Hi Christos, Thanks for providing these information! I will revisit my current code base and think carefully how those two bugs are generated (by keeping your hints in mind). Hopefully I can find the correct place and realize my mistakes. :)
Thanks Charles 2016-07-31 5:54 GMT-07:00 Christos Zoulas <chris...@zoulas.com>: > On Jul 30, 11:42am, charles.cui1...@gmail.com (Charles Cui) wrote: > -- Subject: Re: updates? > > | Yes, it works for me, each time I send you a new patch, I will compile my > | kernel and userland, > | and testing using the program we have written. > | > | Here is a short video that I created from my environment which shows the > | real time signal works for me > | https://www.dropbox.com/s/amds3jmpfev8jz9/realtime_signal.mov?dl=0 > | > | Also, I looked at all files needed in the real time signal patches, > | and attached all files of them, you can replace using these files and see > | whether there are any > | differences with the current settings. > | > | /usr/src/sys/sys/signal.h > | > | /usr/src/include/limits.h > | > | /usr/src/lib/libc/gen/sysconf.c > | > | /usr/src/sys/kern/kern_sig.c > | > | /usr/src/sys/kern/sys_sig.c > | > | /usr/src/sys/sys/signalvar.h > | > | /usr/src/sys/sys/unistd.h > | > | 2016-07-30 9:36 GMT-07:00 Christos Zoulas <chris...@zoulas.com>: > > There are two issues with the code that I found and I have everything > working now. It is not enough to test with the test program (which > works); for example as I mentioned before, booting with a kernel > that has all your patches fails on sshd (you can't login remotely) > and an interrupted make does not work properly. If you want I can > show you what's wrong with the changes, but I think it is better > if you find the problems yourself. The idea here is to make sure > that the code works exactly like before in the absense of real-time > signals, which is not the case currently. Both bugs are in kern_sig.c. > One of them is the critical one that breaks signal delivery, the other > is more of a clarification of what happens when we want to maintain > the pending signal mask. > > Best, > > christos >
