public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* x86/smpboot: Question regarding native_play_dead() __noreturn warning
@ 2025-10-27 12:23 Thorsten Blum
  2025-10-27 12:50 ` Peter Zijlstra
  0 siblings, 1 reply; 4+ messages in thread
From: Thorsten Blum @ 2025-10-27 12:23 UTC (permalink / raw)
  To: Waiman Long, Thomas Gleixner, Ingo Molnar, Borislav Petkov,
	Dave Hansen, H. Peter Anvin
  Cc: linux-kernel, x86

Hi,

I just came across this comment in arch/x86/kernel/smpboot.c:

/*
* native_play_dead() is essentially a __noreturn function, but it can't
* be marked as such as the compiler may complain about it.
*/
void native_play_dead(void) {
	...
}

and when I mark native_play_dead() as __noreturn, neither gcc nor clang
complain about it.

The commit message 2743fe89d4d4 ("x86/idle: Disable IBRS when CPU is
offline to improve single-threaded performance") says:

"Add a comment to say that native_play_dead() is a __noreturn function,
but it can't be marked as such to avoid confusion about the missing
MSR restoration code."

Unfortunately, that doesn't really help me either. Can someone explain
what the issue was and if the comment is still valid? Otherwise, I'd
like to submit a patch adding __noreturn and removing the comment.

Thanks,
Thorsten


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: x86/smpboot: Question regarding native_play_dead() __noreturn warning
  2025-10-27 12:23 x86/smpboot: Question regarding native_play_dead() __noreturn warning Thorsten Blum
@ 2025-10-27 12:50 ` Peter Zijlstra
  2025-10-27 15:48   ` Waiman Long
  0 siblings, 1 reply; 4+ messages in thread
From: Peter Zijlstra @ 2025-10-27 12:50 UTC (permalink / raw)
  To: Thorsten Blum
  Cc: Waiman Long, Thomas Gleixner, Ingo Molnar, Borislav Petkov,
	Dave Hansen, H. Peter Anvin, linux-kernel, x86

On Mon, Oct 27, 2025 at 01:23:02PM +0100, Thorsten Blum wrote:
> Hi,
> 
> I just came across this comment in arch/x86/kernel/smpboot.c:
> 
> /*
> * native_play_dead() is essentially a __noreturn function, but it can't
> * be marked as such as the compiler may complain about it.
> */
> void native_play_dead(void) {
> 	...
> }
> 
> and when I mark native_play_dead() as __noreturn, neither gcc nor clang
> complain about it.
> 
> The commit message 2743fe89d4d4 ("x86/idle: Disable IBRS when CPU is
> offline to improve single-threaded performance") says:
> 
> "Add a comment to say that native_play_dead() is a __noreturn function,
> but it can't be marked as such to avoid confusion about the missing
> MSR restoration code."
> 
> Unfortunately, that doesn't really help me either. Can someone explain
> what the issue was and if the comment is still valid? Otherwise, I'd
> like to submit a patch adding __noreturn and removing the comment.

I'm not sure either, it wasn't there in v2 but appeared in v3.

v2: 20230620140625.1001886-3-longman@redhat.com
v3: 20230622003603.1188364-2-longman@redhat.com

The difference is that v2 tried to restore the msr after 'play_dead'
which is silly, since it would never reach that code. v3 removed that
dead restore code and added the confusing comment.

There is a clue here though:

  20230622054053.uy577qezu5a65buc@treble

Josh suggests play_dead() should be marked noreturn (which it is in
current kernels).

Waiman then replies:

  921e1b98-af36-1f51-5abe-dea36425b706@redhat.com

which is utterly confused again.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: x86/smpboot: Question regarding native_play_dead() __noreturn warning
  2025-10-27 12:50 ` Peter Zijlstra
@ 2025-10-27 15:48   ` Waiman Long
  2025-10-27 16:23     ` Thorsten Blum
  0 siblings, 1 reply; 4+ messages in thread
From: Waiman Long @ 2025-10-27 15:48 UTC (permalink / raw)
  To: Peter Zijlstra, Thorsten Blum
  Cc: Thomas Gleixner, Ingo Molnar, Borislav Petkov, Dave Hansen,
	H. Peter Anvin, linux-kernel, x86

