Re: Re: [PATCH v2 3/3] i387: support lazy restore of FPU state

Re: Re: [PATCH v2 3/3] i387: support lazy restore of FPU state

[Jongman Heo]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=euc-kr, Size: 4352 bytes --]


> Sender : Josh Boyer<jwboyer@gmail.com>
> Date : 2012-02-21 10:50 (GMT+09:00)
> Title : Re: [PATCH v2 3/3] i387: support lazy restore of FPU state
>
> > On Mon, Feb 20, 2012 at 2:48 PM, Linus Torvalds
> > wrote:
> >
> > From: Linus Torvalds 
> > Date: Sun, 19 Feb 2012 13:27:00 -0800
> > Subject: [PATCH v2 3/3] i387: support lazy restore of FPU state
> > 
> > This makes us recognize when we try to restore FPU state that matches
> > what we already have in the FPU on this CPU, and avoids the restore
> > entirely if so.
> >
> > To do this, we add two new data fields:
> >
> >  - a percpu 'fpu_owner_task' variable that gets written any time we
> >   update the "has_fpu" field, and thus acts as a kind of back-pointer
> >   to the task that owns the CPU.  The exception is when we save the FPU
> >   state as part of a context switch - if the save can keep the FPU
> >   state around, we leave the 'fpu_owner_task' variable pointing at the
> >   task whose FP state still remains on the CPU.
> >
> >  - a per-thread 'last_cpu' field, that indicates which CPU that thread
> >   used its FPU on last.  We update this on every context switch
> >   (writing an invalid CPU number if the last context switch didn't
> >   leave the FPU in a lazily usable state), so we know that *that*
> >   thread has done nothing else with the FPU since.
> >
> > These two fields together can be used when next switching back to the
> > task to see if the CPU still matches: if 'fpu_owner_task' matches the
> > task we are switching to, we know that no other task (or kernel FPU
> > usage) touched the FPU on this CPU in the meantime, and if the current
> > CPU number matches the 'last_cpu' field, we know that this thread did no
> > other FP work on any other CPU, so the FPU state on the CPU must match
> > what was saved on last context switch.
> >
> > In that case, we can avoid the 'f[x]rstor' entirely, and just clear the
> > CR0.TS bit.
> >
> > Signed-off-by: Linus Torvalds 
> 
> I haven't tried really figuring this out yet, but building the Fedora kernel
> on x86_64 with your latest tree results in:
> 
> ERROR: "fpu_owner_task" [lib/raid6/raid6_pq.ko] undefined!
> ERROR: "fpu_owner_task" [arch/x86/kvm/kvm.ko] undefined!
> ERROR: "fpu_owner_task" [arch/x86/crypto/sha1-ssse3.ko] undefined!
> ERROR: "fpu_owner_task" [arch/x86/crypto/serpent-sse2-x86_64.ko] undefined!
> ERROR: "fpu_owner_task" [arch/x86/crypto/ghash-clmulni-intel.ko] undefined!
> make[1]: *** [__modpost] Error 1
> make: *** [modules] Error 2
> + exit 1
> 
> Since this patch went in as 7e16838d94b566a1, I'm guessing it's at least
> related.
> 
> I'm building again with more verbose output but I thought I'd send this out
> quickly.
> 
> josh
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Similar here, with 32bit x86 build.

[snip]
  LD      net/built-in.o
  LD      vmlinux.o
  MODPOST vmlinux.o
  GEN     .version
  CHK     include/generated/compile.h
  UPD     include/generated/compile.h
  CC      init/version.o
  LD      init/built-in.o
  LD      .tmp_vmlinux1
arch/x86/built-in.o: In function `__thread_clear_has_fpu':
/usr/src/linux/arch/x86/include/asm/i387.h:300: undefined reference to `fpu_owner_task'
arch/x86/built-in.o: In function `__thread_set_has_fpu':
/usr/src/linux/arch/x86/include/asm/i387.h:307: undefined reference to `fpu_owner_task'
arch/x86/built-in.o: In function `fpu_lazy_restore':
/usr/src/linux/arch/x86/include/asm/i387.h:354: undefined reference to `fpu_owner_task'
arch/x86/built-in.o: In function `__thread_set_has_fpu':
/usr/src/linux/arch/x86/include/asm/i387.h:307: undefined reference to `fpu_owner_task'
arch/x86/built-in.o: In function `__thread_clear_has_fpu':
/usr/src/linux/arch/x86/include/asm/i387.h:300: undefined reference to `fpu_owner_task'
arch/x86/built-in.o:/usr/src/linux/arch/x86/include/asm/i387.h:300: more undefined references to `fpu_owner_task' follow
make: *** [.tmp_vmlinux1] Error 


In case you need my .config, please let me know~.

Jongman Heo.
ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥

Re: Re: [PATCH v2 3/3] i387: support lazy restore of FPU state

[Jongman Heo]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=euc-kr, Size: 1380 bytes --]


> Sender : Linus Torvalds<torvalds@linux-foundation.org>
> Date : 2012-02-21 11:18 (GMT+09:00)
> Title : Re: [PATCH v2 3/3] i387: support lazy restore of FPU state
> 
> On Mon, Feb 20, 2012 at 6:10 PM, Linus Torvalds
> wrote:
> >
> > The attached trivial patch fixes it, I bet.
> 
> Actually, it doesn't fix it on x86-32, because we actually have an
> #ifdef CONFIG_X86_64 around the "current_task" definition due to
> pointless differences in how we do that on x86-64 and x86-32.
> 
> So much for the "common" part of "arch/x86/kernel/cpu/common.c"
> 
> > Although I do wonder if we should just make kernel_fpu_begin() be a
> > real function instead of inlining it. I'm not sure it makes sense to
> > inline that thing, and it might be better to export that one instead.
> 
> I do think that would be better in the long run, but for now here's an
> updated "trivial" patch to fix it.
> 
> I want the fpu_owner_task to be declared next to the cache-hot
> task-switching stuff, and since they are different on 32-bit and
> 64-bit (for no really good reason), that gets duplicated too. Sad.
> 
>                         Linus
> 
>                       Linus

Yeah, this patch fixes my x86-32 build.

Thanks,
Jongman Heo.ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