2016-07-21 10:16 GMT-07:00 Christos Zoulas <chris...@astron.com>: > In article < > ca+sxe9uj7lg40xyt0g1v-cznxwkrnk79t0xcd4d2-qdlrma...@mail.gmail.com>, > Charles Cui <charles.cui1...@gmail.com> wrote: > > >I did not quite understand it. > > > https://github.com/ycui1984/posixtestsuite/blob/master/patches/REALTIME_SIGNAL/0004-improve-error-checking.patch > >This patch removes all ret = -1 logics, and set error at kpsignal2, > >did you want to set error code inside sigput? > > Yes sigput should return EAGAIN, the ret variable should be called error, > and all errors from sigput() should goto discard (unless there is some > cleanup that needs to be done). > Got it, I will fix it.
> > >The handling order of realtime signal and standard signals are not defined > >by POSIX. > >Some OSes handle realtime signal in FIFO order, but some select the > >realtime signal with lowest number first. > >It's still good to know the real results, though. > > Yes, and it is good to verify that the ordering for realtime signals is > correct too. > OK, I can write a program as you designed. Right now I have netbsd and Linux environments, and can verify the ordering in these platforms. > > christos > >
