_ctx *ctx) mutex_unlock(&ctx->wq->mutex); }
-static void apply_wqattrs_lock(void) -{ - /* CPUs should stay stable across pwq creations and installations */ - cpus_read_lock(); - mutex_lock(&wq_pool_mutex); -} - -static void apply_wqattrs_unlock(void) -{ - mutex_unlock(&wq_pool_mutex); - cpus_read_unlock(); -} - static int apply_workqueue_attrs_locked(struct workqueue_struct *wq, const struct workqueue_attrs *attrs) { @@ -5784,7 +5771,7 @@ void thaw_workqueues(void) } #endif /* CONFIG_FREEZER */
-static int workqueue_apply_unbound_cpumask(const cpumask_var_t unbound_cpumask) +static int __maybe_unused workqueue_apply_unbound_cpumask(const cpumask_var_t unbound_cpumask) { LIST_HEAD(ctxs); int ret = 0; @@ -5827,43 +5814,6 @@ static int workqueue_apply_unbound_cpumask(const cpumask_var_t unbound_cpumask) return ret; }
-/** - * workqueue_set_unbound_cpumask - Set the low-level unbound cpumask - * @cpumask: the cpumask to set - * - * The low-level workqueues cpumask is a global cpumask that limits - * the affinity of all unbound workqueues. This function check the @cpumask - * and apply it to all unbound workqueues and updates all pwqs of them. - * - * Return: 0 - Success - * -EINVAL - Invalid @cpumask - * -ENOMEM - Failed to allocate memory for attrs or pwqs. - */ -int workqueue_set_unbound_cpumask(cpumask_var_t cpumask) -{ - int ret = -EINVAL; - - /* - * Not excluding isolated cpus on purpose. - * If the user wishes to include them, we allow that. - */ - cpumask_and(cpumask, cpumask, cpu_possible_mask); - if (!cpumask_empty(cpumask)) { - apply_wqattrs_lock(); - if (cpumask_equal(cpumask, wq_unbound_cpumask)) { - ret = 0; - goto out_unlock; - } - - ret = workqueue_apply_unbound_cpumask(cpumask); - -out_unlock: - apply_wqattrs_unlock(); - } - - return ret; -} - static int parse_affn_scope(const char *val) { int i; @@ -5982,6 +5932,19 @@ static struct attribute *wq_sysfs_attrs[] = { }; ATTRIBUTE_GROUPS(wq_sysfs);
+static void apply_wqattrs_lock(void) +{ + /* CPUs should stay stable across pwq creations and installations */ + cpus_read_lock(); + mutex_lock(&wq_pool_mutex); +} + +static void apply_wqattrs_unlock(void) +{ + mutex_unlock(&wq_pool_mutex); + cpus_read_unlock(); +} + static ssize_t wq_nice_show(struct device *dev, struct device_attribute *attr, char *buf) { @@ -6158,6 +6121,43 @@ static struct bus_type wq_subsys = { .dev_groups = wq_sysfs_groups, };
+/** + * workqueue_set_unbound_cpumask - Set the low-level unbound cpumask + * @cpumask: the cpumask to set + * + * The low-level workqueues cpumask is a global cpumask that limits + * the affinity of all unbound workqueues. This function check the @cpumask + * and apply it to all unbound workqueues and updates all pwqs of them. + * + * Return: 0 - Success + * -EINVAL - Invalid @cpumask + * -ENOMEM - Failed to allocate memory for attrs or pwqs. + */ +static int workqueue_set_unbound_cpumask(cpumask_var_t cpumask) +{ + int ret = -EINVAL; + + /* + * Not excluding isolated cpus on purpose. + * If the user wishes to include them, we allow that. + */ + cpumask_and(cpumask, cpumask, cpu_possible_mask); + if (!cpumask_empty(cpumask)) { + apply_wqattrs_lock(); + if (cpumask_equal(cpumask, wq_unbound_cpumask)) { + ret = 0; + goto out_unlock; + } + + ret = workqueue_apply_unbound_cpumask(cpumask); + +out_unlock: + apply_wqattrs_unlock(); + } + + return ret; +} + static ssize_t wq_unbound_cpumask_show(struct device *dev, struct device_attribute *attr, char *buf) {
When the "isolcpus" boot command line option is used to add a set of isolated CPUs, those CPUs will be excluded automatically from wq_unbound_cpumask to avoid running work functions from unbound workqueues.
Recently cpuset has been extended to allow the creation of partitions of isolated CPUs dynamically. To make it closer to the "isolcpus" in functionality, the CPUs in those isolated cpuset partitions should be excluded from wq_unbound_cpumask as well. This can be done currently by explicitly writing to the workqueue's cpumask sysfs file after creating the isolated partitions. However, this process can be error prone.
Ideally, the cpuset code should be allowed to request the workqueue code to exclude those isolated CPUs from wq_unbound_cpumask so that this operation can be done automatically and the isolated CPUs will be returned back to wq_unbound_cpumask after the destructions of the isolated cpuset partitions.
This patch adds a new workqueue_unbound_exclude_cpumask() function to enable that. This new function will exclude the specified isolated CPUs from wq_unbound_cpumask. To be able to restore those isolated CPUs back after the destruction of isolated cpuset partitions, a new wq_requested_unbound_cpumask is added to store the user provided unbound cpumask either from the boot command line options or from writing to the cpumask sysfs file. This new cpumask provides the basis for CPU exclusion.
To enable users to understand how the wq_unbound_cpumask is being modified internally, this patch also exposes the newly introduced wq_requested_unbound_cpumask as well as a wq_isolated_cpumask to store the cpumask to be excluded from wq_unbound_cpumask as read-only sysfs files.