On 10/27/25 8:50 AM, Peter Zijlstra wrote:
> On Mon, Oct 27, 2025 at 01:23:02PM +0100, Thorsten Blum wrote:
>> Hi,
>>
>> I just came across this comment in arch/x86/kernel/smpboot.c:
>>
>> /*
>> * native_play_dead() is essentially a __noreturn function, but it can't
>> * be marked as such as the compiler may complain about it.
>> */
>> void native_play_dead(void) {
>> 	...
>> }
>>
>> and when I mark native_play_dead() as __noreturn, neither gcc nor clang
>> complain about it.
>>
>> The commit message 2743fe89d4d4 ("x86/idle: Disable IBRS when CPU is
>> offline to improve single-threaded performance") says:
>>
>> "Add a comment to say that native_play_dead() is a __noreturn function,
>> but it can't be marked as such to avoid confusion about the missing
>> MSR restoration code."
>>
>> Unfortunately, that doesn't really help me either. Can someone explain
>> what the issue was and if the comment is still valid? Otherwise, I'd
>> like to submit a patch adding __noreturn and removing the comment.
> I'm not sure either, it wasn't there in v2 but appeared in v3.
>
> v2: 20230620140625.1001886-3-longman@redhat.com
> v3: 20230622003603.1188364-2-longman@redhat.com
>
> The difference is that v2 tried to restore the msr after 'play_dead'
> which is silly, since it would never reach that code. v3 removed that
> dead restore code and added the confusing comment.
>
> There is a clue here though:
>
>    20230622054053.uy577qezu5a65buc@treble
>
> Josh suggests play_dead() should be marked noreturn (which it is in
> current kernels).
>
> Waiman then replies:
>
>    921e1b98-af36-1f51-5abe-dea36425b706@redhat.com
>
> which is utterly confused again.

I don't remember exactly how I got the warning when __noreturn is added 
to native_play_dead(). It may be a limitation of the objtool or gcc that 
I was using at that time. If Thorsten doesn't have problem adding 
__noreturn, I won't mind him doing that and taking out the the comment. 
We can see if there is other issue coming up in the future.

Cheers,
Longman


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: x86/smpboot: Question regarding native_play_dead() __noreturn warning
  2025-10-27 15:48   ` Waiman Long
@ 2025-10-27 16:23     ` Thorsten Blum
  0 siblings, 0 replies; 4+ messages in thread
From: Thorsten Blum @ 2025-10-27 16:23 UTC (permalink / raw)
  To: Waiman Long
  Cc: Peter Zijlstra, Thomas Gleixner, Ingo Molnar, Borislav Petkov,
	Dave Hansen, H. Peter Anvin, linux-kernel, x86

On 27. Oct 2025, at 16:48, Waiman Long wrote:
> On 10/27/25 8:50 AM, Peter Zijlstra wrote:
>> On Mon, Oct 27, 2025 at 01:23:02PM +0100, Thorsten Blum wrote:
>>> [...]
>> I'm not sure either, it wasn't there in v2 but appeared in v3.
>> 
>> v2: 20230620140625.1001886-3-longman@redhat.com
>> v3: 20230622003603.1188364-2-longman@redhat.com
>> 
>> The difference is that v2 tried to restore the msr after 'play_dead'
>> which is silly, since it would never reach that code. v3 removed that
>> dead restore code and added the confusing comment.
>> 
>> There is a clue here though:
>> 
>>  20230622054053.uy577qezu5a65buc@treble
>> 
>> Josh suggests play_dead() should be marked noreturn (which it is in
>> current kernels).
>> 
>> Waiman then replies:
>> 
>>  921e1b98-af36-1f51-5abe-dea36425b706@redhat.com
>> 
>> which is utterly confused again.
> 
> I don't remember exactly how I got the warning when __noreturn is added to native_play_dead(). It may be a limitation of the objtool or gcc that I was using at that time. If Thorsten doesn't have problem adding __noreturn, I won't mind him doing that and taking out the the comment. We can see if there is other issue coming up in the future.

Thanks for the quick replies!

I submitted the patch adding __noreturn here:

https://lore.kernel.org/lkml/20251027155107.183136-1-thorsten.blum@linux.dev/


^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2025-10-27 16:23 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-10-27 12:23 x86/smpboot: Question regarding native_play_dead() __noreturn warning Thorsten Blum
2025-10-27 12:50 ` Peter Zijlstra
2025-10-27 15:48   ` Waiman Long
2025-10-27 16:23     ` Thorsten Blum

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox