when does the system send SIGKLILL
in [Unix Programming]
|
Prev: Suggestions needed
Next: A tool that suggests optimized logic for a piece of code/module/function
From: cerr on 15 Jan 2010 18:55 On Jan 15, 1:32 pm, sc...(a)slp53.sl.home (Scott Lurndal) wrote: > cerr <ron.egg...(a)gmail.com> writes: > >On Jan 15, 9:12=A0am, David Schwartz <dav...(a)webmaster.com> wrote: > >> On Jan 15, 8:49=A0am, cerr <ron.egg...(a)gmail.com> wrote: > > >> > Exactly that's where one of the problems was: > >> > fre(var); > >> > without > >> > var=3DNULL; > >> > which leads to a stale pointer that still has a "seemingly" valid > >> > value but is wrong! > >> > Now I'm off to huntting the next seg fault... :o - that I - > >> > unfortunately - don't really know where it's happening cause gdb only > >> > reports the SIGKILL but with strace I see that a SIGSEGV is leading to > >> > SIGKILLing the other threads.... no idea why i can't see the SIGSEGV > >> > in gdb... :( > > >> Catch the SIGSEGV and call 'abort'. > >Huh, IO don't understand. Where would I call abort(). > > #include <assert.h> > #include <signal.h> > #include <stdlib.h> > > void > catch_sigsegv(int signo, siginfo_t *sip, void *ucp) > { > abort(); > > } > > main(...) > > ... > > struct sigaction sa; > stack_t ss; > > ss.ss_sp = malloc(SIGSTKSZ); > assert(ss.ss_sp != NULL); > ss.ss_size = SIGSTKSZ; > ss.ss_flags = 0; > diag = sigaltstack(&ss, NULL); > if (diag == -1) { > //handle error > } > > sigemptyset(&sa.sa_mask); > sa.sa_flags = SA_ONSTACK; > sa.sigaction = catch_sigsegv; > diag = sigaction(SIGSEGV, &sa, NULL); > if (diag == -1) { > // handle error > } > Hi Scott, Okay, thanks for that, but what am i doing here with that? I thought I'd need to use signal() to trigger action like calling a function upon reception of such, no? How do I read the stack from the thread the SIGSEGV happened in after it happened? Doesn't the stack of that thread get ereased immediately? Also how can i read where it happened/which function the SIGSEGV happened in? Thanks, Ron
From: Scott Lurndal on 15 Jan 2010 19:08 cerr <ron.eggler(a)gmail.com> writes: >On Jan 15, 1:32=A0pm, sc...(a)slp53.sl.home (Scott Lurndal) wrote: >> #include <assert.h> >> #include <signal.h> >> #include <stdlib.h> >> >> void >> catch_sigsegv(int signo, siginfo_t *sip, void *ucp) >> { >> =A0 =A0 =A0 =A0 abort(); >> >> } >> >> main(...) >> >> =A0 ... >> >> =A0 struct sigaction sa; >> =A0 stack_t =A0 =A0 =A0 =A0 =A0ss; >> >> =A0 ss.ss_sp =3D malloc(SIGSTKSZ); >> =A0 assert(ss.ss_sp !=3D NULL); >> =A0 ss.ss_size =3D SIGSTKSZ; >> =A0 ss.ss_flags =3D 0; >> =A0 diag =3D sigaltstack(&ss, NULL); >> =A0 if (diag =3D=3D -1) { >> =A0 =A0 =A0//handle error >> =A0 } >> >> =A0 sigemptyset(&sa.sa_mask); >> =A0 sa.sa_flags =3D SA_ONSTACK; >> =A0 sa.sigaction =3D catch_sigsegv; >> =A0 diag =3D sigaction(SIGSEGV, &sa, NULL); >> =A0 if (diag =3D=3D -1) { >> =A0 =A0 =A0// handle error >> =A0 } >> >Hi Scott, > >Okay, thanks for that, but what am i doing here with that? I thought >I'd need to use signal() to trigger action like calling a function signal(2) is deprecated in favor of sigaction(2) (as used above). >upon reception of such, no? >How do I read the stack from the thread the SIGSEGV happened in after In the handler, you can: void catch_sigsegv(int signo, siginfo_t *sip, void *ucp) { ucontext_t *uconp = (ucontext_t *)ucp; /* Instruction pointer at failure (x86_64) */ printf("failure at 0x%lx\n", uconp->uc_mcontext.gregs[REG_RIP]); /* For i386 */ printf("failure at 0x%lx\n", uconp->uc_mcontext.gregs[REG_EIP]); /* for other architectures, see /usr/include/sys/ucontext.h */ } Take the address printed above and feed it into the addr2line command along with your executable, and if you're compiled with -g, addr2line will tell you what line in what source file corresponds to the instruction pointer at the time of the SIGSEGV. scott
From: Ersek, Laszlo on 15 Jan 2010 20:18 In article <x954n.11461$0X.9105(a)news.usenetserver.com>, scott(a)slp53.sl.home (Scott Lurndal) writes: > sa.sa_flags = SA_ONSTACK; > sa.sigaction = catch_sigsegv; I'd risk sa.sa_flags = SA_ONSTACK | SA_SIGINFO; sa.sa_sigaction = catch_sigsegv; Cheers, lacos
From: Scott Lurndal on 15 Jan 2010 21:04 lacos(a)ludens.elte.hu (Ersek, Laszlo) writes: >In article <x954n.11461$0X.9105(a)news.usenetserver.com>, scott(a)slp53.sl.home (Scott Lurndal) writes: > >> sa.sa_flags = SA_ONSTACK; >> sa.sigaction = catch_sigsegv; > >I'd risk > > sa.sa_flags = SA_ONSTACK | SA_SIGINFO; > sa.sa_sigaction = catch_sigsegv; > >Cheers, >lacos yes, good point (although the sa_sigaction and sa_handler fields are usually in a union, and it is ok for the implementation to pass the siginfo_t and ucontext_t pointers even when SA_SIGINFO isn't set. scott
From: cerr on 15 Jan 2010 22:19
On Jan 15, 5:18 pm, la...(a)ludens.elte.hu (Ersek, Laszlo) wrote: > In article <x954n.11461$0X.9...(a)news.usenetserver.com>, sc...(a)slp53.sl.home (Scott Lurndal) writes: > > sa.sa_flags = SA_ONSTACK; > > sa.sigaction = catch_sigsegv; > > I'd risk > > sa.sa_flags = SA_ONSTACK | SA_SIGINFO; > sa.sa_sigaction = catch_sigsegv; okay, I'll try this asap - maybe even tonight - depending on how my evening develops ;) However, why are we not doing a signal(SIGSEGV, catch_sigsegv); to catch the signal? Is that cause I'll be missing the stack which I guess I'll get to keep with this: ss.ss_sp = malloc(SIGSTKSZ); assert(ss.ss_sp != NULL); ss.ss_size = SIGSTKSZ; ss.ss_flags = 0; diag = sigaltstack(&ss, NULL); ??? -- roN |