Re: [libvirt] [PATCH 6/6] Dump the debug buffer to libvirtd.log on fatal signal

On Thu, Mar 03, 2011 at 06:22:17PM +0800, Daniel Veillard wrote: > > In case of imminent crash or upon request (signal USR2), > dump the logging buffer to the libvirtd.log file for > post-mortem analysis > * daemon/libvirtd.c: create a sig_fatal() handler connected to > SIGFPE SIGSEGV SIGILL SIGABRT SIGBUS and SIGUSR2, just dumping > the log buffer to the libvirtd.log open file descriptor

> + /* > + * If the signal is fatal, avoid looping over this handler > + * by desactivating it > + */ > + if (sig != SIGUSR2) { > + sig_action.sa_flags = SA_SIGINFO; > + sig_action.sa_handler = SIG_IGN; > + sigaction(sig, &sig_action, NULL); > + } > + errno = origerrno; > +} I think this code should really be part of the logging.c file. eg create an API like: void virLogEmergencyDumpAll(int signum) And have that dump the cached log buffer to all registered log outputs. We'd have to excluding any syslog one which isn't async signal safe, but the 'stderr' or 'file' log outputs can be made async signal safe.

Then, the libvirtd signal handler would just be static void sig_fatal(int sig, siginfo_t * siginfo ATTRIBUTE_UNUSED, void* context ATTRIBUTE_UNUSED) { virLogEmergencyDumpAll(sig);