From: Garrett C. <yan...@gm...> - 2010-01-26 08:25:52
|
On Mon, Jan 25, 2010 at 11:54 PM, Garrett Cooper <yan...@gm...> wrote: > On Mon, Jan 25, 2010 at 9:45 PM, Shi Weihua <sh...@cn...> wrote: >> Hi, Cooper >> >> I noticed that you added a check of platform which not defined >> sa_sigaction in struct sigaction. >> http://ltp.cvs.sourceforge.net/viewvc/ltp/ltp/testcases/kernel/syscalls/rt_sigaction/rt_sigaction01.c?r1=1.6&r2=1.7 >> >> As your expectted, the following message will output on x86_64. >> ------------ >> Your architecture doesn't support this test (no sa_sigaction field in struct sigaction). >> ------------ >> But unfortunately, it still failed on my x86_64. >> >> On my x86_64, signal.h like this: >> ------------ >> #ifndef __ASSEMBLY__ >> >> #ifdef __i386__ >> (sa_sigaction defined here) <-- >> #else /* __i386__ */ >> (no sa_sigaction defined) <-- >> #endif /* !__i386__ */ >> >> #endif /* __ASSEMBLY__ */ >> ------------ >> your check still work ? Sorry, I don't know the check process in >> LTP configure clearly. Could you give me some explanation. > > Shi, > I'm just going to test out the other proposed rt_sigaction, and > revert this change if it doesn't work back to a semi-functioning state > (which means that x86_64 will once again be broken). Files reverted because liubo's patch didn't stick. Thanks, -Garrett |