diff options
| author | Tom Tromey <tromey@adacore.com> | 2020-09-08 10:20:44 -0600 |
|---|---|---|
| committer | Tom Tromey <tromey@adacore.com> | 2020-09-08 10:41:03 -0600 |
| commit | d706b69e48268ccf3e95fd29b5374ac94c3a507b (patch) | |
| tree | d16e0e4a5c4d18ee5c944eaaec0defe30e369a28 /gnulib/import/mbrtowc-impl.h | |
| parent | 03b0a45f15e44da7983dffea921c0c43a19250ff (diff) | |
| download | gdb-d706b69e48268ccf3e95fd29b5374ac94c3a507b.zip gdb-d706b69e48268ccf3e95fd29b5374ac94c3a507b.tar.gz gdb-d706b69e48268ccf3e95fd29b5374ac94c3a507b.tar.bz2 | |
Do not adjust mtime timezone on Windows
PR win32/25302 notes that gdb will crash when trying to "run" even a
simple program on Windows. The essential bug here is that the BFD
cache can easily be corrupted -- I have sent a separate patch for
that.
The particular reason that the cache is corrupted on Windows is that
gnulib overrides "stat" to make it do timezone adjustment -- but BFD
does not use this version of stat. The difference here triggers the
latent cache bug, but can also cause other bugs as well; in particular
it can cause spurious warnings about source files being newer.
This patch simply removes the stat override on mingw, making gnulib
and BFD agree.
I tested this by backing out the local AdaCore changes to work around
this bug and then verifying that I could reproduce it. Then, I
applied this patch and verified that "run" works again.
2020-09-08 Tom Tromey <tromey@adacore.com>
PR win32/25302:
* update-gnulib.sh: Apply stat patch.
* patches/0001-use-windows-stat: New file.
* import/m4/stat.m4: Update.
* configure: Rebuild.
Diffstat (limited to 'gnulib/import/mbrtowc-impl.h')
0 files changed, 0 insertions, 0 deletions