kthread: dynamically allocate memory to store kthread's full name
Subsystems (6)
PATCH v2 1/1 kthread: dynamically allocate memory to store kthread's full name
When I was implementing a new per-cpu kthread cfs_migration, I found the comm of it "cfs_migration/%u" is truncated due to the limitation of TASK_COMM_LEN. For example, the comm of the percpu thread on CPU10~19 are all with the same name "cfs_migration/1", which will confuse the user. This issue is not critical, because we can get the corresponding CPU from the task's Cpus_allowed. But for kthreads correspoinding to other hardware devices, it is not easy to get the detailed device info from task comm, for example,
jbd2/nvme0n1p2-
xfs-reclaim/sdfCurrently there are so many truncated kthreads:
rcu_tasks_kthre
rcu_tasks_rude_
rcu_tasks_trace
poll_mpt3sas0_s
ext4-rsv-conver
xfs-reclaim/sd{a, b, c, ...}
xfs-blockgc/sd{a, b, c, ...}
xfs-inodegc/sd{a, b, c, ...}
audit_send_repl
ecryptfs-kthrea
vfio-irqfd-clea
jbd2/nvme0n1p2-
...We can shorten these names to work around this problem, but it may be not applied to all of the truncated kthreads. Take 'jbd2/nvme0n1p2-' for example, it is a nice name, and it is not a good idea to shorten it.
One possible way to fix this issue is extending the task comm size, but as task->comm is used in lots of places, that may cause some potential buffer overflows. Another more conservative approach is introducing a new pointer to store kthread's full name if it is truncated, which won't introduce too much overhead as it is in the non-critical path. Finally we make a dicision to use the second approach. See also the discussions in this thread:
https://lore.kernel.org/lkml/[email protected]/
After this change, the full name of these truncated kthreads will be displayed via /proc/[pid]/comm:
rcu_tasks_kthread
rcu_tasks_rude_kthread
rcu_tasks_trace_kthread
poll_mpt3sas0_statu
ext4-rsv-conversion
xfs-reclaim/sdf1
xfs-blockgc/sdf1
xfs-inodegc/sdf1
audit_send_reply
ecryptfs-kthread
vfio-irqfd-cleanup
jbd2/nvme0n1p2-8---
Changes since v1:
1. leave it turncated when out of memory (Kees & Petr)
2. do null check in free_kthread_struct (Petr)
--- fs/proc/array.c | 3 +++ include/linux/kthread.h | 1 + kernel/kthread.c | 32 ++++++++++++++++++++++++++++++-- 3 files changed, 34 insertions(+), 2 deletions(-)
fs/proc/array.c +3 -0
@@ -92,6 +92,7 @@#include <linux/string_helpers.h> #include <linux/user_namespace.h> #include <linux/fs_struct.h> +#include <linux/kthread.h> #include <asm/processor.h> #include "internal.h"
@@ -102,6 +103,8 @@ void proc_task_name(struct seq_file *m, struct task_struct *p, bool escape)if (p->flags &PF_WQ_WORKER )wq_worker_comm (tcomm,sizeof (tcomm), p); +else if (p->flags &PF_KTHREAD ) +get_kthread_comm (tcomm,sizeof (tcomm), p);else __get_task_comm (tcomm,sizeof (tcomm), p);
include/linux/kthread.h +1 -0
@@ -33,6 +33,7 @@ struct task_struct *kthread_create_on_cpu(int (*threadfn)(void *data),unsigned int cpu, const char *namefmt); +void get_kthread_comm (char *buf,size_t buf_size,struct task_struct *tsk);void set_kthread_struct (struct task_struct *p);void kthread_set_per_cpu (struct task_struct *k,int cpu);
kernel/kthread.c +30 -2
@@ -60,6 +60,8 @@ struct kthread {#ifdef CONFIG_BLK_CGROUP struct cgroup_subsys_state *blkcg_css;#endif +/* To store the full name if task comm is truncated. */ +char *full_name; };enum KTHREAD_BITS {
@@ -93,6 +95,18 @@ static inline struct kthread *__to_kthread(struct task_struct *p)return kthread; } +void get_kthread_comm (char *buf,size_t buf_size,struct task_struct *tsk) +{ +struct kthread *kthread =to_kthread (tsk); + +if (!kthread || !kthread->full_name) { +__get_task_comm (buf, buf_size, tsk); +return ; + } + +strscpy_pad (buf, kthread->full_name, buf_size); +} +void set_kthread_struct (struct task_struct *p) {struct kthread *kthread;