Signed-off-by: Waiman Long longman@redhat.com --- include/linux/workqueue.h | 1 + kernel/workqueue.c | 91 +++++++++++++++++++++++++++++++++++---- 2 files changed, 84 insertions(+), 8 deletions(-)
diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h index cf49b467bd57..b0b9604b76b8 100644 --- a/include/linux/workqueue.h +++ b/include/linux/workqueue.h @@ -491,6 +491,7 @@ struct workqueue_attrs *alloc_workqueue_attrs(void); void free_workqueue_attrs(struct workqueue_attrs *attrs); int apply_workqueue_attrs(struct workqueue_struct *wq, const struct workqueue_attrs *attrs); +extern int workqueue_unbound_exclude_cpumask(cpumask_var_t cpumask);
extern bool queue_work_on(int cpu, struct workqueue_struct *wq, struct work_struct *work); diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 145159f993e9..2fc585d3d6ca 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -381,6 +381,12 @@ static bool workqueue_freezing; /* PL: have wqs started freezing? */ /* PL&A: allowable cpus for unbound wqs and work items */ static cpumask_var_t wq_unbound_cpumask;
+/* PL: user requested unbound cpumask via sysfs */ +static cpumask_var_t wq_requested_unbound_cpumask; + +/* PL: isolated cpumask to be excluded from unbound cpumask */ +static cpumask_var_t wq_isolated_cpumask; + /* for further constrain wq_unbound_cpumask by cmdline parameter*/ static struct cpumask wq_cmdline_cpumask __initdata;
@@ -5771,7 +5777,7 @@ void thaw_workqueues(void) } #endif /* CONFIG_FREEZER */
-static int __maybe_unused workqueue_apply_unbound_cpumask(const cpumask_var_t unbound_cpumask) +static int workqueue_apply_unbound_cpumask(const cpumask_var_t unbound_cpumask) { LIST_HEAD(ctxs); int ret = 0; @@ -5814,6 +5820,44 @@ static int __maybe_unused workqueue_apply_unbound_cpumask(const cpumask_var_t un return ret; }
+/** + * workqueue_unbound_exclude_cpumask - Exclude given CPUs from unbound cpumask + * @exclude_cpumask: the cpumask to be excluded from wq_unbound_cpumask + * + * This function can be called from cpuset code to provide a set of isolated + * CPUs that should be excluded from wq_unbound_cpumask. The caller must hold + * either cpus_read_lock or cpus_write_lock. + */ +int workqueue_unbound_exclude_cpumask(cpumask_var_t exclude_cpumask) +{ + cpumask_var_t cpumask; + int ret = 0; + + if (!zalloc_cpumask_var(&cpumask, GFP_KERNEL)) + return -ENOMEM; + + lockdep_assert_cpus_held(); + mutex_lock(&wq_pool_mutex); + + /* Save the current isolated cpumask & export it via sysfs */ + cpumask_copy(wq_isolated_cpumask, exclude_cpumask); + + /* + * If the operation fails, it will fall back to + * wq_requested_unbound_cpumask which is initially set to + * (HK_TYPE_WQ ∩ HK_TYPE_DOMAIN) house keeping mask and rewritten + * by any subsequent write to workqueue/cpumask sysfs file. + */ + if (!cpumask_andnot(cpumask, wq_requested_unbound_cpumask, exclude_cpumask)) + cpumask_copy(cpumask, wq_requested_unbound_cpumask); + if (!cpumask_equal(cpumask, wq_unbound_cpumask)) + ret = workqueue_apply_unbound_cpumask(cpumask); + + mutex_unlock(&wq_pool_mutex); + free_cpumask_var(cpumask); + return ret; +} + static int parse_affn_scope(const char *val) { int i; @@ -6144,6 +6188,7 @@ static int workqueue_set_unbound_cpumask(cpumask_var_t cpumask) cpumask_and(cpumask, cpumask, cpu_possible_mask); if (!cpumask_empty(cpumask)) { apply_wqattrs_lock(); + cpumask_copy(wq_requested_unbound_cpumask, cpumask); if (cpumask_equal(cpumask, wq_unbound_cpumask)) { ret = 0; goto out_unlock; @@ -6158,19 +6203,36 @@ static int workqueue_set_unbound_cpumask(cpumask_var_t cpumask) return ret; }
-static ssize_t wq_unbound_cpumask_show(struct device *dev, - struct device_attribute *attr, char *buf) +static ssize_t __wq_cpumask_show(struct device *dev, + struct device_attribute *attr, char *buf, cpumask_var_t mask) { int written;
mutex_lock(&wq_pool_mutex); - written = scnprintf(buf, PAGE_SIZE, "%*pb\n", - cpumask_pr_args(wq_unbound_cpumask)); + written = scnprintf(buf, PAGE_SIZE, "%*pb\n", cpumask_pr_args(mask)); mutex_unlock(&wq_pool_mutex);
return written; }
+static ssize_t wq_unbound_cpumask_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + return __wq_cpumask_show(dev, attr, buf, wq_unbound_cpumask); +} + +static ssize_t wq_requested_cpumask_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + return __wq_cpumask_show(dev, attr, buf, wq_requested_unbound_cpumask); +} + +static ssize_t wq_isolated_cpumask_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + return __wq_cpumask_show(dev, attr, buf, wq_isolated_cpumask); +} + static ssize_t wq_unbound_cpumask_store(struct device *dev, struct device_attribute *attr, const char *buf, size_t count) { @@ -6188,9 +6250,13 @@ static ssize_t wq_unbound_cpumask_store(struct device *dev, return ret ? ret : count; }
-static struct device_attribute wq_sysfs_cpumask_attr = +static struct device_attribute wq_sysfs_cpumask_attrs[] = { __ATTR(cpumask, 0644, wq_unbound_cpumask_show, - wq_unbound_cpumask_store); + wq_unbound_cpumask_store), + __ATTR(cpumask_requested, 0444, wq_requested_cpumask_show, NULL), + __ATTR(cpumask_isolated, 0444, wq_isolated_cpumask_show, NULL), + __ATTR_NULL, +};
static int __init wq_sysfs_init(void) { @@ -6203,7 +6269,13 @@ static int __init wq_sysfs_init(void)
dev_root = bus_get_dev_root(&wq_subsys); if (dev_root) { - err = device_create_file(dev_root, &wq_sysfs_cpumask_attr); + struct device_attribute *attr; + + for (attr = wq_sysfs_cpumask_attrs; attr->attr.name; attr++) { + err = device_create_file(dev_root, attr); + if (err) + break; + } put_device(dev_root); } return err; @@ -6534,11 +6606,14 @@ void __init workqueue_init_early(void) BUILD_BUG_ON(__alignof__(struct pool_workqueue) < __alignof__(long long));
BUG_ON(!alloc_cpumask_var(&wq_unbound_cpumask, GFP_KERNEL)); + BUG_ON(!alloc_cpumask_var(&wq_requested_unbound_cpumask, GFP_KERNEL)); + BUG_ON(!zalloc_cpumask_var(&wq_isolated_cpumask, GFP_KERNEL)); cpumask_copy(wq_unbound_cpumask, housekeeping_cpumask(HK_TYPE_WQ)); cpumask_and(wq_unbound_cpumask, wq_unbound_cpumask, housekeeping_cpumask(HK_TYPE_DOMAIN));
if (!cpumask_empty(&wq_cmdline_cpumask)) cpumask_and(wq_unbound_cpumask, wq_unbound_cpumask, &wq_cmdline_cpumask); + cpumask_copy(wq_requested_unbound_cpumask, wq_unbound_cpumask);
pwq_cache = KMEM_CACHE(pool_workqueue, SLAB_PANIC);
Minor cleanup of test matrix and relocation of test_isolated() function to prepare for the next patch. There is no functional change.
