Re: Panic when loading vboxdrv module
On 07/25/16 03:27 PM, Jung-uk Kim wrote: > On 07/25/16 03:13 PM, Randy Westlund wrote: >> I'm running 12-CURRENT r303286 and virtualbox-ose-kmod-5.0.26. I just >> rebuilt the module, so I know they're in sync. When I boot with >> vboxdrv_load="YES" in /etc/loader.conf, I get this panic during boot: >> >>> panic: mutex Giant owned at /usr/src/sys/kern/kern_sync.c:406 >>> cpuid = 0 >>> KDB: stack backtrace: >>> db_trace_self_wrapper() at db_trace_self_wrapper+0x2b >>> vpanic() at vpanic+0x182 >>> panic() at panic+0x43 >>> __mtx_assert() at __mtx_assert+0xd1 >>> mi_switch() at mi_switch+0x7b >>> sleepq_switch() at sleepq_switch+0x7e >>> sleepq_timedwait() at sleepq_timedwait+0x43 >>> rtR0SemEventWait() at rtR0SemEventWait+0x267 >>> supdrvGipCreate() at supdrvGipCreate+0x7f6 >>> supdrvInitDexExt() at supdrvInitDexExt+0x18f >>> VBoxDrvFreeBSDModuleEvent() at VBoxDrvFreeBSDModuleEvent+0x49 >>> module_register_init() at module_register_init+0xb0 >>> mi_startup() at mi_startup+0x118 >>> btext() at btext+0x2c >>> KDB: enter: panic >>> [ thread pid 0 tid 10 ] >>> Stopped at kdb_enter+0x3b: movq$0,kdb_why >> >> I'm not sure whether this is a problem with the base system or the port, >> but this happens every time I boot with the module enabled. > > emulators/virtualbox-ose-kmod is causing the trouble and this problem is > well known. To work around it, remove WITNESS option from kernel > configuration for now. Oops, I meant INVARIANTS. Anyway, it should be fixed in r419083. I understand you have a panic with aio(4) but that's a separate issue. Jung-uk Kim ___ freebsd-current@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"
Re: Panic when loading vboxdrv module
On Tue, Jul 26, 2016 at 09:22:47AM -0700, John Baldwin wrote: > On Monday, July 25, 2016 11:08:15 PM Randy Westlund wrote: > > On Mon, Jul 25, 2016 at 07:43:32PM -0400, Jung-uk Kim wrote: > > > Oops, I meant INVARIANTS. Anyway, it should be fixed in r419083. I > > > understand you have a panic with aio(4) but that's a separate issue. > > > > > > Jung-uk Kim > > > > Should I create a new PR for virtualbox and aio, or is this already > > known? > > Please. cc me on it since I've hacked on aio in 11/HEAD. > > -- > John Baldwin Done. https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=211390 signature.asc Description: PGP signature
Re: Panic when loading vboxdrv module
On Monday, July 25, 2016 11:08:15 PM Randy Westlund wrote: > On Mon, Jul 25, 2016 at 07:43:32PM -0400, Jung-uk Kim wrote: > > Oops, I meant INVARIANTS. Anyway, it should be fixed in r419083. I > > understand you have a panic with aio(4) but that's a separate issue. > > > > Jung-uk Kim > > Should I create a new PR for virtualbox and aio, or is this already > known? Please. cc me on it since I've hacked on aio in 11/HEAD. -- John Baldwin ___ freebsd-current@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"
Re: Panic when loading vboxdrv module
On Mon, Jul 25, 2016 at 07:43:32PM -0400, Jung-uk Kim wrote: > Oops, I meant INVARIANTS. Anyway, it should be fixed in r419083. I > understand you have a panic with aio(4) but that's a separate issue. > > Jung-uk Kim Should I create a new PR for virtualbox and aio, or is this already known? signature.asc Description: PGP signature
