From: Jeff Law <jeffreyalaw@gmail.com>
To: Richard Biener <rguenther@suse.de>
Cc: Robin Dapp <rdapp.gcc@gmail.com>,
gcc-patches@gcc.gnu.org, "Li, Pan2" <pan2.li@intel.com>
Subject: Re: [PATCH][RFC] middle-end/110237 - wrong MEM_ATTRs for partial loads/stores
Date: Tue, 28 Nov 2023 08:00:28 -0700 [thread overview]
Message-ID: <d6854409-36cb-4224-a0c2-e6b6cd421bcd@gmail.com> (raw)
In-Reply-To: <nycvar.YFH.7.77.849.2311280735280.21409@jbgna.fhfr.qr>
On 11/28/23 00:50, Richard Biener wrote:
>
> There's no way to distinguish a partial vs. non-partial MEM on RTL and
> while without the bogus MEM_ATTR the alias oracle pieces that
> miscompiled the original case are fended off we still see the load/store
> as full given they have a mode with a size - that for example means
> that DSE can elide a previous store to a masked part. Eventually
> that's fended off by using an UNSPEC, but whether the RTL IL has
> the correct semantics is questionable.
>
> That said, I did propose scrapping the MEM_EXPR which I think is
> the correct thing to do unless we want to put a CALL_EXPR into it
> (nothing would use that at the moment) or re-do MEM_EXPR and instead
> have an ao_ref (or sth slightly more complete) instead of the current
> MEM_ATTRs - but that would be a lot of work.
>
> This leaves the question wrt. semantics of for example x86 mask_store:
>
> (insn 23 22 24 5 (set (mem:V4DF (plus:DI (reg/v/f:DI 106 [ x ])
> (reg:DI 101 [ ivtmp.15 ])) [2 MEM <vector(4) double>
> [(double *)x_11(D) + ivtmp.15_33 * 1]+0 S32 A64])
> (unspec:V4DF [
> (reg:V4DI 104 [ mask__16.8 ])
> (reg:V4DF 105 [ vect_cst__42 ])
> (mem:V4DF (plus:DI (reg/v/f:DI 106 [ x ])
> (reg:DI 101 [ ivtmp.15 ])) [2 MEM <vector(4)
> double> [(double *)x_11(D) + ivtmp.15_33 * 1]+0 S32 A64])
> ] UNSPEC_MASKMOV)) "t.c":5:12 8523 {avx_maskstorepd256}
> (nil))
>
> it uses a read-modify-write which makes it safe for DSE.
Agreed.
mask_load
> looks like
>
> (insn 28 27 29 6 (set (reg:V4DF 115 [ vect__7.11 ])
> (unspec:V4DF [
> (reg:V4DI 114 [ mask__8.8 ])
> (mem:V4DF (plus:DI (reg/v/f:DI 118 [ val ])
> (reg:DI 103 [ ivtmp.29 ])) [2 MEM <vector(4)
> double> [(double *)val_13(D) + ivtmp.29_22 * 1]+0 S32 A64])
> ] UNSPEC_MASKMOV)) "t.c":5:17 8515 {avx_maskloadpd256}
> (nil))
So with the mem:V4DF inside the unspec, ISTM we must treat that as a
potential full read, but we can't rely on it being a full read. I don't
think UNSPEC semantics are that it must read/consume all its operands in
full, just that it might. That might be worth a documentation
clarification.
>
> both have (as operand of the UNSPEC) a MEM with V4DFmode (and a
> MEM_EXPR with a similarly bougs MEM_EXPR) indicating the loads
> are _not_ partial. That means the disambiguation against a store
> to an object that's smaller than V4DF is still possible.
> Setting MEM_SIZE to UNKNOWN doesn't help - that just asks to look
> at the mode. As discussed using a BLKmode MEM _might_ be a way
> out but I didn't try what will happen then (patterns would need to
> be adjusted I guess).
>
> That said, I'm happy to commit the partial fix, scrapping the
> bogus MEM_EXPRs.
>
> OK for that?
Works for me.
jeff
next prev parent reply other threads:[~2023-11-28 15:00 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20230621075019.7CA813858033@sourceware.org>
2023-11-27 12:39 ` Robin Dapp
2023-11-27 15:45 ` Jeff Law
2023-11-28 7:50 ` Richard Biener
2023-11-28 10:31 ` Richard Sandiford
2023-11-28 11:21 ` Richard Biener
2023-11-28 11:32 ` Richard Sandiford
2023-11-28 12:17 ` Richard Biener
2023-11-28 15:00 ` Jeff Law [this message]
2023-11-29 7:16 ` Richard Biener
[not found] <20230621074956.1174B3858288@sourceware.org>
2023-06-26 8:29 ` Hongtao Liu
2023-06-26 8:41 ` Richard Biener
[not found] <20230621074951.F3C3C3858433@sourceware.org>
2023-06-21 15:29 ` Jeff Law
2023-06-22 6:39 ` Richard Biener
2023-06-24 14:32 ` Jeff Law
2023-06-21 7:49 Richard Biener
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=d6854409-36cb-4224-a0c2-e6b6cd421bcd@gmail.com \
--to=jeffreyalaw@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=pan2.li@intel.com \
--cc=rdapp.gcc@gmail.com \
--cc=rguenther@suse.de \
/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).