diff mbox series

[v5] ring-buffer: Ensure proper resetting of atomic variables in ring_buffer_reset_online_cpus

Message ID 20230426010446.10753-1-Tze-nan.Wu@mediatek.com (mailing list archive)
State New, archived
Headers show
Series [v5] ring-buffer: Ensure proper resetting of atomic variables in ring_buffer_reset_online_cpus | expand

Commit Message

Tze-nan Wu April 26, 2023, 1:04 a.m. UTC 
From: "Tze-nan Wu" <Tze-nan.Wu@mediatek.com>

In ring_buffer_reset_online_cpus, the buffer_size_kb write operation
may permanently fail if the cpu_online_mask changes between two
for_each_online_buffer_cpu loops. The number of increases and decreases
on both cpu_buffer->resize_disabled and cpu_buffer->record_disabled may be
inconsistent, causing some CPUs to have non-zero values for these atomic
variables after the function returns.

This issue can be reproduced by "echo 0 > trace" while hotplugging cpu.
After reproducing success, we can find out buffer_size_kb will not be
functional anymore.

To prevent leaving 'resize_disabled' and 'record_disabled' non-zero after
ring_buffer_reset_online_cpus returns, we ensure that each atomic variable
has been set up before atomic_sub() to it.

Cc: stable@vger.kernel.org
Cc: npiggin@gmail.com
Fixes: b23d7a5f4a07 ("ring-buffer: speed up buffer resets by avoiding synchronize_rcu for each CPU")
Reviewed-by: Cheng-Jui Wang <cheng-jui.wang@mediatek.com>
Signed-off-by: Tze-nan Wu <Tze-nan.Wu@mediatek.com>
---
Changes from v4 to v5: https://lore.kernel.org/lkml/20230412112401.25081-1-Tze-nan.Wu@mediatek.com/
  - Move the define before the function
---
 kernel/trace/ring_buffer.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

Comments

Steven Rostedt April 26, 2023, 1:17 a.m. UTC | #1 
For some reason, this email did not make it to
linux-trace-kernel@vger.kernel.org, and therefore did not make it into
patchwork?

John?

-- Steve


On Wed, 26 Apr 2023 09:04:44 +0800
Tze-nan.Wu <Tze-nan.Wu@mediatek.com> wrote:

> From: "Tze-nan Wu" <Tze-nan.Wu@mediatek.com>
> 
> In ring_buffer_reset_online_cpus, the buffer_size_kb write operation
> may permanently fail if the cpu_online_mask changes between two
> for_each_online_buffer_cpu loops. The number of increases and decreases
> on both cpu_buffer->resize_disabled and cpu_buffer->record_disabled may be
> inconsistent, causing some CPUs to have non-zero values for these atomic
> variables after the function returns.
> 
> This issue can be reproduced by "echo 0 > trace" while hotplugging cpu.
> After reproducing success, we can find out buffer_size_kb will not be
> functional anymore.
> 
> To prevent leaving 'resize_disabled' and 'record_disabled' non-zero after
> ring_buffer_reset_online_cpus returns, we ensure that each atomic variable
> has been set up before atomic_sub() to it.
> 
> Cc: stable@vger.kernel.org
> Cc: npiggin@gmail.com
> Fixes: b23d7a5f4a07 ("ring-buffer: speed up buffer resets by avoiding synchronize_rcu for each CPU")
> Reviewed-by: Cheng-Jui Wang <cheng-jui.wang@mediatek.com>
> Signed-off-by: Tze-nan Wu <Tze-nan.Wu@mediatek.com>
> ---
> Changes from v4 to v5: https://lore.kernel.org/lkml/20230412112401.25081-1-Tze-nan.Wu@mediatek.com/
>   - Move the define before the function
> ---
>  kernel/trace/ring_buffer.c | 16 +++++++++++++---
>  1 file changed, 13 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
> index 76a2d91eecad..253ef85a9ec3 100644
> --- a/kernel/trace/ring_buffer.c
> +++ b/kernel/trace/ring_buffer.c
> @@ -5345,6 +5345,9 @@ void ring_buffer_reset_cpu(struct trace_buffer *buffer, int cpu)
>  }
>  EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu);
>  
> +/* Flag to ensure proper resetting of atomic variables */
> +#define RESET_BIT	(1 << 30)
> +
>  /**
>   * ring_buffer_reset_online_cpus - reset a ring buffer per CPU buffer
>   * @buffer: The ring buffer to reset a per cpu buffer of
> @@ -5361,20 +5364,27 @@ void ring_buffer_reset_online_cpus(struct trace_buffer *buffer)
>  	for_each_online_buffer_cpu(buffer, cpu) {
>  		cpu_buffer = buffer->buffers[cpu];
>  
> -		atomic_inc(&cpu_buffer->resize_disabled);
> +		atomic_add(RESET_BIT, &cpu_buffer->resize_disabled);
>  		atomic_inc(&cpu_buffer->record_disabled);
>  	}
>  
>  	/* Make sure all commits have finished */
>  	synchronize_rcu();
>  
> -	for_each_online_buffer_cpu(buffer, cpu) {
> +	for_each_buffer_cpu(buffer, cpu) {
>  		cpu_buffer = buffer->buffers[cpu];
>  
> +		/*
> +		 * If a CPU came online during the synchronize_rcu(), then
> +		 * ignore it.
> +		 */
> +		if (!(atomic_read(&cpu_buffer->resize_disabled) & RESET_BIT))
> +			continue;
> +
>  		reset_disabled_cpu_buffer(cpu_buffer);
>  
>  		atomic_dec(&cpu_buffer->record_disabled);
> -		atomic_dec(&cpu_buffer->resize_disabled);
> +		atomic_sub(RESET_BIT, &cpu_buffer->resize_disabled);
>  	}
>  
>  	mutex_unlock(&buffer->mutex);
Steven Rostedt April 26, 2023, 1:21 a.m. UTC | #2 
On Tue, 25 Apr 2023 21:17:37 -0400
Steven Rostedt <rostedt@goodmis.org> wrote:

