public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "msebor at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug middle-end/95667] New: warning for memset writing across multiple members Date: Sat, 13 Jun 2020 23:36:37 +0000 [thread overview] Message-ID: <bug-95667-4@http.gcc.gnu.org/bugzilla/> (raw) https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95667 Bug ID: 95667 Summary: warning for memset writing across multiple members Product: gcc Version: 10.0 Status: UNCONFIRMED Severity: normal Priority: P3 Component: middle-end Assignee: unassigned at gcc dot gnu.org Reporter: msebor at gcc dot gnu.org Target Milestone: --- Although strictly correct, GCC usually avoids warning for calls to memcpy that write into multiple members at the same time, up to the size of the complete object. This is done because the Linux kernel does these things in a few places. As dangerous as the practice is, the change committed to resolve pr95353 (r11-1183) tried to preserve this behavior for now (in the future I'd like to add a new warning option to control it), but the test case below shows it didn't preserve the GCC 10 and prior behavior in this case (same with memcpy). $ cat z.c && gcc -O2 -S -Wall -Wextra -fdump-tree-optimized=/dev/stdout z.c struct S { char a[3], b[5]; }; void* f (void) { struct S *p = __builtin_malloc (sizeof *p); char s[] = "1234567"; __builtin_strcpy (p->a, s); // warning (good) return p; } void* g (void) { struct S *p = __builtin_malloc (sizeof *p); __builtin_memset (p->a, 0, 8); // warning not expected (yet) return p; } z.c: In function ‘f’: z.c:7:3: warning: ‘__builtin_strcpy’ writing 8 bytes into a region of size 3 [-Wstringop-overflow=] 7 | __builtin_strcpy (p->a, s); // warning (good) | ^~~~~~~~~~~~~~~~~~~~~~~~~~ z.c:1:17: note: at offset 0 to object ‘a’ with size 3 declared here 1 | struct S { char a[3], b[5]; }; | ^ ;; Function f (f, funcdef_no=0, decl_uid=1934, cgraph_uid=1, symbol_order=0) f () { char s[8]; struct S * p; long unsigned int _8; <bb 2> [local count: 1073741824]: p_4 = __builtin_malloc (8); s = "1234567"; _8 = MEM <long unsigned int> [(char * {ref-all})&s]; MEM <long unsigned int> [(char * {ref-all})p_4] = _8; s ={v} {CLOBBER}; return p_4; } ;; Function g (g, funcdef_no=1, decl_uid=1939, cgraph_uid=2, symbol_order=1) g () { struct S * p; char[3] * _1; <bb 2> [local count: 1073741824]: p_4 = __builtin_malloc (8); _1 = &p_4->a; __builtin_memset (_1, 0, 8); return p_4; } z.c: In function ‘g’: z.c:14:3: warning: ‘__builtin_memset’ writing 8 bytes into a region of size 3 overflows the destination [-Wstringop-overflow=] 14 | __builtin_memset (p->a, 0, 8); // warning not expected (yet) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~ z.c:1:17: note: destination object ‘a’ 1 | struct S { char a[3], b[5]; }; | ^
next reply other threads:[~2020-06-13 23:36 UTC|newest] Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-06-13 23:36 msebor at gcc dot gnu.org [this message] 2020-06-13 23:38 ` [Bug middle-end/95667] [11 Regression] unintended " msebor at gcc dot gnu.org 2020-06-15 6:49 ` rguenth at gcc dot gnu.org 2020-06-18 15:32 ` msebor at gcc dot gnu.org 2020-06-18 18:02 ` cvs-commit at gcc dot gnu.org 2020-08-20 16:37 ` msebor 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-95667-4@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).