Re: [tip:sched/core] sched, x86: Optimize branch hint in __switch_to()

From: Brian Gerst
Date: Wed Nov 25 2009 - 10:54:06 EST


On Wed, Nov 25, 2009 at 4:03 AM, Tim Blechmann <tim@xxxxxxxxxx> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 11/24/2009 06:28 PM, Brian Gerst wrote:
>> On Tue, Nov 24, 2009 at 11:57 AM, tip-bot for Tim Blechmann
>> <tim@xxxxxxxxxx> wrote:
>>> Commit-ID: Âa3a1de0c34de6f5f8332cd6151c46af7813c0fcb
>>> Gitweb: Â Â http://git.kernel.org/tip/a3a1de0c34de6f5f8332cd6151c46af7813c0fcb
>>> Author: Â Â Tim Blechmann <tim@xxxxxxxxxx>
>>> AuthorDate: Tue, 24 Nov 2009 11:55:15 +0100
>>> Committer: ÂIngo Molnar <mingo@xxxxxxx>
>>> CommitDate: Tue, 24 Nov 2009 12:20:04 +0100
>>>
>>> sched, x86: Optimize branch hint in __switch_to()
>>>
>>> Branch hint profiling on my nehalem machine showed 96%
>>> incorrect branch hints:
>>>
>>> Â6548732 174664120 Â96 __switch_to          Âprocess_64.c
>>> Â Â406
>>> Â6548745 174565593 Â96 __switch_to          Âprocess_64.c
>>> Â Â410
>>>
>>> Signed-off-by: Tim Blechmann <tim@xxxxxxxxxx>
>>> Cc: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx>
>>> Cc: Mike Galbraith <efault@xxxxxx>
>>> Cc: Paul Mackerras <paulus@xxxxxxxxx>
>>> Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
>>> Cc: Frederic Weisbecker <fweisbec@xxxxxxxxx>
>>> LKML-Reference: <4B0BBB93.3080307@xxxxxxxxxx>
>>> Signed-off-by: Ingo Molnar <mingo@xxxxxxx>
>>> ---
>>> Âarch/x86/kernel/process_64.c | Â Â5 ++---
>>> Â1 files changed, 2 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
>>> index ad535b6..d9db104 100644
>>> --- a/arch/x86/kernel/process_64.c
>>> +++ b/arch/x86/kernel/process_64.c
>>> @@ -406,11 +406,10 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
>>> Â Â Â Â * This won't pick up thread selector changes, but I guess that is ok.
>>> Â Â Â Â */
>>> Â Â Â Âsavesegment(es, prev->es);
>>> - Â Â Â if (unlikely(next->es | prev->es))
>>> + Â Â Â if (next->es | prev->es)
>>> Â Â Â Â Â Â Â Âloadsegment(es, next->es);
>>> -
>>> Â Â Â Âsavesegment(ds, prev->ds);
>>> - Â Â Â if (unlikely(next->ds | prev->ds))
>>> + Â Â Â if (next->ds | prev->ds)
>>> Â Â Â Â Â Â Â Âloadsegment(ds, next->ds);
>>>
>>>
>>
>> 64-bit tasks should have %ds and %es set to null selectors. ÂThe only
>> time they should be different is for 32-bit tasks.
>
> this doesn't seem to be the case on my machine for 96% of the calls ...
> i am just running 64bit binaries on this machine, no 32bit programs
> (that i know of (ubuntu karmic amd64))
>
> tim

Some early kernel threads inherited KERNEL_DS from the boot process.
Patch coming soon.

--
Brian Gerst
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/