Signed-off-by: Waiman Long longman@redhat.com --- .../selftests/cgroup/test_cpuset_prs.sh | 142 +++++++++--------- 1 file changed, 71 insertions(+), 71 deletions(-)
diff --git a/tools/testing/selftests/cgroup/test_cpuset_prs.sh b/tools/testing/selftests/cgroup/test_cpuset_prs.sh index a6e9848189d6..2b825019f806 100755 --- a/tools/testing/selftests/cgroup/test_cpuset_prs.sh +++ b/tools/testing/selftests/cgroup/test_cpuset_prs.sh @@ -146,71 +146,6 @@ test_add_proc() echo $$ > $CGROUP2/cgroup.procs # Move out the task }
-# -# Testing the new "isolated" partition root type -# -test_isolated() -{ - cd $CGROUP2/test - echo 2-3 > cpuset.cpus - TYPE=$(cat cpuset.cpus.partition) - [[ $TYPE = member ]] || echo member > cpuset.cpus.partition - - console_msg "Change from member to root" - test_partition root - - console_msg "Change from root to isolated" - test_partition isolated - - console_msg "Change from isolated to member" - test_partition member - - console_msg "Change from member to isolated" - test_partition isolated - - console_msg "Change from isolated to root" - test_partition root - - console_msg "Change from root to member" - test_partition member - - # - # Testing partition root with no cpu - # - console_msg "Distribute all cpus to child partition" - echo +cpuset > cgroup.subtree_control - test_partition root - - mkdir A1 - cd A1 - echo 2-3 > cpuset.cpus - test_partition root - test_effective_cpus 2-3 - cd .. - test_effective_cpus "" - - console_msg "Moving task to partition test" - test_add_proc "No space left" - cd A1 - test_add_proc "" - cd .. - - console_msg "Shrink and expand child partition" - cd A1 - echo 2 > cpuset.cpus - cd .. - test_effective_cpus 3 - cd A1 - echo 2-3 > cpuset.cpus - cd .. - test_effective_cpus "" - - # Cleaning up - console_msg "Cleaning up" - echo $$ > $CGROUP2/cgroup.procs - [[ -d A1 ]] && rmdir A1 -} - # # Cpuset controller state transition test matrix. # @@ -304,7 +239,7 @@ TEST_MATRIX=( A1:P0,A2:P2,A3:P1 2-4" " C0-4:X2-4:S+ C1-4:X2-4:S+:P2 C2-4:X4:P1 \ . . X5 . . 0 A1:0-4,A2:1-4,A3:2-4 \ - A1:P0,A2:P-2,A3:P-1 ." + A1:P0,A2:P-2,A3:P-1" " C0-4:X2-4:S+ C1-4:X2-4:S+:P2 C2-4:X4:P1 \ . . . X1 . 0 A1:0-1,A2:2-4,A3:2-4 \ A1:P0,A2:P2,A3:P-1 2-4" @@ -347,10 +282,10 @@ TEST_MATRIX=( # cpus_allowed/exclusive_cpus update tests " C0-3:X2-3:S+ C1-3:X2-3:S+ C2-3:X2-3 \ . C4 . P2 . 0 A1:4,A2:4,XA2:,XA3:,A3:4 \ - A1:P0,A3:P-2 ." + A1:P0,A3:P-2" " C0-3:X2-3:S+ C1-3:X2-3:S+ C2-3:X2-3 \ . X1 . P2 . 0 A1:0-3,A2:1-3,XA1:1,XA2:,XA3:,A3:2-3 \ - A1:P0,A3:P-2 ." + A1:P0,A3:P-2" " C0-3:X2-3:S+ C1-3:X2-3:S+ C2-3:X2-3 \ . . C3 P2 . 0 A1:0-2,A2:0-2,XA2:3,XA3:3,A3:3 \ A1:P0,A3:P2 3" @@ -359,13 +294,13 @@ TEST_MATRIX=( A1:P0,A3:P2 3" " C0-3:X2-3:S+ C1-3:X2-3:S+ C2-3:X2-3:P2 \ . . X3 . . 0 A1:0-3,A2:1-3,XA2:3,XA3:3,A3:2-3 \ - A1:P0,A3:P-2 ." + A1:P0,A3:P-2" " C0-3:X2-3:S+ C1-3:X2-3:S+ C2-3:X2-3:P2 \ . . C3 . . 0 A1:0-3,A2:3,XA2:3,XA3:3,A3:3 \ - A1:P0,A3:P-2 ." + A1:P0,A3:P-2" " C0-3:X2-3:S+ C1-3:X2-3:S+ C2-3:X2-3:P2 \ . C4 . . . 0 A1:4,A2:4,A3:4,XA1:,XA2:,XA3 \ - A1:P0,A3:P-2 ." + A1:P0,A3:P-2"
# old-A1 old-A2 old-A3 old-B1 new-A1 new-A2 new-A3 new-B1 fail ECPUs Pstate ISOLCPUS # ------ ------ ------ ------ ------ ------ ------ ------ ---- ----- ------ -------- @@ -804,6 +739,71 @@ run_state_test() echo "All $I tests of $TEST PASSED." }
+# +# Testing the new "isolated" partition root type +# +test_isolated() +{ + cd $CGROUP2/test + echo 2-3 > cpuset.cpus + TYPE=$(cat cpuset.cpus.partition) + [[ $TYPE = member ]] || echo member > cpuset.cpus.partition + + console_msg "Change from member to root" + test_partition root + + console_msg "Change from root to isolated" + test_partition isolated + + console_msg "Change from isolated to member" + test_partition member + + console_msg "Change from member to isolated" + test_partition isolated + + console_msg "Change from isolated to root" + test_partition root + + console_msg "Change from root to member" + test_partition member + + # + # Testing partition root with no cpu + # + console_msg "Distribute all cpus to child partition" + echo +cpuset > cgroup.subtree_control + test_partition root + + mkdir A1 + cd A1 + echo 2-3 > cpuset.cpus + test_partition root + test_effective_cpus 2-3 + cd .. + test_effective_cpus "" + + console_msg "Moving task to partition test" + test_add_proc "No space left" + cd A1 + test_add_proc "" + cd .. + + console_msg "Shrink and expand child partition" + cd A1 + echo 2 > cpuset.cpus + cd .. + test_effective_cpus 3 + cd A1 + echo 2-3 > cpuset.cpus + cd .. + test_effective_cpus "" + + # Cleaning up + console_msg "Cleaning up" + echo $$ > $CGROUP2/cgroup.procs + [[ -d A1 ]] && rmdir A1 +} + # # Wait for inotify event for the given file and read it # $1: cgroup file to wait for
Add a new internal isolated_cpus mask to keep track of the CPUs that are in isolated partitions. Expose that new cpumask as a new root-only control file "cpuset.cpus.isolated".
Signed-off-by: Waiman Long longman@redhat.com --- kernel/cgroup/cpuset.c | 190 +++++++++++++++++++++++++++-------------- 1 file changed, 127 insertions(+), 63 deletions(-)
diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 615daaf87f1f..a265e559f3fa 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -204,6 +204,11 @@ struct cpuset { */ static cpumask_var_t subpartitions_cpus;
+/* + * Exclusive CPUs in isolated partitions + */ +static cpumask_var_t isolated_cpus; + /* List of remote partition root children */ static struct list_head remote_children;
@@ -1317,6 +1322,7 @@ static void compute_effective_cpumask(struct cpumask *new_cpus, */ enum partition_cmd { partcmd_enable, /* Enable partition root */ + partcmd_enablei, /* Enable isolated partition root */ partcmd_disable, /* Disable partition root */ partcmd_update, /* Update parent's effective_cpus */ partcmd_invalidate, /* Make partition invalid */ @@ -1418,6 +1424,74 @@ static void reset_partition_data(struct cpuset *cs) } }
+/* + * partition_xcpus_newstate - Exclusive CPUs state change + * @old_prs: old partition_root_state + * @new_prs: new partition_root_state + * @xcpus: exclusive CPUs with state change + */ +static void partition_xcpus_newstate(int old_prs, int new_prs, struct cpumask *xcpus) +{ + WARN_ON_ONCE(old_prs == new_prs); + if (new_prs == PRS_ISOLATED) + cpumask_or(isolated_cpus, isolated_cpus, xcpus); + else + cpumask_andnot(isolated_cpus, isolated_cpus, xcpus); +} + +/* + * partition_xcpus_add - Add new exclusive CPUs to partition + * @new_prs: new partition_root_state + * @parent: parent cpuset + * @xcpus: exclusive CPUs to be added + * + * Remote partition if parent == NULL + */ +static void partition_xcpus_add(int new_prs, struct cpuset *parent, + struct cpumask *xcpus) +{ + WARN_ON_ONCE(new_prs < 0); + lockdep_assert_held(&callback_lock); + if (!parent) + parent = &top_cpuset; + + if (parent == &top_cpuset) + cpumask_or(subpartitions_cpus, subpartitions_cpus, xcpus); + + if (new_prs != parent->partition_root_state) + partition_xcpus_newstate(parent->partition_root_state, new_prs, + xcpus); + + cpumask_andnot(parent->effective_cpus, parent->effective_cpus, xcpus); +} + +/* + * partition_xcpus_del - Remove exclusive CPUs from partition + * @old_prs: old partition_root_state + * @parent: parent cpuset + * @xcpus: exclusive CPUs to be removed + * + * Remote partition if parent == NULL + */ +static void partition_xcpus_del(int old_prs, struct cpuset *parent, + struct cpumask *xcpus) +{ + WARN_ON_ONCE(old_prs < 0); + lockdep_assert_held(&callback_lock); + if (!parent) + parent = &top_cpuset; + + if (parent == &top_cpuset) + cpumask_andnot(subpartitions_cpus, subpartitions_cpus, xcpus); + + if (old_prs != parent->partition_root_state) + partition_xcpus_newstate(old_prs, parent->partition_root_state, + xcpus); + + cpumask_and(xcpus, xcpus, cpu_active_mask); + cpumask_or(parent->effective_cpus, parent->effective_cpus, xcpus); +} + /* * compute_effective_exclusive_cpumask - compute effective exclusive CPUs * @cs: cpuset @@ -1456,13 +1530,15 @@ static inline bool is_local_partition(struct cpuset *cs) /* * remote_partition_enable - Enable current cpuset as a remote partition root * @cs: the cpuset to update + * @new_prs: new partition_root_state * @tmp: temparary masks * Return: 1 if successful, 0 if error * * Enable the current cpuset to become a remote partition root taking CPUs * directly from the top cpuset. cpuset_mutex must be held by the caller. */ -static int remote_partition_enable(struct cpuset *cs, struct tmpmasks *tmp) +static int remote_partition_enable(struct cpuset *cs, int new_prs, + struct tmpmasks *tmp) { /* * The user must have sysadmin privilege. @@ -1485,18 +1561,14 @@ static int remote_partition_enable(struct cpuset *cs, struct tmpmasks *tmp) return 0;
spin_lock_irq(&callback_lock); - cpumask_andnot(top_cpuset.effective_cpus, - top_cpuset.effective_cpus, tmp->new_cpus); - cpumask_or(subpartitions_cpus, - subpartitions_cpus, tmp->new_cpus); - + partition_xcpus_add(new_prs, NULL, tmp->new_cpus); + list_add(&cs->remote_sibling, &remote_children); if (cs->use_parent_ecpus) { struct cpuset *parent = parent_cs(cs);
cs->use_parent_ecpus = false; parent->child_ecpus_count--; } - list_add(&cs->remote_sibling, &remote_children); spin_unlock_irq(&callback_lock);
/* @@ -1524,13 +1596,8 @@ static void remote_partition_disable(struct cpuset *cs, struct tmpmasks *tmp) WARN_ON_ONCE(!cpumask_subset(tmp->new_cpus, subpartitions_cpus));
spin_lock_irq(&callback_lock); - cpumask_andnot(subpartitions_cpus, - subpartitions_cpus, tmp->new_cpus); - cpumask_and(tmp->new_cpus, - tmp->new_cpus, cpu_active_mask); - cpumask_or(top_cpuset.effective_cpus, - top_cpuset.effective_cpus, tmp->new_cpus); list_del_init(&cs->remote_sibling); + partition_xcpus_del(cs->partition_root_state, NULL, tmp->new_cpus); cs->partition_root_state = -cs->partition_root_state; if (!cs->prs_err) cs->prs_err = PERR_INVCPUS; @@ -1557,6 +1624,7 @@ static void remote_cpus_update(struct cpuset *cs, struct cpumask *newmask, struct tmpmasks *tmp) { bool adding, deleting; + int prs = cs->partition_root_state;
if (WARN_ON_ONCE(!is_remote_partition(cs))) return; @@ -1580,20 +1648,10 @@ static void remote_cpus_update(struct cpuset *cs, struct cpumask *newmask, goto invalidate;
spin_lock_irq(&callback_lock); - if (adding) { - cpumask_or(subpartitions_cpus, - subpartitions_cpus, tmp->addmask); - cpumask_andnot(top_cpuset.effective_cpus, - top_cpuset.effective_cpus, tmp->addmask); - } - if (deleting) { - cpumask_andnot(subpartitions_cpus, - subpartitions_cpus, tmp->delmask); - cpumask_and(tmp->delmask, - tmp->delmask, cpu_active_mask); - cpumask_or(top_cpuset.effective_cpus, - top_cpuset.effective_cpus, tmp->delmask); - } + if (adding) + partition_xcpus_add(prs, NULL, tmp->addmask); + if (deleting) + partition_xcpus_del(prs, NULL, tmp->delmask); spin_unlock_irq(&callback_lock);
