From: Uros Bizjak <ubizjak@gmail.com>
To: "H.J. Lu" <hjl.tools@gmail.com>
Cc: gcc-patches@gcc.gnu.org
Subject: Re: PATCH: Properly check the end of basic block
Date: Thu, 18 Nov 2010 10:47:00 -0000 [thread overview]
Message-ID: <AANLkTi=FsHzQinx2HVrmf94LppvQ3gCd+-8tBrruXGwg@mail.gmail.com> (raw)
In-Reply-To: <AANLkTik8TA3or7jM0SxaPcNLQ5Gk1m=htoAsKb20udPq@mail.gmail.com>
On Thu, Nov 18, 2010 at 12:36 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> Here is the patch for
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46519
>
> We have 2 blocks pointing to each others. This patch first scans
> all blocks without moving vzeroupper so that we can have accurate
> information about upper 128bits at block entry.
This introduces another insn scanning pass, almost the same as
existing vzeroupper pass (modulo CALL_INSN/JUMP_INSN handling).
So, if I understand correctly:
- The patch removes the detection if the function ever touches AVX registers.
- Due to this, all call_insn RTXes have to be decorated with
CALL_NEEDS_VZEROUPPER.
- A new pre-pass is required that scans all functions in order to
detect functions with live AVX registers at exit, and at the same time
marks the functions that *do not* use AVX registers.
- Existing pass then re-scans everything to again detect functions
with live AVX registers at exit and handles vzeroupper emission.
I don't think this approach is acceptable. Maybe a LCM infrastructure
can be used to handle this case?
Uros.
next prev parent reply other threads:[~2010-11-18 9:12 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-11-17 8:34 H.J. Lu
2010-11-17 9:24 ` Uros Bizjak
2010-11-17 15:21 ` H.J. Lu
2010-11-17 20:26 ` Uros Bizjak
2010-11-18 0:49 ` H.J. Lu
2010-11-18 7:06 ` H.J. Lu
2010-11-18 9:23 ` Uros Bizjak
2010-11-18 10:47 ` Uros Bizjak [this message]
2010-11-18 18:38 ` H.J. Lu
2010-11-18 19:27 ` Uros Bizjak
2010-11-18 19:51 ` H.J. Lu
2010-11-18 20:25 ` Uros Bizjak
2010-11-18 20:44 ` H.J. Lu
2010-11-18 22:30 ` Uros Bizjak
2010-11-18 22:33 ` Andrew Pinski
2010-11-18 22:36 ` Jakub Jelinek
2010-11-18 22:55 ` Uros Bizjak
2010-11-18 22:50 ` Richard Henderson
2010-11-18 23:48 ` Uros Bizjak
2010-11-19 0:56 ` Richard Henderson
2010-11-19 1:15 ` Uros Bizjak
2010-11-19 9:56 ` Jan Hubicka
2010-11-18 23:09 Steven Bosscher
2010-11-18 23:37 ` Uros Bizjak
2010-11-23 12:42 ` Richard Sandiford
2010-11-23 13:28 ` Steven Bosscher
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='AANLkTi=FsHzQinx2HVrmf94LppvQ3gCd+-8tBrruXGwg@mail.gmail.com' \
--to=ubizjak@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=hjl.tools@gmail.com \
/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 read-only IMAP folder(s) and NNTP newsgroup(s).