Messages in this thread | | | Date | Tue, 28 Oct 2008 15:20:35 -0500 | From | "Michael Kerrisk" <> | Subject | Re: [PATCH] cleanup paccept |
| |
>> The actual patch which is in the kernel is different from >> the patch in the mail: only the signal mask handling has been disabled. >> This is why this is only a cleanup patch. > > And that does not explain to the world at large what this > to-be-enabled system call does, or why it's needed.
And one further point: what is the userspace interface going to look like for this system call?
Are we to see a repeat of the unsafe userspace implementation of pselect() that was present in glibc before the kernel finally added a pselect() system call?
In other words: is the userspace interface for accept4() going to be the thinnest of wrappers, or will glibc add an unsafe paccept()-with-signal-mask in userspace?
| |