/* @@ -1676,11 +1734,11 @@ static bool prstate_housekeeping_conflict(int prstate, struct cpumask *new_cpus) * @tmp: Temporary addmask and delmask * Return: 0 or a partition root state error code * - * For partcmd_enable, the cpuset is being transformed from a non-partition - * root to a partition root. The effective_xcpus (cpus_allowed if effective_xcpus - * not set) mask of the given cpuset will be taken away from parent's - * effective_cpus. The function will return 0 if all the CPUs listed in - * effective_xcpus can be granted or an error code will be returned. + * For partcmd_enable*, the cpuset is being transformed from a non-partition + * root to a partition root. The effective_xcpus (cpus_allowed if + * effective_xcpus not set) mask of the given cpuset will be taken away from + * parent's effective_cpus. The function will return 0 if all the CPUs listed + * in effective_xcpus can be granted or an error code will be returned. * * For partcmd_disable, the cpuset is being transformed from a partition * root back to a non-partition root. Any CPUs in effective_xcpus will be @@ -1695,7 +1753,7 @@ static bool prstate_housekeeping_conflict(int prstate, struct cpumask *new_cpus) * * For partcmd_invalidate, the current partition will be made invalid. * - * The partcmd_enable and partcmd_disable commands are used by + * The partcmd_enable* and partcmd_disable commands are used by * update_prstate(). An error code may be returned and the caller will check * for error. * @@ -1760,7 +1818,7 @@ static int update_parent_effective_cpumask(struct cpuset *cs, int cmd,
nocpu = tasks_nocpu_error(parent, cs, xcpus);
- if (cmd == partcmd_enable) { + if ((cmd == partcmd_enable) || (cmd == partcmd_enablei)) { /* * Enabling partition root is not allowed if its * effective_xcpus is empty or doesn't overlap with @@ -1783,6 +1841,7 @@ static int update_parent_effective_cpumask(struct cpuset *cs, int cmd, cpumask_copy(tmp->delmask, xcpus); deleting = true; subparts_delta++; + new_prs = (cmd == partcmd_enable) ? PRS_ROOT : PRS_ISOLATED; } else if (cmd == partcmd_disable) { /* * May need to add cpus to parent's effective_cpus for @@ -1792,6 +1851,7 @@ static int update_parent_effective_cpumask(struct cpuset *cs, int cmd, cpumask_and(tmp->addmask, xcpus, parent->effective_xcpus); if (adding) subparts_delta--; + new_prs = PRS_MEMBER; } else if (newmask) { /* * Empty cpumask is not allowed @@ -1940,37 +2000,24 @@ static int update_parent_effective_cpumask(struct cpuset *cs, int cmd, * newly deleted ones will be added back to effective_cpus. */ spin_lock_irq(&callback_lock); - if (adding) { - if (parent == &top_cpuset) - cpumask_andnot(subpartitions_cpus, - subpartitions_cpus, tmp->addmask); - /* - * Some of the CPUs in effective_xcpus might have been offlined. - */ - cpumask_or(parent->effective_cpus, - parent->effective_cpus, tmp->addmask); - cpumask_and(parent->effective_cpus, - parent->effective_cpus, cpu_active_mask); - } - if (deleting) { - if (parent == &top_cpuset) - cpumask_or(subpartitions_cpus, - subpartitions_cpus, tmp->delmask); - cpumask_andnot(parent->effective_cpus, - parent->effective_cpus, tmp->delmask); - } - - if (is_partition_valid(parent)) { - parent->nr_subparts += subparts_delta; - WARN_ON_ONCE(parent->nr_subparts < 0); - } - if (old_prs != new_prs) { cs->partition_root_state = new_prs; if (new_prs <= 0) cs->nr_subparts = 0; } + /* + * Adding to parent's effective_cpus means deletion CPUs from cs + * and vice versa. + */ + if (adding) + partition_xcpus_del(old_prs, parent, tmp->addmask); + if (deleting) + partition_xcpus_add(new_prs, parent, tmp->delmask);
+ if (is_partition_valid(parent)) { + parent->nr_subparts += subparts_delta; + WARN_ON_ONCE(parent->nr_subparts < 0); + } spin_unlock_irq(&callback_lock);
if ((old_prs != new_prs) && (cmd == partcmd_update)) @@ -2948,6 +2995,7 @@ static int update_prstate(struct cpuset *cs, int new_prs) int err = PERR_NONE, old_prs = cs->partition_root_state; struct cpuset *parent = parent_cs(cs); struct tmpmasks tmpmask; + bool new_xcpus_state = false;
if (old_prs == new_prs) return 0; @@ -2977,6 +3025,9 @@ static int update_prstate(struct cpuset *cs, int new_prs) goto out;
if (!old_prs) { + enum partition_cmd cmd = (new_prs == PRS_ROOT) + ? partcmd_enable : partcmd_enablei; + /* * cpus_allowed cannot be empty. */ @@ -2985,19 +3036,18 @@ static int update_prstate(struct cpuset *cs, int new_prs) goto out; }
- err = update_parent_effective_cpumask(cs, partcmd_enable, - NULL, &tmpmask); + err = update_parent_effective_cpumask(cs, cmd, NULL, &tmpmask); /* * If an attempt to become local partition root fails, * try to become a remote partition root instead. */ - if (err && remote_partition_enable(cs, &tmpmask)) + if (err && remote_partition_enable(cs, new_prs, &tmpmask)) err = 0; } else if (old_prs && new_prs) { /* * A change in load balance state only, no change in cpumasks. */ - ; + new_xcpus_state = true; } else { /* * Switching back to member is always allowed even if it @@ -3029,6 +3079,8 @@ static int update_prstate(struct cpuset *cs, int new_prs) WRITE_ONCE(cs->prs_err, err); if (!is_partition_valid(cs)) reset_partition_data(cs); + else if (new_xcpus_state) + partition_xcpus_newstate(old_prs, new_prs, cs->effective_xcpus); spin_unlock_irq(&callback_lock);
/* Force update if switching back to member */ @@ -3386,6 +3438,7 @@ typedef enum { FILE_SUBPARTS_CPULIST, FILE_EXCLUSIVE_CPULIST, FILE_EFFECTIVE_XCPULIST, + FILE_ISOLATED_CPULIST, FILE_CPU_EXCLUSIVE, FILE_MEM_EXCLUSIVE, FILE_MEM_HARDWALL, @@ -3582,6 +3635,9 @@ static int cpuset_common_seq_show(struct seq_file *sf, void *v) case FILE_SUBPARTS_CPULIST: seq_printf(sf, "%*pbl\n", cpumask_pr_args(subpartitions_cpus)); break; + case FILE_ISOLATED_CPULIST: + seq_printf(sf, "%*pbl\n", cpumask_pr_args(isolated_cpus)); + break; default: ret = -EINVAL; } @@ -3875,6 +3931,13 @@ static struct cftype dfl_files[] = { .flags = CFTYPE_ONLY_ON_ROOT | CFTYPE_DEBUG, },
+ { + .name = "cpus.isolated", + .seq_show = cpuset_common_seq_show, + .private = FILE_ISOLATED_CPULIST, + .flags = CFTYPE_ONLY_ON_ROOT, + }, + { } /* terminate */ };
@@ -4194,6 +4257,7 @@ int __init cpuset_init(void) BUG_ON(!alloc_cpumask_var(&top_cpuset.effective_xcpus, GFP_KERNEL)); BUG_ON(!alloc_cpumask_var(&top_cpuset.exclusive_cpus, GFP_KERNEL)); BUG_ON(!zalloc_cpumask_var(&subpartitions_cpus, GFP_KERNEL)); + BUG_ON(!zalloc_cpumask_var(&isolated_cpus, GFP_KERNEL));
cpumask_setall(top_cpuset.cpus_allowed); nodes_setall(top_cpuset.mems_allowed);
To make CPUs in isolated cpuset partition closer in isolation to the boot time isolated CPUs specified in the "isolcpus" boot command line option, we need to take those CPUs out of the workqueue unbound cpumask so that work functions from the unbound workqueues won't run on those CPUs. Otherwise, they will interfere the user tasks running on those isolated CPUs.
With the introduction of the workqueue_unbound_exclude_cpumask() helper function in an earlier commit, those isolated CPUs can now be taken out from the workqueue unbound cpumask.
This patch also updates cgroup-v2.rst to mention that isolated CPUs will be excluded from unbound workqueue cpumask as well as updating test_cpuset_prs.sh to verify the correctness of the new *cpuset.cpus.isolated file, if available via cgroup_debug option.
