[PATCH 0/2] Fix vm.dirtytime_expire_seconds=0 causing 100% CPU

[PATCH 0/2] Fix vm.dirtytime_expire_seconds=0 causing 100% CPU

[Laveesh Bansal]

Setting vm.dirtytime_expire_seconds to 0 causes wakeup_dirtytime_writeback()
to reschedule itself with a delay of 0, creating an infinite busy loop that
spins kworker at 100% CPU.

This series:
- Patch 1: Fixes the bug by handling interval=0 as "disable writeback"
           (consistent with dirty_writeback_centisecs behavior)
- Patch 2: Documents that setting the value to 0 disables writeback

Tested by booting kernels in QEMU with virtme-ng:
- Buggy kernel: kworker CPU spikes to ~73% when interval set to 0
- Fixed kernel: CPU remains normal, writeback correctly disabled
- Re-enabling (0 -> non-zero): writeback resumes correctly

Laveesh Bansal (2):
  writeback: fix 100% CPU usage when dirtytime_expire_interval is 0
  docs: clarify that dirtytime_expire_seconds=0 disables writeback

 Documentation/admin-guide/sysctl/vm.rst |  2 ++
 fs/fs-writeback.c                       | 14 ++++++++++----
 2 files changed, 12 insertions(+), 4 deletions(-)

--
2.43.0

[PATCH 1/2] writeback: fix 100% CPU usage when dirtytime_expire_interval is 0

[Laveesh Bansal]

When vm.dirtytime_expire_seconds is set to 0, wakeup_dirtytime_writeback()
schedules delayed work with a delay of 0, causing immediate execution.
The function then reschedules itself with 0 delay again, creating an
infinite busy loop that causes 100% kworker CPU usage.

Fix by:
- Only scheduling delayed work in wakeup_dirtytime_writeback() when
  dirtytime_expire_interval is non-zero
- Cancelling the delayed work in dirtytime_interval_handler() when
  the interval is set to 0
- Adding a guard in start_dirtytime_writeback() for defensive coding

Tested by booting kernel in QEMU with virtme-ng:
- Before fix: kworker CPU spikes to ~73%
- After fix: CPU remains at normal levels
- Setting interval back to non-zero correctly resumes writeback

Fixes: a2f4870697a5 ("fs: make sure the timestamps for lazytime inodes eventually get written")
Cc: stable@vger.kernel.org
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220227
Signed-off-by: Laveesh Bansal <laveeshb@laveeshbansal.com>
---
 fs/fs-writeback.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 6800886c4d10..cd21c74cd0e5 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -2492,7 +2492,8 @@ static void wakeup_dirtytime_writeback(struct work_struct *w)
 				wb_wakeup(wb);
 	}
 	rcu_read_unlock();
-	schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
+	if (dirtytime_expire_interval)
+		schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
 }
 
 static int dirtytime_interval_handler(const struct ctl_table *table, int write,
@@ -2501,8 +2502,12 @@ static int dirtytime_interval_handler(const struct ctl_table *table, int write,
 	int ret;
 
 	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
-	if (ret == 0 && write)
-		mod_delayed_work(system_percpu_wq, &dirtytime_work, 0);
+	if (ret == 0 && write) {
+		if (dirtytime_expire_interval)
+			mod_delayed_work(system_percpu_wq, &dirtytime_work, 0);
+		else
+			cancel_delayed_work_sync(&dirtytime_work);
+	}
 	return ret;
 }
 
@@ -2519,7 +2524,8 @@ static const struct ctl_table vm_fs_writeback_table[] = {
 
 static int __init start_dirtytime_writeback(void)
 {
-	schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
+	if (dirtytime_expire_interval)
+		schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
 	register_sysctl_init("vm", vm_fs_writeback_table);
 	return 0;
 }
-- 
2.43.0

[PATCH 2/2] docs: clarify that dirtytime_expire_seconds=0 disables writeback

[Laveesh Bansal]

Document that setting vm.dirtytime_expire_seconds to zero disables
periodic dirtytime writeback, matching the behavior of the related
dirty_writeback_centisecs sysctl which already documents this.

Signed-off-by: Laveesh Bansal <laveeshb@laveeshbansal.com>
---
 Documentation/admin-guide/sysctl/vm.rst | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/Documentation/admin-guide/sysctl/vm.rst b/Documentation/admin-guide/sysctl/vm.rst
index 4d71211fdad8..e2fdbc521033 100644
--- a/Documentation/admin-guide/sysctl/vm.rst
+++ b/Documentation/admin-guide/sysctl/vm.rst
@@ -231,6 +231,8 @@ eventually gets pushed out to disk.  This tunable is used to define when dirty
 inode is old enough to be eligible for writeback by the kernel flusher threads.
 And, it is also used as the interval to wakeup dirtytime_writeback thread.
 