> For some reason, this email did not make it to
> linux-trace-kernel@vger.kernel.org, and therefore did not make it into
> patchwork?
> 

And the email was definitely Cc'd properly, as my reply made it to lore,
but not the original email.

 https://lore.kernel.org/linux-trace-kernel/20230425211737.757208b3@gandalf.local.home/

-- Steve
John 'Warthog9' Hawley April 26, 2023, 2:58 a.m. UTC | #3 
Ran afoul of the taboo filters:

Illegal-Object: Syntax error in From: address found on vger.kernel.org:
         From:   Tze-nan.Wu<Tze-nan.Wu@mediatek.com>
                                   ^-missing end of mailbox

I've seen a few of these but I've been unable to replicate the problem. 
I've a suspicion but so far been unable to prove the theory on what's wrong.

- John

On 4/25/23 18:17, Steven Rostedt wrote:
> 
> For some reason, this email did not make it to
> linux-trace-kernel@vger.kernel.org, and therefore did not make it into
> patchwork?
> 
> John?
> 
> -- Steve
> 
> 
> On Wed, 26 Apr 2023 09:04:44 +0800
> Tze-nan.Wu <Tze-nan.Wu@mediatek.com> wrote:
> 
>> From: "Tze-nan Wu" <Tze-nan.Wu@mediatek.com>
>>
>> In ring_buffer_reset_online_cpus, the buffer_size_kb write operation
>> may permanently fail if the cpu_online_mask changes between two
>> for_each_online_buffer_cpu loops. The number of increases and decreases
>> on both cpu_buffer->resize_disabled and cpu_buffer->record_disabled may be
>> inconsistent, causing some CPUs to have non-zero values for these atomic
>> variables after the function returns.
>>
>> This issue can be reproduced by "echo 0 > trace" while hotplugging cpu.
>> After reproducing success, we can find out buffer_size_kb will not be
>> functional anymore.
>>
>> To prevent leaving 'resize_disabled' and 'record_disabled' non-zero after
>> ring_buffer_reset_online_cpus returns, we ensure that each atomic variable
>> has been set up before atomic_sub() to it.
>>
>> Cc: stable@vger.kernel.org
>> Cc: npiggin@gmail.com
>> Fixes: b23d7a5f4a07 ("ring-buffer: speed up buffer resets by avoiding synchronize_rcu for each CPU")
>> Reviewed-by: Cheng-Jui Wang <cheng-jui.wang@mediatek.com>
>> Signed-off-by: Tze-nan Wu <Tze-nan.Wu@mediatek.com>
>> ---
>> Changes from v4 to v5: https://lore.kernel.org/lkml/20230412112401.25081-1-Tze-nan.Wu@mediatek.com/
>>    - Move the define before the function
>> ---
>>   kernel/trace/ring_buffer.c | 16 +++++++++++++---
>>   1 file changed, 13 insertions(+), 3 deletions(-)
>>
>> diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
>> index 76a2d91eecad..253ef85a9ec3 100644
>> --- a/kernel/trace/ring_buffer.c
>> +++ b/kernel/trace/ring_buffer.c
>> @@ -5345,6 +5345,9 @@ void ring_buffer_reset_cpu(struct trace_buffer *buffer, int cpu)
>>   }
>>   EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu);
>>   
>> +/* Flag to ensure proper resetting of atomic variables */
>> +#define RESET_BIT	(1 << 30)
>> +
>>   /**
>>    * ring_buffer_reset_online_cpus - reset a ring buffer per CPU buffer
>>    * @buffer: The ring buffer to reset a per cpu buffer of
>> @@ -5361,20 +5364,27 @@ void ring_buffer_reset_online_cpus(struct trace_buffer *buffer)
>>   	for_each_online_buffer_cpu(buffer, cpu) {
>>   		cpu_buffer = buffer->buffers[cpu];
>>   
>> -		atomic_inc(&cpu_buffer->resize_disabled);
>> +		atomic_add(RESET_BIT, &cpu_buffer->resize_disabled);
>>   		atomic_inc(&cpu_buffer->record_disabled);
>>   	}
>>   
>>   	/* Make sure all commits have finished */
>>   	synchronize_rcu();
>>   
>> -	for_each_online_buffer_cpu(buffer, cpu) {
>> +	for_each_buffer_cpu(buffer, cpu) {
>>   		cpu_buffer = buffer->buffers[cpu];
>>   
>> +		/*
>> +		 * If a CPU came online during the synchronize_rcu(), then
>> +		 * ignore it.
>> +		 */
>> +		if (!(atomic_read(&cpu_buffer->resize_disabled) & RESET_BIT))
>> +			continue;
>> +
>>   		reset_disabled_cpu_buffer(cpu_buffer);
>>   
>>   		atomic_dec(&cpu_buffer->record_disabled);
>> -		atomic_dec(&cpu_buffer->resize_disabled);
>> +		atomic_sub(RESET_BIT, &cpu_buffer->resize_disabled);
>>   	}
>>   
>>   	mutex_unlock(&buffer->mutex);
>
Tze-nan Wu April 26, 2023, 6:39 a.m. UTC | #4 
On Tue, 2023-04-25 at 19:58 -0700, John 'Warthog9' Hawley wrote:
> External email : Please do not click links or open attachments until
> you have verified the sender or the content.
> 
> 
> Ran afoul of the taboo filters:
> 
> Illegal-Object: Syntax error in From: address found on
> vger.kernel.org:
>          From:   Tze-nan.Wu<Tze-nan.Wu@mediatek.com>
>                                    ^-missing end of mailbox
> 
> I've seen a few of these but I've been unable to replicate the
> problem.
> I've a suspicion but so far been unable to prove the theory on what's
> wrong.
> 
> - John
> 
  So I sent the patch again, but this time I changed the name field in
.gitconfig from "Tze-nan.Wu"(internal-use) to "Tze-nan Wu".
This can avoid the line "From:  Tze-nan.Wu<Tze-nan.Wu@mediatek.com>"
being added to the mail when we execute "git send-email", and it works,
the last patch I sent is not filter out anymore.

---
Here is the magore difference between the patches:
old patch(fail):
  From: "Tze-nan Wu" <Tze-nan.Wu@mediatek.com>   <== (modify by editor
from Tze-nan.Wu to Tze-nan Wu)
---
new patch(success):
  From: Tze-nan Wu <Tze-nan.Wu@mediatek.com>
---

