Igor Sysoev
2004-Oct-22 13:35 UTC
panic caused by EVFILT_SIGNAL detaching in rfork()ed thread
Here is more correct patch to fix the panic in 4.x reported in http://freebsd.rambler.ru/bsdmail/freebsd-hackers_2004/msg02732.html ------------------------- --- src/sys/kern/kern_event.c Sun Oct 10 12:17:55 2004 +++ src/sys/kern/kern_event.c Sun Oct 10 12:19:29 2004 @@ -794,7 +794,8 @@ while (kn != NULL) { kn0 = SLIST_NEXT(kn, kn_link); if (kq == kn->kn_kq) { - kn->kn_fop->f_detach(kn); + if (!(kn->kn_status & KN_DETACHED)) + kn->kn_fop->f_detach(kn); /* XXX non-fd release of kn->kn_ptr */ knote_free(kn); *knp = kn0; ------------------------- The patch based on the fix for FreeBSD 5.x: http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/kern/kern_event.c.diff?r1=1.79&r2=1.80 For more information see the thread started in http://freebsd.rambler.ru/bsdmail/freebsd-current_2004/msg18389.html To reproduce the panic in 4.x you could download http://sysoev.ru/nginx/nginx-0.1.2.tar.gz and build it without the installation: tar zxf nginx-0.1.2.tar.gz cd nginx-0.1.2 ./configure --with-threads=rfork \ --without-http_rewrite_module \ --prefix=$PWD \ --pid-path=nginx.pid \ --http-log-path=access.log \ --error-log-path=stderr make Now you have to edit ./conf/nginx.conf. Add the line "daemon off;" in the start of the file. Change the lines "worker_processes 3;" to "worker_threads 3;" and "listen 80;" to "listen 8000;". Now you could run ./nginx as non-privileged user. If you run in another console ps ax -o pid,ppid,%cpu,vsz,wchan,command|egrep '(nginx:|PID)' you will see something like this: PID PPID %CPU VSZ WCHAN COMMAND 3737 238 0.0 1340 pause nginx: master process ./nginx 3738 3737 0.0 1844 kqread nginx: worker process (nginx) 3739 3738 0.0 1844 kqread nginx: worker thread (nginx) 3740 3738 0.0 1844 kqread nginx: worker thread (nginx) 3741 3738 0.0 1844 kqread nginx: worker thread (nginx) Now stop nginx by pressing ^C. Your system may panic. Igor Sysoev http://sysoev.ru/en/
Uwe Doering
2004-Oct-23 06:57 UTC
panic caused by EVFILT_SIGNAL detaching in rfork()ed thread
Igor Sysoev wrote:> Here is more correct patch to fix the panic in 4.x reported in > http://freebsd.rambler.ru/bsdmail/freebsd-hackers_2004/msg02732.html > > ------------------------- > --- src/sys/kern/kern_event.c Sun Oct 10 12:17:55 2004 > +++ src/sys/kern/kern_event.c Sun Oct 10 12:19:29 2004 > @@ -794,7 +794,8 @@ > while (kn != NULL) { > kn0 = SLIST_NEXT(kn, kn_link); > if (kq == kn->kn_kq) { > - kn->kn_fop->f_detach(kn); > + if (!(kn->kn_status & KN_DETACHED)) > + kn->kn_fop->f_detach(kn); > /* XXX non-fd release of kn->kn_ptr */ > knote_free(kn); > *knp = kn0; > -------------------------Your patch appears to be an excerpt from the fix to RELENG_5. May I suggest a different approach for RELENG_4? My reasoning is that the implementation of kevents differs between RELENG_4 and RELENG_5. In RELENG_5 the flag KN_DETACHED is used in a more general way. It gets set by knlist_add() and cleared by knlist_remove(), in sync with list insertion and removal. As far as I can tell these routines have originally been introduced in order to centralize the locking for kevent list manipulations, and they don't exist in RELENG_4. Now, the proper way to MFC the RELENG_5 fix to RELENG_4 more or less unchanged would be to MFC the whole knlist_add()/knlist_remove() business as well (w/o the locking stuff), which, however, would be overkill for RELENG_4's single threaded kernel. In RELENG_4's implementation of kevents, the only case in which KN_DETACHED gets set is when a process exits and posts a NOTE_EXIT event. That is, the meaning of KN_DETACHED is much narrower than in RELENG_5. For this reason I believe the most appropriate fix would be to check for KN_DETACHED in filt_sigdetach() in the same way it is already done in filt_procdetach(). In fact, if you compare the two routines it becomes pretty obvious that they should