From: Richard Sandiford <rdsandiford@googlemail.com>
To: Eric Botcazou <ebotcazou@adacore.com>
Cc: Jakub Jelinek <jakub@redhat.com>,
gcc-patches@gcc.gnu.org, Hans-Peter Nilsson <hp@bitrange.com>,
Alexandre Oliva <aoliva@redhat.com>
Subject: Re: [RFA:] fix PR55030, wrong code from __builtin_setjmp_receiver
Date: Wed, 28 Nov 2012 20:10:00 -0000 [thread overview]
Message-ID: <87k3t5mqun.fsf@talisman.default> (raw)
In-Reply-To: <20520021.fBU7p54CN9@polaris> (Eric Botcazou's message of "Tue, 27 Nov 2012 18:35:52 +0100")
Eric Botcazou <ebotcazou@adacore.com> writes:
>> I strongly disagree with this. Outputs and clobbers have significant
>> meaning even on volatile asms, asm volatile doesn't mean all registers and
>> all memory are supposed to be considered clobbered. For memory you have
>> "memory" clobber for that, for registers it is users responsibility to
>> describe exactly what changes, either in clobbers or in outputs.
>> The difference between non-volatile and volatile asm is, as the
>> documentation states:
>>
>> The `volatile' keyword indicates that the instruction has important
>> side-effects. GCC will not delete a volatile `asm' if it is reachable.
>>
>> Volatile asm acts as an optimization barrier to some extent, but it really
>> can't modify registers or memory that aren't described as modified in the
>> asm pattern. The important side-effects are of some other kind than
>> modifying registers or memory visible from the current function.
>> Ditto for UNSPEC_VOLATILE.
>
> Well, the last sentence would essentially void the entire argument I
> think. It's well established in the RTL middle-end that
> UNSPEC_VOLATILE can do pretty much everything behind the back of the
> compiler.
As always when jumping in the middle of thread, I might well be missing
the point sorry, but this sounded a bit extreme if taken literally.
An UNSPEC_VOLATILE doesn't in itself force the function to save all
call-saved registers, so an UNSPEC_VOLATILE that modifies those registers
behind the back of the compiler would lead to us generating wrong code.
And I thought UNSPEC_VOLATILEs that also clobber visible memory in an
unpredictable way had to have something like (clobber (mem:BLK (scratch))).
I thought Jakub's "the important side-effects are of some other
kind than modifying registers or memory visible from the current
function" applied to UNSPEC_VOLATILE too.
Richard
next prev parent reply other threads:[~2012-11-28 20:10 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-11-12 4:49 Hans-Peter Nilsson
2012-11-12 8:28 ` Eric Botcazou
2012-11-19 3:54 ` Hans-Peter Nilsson
2012-11-19 9:35 ` Eric Botcazou
2012-11-26 3:30 ` Hans-Peter Nilsson
2012-11-27 11:44 ` Hans-Peter Nilsson
2012-11-27 12:04 ` Jakub Jelinek
2012-11-27 12:27 ` Hans-Peter Nilsson
2012-11-27 13:32 ` Jakub Jelinek
2012-11-27 17:39 ` Eric Botcazou
2012-11-27 18:20 ` Jakub Jelinek
2012-12-01 0:03 ` Eric Botcazou
2012-11-28 20:10 ` Richard Sandiford [this message]
2012-11-28 20:46 ` Richard Henderson
2012-11-28 13:44 ` Hans-Peter Nilsson
2012-11-28 13:56 ` Jakub Jelinek
2012-11-28 20:53 ` Richard Henderson
2012-11-27 19:30 Uros Bizjak
2012-11-28 9:50 ` Uros Bizjak
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=87k3t5mqun.fsf@talisman.default \
--to=rdsandiford@googlemail.com \
--cc=aoliva@redhat.com \
--cc=ebotcazou@adacore.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=hp@bitrange.com \
--cc=jakub@redhat.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).