On Sat, Nov 27, 2010 at 12:05:43PM +0100, Maciek Borzecki wrote: >I was hunting down for a nasty bug with a messed up >SIGCHDL/waitpid/pselect sequence and it seems that the signal may be >lost, with select not being interrupted as expected. It was suspected >that the signal may be somehow lost, what made me look at the >pselect() implementation in uClibc (version I used was 0.9.29 on a >XScale host with 2.6.21 & 2.6.29 kernels). >Can someone confirm that for the actual pselect() implementation >(libc/sysdeps/linux/common/pselect.c) does not use the pselect syscall >(__NR_pselect6)? Also when looking at the implementation of select >(libc/sysdeps/linux/common/select.c), a symbol __libc_pselect6 (which >does the syscall) is defined just for use inside select(). > >Is this intentional, or was that just overlooked? I'd be happy to >provide a patch if needed.
Sounds like it was just overlooked. It would be great if you could provide a patch. cheers, _______________________________________________ uClibc mailing list [email protected] http://lists.busybox.net/mailman/listinfo/uclibc