 - Tzenan
> On 4/25/23 18:17, Steven Rostedt wrote:
> > 
> > For some reason, this email did not make it to
> > linux-trace-kernel@vger.kernel.org, and therefore did not make it
> > into
> > patchwork?
> > 
> > John?
> > 
> > -- Steve
> > 
> > 
> > On Wed, 26 Apr 2023 09:04:44 +0800
> > Tze-nan.Wu <Tze-nan.Wu@mediatek.com> wrote:
> > 
> > > From: "Tze-nan Wu" <Tze-nan.Wu@mediatek.com>
> > > 
> > > In ring_buffer_reset_online_cpus, the buffer_size_kb write
> > > operation
> > > may permanently fail if the cpu_online_mask changes between two
> > > for_each_online_buffer_cpu loops. The number of increases and
> > > decreases
> > > on both cpu_buffer->resize_disabled and cpu_buffer-
> > > >record_disabled may be
> > > inconsistent, causing some CPUs to have non-zero values for these
> > > atomic
> > > variables after the function returns.
> > > 
> > > This issue can be reproduced by "echo 0 > trace" while
> > > hotplugging cpu.
> > > After reproducing success, we can find out buffer_size_kb will
> > > not be
> > > functional anymore.
> > > 
> > > To prevent leaving 'resize_disabled' and 'record_disabled' non-
> > > zero after
> > > ring_buffer_reset_online_cpus returns, we ensure that each atomic
> > > variable
> > > has been set up before atomic_sub() to it.
> > > 
> > > Cc: stable@vger.kernel.org
> > > Cc: npiggin@gmail.com
> > > Fixes: b23d7a5f4a07 ("ring-buffer: speed up buffer resets by
> > > avoiding synchronize_rcu for each CPU")
> > > Reviewed-by: Cheng-Jui Wang <cheng-jui.wang@mediatek.com>
> > > Signed-off-by: Tze-nan Wu <Tze-nan.Wu@mediatek.com>
> > > ---
> > > Changes from v4 to v5: 
> > > https://lore.kernel.org/lkml/20230412112401.25081-1-Tze-nan.Wu@mediatek.com/
> > >    - Move the define before the function
> > > ---
> > >   kernel/trace/ring_buffer.c | 16 +++++++++++++---
> > >   1 file changed, 13 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/kernel/trace/ring_buffer.c
> > > b/kernel/trace/ring_buffer.c
> > > index 76a2d91eecad..253ef85a9ec3 100644
> > > --- a/kernel/trace/ring_buffer.c
> > > +++ b/kernel/trace/ring_buffer.c
> > > @@ -5345,6 +5345,9 @@ void ring_buffer_reset_cpu(struct
> > > trace_buffer *buffer, int cpu)
> > >   }
> > >   EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu);
> > > 
> > > +/* Flag to ensure proper resetting of atomic variables */
> > > +#define RESET_BIT   (1 << 30)
> > > +
> > >   /**
> > >    * ring_buffer_reset_online_cpus - reset a ring buffer per CPU
> > > buffer
> > >    * @buffer: The ring buffer to reset a per cpu buffer of
> > > @@ -5361,20 +5364,27 @@ void ring_buffer_reset_online_cpus(struct
> > > trace_buffer *buffer)