Signed-off-by: Waiman Long longman@redhat.com --- Documentation/admin-guide/cgroup-v2.rst | 10 +- kernel/cgroup/cpuset.c | 116 +++++++++++++++--- .../selftests/cgroup/test_cpuset_prs.sh | 74 +++++++++-- 3 files changed, 166 insertions(+), 34 deletions(-)
diff --git a/Documentation/admin-guide/cgroup-v2.rst b/Documentation/admin-guide/cgroup-v2.rst index 3f85254f3cef..cf5651a11df8 100644 --- a/Documentation/admin-guide/cgroup-v2.rst +++ b/Documentation/admin-guide/cgroup-v2.rst @@ -2358,11 +2358,11 @@ Cpuset Interface Files partition or scheduling domain. The set of exclusive CPUs is determined by the value of its "cpuset.cpus.exclusive.effective".
- When set to "isolated", the CPUs in that partition will - be in an isolated state without any load balancing from the - scheduler. Tasks placed in such a partition with multiple - CPUs should be carefully distributed and bound to each of the - individual CPUs for optimal performance. + When set to "isolated", the CPUs in that partition will be in + an isolated state without any load balancing from the scheduler + and excluded from the unbound workqueues. Tasks placed in such + a partition with multiple CPUs should be carefully distributed + and bound to each of the individual CPUs for optimal performance.
A partition root ("root" or "isolated") can be in one of the two possible states - valid or invalid. An invalid partition diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index a265e559f3fa..2a16df86c55c 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -25,6 +25,7 @@ #include <linux/cpu.h> #include <linux/cpumask.h> #include <linux/cpuset.h> +#include <linux/delay.h> #include <linux/init.h> #include <linux/interrupt.h> #include <linux/kernel.h> @@ -43,6 +44,7 @@ #include <linux/sched/isolation.h> #include <linux/cgroup.h> #include <linux/wait.h> +#include <linux/workqueue.h>
DEFINE_STATIC_KEY_FALSE(cpusets_pre_enable_key); DEFINE_STATIC_KEY_FALSE(cpusets_enabled_key); @@ -1444,25 +1446,31 @@ static void partition_xcpus_newstate(int old_prs, int new_prs, struct cpumask *x * @new_prs: new partition_root_state * @parent: parent cpuset * @xcpus: exclusive CPUs to be added + * Return: true if isolated_cpus modified, false otherwise * * Remote partition if parent == NULL */ -static void partition_xcpus_add(int new_prs, struct cpuset *parent, +static bool partition_xcpus_add(int new_prs, struct cpuset *parent, struct cpumask *xcpus) { + bool isolcpus_updated; + WARN_ON_ONCE(new_prs < 0); lockdep_assert_held(&callback_lock); if (!parent) parent = &top_cpuset;
+ if (parent == &top_cpuset) cpumask_or(subpartitions_cpus, subpartitions_cpus, xcpus);
- if (new_prs != parent->partition_root_state) + isolcpus_updated = (new_prs != parent->partition_root_state); + if (isolcpus_updated) partition_xcpus_newstate(parent->partition_root_state, new_prs, xcpus);
cpumask_andnot(parent->effective_cpus, parent->effective_cpus, xcpus); + return isolcpus_updated; }
/* @@ -1470,12 +1478,15 @@ static void partition_xcpus_add(int new_prs, struct cpuset *parent, * @old_prs: old partition_root_state * @parent: parent cpuset * @xcpus: exclusive CPUs to be removed + * Return: true if isolated_cpus modified, false otherwise * * Remote partition if parent == NULL */ -static void partition_xcpus_del(int old_prs, struct cpuset *parent, +static bool partition_xcpus_del(int old_prs, struct cpuset *parent, struct cpumask *xcpus) { + bool isolcpus_updated; + WARN_ON_ONCE(old_prs < 0); lockdep_assert_held(&callback_lock); if (!parent) @@ -1484,12 +1495,27 @@ static void partition_xcpus_del(int old_prs, struct cpuset *parent, if (parent == &top_cpuset) cpumask_andnot(subpartitions_cpus, subpartitions_cpus, xcpus);
- if (old_prs != parent->partition_root_state) + isolcpus_updated = (old_prs != parent->partition_root_state); + if (isolcpus_updated) partition_xcpus_newstate(old_prs, parent->partition_root_state, xcpus);
cpumask_and(xcpus, xcpus, cpu_active_mask); cpumask_or(parent->effective_cpus, parent->effective_cpus, xcpus); + return isolcpus_updated; +} + +static void update_unbound_workqueue_cpumask(bool isolcpus_updated) +{ + int ret; + + lockdep_assert_cpus_held(); + + if (!isolcpus_updated) + return; + + ret = workqueue_unbound_exclude_cpumask(isolated_cpus); + WARN_ON_ONCE(ret < 0); }
/* @@ -1540,6 +1566,8 @@ static inline bool is_local_partition(struct cpuset *cs) static int remote_partition_enable(struct cpuset *cs, int new_prs, struct tmpmasks *tmp) { + bool isolcpus_updated; + /* * The user must have sysadmin privilege. */ @@ -1561,7 +1589,7 @@ static int remote_partition_enable(struct cpuset *cs, int new_prs, return 0;
spin_lock_irq(&callback_lock); - partition_xcpus_add(new_prs, NULL, tmp->new_cpus); + isolcpus_updated = partition_xcpus_add(new_prs, NULL, tmp->new_cpus); list_add(&cs->remote_sibling, &remote_children); if (cs->use_parent_ecpus) { struct cpuset *parent = parent_cs(cs); @@ -1570,13 +1598,13 @@ static int remote_partition_enable(struct cpuset *cs, int new_prs, parent->child_ecpus_count--; } spin_unlock_irq(&callback_lock); + update_unbound_workqueue_cpumask(isolcpus_updated);
/* * Proprogate changes in top_cpuset's effective_cpus down the hierarchy. */ update_tasks_cpumask(&top_cpuset, tmp->new_cpus); update_sibling_cpumasks(&top_cpuset, NULL, tmp); - return 1; }
@@ -1591,18 +1619,22 @@ static int remote_partition_enable(struct cpuset *cs, int new_prs, */ static void remote_partition_disable(struct cpuset *cs, struct tmpmasks *tmp) { + bool isolcpus_updated; + compute_effective_exclusive_cpumask(cs, tmp->new_cpus); WARN_ON_ONCE(!is_remote_partition(cs)); WARN_ON_ONCE(!cpumask_subset(tmp->new_cpus, subpartitions_cpus));
spin_lock_irq(&callback_lock); list_del_init(&cs->remote_sibling); - partition_xcpus_del(cs->partition_root_state, NULL, tmp->new_cpus); + isolcpus_updated = partition_xcpus_del(cs->partition_root_state, + NULL, tmp->new_cpus); cs->partition_root_state = -cs->partition_root_state; if (!cs->prs_err) cs->prs_err = PERR_INVCPUS; reset_partition_data(cs); spin_unlock_irq(&callback_lock); + update_unbound_workqueue_cpumask(isolcpus_updated);
/* * Proprogate changes in top_cpuset's effective_cpus down the hierarchy. @@ -1625,6 +1657,7 @@ static void remote_cpus_update(struct cpuset *cs, struct cpumask *newmask, { bool adding, deleting; int prs = cs->partition_root_state; + int isolcpus_updated = 0;
if (WARN_ON_ONCE(!is_remote_partition(cs))) return; @@ -1649,10 +1682,11 @@ static void remote_cpus_update(struct cpuset *cs, struct cpumask *newmask,
spin_lock_irq(&callback_lock); if (adding) - partition_xcpus_add(prs, NULL, tmp->addmask); + isolcpus_updated += partition_xcpus_add(prs, NULL, tmp->addmask); if (deleting) - partition_xcpus_del(prs, NULL, tmp->delmask); + isolcpus_updated += partition_xcpus_del(prs, NULL, tmp->delmask); spin_unlock_irq(&callback_lock); + update_unbound_workqueue_cpumask(isolcpus_updated);
/* * Proprogate changes in top_cpuset's effective_cpus down the hierarchy. @@ -1774,6 +1808,7 @@ static int update_parent_effective_cpumask(struct cpuset *cs, int cmd, int part_error = PERR_NONE; /* Partition error? */ int subparts_delta = 0; struct cpumask *xcpus; /* cs effective_xcpus */ + int isolcpus_updated = 0; bool nocpu;
lockdep_assert_held(&cpuset_mutex); @@ -2010,15 +2045,18 @@ static int update_parent_effective_cpumask(struct cpuset *cs, int cmd, * and vice versa. */ if (adding) - partition_xcpus_del(old_prs, parent, tmp->addmask); + isolcpus_updated += partition_xcpus_del(old_prs, parent, + tmp->addmask); if (deleting) - partition_xcpus_add(new_prs, parent, tmp->delmask); + isolcpus_updated += partition_xcpus_add(new_prs, parent, + tmp->delmask);
if (is_partition_valid(parent)) { parent->nr_subparts += subparts_delta; WARN_ON_ONCE(parent->nr_subparts < 0); } spin_unlock_irq(&callback_lock); + update_unbound_workqueue_cpumask(isolcpus_updated);
if ((old_prs != new_prs) && (cmd == partcmd_update)) update_partition_exclusive(cs, new_prs); @@ -3082,6 +3120,7 @@ static int update_prstate(struct cpuset *cs, int new_prs) else if (new_xcpus_state) partition_xcpus_newstate(old_prs, new_prs, cs->effective_xcpus); spin_unlock_irq(&callback_lock); + update_unbound_workqueue_cpumask(new_xcpus_state);
/* Force update if switching back to member */ update_cpumasks_hier(cs, &tmpmask, !new_prs ? HIER_CHECKALL : 0); @@ -4370,6 +4409,30 @@ void cpuset_force_rebuild(void) force_rebuild = true; }
+/* + * Attempt to acquire a cpus_read_lock while a hotplug operation may be in + * progress. + * Return: true if successful, false otherwise + * + * To avoid circular lock dependency between cpuset_mutex and cpus_read_lock, + * cpus_read_trylock() is used here to acquire the lock. + */ +static bool cpuset_hotplug_cpus_read_trylock(void) +{ + int retries = 0; + + while (!cpus_read_trylock()) { + /* + * CPU hotplug still in progress. Retry 5 times + * with a 10ms wait before bailing out. + */ + if (++retries > 5) + return false; + msleep(10); + } + return true; +} + /** * cpuset_hotplug_update_tasks - update tasks in a cpuset for hotunplug * @cs: cpuset in interest @@ -4386,6 +4449,7 @@ static void cpuset_hotplug_update_tasks(struct cpuset *cs, struct tmpmasks *tmp) bool cpus_updated; bool mems_updated; bool remote; + int partcmd = -1; struct cpuset *parent; retry: wait_event(cpuset_attach_wq, cs->attach_in_progress == 0); @@ -4417,11 +4481,13 @@ static void cpuset_hotplug_update_tasks(struct cpuset *cs, struct tmpmasks *tmp) compute_partition_effective_cpumask(cs, &new_cpus);
if (remote && cpumask_empty(&new_cpus) && - partition_is_populated(cs, NULL)) { + partition_is_populated(cs, NULL) && + cpuset_hotplug_cpus_read_trylock()) { remote_partition_disable(cs, tmp); compute_effective_cpumask(&new_cpus, cs, parent); remote = false; cpuset_force_rebuild(); + cpus_read_unlock(); }
/* @@ -4432,18 +4498,28 @@ static void cpuset_hotplug_update_tasks(struct cpuset *cs, struct tmpmasks *tmp) * partitions. */ if (is_local_partition(cs) && (!is_partition_valid(parent) || - tasks_nocpu_error(parent, cs, &new_cpus))) { - update_parent_effective_cpumask(cs, partcmd_invalidate, NULL, tmp); - compute_effective_cpumask(&new_cpus, cs, parent); - cpuset_force_rebuild(); - } + tasks_nocpu_error(parent, cs, &new_cpus))) + partcmd = partcmd_invalidate; /* * On the other hand, an invalid partition root may be transitioned * back to a regular one. */ - else if (is_partition_valid(parent) && is_partition_invalid(cs)) { - update_parent_effective_cpumask(cs, partcmd_update, NULL, tmp); - if (is_partition_valid(cs)) { + else if (is_partition_valid(parent) && is_partition_invalid(cs)) + partcmd = partcmd_update; + + /* + * cpus_read_lock needs to be held before calling + * update_parent_effective_cpumask(). To avoid circular lock + * dependency between cpuset_mutex and cpus_read_lock, + * cpus_read_trylock() is used here to acquire the lock. + */ + if (partcmd >= 0) { + if (!cpuset_hotplug_cpus_read_trylock()) + goto update_tasks; + + update_parent_effective_cpumask(cs, partcmd, NULL, tmp); + cpus_read_unlock(); + if ((partcmd == partcmd_invalidate) || is_partition_valid(cs)) { compute_partition_effective_cpumask(cs, &new_cpus); cpuset_force_rebuild(); } diff --git a/tools/testing/selftests/cgroup/test_cpuset_prs.sh b/tools/testing/selftests/cgroup/test_cpuset_prs.sh index 2b825019f806..e31c2dcdade7 100755 --- a/tools/testing/selftests/cgroup/test_cpuset_prs.sh +++ b/tools/testing/selftests/cgroup/test_cpuset_prs.sh @@ -232,11 +232,11 @@ TEST_MATRIX=( " C0-3:S+ C1-3:S+ C2-3 C4-5 X2-3 X2-3:P1 P2 P1 0 A1:0-1,A2:,A3:2-3,B1:4-5 \ A1:P0,A2:P1,A3:P2,B1:P1 2-3" " C0-3:S+ C1-3:S+ C2-3 C4 X2-3 X2-3:P1 P2 P1 0 A1:0-1,A2:,A3:2-3,B1:4 \ - A1:P0,A2:P1,A3:P2,B1:P1 2-4" + A1:P0,A2:P1,A3:P2,B1:P1 2-4,2-3" " C0-3:S+ C1-3:S+ C3 C4 X2-3 X2-3:P1 P2 P1 0 A1:0-1,A2:2,A3:3,B1:4 \ - A1:P0,A2:P1,A3:P2,B1:P1 2-4" + A1:P0,A2:P1,A3:P2,B1:P1 2-4,3" " C0-4:S+ C1-4:S+ C2-4 . X2-4 X2-4:P2 X4:P1 . 0 A1:0-1,A2:2-3,A3:4 \ - A1:P0,A2:P2,A3:P1 2-4" + A1:P0,A2:P2,A3:P1 2-4,2-3" " C0-4:X2-4:S+ C1-4:X2-4:S+:P2 C2-4:X4:P1 \ . . X5 . . 0 A1:0-4,A2:1-4,A3:2-4 \ A1:P0,A2:P-2,A3:P-1" @@ -248,7 +248,7 @@ TEST_MATRIX=( " C0-3:S+ C1-3:S+ C2-3 . X2-3 X2-3 X2-3:P2:O2=0 . 0 A1:0-1,A2:1,A3:3 A1:P0,A3:P2 2-3" " C0-3:S+ C1-3:S+ C2-3 . X2-3 X2-3 X2-3:P2:O2=0 O2=1 0 A1:0-1,A2:1,A3:2-3 A1:P0,A3:P2 2-3" " C0-3:S+ C1-3:S+ C3 . X2-3 X2-3 P2:O3=0 . 0 A1:0-2,A2:1-2,A3: A1:P0,A3:P2 3" - " C0-3:S+ C1-3:S+ C3 . X2-3 X2-3 T:P2:O3=0 . 0 A1:0-2,A2:1-2,A3:1-2 A1:P0,A3:P-2 3" + " C0-3:S+ C1-3:S+ C3 . X2-3 X2-3 T:P2:O3=0 . 0 A1:0-2,A2:1-2,A3:1-2 A1:P0,A3:P-2 3,"
# An invalidated remote partition cannot self-recover from hotplug " C0-3:S+ C1-3:S+ C2 . X2-3 X2-3 T:P2:O2=0 O2=1 0 A1:0-3,A2:1-3,A3:2 A1:P0,A3:P-2" @@ -376,7 +376,7 @@ write_cpu_online() } fi echo $VAL > $CPUFILE - pause 0.01 + pause 0.05 }
# @@ -508,12 +508,14 @@ dump_states() XECPUS=$DIR/cpuset.cpus.exclusive.effective PRS=$DIR/cpuset.cpus.partition PCPUS=$DIR/.__DEBUG__.cpuset.cpus.subpartitions + ISCPUS=$DIR/cpuset.cpus.isolated [[ -e $CPUS ]] && echo "$CPUS: $(cat $CPUS)" [[ -e $XCPUS ]] && echo "$XCPUS: $(cat $XCPUS)" [[ -e $ECPUS ]] && echo "$ECPUS: $(cat $ECPUS)" [[ -e $XECPUS ]] && echo "$XECPUS: $(cat $XECPUS)" [[ -e $PRS ]] && echo "$PRS: $(cat $PRS)" [[ -e $PCPUS ]] && echo "$PCPUS: $(cat $PCPUS)" + [[ -e $ISCPUS ]] && echo "$ISCPUS: $(cat $ISCPUS)" done }
@@ -591,11 +593,17 @@ check_cgroup_states()
# # Get isolated (including offline) CPUs by looking at -# /sys/kernel/debug/sched/domains and compare that with the expected value. +# /sys/kernel/debug/sched/domains and *cpuset.cpus.isolated control file, +# if available, and compare that with the expected value. # -# Note that a sched domain of just 1 CPU will be considered isolated. +# Note that isolated CPUs from the sched/domains context include offline +# CPUs as well as CPUs in non-isolated 1-CPU partition. Those CPUs may +# not be included in the *cpuset.cpus.isolated control file which contains +# only CPUs in isolated partitions. # -# $1 - expected isolated cpu list +# $1 - expected isolated cpu list(s) <isolcpus1>{,<isolcpus2>} +# <isolcpus1> - expected sched/domains value +# <isolcpus2> - *cpuset.cpus.isolated value = <isolcpus1> if not defined # check_isolcpus() { @@ -603,8 +611,38 @@ check_isolcpus() ISOLCPUS= LASTISOLCPU= SCHED_DOMAINS=/sys/kernel/debug/sched/domains + ISCPUS=${CGROUP2}/cpuset.cpus.isolated + if [[ $EXPECT_VAL = . ]] + then + EXPECT_VAL= + EXPECT_VAL2= + elif [[ $(expr $EXPECT_VAL : ".*,.*") > 0 ]] + then + set -- $(echo $EXPECT_VAL | sed -e "s/,/ /g") + EXPECT_VAL=$1 + EXPECT_VAL2=$2 + else + EXPECT_VAL2=$EXPECT_VAL + fi + + # + # Check the debug isolated cpumask, if present + # + [[ -f $ISCPUS ]] && { + ISOLCPUS=$(cat $ISCPUS) + [[ "$EXPECT_VAL2" != "$ISOLCPUS" ]] && { + # Take a 50ms pause and try again + pause 0.05 + ISOLCPUS=$(cat $ISCPUS) + } + [[ "$EXPECT_VAL2" != "$ISOLCPUS" ]] && return 1 + ISOLCPUS= + } + + # + # Use the sched domain in debugfs to check isolated CPUs, if available + # [[ -d $SCHED_DOMAINS ]] || return 0 - [[ $EXPECT_VAL = . ]] && EXPECT_VAL=
for ((CPU=0; CPU < $NR_CPUS; CPU++)) do @@ -648,6 +686,22 @@ test_fail() exit 1 }
+# +# Check to see if there are unexpected isolated CPUs left +# +null_isolcpus_check() +{ + [[ $VERBOSE -gt 0 ]] || return 0 + pause 0.02 + check_isolcpus "." + if [[ $? -ne 0 ]] + then + echo "Unexpected isolated CPUs: $ISOLCPUS" + dump_states + exit 1 + fi +} + # # Run cpuset state transition test # $1 - test matrix name @@ -733,6 +787,7 @@ run_state_test() echo "Effective cpus changed to $NEWLIST after test $I!" exit 1 } + null_isolcpus_check [[ $VERBOSE -gt 0 ]] && echo "Test $I done." ((I++)) done @@ -802,6 +857,7 @@ test_isolated() console_msg "Cleaning up" echo $$ > $CGROUP2/cgroup.procs [[ -d A1 ]] && rmdir A1 + null_isolcpus_check }
#
On Wed, Nov 15, 2023 at 10:34:00PM -0500, Waiman Long wrote:
v4:
- Update patch 1 to move apply_wqattrs_lock() and apply_wqattrs_unlock() down into CONFIG_SYSFS block to avoid compilation warnings.
I already applied v3 to cgroup/for-6.8. Can you please send the fix up patch against that branch?
Thanks.
On 11/19/23 10:23, Tejun Heo wrote:
On Wed, Nov 15, 2023 at 10:34:00PM -0500, Waiman Long wrote:
v4:
- Update patch 1 to move apply_wqattrs_lock() and apply_wqattrs_unlock() down into CONFIG_SYSFS block to avoid compilation warnings.
I already applied v3 to cgroup/for-6.8. Can you please send the fix up patch against that branch?
Sure. I will post another fixup patch.
Thanks, Longman
linux-kselftest-mirror@lists.linaro.org
-
Tejun Heo -
Waiman Long