From: "Christopher M. Riedl" <cmr@codefail.de>
To: "Christophe Leroy" <christophe.leroy@csgroup.eu>,
<linuxppc-dev@lists.ozlabs.org>
Subject: Re: [PATCH 2/8] powerpc/signal: Add unsafe_copy_{vsx,fpr}_from_user()
Date: Sat, 06 Feb 2021 11:39:13 -0600 [thread overview]
Message-ID: <C92MQRHFCFEA.37OV051PFFY6@geist> (raw)
In-Reply-To: <fce0b2d0-58a3-a94d-a8e9-d104fc2b3058@csgroup.eu>
On Sat Feb 6, 2021 at 10:32 AM CST, Christophe Leroy wrote:
>
>
> Le 20/10/2020 à 04:01, Christopher M. Riedl a écrit :
> > On Fri Oct 16, 2020 at 10:48 AM CDT, Christophe Leroy wrote:
> >>
> >>
> >> Le 15/10/2020 à 17:01, Christopher M. Riedl a écrit :
> >>> Reuse the "safe" implementation from signal.c except for calling
> >>> unsafe_copy_from_user() to copy into a local buffer. Unlike the
> >>> unsafe_copy_{vsx,fpr}_to_user() functions the "copy from" functions
> >>> cannot use unsafe_get_user() directly to bypass the local buffer since
> >>> doing so significantly reduces signal handling performance.
> >>
> >> Why can't the functions use unsafe_get_user(), why does it significantly
> >> reduces signal handling
> >> performance ? How much significant ? I would expect that not going
> >> through an intermediate memory
> >> area would be more efficient
> >>
> >
> > Here is a comparison, 'unsafe-signal64-regs' avoids the intermediate buffer:
> >
> > | | hash | radix |
> > | -------------------- | ------ | ------ |
> > | linuxppc/next | 289014 | 158408 |
> > | unsafe-signal64 | 298506 | 253053 |
> > | unsafe-signal64-regs | 254898 | 220831 |
> >
> > I have not figured out the 'why' yet. As you mentioned in your series,
> > technically calling __copy_tofrom_user() is overkill for these
> > operations. The only obvious difference between unsafe_put_user() and
> > unsafe_get_user() is that we don't have asm-goto for the 'get' variant.
> > Instead we wrap with unsafe_op_wrap() which inserts a conditional and
> > then goto to the label.
> >
> > Implemenations:
> >
> > #define unsafe_copy_fpr_from_user(task, from, label) do { \
> > struct task_struct *__t = task; \
> > u64 __user *buf = (u64 __user *)from; \
> > int i; \
> > \
> > for (i = 0; i < ELF_NFPREG - 1; i++) \
> > unsafe_get_user(__t->thread.TS_FPR(i), &buf[i], label); \
> > unsafe_get_user(__t->thread.fp_state.fpscr, &buf[i], label); \
> > } while (0)
> >
> > #define unsafe_copy_vsx_from_user(task, from, label) do { \
> > struct task_struct *__t = task; \
> > u64 __user *buf = (u64 __user *)from; \
> > int i; \
> > \
> > for (i = 0; i < ELF_NVSRHALFREG ; i++) \
> > unsafe_get_user(__t->thread.fp_state.fpr[i][TS_VSRLOWOFFSET], \
> > &buf[i], label); \
> > } while (0)
> >
>
> Do you have CONFIG_PROVE_LOCKING or CONFIG_DEBUG_ATOMIC_SLEEP enabled in
> your config ?
I don't have these set in my config (ppc64le_defconfig). I think I
figured this out - the reason for the lower signal throughput is the
barrier_nospec() in __get_user_nocheck(). When looping we incur that
cost on every iteration. Commenting it out results in signal performance
of ~316K w/ hash on the unsafe-signal64-regs branch. Obviously the
barrier is there for a reason but it is quite costly.
