On Mon, Aug 18, 2008 at 01:55:20PM +0200, Christian Franke wrote:
>During testing of my packages on 1.7, I got the following error message
>from the grub-mkrescue script from package grub:
>
>$ cat boot.img /dev/zero | dd bs=1024 count=1440 > floppy.img
>cat: write error: No space left on device
>
>But due to a hack in coreutils*/lib/full_write.c, the error message is
>misleading.
>
>
>The following test program demonstrates the root of the problem:
>
>===== testpipe.c =====
>#include <stdio.h>
>#include <unistd.h>
>#include <signal.h>
>#include <errno.h>
>
>void sigpipe_handler()
>{
> write(2, "[SIGPIPE!]", 10);
>}
>
>int main()
>{
> static char buf[1024*1024];
> signal(SIGPIPE, sigpipe_handler);
> errno = 0;
> int n = write(1, buf, sizeof(buf));
> fprintf(stderr, "write = %d, errno = %d\n", n, errno);
> return 0;
>}
>======================
>
>
>Result on 1.5.25-15:
>
>$ ./testpipe | dd bs=1 count=1 >/dev/null 2>/dev/null
>[SIGPIPE!]write = -1, errno = 32
>
>Result on 1.7.0-28:
>
>$ ./testpipe | dd bs=1 count=1 >/dev/null 2>/dev/null
>write = 0, errno = 32
>
>
>According to strace, errno=EPIPE is set in
>fhandler_base::wait_overlapped() due to a failing GetOverlappedResult().
>
>Missing raise(SIGPIPE) call in fhandler_base::write_overlapped() ?
Thanks for the test case. I'll look at this when I get back to a
windows system. Right now my vmware workstation is broken and my vnc
access to my windows system is down. Quite irritating.
cgf
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