@@ -118,9 +132,13 @@ void free_kthread_struct(struct task_struct *k)* or if kmalloc() in kthread() failed. */ kthread =to_kthread (k); +if (!kthread) +return ; +#ifdef CONFIG_BLK_CGROUP -WARN_ON_ONCE (kthread && kthread->blkcg_css); +WARN_ON_ONCE (kthread->blkcg_css);#endif +kfree (kthread->full_name);kfree (kthread); }
@@ -406,12 +424,22 @@ struct task_struct *__kthread_create_on_node(int (*threadfn)(void *data), task = create->result;if (!IS_ERR (task)) {char name[TASK_COMM_LEN ]; +va_list aq; +int len;/* * task is already visible to other tasks, so updating * COMM must be protected. */ -vsnprintf (name,sizeof (name), namefmt, args); +va_copy (aq, args); + len =vsnprintf (name,sizeof (name), namefmt, aq); +va_end (aq); +if (len >=TASK_COMM_LEN ) { +struct kthread *kthread =to_kthread (task); + +/* leave it truncated when out of memory. */ + kthread->full_name =kvasprintf (GFP_KERNEL , namefmt, args); + }set_task_comm (task, name); }kfree (create);
Signature · 1 line
2.17.1
Quoted 13 lines · Yafang Shao
When I was implementing a new per-cpu kthread
cfs_migration, I found the comm of it "cfs_migration/%u" is truncated due to the limitation ofTASK_COMM_LEN.One possible way to fix this issue is extending the task comm size, but as
task->commis used in lots of places, that may cause some potential buffer overflows. Another more conservative approach is introducing a new pointer to store kthread's full name if it is truncated, which won't introduce too much overhead as it is in the non-critical path. Finally we make a dicision to use the second approach. See also the discussions in this thread:
https://lore.kernel.org/lkml/[email protected]/
Quoted 4 lines
After this change, the full name of these truncated kthreads will be displayed via /proc/[pid]/comm:
Signed-off-by Yafang Shao <[email protected]>
Looks good to me:
Best Regards,
Petr
Quoted 44 lines · Yafang Shao
When I was implementing a new per-cpu kthread
cfs_migration, I found the comm of it "cfs_migration/%u" is truncated due to the limitation ofTASK_COMM_LEN. For example, the comm of the percpu thread on CPU10~19 are all with the same name "cfs_migration/1", which will confuse the user. This issue is not critical, because we can get the corresponding CPU from the task'sCpus_allowed. But for kthreads correspoinding to other hardware devices, it is not easy to get the detailed device info from task comm, for example,jbd2/nvme0n1p2- xfs-reclaim/sdfCurrently there are so many truncated kthreads:
rcu_tasks_kthre rcu_tasks_rude_ rcu_tasks_trace poll_mpt3sas0_s ext4-rsv-conver xfs-reclaim/sd{a, b, c, ...} xfs-blockgc/sd{a, b, c, ...} xfs-inodegc/sd{a, b, c, ...} audit_send_repl ecryptfs-kthrea vfio-irqfd-clea jbd2/nvme0n1p2- ...We can shorten these names to work around this problem, but it may be not applied to all of the truncated kthreads. Take 'jbd2/nvme0n1p2-' for example, it is a nice name, and it is not a good idea to shorten it.
One possible way to fix this issue is extending the task comm size, but as
task->commis used in lots of places, that may cause some potential buffer overflows. Another more conservative approach is introducing a new pointer to store kthread's full name if it is truncated, which won't introduce too much overhead as it is in the non-critical path. Finally we make a dicision to use the second approach. See also the discussions in this thread:
https://lore.kernel.org/lkml/[email protected]/After this change, the full name of these truncated kthreads will be displayed via /proc/[pid]/comm:
Quoted 12 lines
rcu_tasks_kthread
rcu_tasks_rude_kthread
rcu_tasks_trace_kthread
poll_mpt3sas0_statu
ext4-rsv-conversion
xfs-reclaim/sdf1
xfs-blockgc/sdf1
xfs-inodegc/sdf1
audit_send_reply
ecryptfs-kthread
vfio-irqfd-cleanup
jbd2/nvme0n1p2-8I do wonder if that could break some user space that assumes these names have maximum length ..
