On 27/09/11 03:42, Matt Turner wrote: > On Mon, Sep 26, 2011 at 3:41 AM, Michael Cree <[email protected]> wrote: >> And glibc source has a comment in the file >> ./sysdeps/unix/sysv/linux/alpha/kernel-features.h that the accept4 >> syscall is in the Alpha kernel since version 2.6.33-rc1. The author of >> that comment (who will know who he is) was sadly misinformed. > > Strange. I'm looking through git for some indication of why I'd have > thought that, and I can't find anything.
Possibly because you did the same thing I did and noted the absence of any output from the checksyscalls script (which is run at the start of the kernel build) and thought that indicated we were completely up to date with syscall implementations. I now see that the checksyscalls script compares the syscalls of an arch against those of x86 32bit syscalls. If the syscall is not implemented in x86 32bit (which is the case for accept4 as it is provided via a different syscall on x86 32bit) then no warning is generated if it is missing! Cheers Michael. -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected] Archive: http://lists.debian.org/[email protected]
