I found a couple of theoretical integer overflows in cat.c: it
misbehaves if you give it 2**32 options, and it screws up if a file
block size is greater than INT_MAX. While fixing these minor glitches
and doing the usual "use bool for values that are booleans" fixes I
noticed also that cat uses su
I noticed that dcgen generated C code with wider integers than it had
to, so I cleaned that up, and while I was at it I modified it to
remove comments, trailing white space, and empty lines.
2004-07-28 Paul Eggert <[EMAIL PROTECTED]>
* src/dcgen: Remove comments, trailing white space, a
This is related to the previous patch: it cleans up chown/chgrp
themselves. The only user-visible change should be that "chgrp 010
file" works correctly instead of misparsing the number as octal.
2004-07-28 Paul Eggert <[EMAIL PROTECTED]>
* src/chgrp.c (parse_group): Require base 10 wh
coreutils chown and chgrp mishandle some numeric uids and gids that
have leading zeros. For example, "chown 010 file" is treated as
"chown 8 file", where it should be like "chown 10 file". I discovered
this while auditing for integer arithmetic problems, and installed the
following patch to fix t
I installed these patches, to remove the assumption that sizeof (long)
must be either 4 or 8, and also to remove the assumption that a
pointer must be castable to an integer in a useful way.
2004-07-28 Paul Eggert <[EMAIL PROTECTED]>
* lib/memchr.c: Include , not and .
(LONG_MA
Continuing in our series of integer-arithmetic cleanups for coreutils,
here is a patch to make md5.c and sha1.c portable to hosts (like the
AS/400) which do not let you cast pointers to integers in any
meaningful way. This also modernizes the integer terminology a bit
(using C99 names) and fixes t
I installed this patch for a very unlikely problem: more than INT_MAX groups.
2004-07-28 Paul Eggert <[EMAIL PROTECTED]>
* getugroups.c: Include .
(EOVERFLOW): Define if not defined.
(getgroups): Return -1 with errno=EOVERFLOW if an integer overflow
occurs.
Inde
I installed this patch to fix a test-case bug encountered when running
"make test" under the Solaris automounter.
2004-07-28 Paul Eggert <[EMAIL PROTECTED]>
* tests/misc/pwd: Don't assume that Perl's getpwd agrees with our
pwd when there are multiple names for the working direct
"make check" failed on Solaris 8 sparc because Solaris 8 'make'
wouldn't build the groups file: it saw the line "groups: Makefile" and
decided that was enough. I also noticed a lot of needless rebuilding
of localedir.h, which I think I've seen on other platforms too: that
makefile rule is well-int
CVS coreutils rm failed "make check" due to the following scenario
in tests/rm/inaccessible:
can't get the working directory initially;
...
chdir ("/foo/bar/abs1");
... opendir, readdir, remove all entries here; then: ...
chdir ("."); // because we couldn't get the working directory
I installed this fix encountered when porting coreutils CVS to Solaris
8 + 32-bit GCC. This was a portability bug that I introduced in the
2004-07-06 change to printf.c; sorry about that.
2004-07-27 Paul Eggert <[EMAIL PROTECTED]>
* src/printf.c (strtiomax, strtoumax): Declare if not a
11 matches
Mail list logo