Re: Panic when loading vboxdrv module
On Mon, Jul 25, 2016 at 04:09:58PM -0400, Jung-uk Kim wrote: > If you feel adventurous, please drop the attached patch in > *emulators/virtualbox-ose/files* directory, and rebuild > *emulators/virtualbox-ose-kmod*. > > JK > > * Note the patch directory is emulators/virtualbox-ose/files, not > emulators/virtualbox-ose-kmod/files. > --- src/VBox/HostDrivers/Support/SUPDrvInternal.h.orig2016-07-18 > 11:56:19 UTC > +++ src/VBox/HostDrivers/Support/SUPDrvInternal.h > @@ -235,7 +235,7 @@ > # define SUPDRV_WITHOUT_MSR_PROBER > #endif > > -#if 1 > +#if 0 > /** @def SUPDRV_USE_TSC_DELTA_THREAD > * Use a dedicated kernel thread to service TSC-delta measurement requests. > * @todo Test on servers with many CPUs and sockets. */ With the patch, vboxdrv loads successfully. But when I start a VM, I get the following panic (after rebuilding virtualbox-ose as well): > vboxdrv: 82973020 VMMR0.r0 > vboxdrv: 82a92020 VBoxDDR0.r0 > vboxdrv: 82aaf020 VBoxDD2R0.r0 > panic: _mtx_lock_sleep: recursed on non-recursive mutex aiomtx @ > /usr/src/sys/kern/vfs_aio.c:996 > > cpuid = 1 > KBD: stack backtrace: > db_trace_wrapper() at db_trace_wrapper+0x2b > vpanic() at vpanic+0x182 > kassert_panic() at kassert_panic+0x126 > __mtx_lock_sleep() at __mtx_lock_sleep+0x228 > __mtx_lock_flags() at __mtx_lock_flags+0x10d > aio_queue() at aio_queue+0x9d6 > amd64_syscall() at amd64_syscall+0x2db > Xfast_syscall() at Xfast_syscall+0xfb > --- syscall (465, FreeBSD ELF64, sys_aio_fsync), rip = 0x80119fa0a, rsp = > 0x7fffdf2abc98, rbp = 0x7fffdf2abcd0 --- > KDB: enter: panic > [ thread pid 14588 tid 101813 ] > Stopped at kdb_enter+0x3b: movq$0,kdb_why signature.asc Description: PGP signature
Re: Panic when loading vboxdrv module
On 07/25/16 03:13 PM, Randy Westlund wrote: > I'm running 12-CURRENT r303286 and virtualbox-ose-kmod-5.0.26. I just > rebuilt the module, so I know they're in sync. When I boot with > vboxdrv_load="YES" in /etc/loader.conf, I get this panic during boot: > >> panic: mutex Giant owned at /usr/src/sys/kern/kern_sync.c:406 >> cpuid = 0 >> KDB: stack backtrace: >> db_trace_self_wrapper() at db_trace_self_wrapper+0x2b >> vpanic() at vpanic+0x182 >> panic() at panic+0x43 >> __mtx_assert() at __mtx_assert+0xd1 >> mi_switch() at mi_switch+0x7b >> sleepq_switch() at sleepq_switch+0x7e >> sleepq_timedwait() at sleepq_timedwait+0x43 >> rtR0SemEventWait() at rtR0SemEventWait+0x267 >> supdrvGipCreate() at supdrvGipCreate+0x7f6 >> supdrvInitDexExt() at supdrvInitDexExt+0x18f >> VBoxDrvFreeBSDModuleEvent() at VBoxDrvFreeBSDModuleEvent+0x49 >> module_register_init() at module_register_init+0xb0 >> mi_startup() at mi_startup+0x118 >> btext() at btext+0x2c >> KDB: enter: panic >> [ thread pid 0 tid 10 ] >> Stopped at kdb_enter+0x3b: movq$0,kdb_why > > I'm not sure whether this is a problem with the base system or the port, > but this happens every time I boot with the module enabled. emulators/virtualbox-ose-kmod is causing the trouble and this problem is well known. To work around it, remove WITNESS option from kernel configuration for now. Jung-uk Kim signature.asc Description: OpenPGP digital signature