have been identical in the first place, and that the absence of said check from filt_sigdetach() is most likely just an oversight. Therefore, I suggest to adopt the attached patch and leave the rest of RELENG_4's kevent code alone. I checked the kernel sources and found that filt_procdetach() and filt_sigdetach() are in fact the only f_detach() routines that deal with a process' p_klist field, and therefore need this kind of safeguard. Also, it would probably be a good idea to fix RELENG_4 swiftly (and possibly release a security advisory) because this flaw is certainly a great DoS opportunity for maliciously minded shell users ... Uwe -- Uwe Doering | EscapeBox - Managed On-Demand UNIX Servers gemini@geminix.org | http://www.escapebox.net -------------- next part -------------- --- src/sys/kern/kern_sig.c.orig Thu Feb 5 23:26:48 2004 +++ src/sys/kern/kern_sig.c Sat Oct 23 11:15:30 2004 @@ -1739,6 +1739,10 @@ { struct proc *p = kn->kn_ptr.p_proc; + if (kn->kn_status & KN_DETACHED) + return; + + /* XXX locking? this might modify another process. */ SLIST_REMOVE(&p->p_klist, kn, knote, kn_selnext); }
Uwe Doering
2004-Oct-23 07:54 UTC
panic caused by EVFILT_SIGNAL detaching in rfork()ed thread
Uwe Doering wrote:> Igor Sysoev wrote: > >> Here is more correct patch to fix the panic in 4.x reported in >> http://freebsd.rambler.ru/bsdmail/freebsd-hackers_2004/msg02732.html >> >> ------------------------- >> --- src/sys/kern/kern_event.c Sun Oct 10 12:17:55 2004 >> +++ src/sys/kern/kern_event.c Sun Oct 10 12:19:29 2004 >> @@ -794,7 +794,8 @@ >> while (kn != NULL) { >> kn0 = SLIST_NEXT(kn, kn_link); >> if (kq == kn->kn_kq) { >> - kn->kn_fop->f_detach(kn); >> + if (!(kn->kn_status & KN_DETACHED)) >> + kn->kn_fop->f_detach(kn); >> /* XXX non-fd release of kn->kn_ptr */ >> knote_free(kn); >> *knp = kn0; >> ------------------------- > > > Your patch appears to be an excerpt from the fix to RELENG_5. May I > suggest a different approach for RELENG_4? My reasoning is that the > implementation of kevents differs between RELENG_4 and RELENG_5. > > In RELENG_5 the flag KN_DETACHED is used in a more general way. It gets > set by knlist_add() and cleared by knlist_remove(), in sync with list > insertion and removal. > [...]Slight correction: The logic is of course the other way round. That is, the text is supposed to read: It gets cleared by knlist_add() and set by knlist_remove() ... Sorry about that. :-) Uwe -- Uwe Doering | EscapeBox - Managed On-Demand UNIX Servers gemini@geminix.org | http://www.escapebox.net
Igor Sysoev
2004-Oct-23 09:56 UTC
panic caused by EVFILT_SIGNAL detaching in rfork()ed thread
On Sat, 23 Oct 2004, Uwe Doering wrote:> Igor Sysoev wrote: > > Here is more correct patch to fix the panic in 4.x reported in > > http://freebsd.rambler.ru/bsdmail/freebsd-hackers_2004/msg02732.html > > > > ------------------------- > > --- src/sys/kern/kern_event.c Sun Oct 10 12:17:55 2004 > > +++ src/sys/kern/kern_event.c Sun Oct 10 12:19:29 2004 > > @@ -794,7 +794,8 @@ > > while (kn != NULL) { > > kn0 = SLIST_NEXT(kn, kn_link); > > if (kq == kn->kn_kq) { > > - kn->kn_fop->f_detach(kn); > > + if (!(kn->kn_status & KN_DETACHED)) > > + kn->kn_fop->f_detach(kn); > > /* XXX non-fd release of kn->kn_ptr */ > > knote_free(kn); > > *knp = kn0; > > ------------------------- > > Your patch appears to be an excerpt from the fix to RELENG_5. May I > suggest a different approach for RELENG_4? My reasoning is that the > implementation of kevents differs between RELENG_4 and RELENG_5.I agree with your patch. It is similar to my early patch, however, I think that is more correctly to check (kn->kn_status & KN_DETACHED) instead of !SLIST_EMPTY(&p->p_klist).> Also, it would probably be a good idea to fix RELENG_4 swiftly (and > possibly release a security advisory) because this flaw is certainly a > great DoS opportunity for maliciously minded shell users ...Yes. Igor Sysoev http://sysoev.ru/en/