But LGTM
Signature · 3 lines
Thanks, David / dhildenb
Quoted 58 lines · David Hildenbrand
Quoted 44 lines · Yafang Shao
When I was implementing a new per-cpu kthread
cfs_migration, I found the comm of it "cfs_migration/%u" is truncated due to the limitation ofTASK_COMM_LEN. For example, the comm of the percpu thread on CPU10~19 are all with the same name "cfs_migration/1", which will confuse the user. This issue is not critical, because we can get the corresponding CPU from the task'sCpus_allowed. But for kthreads correspoinding to other hardware devices, it is not easy to get the detailed device info from task comm, for example,jbd2/nvme0n1p2- xfs-reclaim/sdfCurrently there are so many truncated kthreads:
rcu_tasks_kthre rcu_tasks_rude_ rcu_tasks_trace poll_mpt3sas0_s ext4-rsv-conver xfs-reclaim/sd{a, b, c, ...} xfs-blockgc/sd{a, b, c, ...} xfs-inodegc/sd{a, b, c, ...} audit_send_repl ecryptfs-kthrea vfio-irqfd-clea jbd2/nvme0n1p2- ...We can shorten these names to work around this problem, but it may be not applied to all of the truncated kthreads. Take 'jbd2/nvme0n1p2-' for example, it is a nice name, and it is not a good idea to shorten it.
One possible way to fix this issue is extending the task comm size, but as
task->commis used in lots of places, that may cause some potential buffer overflows. Another more conservative approach is introducing a new pointer to store kthread's full name if it is truncated, which won't introduce too much overhead as it is in the non-critical path. Finally we make a dicision to use the second approach. See also the discussions in this thread:
https://lore.kernel.org/lkml/[email protected]/After this change, the full name of these truncated kthreads will be displayed via /proc/[pid]/comm:
Quoted 12 lines
rcu_tasks_kthread rcu_tasks_rude_kthread rcu_tasks_trace_kthread poll_mpt3sas0_statu ext4-rsv-conversion xfs-reclaim/sdf1 xfs-blockgc/sdf1 xfs-inodegc/sdf1 audit_send_reply ecryptfs-kthread vfio-irqfd-cleanup jbd2/nvme0n1p2-8
Quoted 2 lines
I do wonder if that could break some user space that assumes these names have maximum length ..
There is high chance that we will be on the safe side. Workqueue kthreads already provided longer names. They are even dynamic because the currently handled workqueue name is part of the name, see wq_worker_comm().
Best Regards,
Petr
Quoted 59 lines · Petr Mladek
Quoted 58 lines · David Hildenbrand
Quoted 56 lines · Yafang Shao
When I was implementing a new per-cpu kthread
cfs_migration, I found the comm of it "cfs_migration/%u" is truncated due to the limitation ofTASK_COMM_LEN. For example, the comm of the percpu thread on CPU10~19 are all with the same name "cfs_migration/1", which will confuse the user. This issue is not critical, because we can get the corresponding CPU from the task'sCpus_allowed. But for kthreads correspoinding to other hardware devices, it is not easy to get the detailed device info from task comm, for example,jbd2/nvme0n1p2- xfs-reclaim/sdfCurrently there are so many truncated kthreads:
rcu_tasks_kthre rcu_tasks_rude_ rcu_tasks_trace poll_mpt3sas0_s ext4-rsv-conver xfs-reclaim/sd{a, b, c, ...} xfs-blockgc/sd{a, b, c, ...} xfs-inodegc/sd{a, b, c, ...} audit_send_repl ecryptfs-kthrea vfio-irqfd-clea jbd2/nvme0n1p2- ...We can shorten these names to work around this problem, but it may be not applied to all of the truncated kthreads. Take 'jbd2/nvme0n1p2-' for example, it is a nice name, and it is not a good idea to shorten it.
One possible way to fix this issue is extending the task comm size, but as
task->commis used in lots of places, that may cause some potential buffer overflows. Another more conservative approach is introducing a new pointer to store kthread's full name if it is truncated, which won't introduce too much overhead as it is in the non-critical path. Finally we make a dicision to use the second approach. See also the discussions in this thread:
https://lore.kernel.org/lkml/[email protected]/After this change, the full name of these truncated kthreads will be displayed via /proc/[pid]/comm:
rcu_tasks_kthread rcu_tasks_rude_kthread rcu_tasks_trace_kthread poll_mpt3sas0_statu ext4-rsv-conversion xfs-reclaim/sdf1 xfs-blockgc/sdf1 xfs-inodegc/sdf1 audit_send_reply ecryptfs-kthread vfio-irqfd-cleanup jbd2/nvme0n1p2-8
Quoted 7 lines
Quoted 2 lines
I do wonder if that could break some user space that assumes these names have maximum length ..
There is high chance that we will be on the safe side. Workqueue kthreads already provided longer names. They are even dynamic because the currently handled workqueue name is part of the name, see
wq_worker_comm().
Great, thanks!
Signature · 3 lines
Thanks, David / dhildenb