public inbox for gcc-prs@sourceware.org help / color / mirror / Atom feed
From: charles@Provis.com To: gcc-gnats@gcc.gnu.org Cc: sms@provis.com Subject: optimization/6377: Bitfield and Optimizer 2.96 and later, multi-platform Date: Fri, 19 Apr 2002 11:36:00 -0000 [thread overview] Message-ID: <200204191830.g3JIU2C08709@dog.provis.com> (raw) >Number: 6377 >Category: optimization >Synopsis: Optimization (>= -O2) with bitfields yields incorrect result >Confidential: no >Severity: serious >Priority: medium >Responsible: unassigned >State: open >Class: wrong-code >Submitter-Id: net >Arrival-Date: Fri Apr 19 11:36:07 PDT 2002 >Closed-Date: >Last-Modified: >Originator: Charles Rennolet >Release: 3.0.3 >Organization: Provis Corporation >Environment: System: SunOS dog 5.8 Generic_108528-12 sun4u sparc SUNW,Sun-Blade-100 Architecture: sun4 host: sparc-sun-solaris2.8 build: sparc-sun-solaris2.8 target: sparc-sun-solaris2.8 configured with: ../gcc-3.0.3/configure >Description: gcc versions 2.96 and later have a problem with bitfields and optimization. There appears to be a mismatch between where the results are left and where they should go, especially when there is an intermediate cast. The result is that the bitfield structure is left with random data in it. This happens on Solaris (with 3.0.3) and Linux (with 2.96, 3.0.3, and 3.0.4) using optimization levels -O2 or higher. >How-To-Repeat: /* To Repeat: save this example program (self-contained except for the printf at the bottom) and compile with -O3. The result should contain nonzero bits, but doesn't. (The problem is that the result in the getFoo is being put in the stack frame, while the calling program expects the result in edx:eax.) Charles Rennolet (charles@provis.com, clr@thurse.mn.org) Example: (A very short one.) */ typedef struct { unsigned int device : 28; unsigned int stuckat : 2; unsigned int direction: 2; unsigned int pin : 12; unsigned int port : 12; unsigned int status : 4; unsigned int type : 4; } *pFoo; unsigned long long getFoo(int type, int device, int port, int pin, int InOut, int stuckat) { unsigned long long retValue = 0; ((pFoo)&retValue)->status = 0; ((pFoo)&retValue)->type = type; ((pFoo)&retValue)->device = device; ((pFoo)&retValue)->port = port; ((pFoo)&retValue)->pin = pin; ((pFoo)&retValue)->direction = InOut; ((pFoo)&retValue)->stuckat = stuckat; return(retValue); } int main( int argc, char **argv) { unsigned long long foo; unsigned int one = 1; unsigned int two = 2; unsigned int three = 3; unsigned int four = 4; unsigned int five = 5; unsigned int six = 6; unsigned int seven = 7; unsigned int tmp; foo = getFoo( one, two, three, four, five, six); printf("foo = %016llx\n", foo); } /* ============================Example Ends============================= */ Preprocessed input file: # 22 "foo.c" typedef struct { unsigned int device : 28; unsigned int stuckat : 2; unsigned int direction: 2; unsigned int pin : 12; unsigned int port : 12; unsigned int status : 4; unsigned int type : 4; } *pFoo; unsigned long long getFoo(int type, int device, int port, int pin, int InOut, int stuckat) { unsigned long long retValue = 0; ((pFoo)&retValue)->status = 0; ((pFoo)&retValue)->type = type; ((pFoo)&retValue)->device = device; ((pFoo)&retValue)->port = port; ((pFoo)&retValue)->pin = pin; ((pFoo)&retValue)->direction = InOut; ((pFoo)&retValue)->stuckat = stuckat; return(retValue); } int main( int argc, char **argv) { unsigned long long foo; unsigned int one = 1; unsigned int two = 2; unsigned int three = 3; unsigned int four = 4; unsigned int five = 5; unsigned int six = 6; unsigned int seven = 7; unsigned int tmp; foo = getFoo( one, two, three, four, five, six); printf("foo = %016llx\n", foo); } >Fix: Don't compile with -O2 or greater? Yeah, I know it's lame. >Release-Note: >Audit-Trail: >Unformatted:
reply other threads:[~2002-04-19 18:36 UTC|newest] Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
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=200204191830.g3JIU2C08709@dog.provis.com \ --to=charles@provis.com \ --cc=gcc-gnats@gcc.gnu.org \ --cc=sms@provis.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: 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).