Date: Tue, 4 Feb 2020 12:23:33 -0800 From: Conrad Meyer <cem@freebsd.org> To: Dmitry Chagin <dchagin@freebsd.org> Cc: src-committers <src-committers@freebsd.org>, svn-src-all <svn-src-all@freebsd.org>, svn-src-head <svn-src-head@freebsd.org> Subject: Re: svn commit: r357492 - in head/sys: kern sys Message-ID: <CAG6CVpVkdw5t=FYozt7-Q8trRd_0gK7tn0VXdUn-O4t28786fw@mail.gmail.com> In-Reply-To: <202002040525.0145Pppn034466@repo.freebsd.org> References: <202002040525.0145Pppn034466@repo.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Dmitry, This seems to trigger some panics via clock_gettime syscalls (native and 32-bit, reported by Syzkaller): panic: mutex process lock not owned at /syzkaller/managers/main/kernel/sys/kern/kern_time.c:261 cpuid = 0 time = 1580841963 KDB: stack backtrace: db_trace_self_wrapper() at db_trace_self_wrapper+0x47/frame 0xfffffe00244778c0 vpanic() at vpanic+0x1ce/frame 0xfffffe0024477930 panic() at panic+0x43/frame 0xfffffe0024477990 __mtx_assert() at __mtx_assert+0x196/frame 0xfffffe00244779d0 kern_thread_cputime() at kern_thread_cputime+0xaa/frame 0xfffffe0024477a20 kern_clock_gettime() at kern_clock_gettime+0x277/frame 0xfffffe0024477a80 sys_clock_gettime() at sys_clock_gettime+0x25/frame 0xfffffe0024477ab0 amd64_syscall() at amd64_syscall+0x499/frame 0xfffffe0024477bf0 panic: mutex process lock not owned at /syzkaller/managers/i386/kernel/sys/kern/kern_time.c:261 cpuid = 0 time = 1580847200 KDB: stack backtrace: db_trace_self_wrapper() at db_trace_self_wrapper+0x47/frame 0xfffffe00244a68b0 vpanic() at vpanic+0x1ce/frame 0xfffffe00244a6920 panic() at panic+0x43/frame 0xfffffe00244a6980 __mtx_assert() at __mtx_assert+0x196/frame 0xfffffe00244a69c0 kern_thread_cputime() at kern_thread_cputime+0xaa/frame 0xfffffe00244a6a10 kern_clock_gettime() at kern_clock_gettime+0x277/frame 0xfffffe00244a6a70 freebsd32_clock_gettime() at freebsd32_clock_gettime+0x25/frame 0xfffffe00244a6ab0 ia32_syscall() at ia32_syscall+0x48c/frame 0xfffffe00244a6bf0 Best, Conrad On Mon, Feb 3, 2020 at 9:25 PM Dmitry Chagin <dchagin@freebsd.org> wrote: > > Author: dchagin > Date: Tue Feb 4 05:25:51 2020 > New Revision: 357492 > URL: https://svnweb.freebsd.org/changeset/base/357492 > > Log: > For code reuse in Linuxulator rename get_proccess_cputime() > and get_thread_cputime() and add prototypes for it to <sys/syscallsubr.h>. > > As both functions become a public interface add process lock assert > to ensure that the process is not exiting under it. > > Fix whitespace nit while here. > > Reviewed by: kib > Differential Revision: https://reviews.freebsd.org/D23340 > MFC after 2 weeks > > Modified: > head/sys/kern/kern_time.c > head/sys/sys/syscallsubr.h > > Modified: head/sys/kern/kern_time.c > ============================================================================== > --- head/sys/kern/kern_time.c Tue Feb 4 05:23:34 2020 (r357491) > +++ head/sys/kern/kern_time.c Tue Feb 4 05:25:51 2020 (r357492) > @@ -242,7 +242,7 @@ sys_clock_gettime(struct thread *td, struct clock_gett > return (error); > } > > -static inline void > +static inline void > cputick2timespec(uint64_t runtime, struct timespec *ats) > { > runtime = cputick2usec(runtime); > @@ -250,12 +250,15 @@ cputick2timespec(uint64_t runtime, struct timespec *at > ats->tv_nsec = runtime % 1000000 * 1000; > } > > -static void > -get_thread_cputime(struct thread *targettd, struct timespec *ats) > +void > +kern_thread_cputime(struct thread *targettd, struct timespec *ats) > { > uint64_t runtime, curtime, switchtime; > + struct proc *p; > > if (targettd == NULL) { /* current thread */ > + p = curthread->td_proc; > + PROC_LOCK_ASSERT(p, MA_OWNED); > critical_enter(); > switchtime = PCPU_GET(switchtime); > curtime = cpu_ticks(); > @@ -263,6 +266,8 @@ get_thread_cputime(struct thread *targettd, struct tim > critical_exit(); > runtime += curtime - switchtime; > } else { > + p = targettd->td_proc; > + PROC_LOCK_ASSERT(p, MA_OWNED); > thread_lock(targettd); > runtime = targettd->td_runtime; > thread_unlock(targettd); > @@ -270,12 +275,13 @@ get_thread_cputime(struct thread *targettd, struct tim > cputick2timespec(runtime, ats); > } > > -static void > -get_process_cputime(struct proc *targetp, struct timespec *ats) > +void > +kern_process_cputime(struct proc *targetp, struct timespec *ats) > { > uint64_t runtime; > struct rusage ru; > > + PROC_LOCK_ASSERT(targetp, MA_OWNED); > PROC_STATLOCK(targetp); > rufetch(targetp, &ru); > runtime = targetp->p_rux.rux_runtime; > @@ -300,14 +306,14 @@ get_cputime(struct thread *td, clockid_t clock_id, str > td2 = tdfind(tid, p->p_pid); > if (td2 == NULL) > return (EINVAL); > - get_thread_cputime(td2, ats); > + kern_thread_cputime(td2, ats); > PROC_UNLOCK(td2->td_proc); > } else { > pid = clock_id & CPUCLOCK_ID_MASK; > error = pget(pid, PGET_CANSEE, &p2); > if (error != 0) > return (EINVAL); > - get_process_cputime(p2, ats); > + kern_process_cputime(p2, ats); > PROC_UNLOCK(p2); > } > return (0); > @@ -360,11 +366,11 @@ kern_clock_gettime(struct thread *td, clockid_t clock_ > ats->tv_nsec = 0; > break; > case CLOCK_THREAD_CPUTIME_ID: > - get_thread_cputime(NULL, ats); > + kern_thread_cputime(NULL, ats); > break; > case CLOCK_PROCESS_CPUTIME_ID: > PROC_LOCK(p); > - get_process_cputime(p, ats); > + kern_process_cputime(p, ats); > PROC_UNLOCK(p); > break; > default: > > Modified: head/sys/sys/syscallsubr.h > ============================================================================== > --- head/sys/sys/syscallsubr.h Tue Feb 4 05:23:34 2020 (r357491) > +++ head/sys/sys/syscallsubr.h Tue Feb 4 05:25:51 2020 (r357492) > @@ -91,6 +91,8 @@ int kern_clock_nanosleep(struct thread *td, clockid_t > const struct timespec *rqtp, struct timespec *rmtp); > int kern_clock_settime(struct thread *td, clockid_t clock_id, > struct timespec *ats); > +void kern_thread_cputime(struct thread *targettd, struct timespec *ats); > +void kern_process_cputime(struct proc *targetp, struct timespec *ats); > int kern_close(struct thread *td, int fd); > int kern_connectat(struct thread *td, int dirfd, int fd, > struct sockaddr *sa);
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAG6CVpVkdw5t=FYozt7-Q8trRd_0gK7tn0VXdUn-O4t28786fw>