+Setting this to zero disables periodic dirtytime writeback.
+
 
 dirty_writeback_centisecs
 =========================
-- 
2.43.0

Re: [PATCH 1/2] writeback: fix 100% CPU usage when dirtytime_expire_interval is 0

[Jan Kara]

On Fri 02-01-26 20:16:56, Laveesh Bansal wrote:
> When vm.dirtytime_expire_seconds is set to 0, wakeup_dirtytime_writeback()
> schedules delayed work with a delay of 0, causing immediate execution.
> The function then reschedules itself with 0 delay again, creating an
> infinite busy loop that causes 100% kworker CPU usage.
> 
> Fix by:
> - Only scheduling delayed work in wakeup_dirtytime_writeback() when
>   dirtytime_expire_interval is non-zero
> - Cancelling the delayed work in dirtytime_interval_handler() when
>   the interval is set to 0
> - Adding a guard in start_dirtytime_writeback() for defensive coding
> 
> Tested by booting kernel in QEMU with virtme-ng:
> - Before fix: kworker CPU spikes to ~73%
> - After fix: CPU remains at normal levels
> - Setting interval back to non-zero correctly resumes writeback
> 
> Fixes: a2f4870697a5 ("fs: make sure the timestamps for lazytime inodes eventually get written")
> Cc: stable@vger.kernel.org
> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220227
> Signed-off-by: Laveesh Bansal <laveeshb@laveeshbansal.com>

Looks good. Feel free to add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> ---
>  fs/fs-writeback.c | 14 ++++++++++----
>  1 file changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
> index 6800886c4d10..cd21c74cd0e5 100644
> --- a/fs/fs-writeback.c
> +++ b/fs/fs-writeback.c
> @@ -2492,7 +2492,8 @@ static void wakeup_dirtytime_writeback(struct work_struct *w)
>  				wb_wakeup(wb);
>  	}
>  	rcu_read_unlock();
> -	schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
> +	if (dirtytime_expire_interval)
> +		schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
>  }
>  
>  static int dirtytime_interval_handler(const struct ctl_table *table, int write,
> @@ -2501,8 +2502,12 @@ static int dirtytime_interval_handler(const struct ctl_table *table, int write,
>  	int ret;
>  
>  	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
> -	if (ret == 0 && write)
> -		mod_delayed_work(system_percpu_wq, &dirtytime_work, 0);
> +	if (ret == 0 && write) {
> +		if (dirtytime_expire_interval)
> +			mod_delayed_work(system_percpu_wq, &dirtytime_work, 0);
> +		else
> +			cancel_delayed_work_sync(&dirtytime_work);
> +	}
>  	return ret;
>  }
>  
> @@ -2519,7 +2524,8 @@ static const struct ctl_table vm_fs_writeback_table[] = {
>  
>  static int __init start_dirtytime_writeback(void)
>  {
> -	schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
> +	if (dirtytime_expire_interval)
> +		schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
>  	register_sysctl_init("vm", vm_fs_writeback_table);
>  	return 0;
>  }
> -- 
> 2.43.0
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH 2/2] docs: clarify that dirtytime_expire_seconds=0 disables writeback

[Jan Kara]

On Fri 02-01-26 20:16:57, Laveesh Bansal wrote:
> Document that setting vm.dirtytime_expire_seconds to zero disables
> periodic dirtytime writeback, matching the behavior of the related
> dirty_writeback_centisecs sysctl which already documents this.
> 
> Signed-off-by: Laveesh Bansal <laveeshb@laveeshbansal.com>

Looks good. Feel free to add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> ---
>  Documentation/admin-guide/sysctl/vm.rst | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/Documentation/admin-guide/sysctl/vm.rst b/Documentation/admin-guide/sysctl/vm.rst
> index 4d71211fdad8..e2fdbc521033 100644
> --- a/Documentation/admin-guide/sysctl/vm.rst
> +++ b/Documentation/admin-guide/sysctl/vm.rst
> @@ -231,6 +231,8 @@ eventually gets pushed out to disk.  This tunable is used to define when dirty
>  inode is old enough to be eligible for writeback by the kernel flusher threads.
>  And, it is also used as the interval to wakeup dirtytime_writeback thread.
>  
> +Setting this to zero disables periodic dirtytime writeback.
> +
>  
>  dirty_writeback_centisecs
>  =========================
> -- 
> 2.43.0
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR