Message179029
> Here's the patch.
_set_cloexec() is still called whereas it is useless if the OS supports O_CLOEXEC... But the call must be kept because Linux < 2.6.23 just ignores O_CLOEXEC: we would have to check _fcntl.fcntl(fd, _fcntl.F_GETFD, 0) & _fcntl.FD_CLOEXEC to check if the kernel does really support O_CLOEXEC, which is overkill. The possibly useless syscall doesn't hurt.
> (if the file is created with O_CREAT|O_EXCL, then by definition it's not a symlink).
Ah yes, because of O_EXCL.
The patch looks good to me! |
|
Date |
User |
Action |
Args |
2013-01-04 14:35:46 | vstinner | set | recipients:
+ vstinner, neologix |
2013-01-04 14:35:46 | vstinner | set | messageid: <1357310146.03.0.758048650169.issue16860@psf.upfronthosting.co.za> |
2013-01-04 14:35:46 | vstinner | link | issue16860 messages |
2013-01-04 14:35:45 | vstinner | create | |
|