public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "krebbel at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug tree-optimization/108199] Bitfields and storage_order_attribute Date: Thu, 22 Dec 2022 16:24:11 +0000 [thread overview] Message-ID: <bug-108199-4-KrCIqG1LEr@http.gcc.gnu.org/bugzilla/> (raw) In-Reply-To: <bug-108199-4@http.gcc.gnu.org/bugzilla/> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108199 Andreas Krebbel <krebbel at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Target| |x86_64 Build| |x86_64 Keywords| |wrong-code Host| |x86_64 --- Comment #2 from Andreas Krebbel <krebbel at gcc dot gnu.org> --- The testcase does an assigned between two struct with endianess differing from host endianess (assumed to be little). Here the required byteswaps are supposed to cancel each other out. After that a bitfield comparison on the target struct is done. This comparison uses the wrong byte offset into the bitfield: testb $64, 7(%rdi) jne .L11 On a big endian target the first bits in the bitfield are supposed to reside in the first bytes in memory. The problem appears to get introduced when dead store elimination removes the assignment to the target struct in FRE. Before FRE we have the following: _1 = src_6(D)->a; bswap dst$val_9 = _1; bswap _2 = BIT_FIELD_REF <dst$val_9, 8, 0>; bswap _3 = _2 & 64; if (_3 != 0) ... This would result in 3 bswaps chained to each other. However, after FRE we have only two because the dead store to dst$val is removed. _1 = src_6(D)->a; _2 = BIT_FIELD_REF <_1, 8, 0>; _3 = _2 & 64; if (_3 != 0) Now we have only which cancel each other out. Looks like we have to prevent depending stores/loads with different endianess from getting removed - perhaps by making them also volatile? I think we have to keep the number of memory accesses with foreign endianess constant over the optimizations.
next prev parent reply other threads:[~2022-12-22 16:24 UTC|newest] Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top 2022-12-22 16:08 [Bug tree-optimization/108199] New: " krebbel at gcc dot gnu.org 2022-12-22 16:09 ` [Bug tree-optimization/108199] " krebbel at gcc dot gnu.org 2022-12-22 16:24 ` krebbel at gcc dot gnu.org [this message] 2022-12-22 16:29 ` krebbel at gcc dot gnu.org 2022-12-22 16:39 ` [Bug tree-optimization/108199] Bitfields, unions and SRA " pinskia at gcc dot gnu.org 2022-12-22 17:08 ` krebbel at gcc dot gnu.org 2022-12-22 18:29 ` pinskia at gcc dot gnu.org 2022-12-23 7:53 ` krebbel at gcc dot gnu.org 2023-01-09 13:21 ` rguenth at gcc dot gnu.org 2023-01-09 20:15 ` ebotcazou at gcc dot gnu.org 2023-01-09 20:26 ` pinskia at gcc dot gnu.org 2023-01-09 21:46 ` ebotcazou at gcc dot gnu.org 2023-01-10 11:11 ` ebotcazou at gcc dot gnu.org 2023-01-10 11:24 ` ebotcazou at gcc dot gnu.org 2023-01-11 15:03 ` cvs-commit at gcc dot gnu.org 2023-01-11 15:35 ` krebbel at gcc dot gnu.org 2023-01-11 15:41 ` krebbel at gcc dot gnu.org 2023-01-11 16:11 ` ebotcazou at gcc dot gnu.org 2023-01-11 19:46 ` cvs-commit at gcc dot gnu.org 2023-01-11 19:47 ` cvs-commit at gcc dot gnu.org 2023-01-11 19:48 ` ebotcazou at gcc dot gnu.org
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=bug-108199-4-KrCIqG1LEr@http.gcc.gnu.org/bugzilla/ \ --to=gcc-bugzilla@gcc.gnu.org \ --cc=gcc-bugs@gcc.gnu.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: linkBe 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).