> > >      for_each_online_buffer_cpu(buffer, cpu) {
> > >              cpu_buffer = buffer->buffers[cpu];
> > > 
> > > -            atomic_inc(&cpu_buffer->resize_disabled);
> > > +            atomic_add(RESET_BIT, &cpu_buffer->resize_disabled);
> > >              atomic_inc(&cpu_buffer->record_disabled);
> > >      }
> > > 
> > >      /* Make sure all commits have finished */
> > >      synchronize_rcu();
> > > 
> > > -    for_each_online_buffer_cpu(buffer, cpu) {
> > > +    for_each_buffer_cpu(buffer, cpu) {
> > >              cpu_buffer = buffer->buffers[cpu];
> > > 
> > > +            /*
> > > +             * If a CPU came online during the
> > > synchronize_rcu(), then
> > > +             * ignore it.
> > > +             */
> > > +            if (!(atomic_read(&cpu_buffer->resize_disabled) &
> > > RESET_BIT))
> > > +                    continue;
> > > +
> > >              reset_disabled_cpu_buffer(cpu_buffer);
> > > 
> > >              atomic_dec(&cpu_buffer->record_disabled);
> > > -            atomic_dec(&cpu_buffer->resize_disabled);
> > > +            atomic_sub(RESET_BIT, &cpu_buffer->resize_disabled);
> > >      }
> > > 
> > >      mutex_unlock(&buffer->mutex);
diff mbox series

Patch

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 76a2d91eecad..253ef85a9ec3 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -5345,6 +5345,9 @@  void ring_buffer_reset_cpu(struct trace_buffer *buffer, int cpu)
 }
 EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu);
 
+/* Flag to ensure proper resetting of atomic variables */
+#define RESET_BIT	(1 << 30)
+
 /**
  * ring_buffer_reset_online_cpus - reset a ring buffer per CPU buffer
  * @buffer: The ring buffer to reset a per cpu buffer of
@@ -5361,20 +5364,27 @@  void ring_buffer_reset_online_cpus(struct trace_buffer *buffer)
 	for_each_online_buffer_cpu(buffer, cpu) {
 		cpu_buffer = buffer->buffers[cpu];
 
-		atomic_inc(&cpu_buffer->resize_disabled);
+		atomic_add(RESET_BIT, &cpu_buffer->resize_disabled);
 		atomic_inc(&cpu_buffer->record_disabled);
 	}
 
 	/* Make sure all commits have finished */
 	synchronize_rcu();
 
-	for_each_online_buffer_cpu(buffer, cpu) {
+	for_each_buffer_cpu(buffer, cpu) {
 		cpu_buffer = buffer->buffers[cpu];
 
+		/*
+		 * If a CPU came online during the synchronize_rcu(), then
+		 * ignore it.
+		 */
+		if (!(atomic_read(&cpu_buffer->resize_disabled) & RESET_BIT))
+			continue;
+
 		reset_disabled_cpu_buffer(cpu_buffer);
 
 		atomic_dec(&cpu_buffer->record_disabled);
-		atomic_dec(&cpu_buffer->resize_disabled);
+		atomic_sub(RESET_BIT, &cpu_buffer->resize_disabled);
 	}
 
 	mutex_unlock(&buffer->mutex);