Date: Thu, 29 Jan 2009 14:21:28 GMT From: Robert Watson <rwatson@FreeBSD.org> To: Perforce Change Reviews <perforce@freebsd.org> Subject: PERFORCE change 156859 for review Message-ID: <200901291421.n0TELSAx080526@repoman.freebsd.org>
next in thread | raw e-mail | index | archive | help
http://perforce.freebsd.org/chv.cgi?CH=156859 Change 156859 by rwatson@rwatson_freebsd_capabilities on 2009/01/29 14:20:44 Move down the process descriptor exemption in the wait loop, and change the logic slightly: only exempt processes with process descriptors from visibility in wait4() if they aren't being debugged with ptrace(). Otherwise decidedly odd things happen when gdb fails to detach the process and exits, as the parent process becomes init while it still has a process descriptor. Affected files ... .. //depot/projects/trustedbsd/capabilities/src/sys/kern/kern_exit.c#12 edit Differences ... ==== //depot/projects/trustedbsd/capabilities/src/sys/kern/kern_exit.c#12 (text+ko) ==== @@ -872,15 +872,6 @@ } /* - * If a process has a process descriptor, then it won't be - * picked up by wait4(). - */ - if (p->p_procdesc != NULL) { - PROC_UNLOCK(p); - continue; - } - - /* * This special case handles a kthread spawned by linux_clone * (see linux_misc.c). The linux_wait4 and linux_waitpid * functions need to be able to distinguish between waiting @@ -894,6 +885,17 @@ continue; } + /* + * If a process has a process descriptor, then it won't be + * picked up by wait4(). Unless it's being debugged, in + * which case the debugging process will need to manage it + * with waitpid(). + */ + if (p->p_procdesc != NULL && p->p_oppid == 0) { + PROC_UNLOCK(p); + continue; + } + nfound++; PROC_SLOCK(p); if (p->p_state == PRS_ZOMBIE) {
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200901291421.n0TELSAx080526>