First Last Prev Next    No search results available
Details
: [llvm-gcc] miscompilation of 'X = Y = Z' with aggregate v...
Bug#: 221
: tools
: llvm-gcc
Status: RESOLVED
Resolution: FIXED
: All
: All
: 1.0
: P2
: normal
: 1.2

:
: miscompilation
:
:
  Show dependency tree - Show dependency graph
People
Reporter: Brian R. Gaeke <gaeke+bugs@uiuc.edu>
Assigned To: Chris Lattner <sabre@nondot.org>
:

Attachments
Apparently miscompiled portion of bison's parse-gram.c (505 bytes, text/plain)
2004-02-03 01:18, Brian R. Gaeke
Details


Note

You need to log in before you can comment on or make changes to this bug.

Related actions


Description:   Opened: 2004-02-03 01:13

    
------- Comment #1 From Brian R. Gaeke 2004-02-03 01:17:08 -------
Whoops... I hit enter at the "enter bug" screen instead of tab, thus making my
bug's description empty. Sigh.

I am going to attach the part of bison that seems to be getting miscompiled.
You get two different crashes depending on whether you use -Wa,-disable-opt
-Wl,-disable-opt or not, but that doesn't surprise me much if the input code is
itself broken. 


------- Comment #2 From Brian R. Gaeke 2004-02-03 01:18:08 -------
Created an attachment (id=72) [details]
Apparently miscompiled portion of bison's parse-gram.c

Preprocessed source.
------- Comment #3 From Chris Lattner 2004-02-03 01:23:05 -------
Cool, I'm looking into it, though I may not get to it until tommorow.

Thanks Brian!

-Chris
------- Comment #4 From Chris Lattner 2004-02-03 01:45:11 -------
Here's a reduced testcase for the miscompilation:

typedef struct {
    int A;
} boundary;

boundary X, Y;

void func(boundary FOO) {
  X = Y = FOO;
}

-Chris
------- Comment #5 From Chris Lattner 2004-02-03 02:01:11 -------
Fixed.  Testcase here: (SingleSource/Regression/C/2004-02-03-AggregateCopy.c)
http://mail.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20040202/011095.html

Patch here: 
http://mail.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20040202/011096.html

Thanks for tracking this down Brian!

-Chris
------- Comment #6 From Brian R. Gaeke 2004-02-08 02:33:20 -------
*** Bug 163 has been marked as a duplicate of this bug. ***

First Last Prev Next    No search results available