Re: Insta-panic for amd64 on reboot after upgrade from r320307 -> r320324
25.06.2017 16:21, Konstantin Belousov пишет: > On Sun, Jun 25, 2017 at 06:05:21AM -0700, David Wolfskill wrote: >> On Sun, Jun 25, 2017 at 03:52:23PM +0300, Konstantin Belousov wrote: >>> ... > The layout of the struct vm_map_entry was changed, the faulted address > is somewhat consistent with ABI mismatch. Kinky. :-} >>> Do you use any third-party modules ? >> >> On the laptop, I use x11/nvidia-driver-340; on the build machine, no. >> >> I think we should focus on the build machine -- it runs a GENERIC kernel >> without ports (3rd-party) modules. > Ok. > >> >> #cat /etc/src-env.conf >> WITH_META_MODE=yes > > So can you _remove_ all kernel object files and rebuild anew with the > clean build dir, please ? I also use WITH_META_MODE=yes. And full rebuild helped here too. Thank you. -- WBR, bsam ___ 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: Insta-panic for amd64 on reboot after upgrade from r320307 -> r320324
On Sun, Jun 25, 2017 at 04:21:16PM +0300, Konstantin Belousov wrote: > ... > > #cat /etc/src-env.conf > > WITH_META_MODE=yes > > So can you _remove_ all kernel object files and rebuild anew with the > clean build dir, please ? OK; that seems to have resulted in a kernel that boots to multi-user mode without incident: FreeBSD freebeast.catwhisker.org 12.0-CURRENT FreeBSD 12.0-CURRENT #386 r320324M/320326:1200035: Sun Jun 25 06:36:19 PDT 2017 r...@freebeast.catwhisker.org:/common/S4/obj/usr/src/sys/GENERIC amd64 To remove any possible ambiguity, here is what I did. (Note that head is on slice 4; recent stable/11 is on slice 1; when booted from slice 1, slice 4's / is mounted on /S4.) * Rebooted from slice 1 (stable/11). * cd /S4/usr/obj * cp -p usr/src/sys/GENERIC/version /tmp * rm -fr usr/src/sys/GENERIC * mkdir !$ * mv /tmp/version !$/ * cd /S4/boot * cp -pr kernel.old kernel.save * mv kernel{,.panic} * mv kernel{.old,} * Rebooted from slice 4 (head@r320307) * setenv TMPDIR /tmp &&\ id &&\ mount &&\ cd /usr/src &&\ uname -a &&\ date &&\ make -j16 buildkernel &&\ date &&\ rm -fr /boot/modules.old &&\ cp -pr /boot/modules{,.old} &&\ make installkernel &&\ date * shutdown -r now * Sent this message. :-) Thanks! Peace, david -- David H. Wolfskill da...@catwhisker.org Trump (et al.): Hiding information doesn't prove its falsity. See http://www.catwhisker.org/~david/publickey.gpg for my public key. signature.asc Description: PGP signature
Re: Insta-panic for amd64 on reboot after upgrade from r320307 -> r320324
On Sun, Jun 25, 2017 at 06:05:21AM -0700, David Wolfskill wrote: > On Sun, Jun 25, 2017 at 03:52:23PM +0300, Konstantin Belousov wrote: > > ... > > > > The layout of the struct vm_map_entry was changed, the faulted address > > > > is somewhat consistent with ABI mismatch. > > > > > > Kinky. :-} > > Do you use any third-party modules ? > > On the laptop, I use x11/nvidia-driver-340; on the build machine, no. > > I think we should focus on the build machine -- it runs a GENERIC kernel > without ports (3rd-party) modules. Ok. > > #cat /etc/src-env.conf > WITH_META_MODE=yes So can you _remove_ all kernel object files and rebuild anew with the clean build dir, please ? ___ 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: Insta-panic for amd64 on reboot after upgrade from r320307 -> r320324
On Sun, Jun 25, 2017 at 03:52:23PM +0300, Konstantin Belousov wrote: > ... > > > The layout of the struct vm_map_entry was changed, the faulted address > > > is somewhat consistent with ABI mismatch. > > > > Kinky. :-} > Do you use any third-party modules ? On the laptop, I use x11/nvidia-driver-340; on the build machine, no. I think we should focus on the build machine -- it runs a GENERIC kernel without ports (3rd-party) modules. Here are a few relevant files: # cat /etc/src.conf WITHOUT_DEBUG_FILES=1 WITH_ELFCOPY_AS_OBJCOPY=1 # #cat /etc/src-env.conf WITH_META_MODE=yes # #cat /etc/make.conf SENDMAIL_MC=/etc/mail/client.mc # added by use.perl 2009-11-07 21:19:31 PERL_VERSION=5.12.1 WITH_PKGNG= YES # #cat /boot/loader.conf console="comconsole,vidconsole" # A comma separated list of console(s) comconsole_pcidev="4:0:0" comconsole_speed="9600" filemon_load="YES" # Peace, david -- David H. Wolfskill da...@catwhisker.org Trump (et al.): Hiding information doesn't prove its falsity. See http://www.catwhisker.org/~david/publickey.gpg for my public key. signature.asc Description: PGP signature
