[Q] sched: __ARCH_WANT_UNLOCKED_CTXSW on IA64

[Q] sched: __ARCH_WANT_UNLOCKED_CTXSW on IA64

[Kirill Tkhai]

Hi, IA64 Gurus,

I was looking for the reasons of IA64's __ARCH_WANT_UNLOCKED_CTXSW,
and found the historical commit below:

    commit f8efa27662532ad5adb2790bfc3f4c78e019cfad
    Author: Chen, Kenneth W <kenneth.w.chen@intel.com>
    Date:   Thu Jan 26 18:24:59 2006 -0800

    [IA64] remove staled comments in asm/system.h
    
    With the recent optimization made to wrap_mmu_context function,
    we don't hold tasklist_lock anymore when wrapping context id.
    The comments in asm/system.h must fall through the crack earlier.
    Remove staled comments.
    
    I believe it is still beneficial to unlock the runqueue lock
    across context switch. So leave __ARCH_WANT_UNLOCKED_CTXSW on.
    
    Signed-off-by: Ken Chen <kenneth.w.chen@intel.com>
    Signed-off-by: Tony Luck <tony.luck@intel.com>

The comment confuses a reader. It may seem that the #define
is not necessary now, no comment around it at all.

So, could you please point the deadlock we want to avoid
by __ARCH_WANT_UNLOCKED_CTXSW? I've tried to find, but nothing
was found by me. No IA64 machines around, so I can't just
to load kernel w/o this define.

Thanks,
Kirill

Re: [Q] sched: __ARCH_WANT_UNLOCKED_CTXSW on IA64

[Peter Zijlstra]

[-- Attachment #1: Type: text/plain, Size: 1342 bytes --]

On Wed, Aug 06, 2014 at 01:41:30AM +0400, Kirill Tkhai wrote:
> Hi, IA64 Gurus,
> 
> I was looking for the reasons of IA64's __ARCH_WANT_UNLOCKED_CTXSW,
> and found the historical commit below:
> 
>     commit f8efa27662532ad5adb2790bfc3f4c78e019cfad
>     Author: Chen, Kenneth W <kenneth.w.chen@intel.com>
>     Date:   Thu Jan 26 18:24:59 2006 -0800
> 
>     [IA64] remove staled comments in asm/system.h
>     
>     With the recent optimization made to wrap_mmu_context function,
>     we don't hold tasklist_lock anymore when wrapping context id.
>     The comments in asm/system.h must fall through the crack earlier.
>     Remove staled comments.
>     
>     I believe it is still beneficial to unlock the runqueue lock
>     across context switch. So leave __ARCH_WANT_UNLOCKED_CTXSW on.
>     
>     Signed-off-by: Ken Chen <kenneth.w.chen@intel.com>
>     Signed-off-by: Tony Luck <tony.luck@intel.com>
> 
> The comment confuses a reader. It may seem that the #define
> is not necessary now, no comment around it at all.
> 
> So, could you please point the deadlock we want to avoid
> by __ARCH_WANT_UNLOCKED_CTXSW? I've tried to find, but nothing
> was found by me. No IA64 machines around, so I can't just
> to load kernel w/o this define.

see http://marc.info/?l=linux-kernel&m=131418885605812

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]