This also explains why the copy_{fpr,vsx}_to_user() direction does not
suffer from the slowdown because there is no need for barrier_nospec().
>
> If yes, could you try together with the patch from Alexey
> https://patchwork.ozlabs.org/project/linuxppc-dev/patch/20210204121612.32721-1-aik@ozlabs.ru/
> ?
>
> Thanks
> Christophe
next prev parent reply other threads:[~2021-02-06 17:55 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-15 15:01 [PATCH 0/8] Improve signal performance on PPC64 with KUAP Christopher M. Riedl
2020-10-15 15:01 ` [PATCH 1/8] powerpc/uaccess: Add unsafe_copy_from_user Christopher M. Riedl
2020-10-16 6:54 ` Christoph Hellwig
2020-10-16 13:18 ` Christophe Leroy
2020-10-16 13:17 ` Christophe Leroy
2020-10-20 3:00 ` Christopher M. Riedl
2020-10-15 15:01 ` [PATCH 2/8] powerpc/signal: Add unsafe_copy_{vsx,fpr}_from_user() Christopher M. Riedl
2020-10-16 13:48 ` [PATCH 2/8] powerpc/signal: Add unsafe_copy_{vsx, fpr}_from_user() Christophe Leroy
2020-10-20 2:01 ` [PATCH 2/8] powerpc/signal: Add unsafe_copy_{vsx,fpr}_from_user() Christopher M. Riedl
2021-02-06 16:32 ` [PATCH 2/8] powerpc/signal: Add unsafe_copy_{vsx, fpr}_from_user() Christophe Leroy
2021-02-06 17:39 ` Christopher M. Riedl [this message]
2021-02-07 10:12 ` Christophe Leroy
2021-02-08 17:14 ` [PATCH 2/8] powerpc/signal: Add unsafe_copy_{vsx,fpr}_from_user() Christopher M. Riedl
2021-02-08 17:18 ` [PATCH 2/8] powerpc/signal: Add unsafe_copy_{vsx, fpr}_from_user() Christophe Leroy
2020-10-15 15:01 ` [PATCH 3/8] powerpc: Mark functions called inside uaccess blocks w/ 'notrace' Christopher M. Riedl
2020-10-16 6:56 ` Christoph Hellwig
2020-10-16 9:41 ` Peter Zijlstra
2020-10-20 7:34 ` Michael Ellerman
2020-10-16 7:02 ` Christophe Leroy
2020-10-20 1:59 ` Christopher M. Riedl
2020-10-15 15:01 ` [PATCH 4/8] powerpc/signal64: Replace setup_sigcontext() w/ unsafe_setup_sigcontext() Christopher M. Riedl
2020-10-15 15:01 ` [PATCH 5/8] powerpc/signal64: Replace restore_sigcontext() w/ unsafe_restore_sigcontext() Christopher M. Riedl
2020-10-15 15:01 ` [PATCH 6/8] powerpc/signal64: Replace setup_trampoline() w/ unsafe_setup_trampoline() Christopher M. Riedl
2020-10-16 13:56 ` Christophe Leroy
2020-10-20 2:42 ` Christopher M. Riedl
2020-10-20 5:02 ` Christophe Leroy
2020-10-15 15:01 ` [PATCH 7/8] powerpc/signal64: Rewrite handle_rt_signal64() to minimise uaccess switches Christopher M. Riedl
2020-10-16 14:00 ` Christophe Leroy
2020-10-20 2:44 ` Christopher M. Riedl
2020-10-15 15:01 ` [PATCH 8/8] powerpc/signal64: Rewrite rt_sigreturn() " Christopher M. Riedl
2020-10-16 14:07 ` Christophe Leroy
2020-10-20 2:45 ` Christopher M. Riedl
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=C92MQRHFCFEA.37OV051PFFY6@geist \
--to=cmr@codefail.de \
--cc=christophe.leroy@csgroup.eu \
--cc=linuxppc-dev@lists.ozlabs.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).