Re: Insta-panic for amd64 on reboot after upgrade from r320307 -> r320324
On Sun, Jun 25, 2017 at 05:47:48AM -0700, David Wolfskill wrote: > On Sun, Jun 25, 2017 at 03:32:26PM +0300, Konstantin Belousov wrote: > > On Sun, Jun 25, 2017 at 05:07:31AM -0700, David Wolfskill wrote: > > > Fatal trap 12: page fault while in kernel mode > > > cpuid = 0; apic id = 00 > > > fault virtual address = 0x120 > > This is clearly an impossible address. > > > > Did you built the kernel with NO_CLEAN ? If yes, try the full build, > > perhaps even after removing all previous kernel objects. > > No; I stopped using NO_CLEAN by 12 March 2016, in favor of filemon and > WITH_FAST_DEPEND. I have been doing daily builds & smoke-tests of > head/amd64 since then. (Well, I was also doing them before then, > as well) > > > The layout of the struct vm_map_entry was changed, the faulted address > > is somewhat consistent with ABI mismatch. > > Kinky. :-} Do you use any third-party modules ? ___ 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: Insta-panic for amd64 on reboot after upgrade from r320307 -> r320324
On Sun, Jun 25, 2017 at 03:32:26PM +0300, Konstantin Belousov wrote: > On Sun, Jun 25, 2017 at 05:07:31AM -0700, David Wolfskill wrote: > > Fatal trap 12: page fault while in kernel mode > > cpuid = 0; apic id = 00 > > fault virtual address = 0x120 > This is clearly an impossible address. > > Did you built the kernel with NO_CLEAN ? If yes, try the full build, > perhaps even after removing all previous kernel objects. No; I stopped using NO_CLEAN by 12 March 2016, in favor of filemon and WITH_FAST_DEPEND. I have been doing daily builds & smoke-tests of head/amd64 since then. (Well, I was also doing them before then, as well) > The layout of the struct vm_map_entry was changed, the faulted address > is somewhat consistent with ABI mismatch. Kinky. :-} > Peace, david -- David H. Wolfskill da...@catwhisker.org Trump (et al.): Hiding information doesn't prove its falsity. See http://www.catwhisker.org/~david/publickey.gpg for my public key. signature.asc Description: PGP signature
Re: Insta-panic for amd64 on reboot after upgrade from r320307 -> r320324
On Sun, Jun 25, 2017 at 05:07:31AM -0700, David Wolfskill wrote: > Fatal trap 12: page fault while in kernel mode > cpuid = 0; apic id = 00 > fault virtual address = 0x120 This is clearly an impossible address. Did you built the kernel with NO_CLEAN ? If yes, try the full build, perhaps even after removing all previous kernel objects. The layout of the struct vm_map_entry was changed, the faulted address is somewhat consistent with ABI mismatch. > fault code = supervisor read data, page not present > instruction pointer = 0x20:0x80d585a4 > stack pointer = 0x28:0x82290a30 > frame pointer = 0x28:0x82290a60 > code segment= base 0x0, limit 0xf, type 0x1b > = DPL 0, pres 1, long 1, def32 0, gran 1 > processor eflags= resume, IOPL = 0 > current process = 0 () > [ thread pid 0 tid 0 ] > Stopped at vm_map_lookup_entry+0x24: cmpq%r15,0x20(%rbx) > db> bt > Tracing pid 0 tid 0 td 0x81e9e860 > vm_map_lookup_entry() at vm_map_lookup_entry+0x24/frame 0x82290a60 > vm_map_insert() at vm_map_insert+0x10b/frame 0x82290b00 > kmem_init() at kmem_init+0x72/frame 0x82290b30 > vm_mem_init() at vm_mem_init+0x46/frame 0x82290b50 > mi_startup() at mi_startup+0x9c/frame 0x82290b70 > btext() at btext+0x2c > db> > > > (The laptop seems to have repeated the: > > loading required module 'kernel' > module 'kernel' exists but with wrong version > > sequence a few times, based on the video I got of it.) > > > The "saving grace" is that the panic happens before the file systems > are mounted :-} > > Unloading the (r320324) kernel and loading the r320307 kernel > succeeds (at least enough to mount file systems). > > Peace, > david > -- > David H. Wolfskillda...@catwhisker.org > Trump (et al.): Hiding information doesn't prove its falsity. > > See http://www.catwhisker.org/~david/publickey.gpg for my public key. ___ 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"