summaryrefslogtreecommitdiff
path: root/lib/dup2.c
diff options
context:
space:
mode:
authorEric Blake <eblake@redhat.com>2013-09-26 07:07:07 -0600
committerEric Blake <eblake@redhat.com>2013-09-26 13:16:04 -0600
commitfdd1b511c5c12912aae79b58e0c5f6ffa0492d6e (patch)
tree0e969daf7b52ec17b51d962720a5ec66fc9387aa /lib/dup2.c
parent3bd0d48db800994cc10c3340c41d3242390dc68f (diff)
downloadgnulib-fdd1b511c5c12912aae79b58e0c5f6ffa0492d6e.tar.gz
dup2, dup3: work around another cygwin crasher
Cygwin 1.7.25 can crash due to an off-by-one bug on an attempt to duplicate a file into the current RLIMIT_NOFILE soft limit, when that limit is smaller than the hard limit. The intent in the cygwin code was to allow the dup and auto-increase the soft limit, which is itself questionable (and which we work around in the gnulib getdtablesize module); but avoiding the crash is worth doing even if the soft limit semantics are wrong. http://cygwin.com/ml/cygwin/2013-09/msg00397.html http://cygwin.com/ml/cygwin-developers/2013-q3/msg00010.html * m4/dup2.m4 (gl_FUNC_DUP2): Expose the bug. * m4/dup3.m4 (gl_FUNC_DUP3): Likewise. * tests/test-dup2.c (main): Likewise. * lib/dup2.c (rpl_dup2): Use setdtablesize to avoid it. * lib/dup3.c (dup3): Likewise. * doc/posix-functions/dup2.texi (dup2): Document it. * doc/glibc-functions/dup3.texi (dup3): Likewise. Signed-off-by: Eric Blake <eblake@redhat.com>
Diffstat (limited to 'lib/dup2.c')
-rw-r--r--lib/dup2.c4
1 files changed, 4 insertions, 0 deletions
diff --git a/lib/dup2.c b/lib/dup2.c
index 9219eb3823..f128e7a63c 100644
--- a/lib/dup2.c
+++ b/lib/dup2.c
@@ -96,7 +96,11 @@ rpl_dup2 (int fd, int desired_fd)
/* On Linux kernels 2.6.26-2.6.29, dup2 (fd, fd) returns -EBADF.
On Cygwin 1.5.x, dup2 (1, 1) returns 0.
On Cygwin 1.7.17, dup2 (1, -1) dumps core.
+ On Cygwin 1.7.25, dup2 (1, 256) can dump core.
On Haiku, dup2 (fd, fd) mistakenly clears FD_CLOEXEC. */
+# if HAVE_SETDTABLESIZE
+ setdtablesize (desired_fd + 1);
+# endif
if (desired_fd < 0)
fd = desired_fd;
if (fd == desired_fd)