Opened 9 years ago

Closed 8 years ago

#5936 closed bug (invalid)

PANIC: vm_page_fault: unhandled page fault in kernel space at 0x0, ip 0x82dcfbe3

Reported by: tonestone57 Owned by: mmlr
Priority: normal Milestone: R1
Component: System/Kernel Version: R1/Development
Keywords: Cc:
Blocked By: Blocking:
Has a Patch: no Platform: x86

Description

Haiku hrev36666, gcc2 hybrid, trunk build.

I would say that recent changes caused this new crash. I was able to boot Alpha2-rc on this machine without Haiku crashing (when I disabled ACPI).

Screenshot of stack trace attached. Let me know what else you need.

Attachments (4)

vm_page_fault.jpg (466.1 KB ) - added by tonestone57 9 years ago.
r36747-bt.jpg (332.9 KB ) - added by tonestone57 9 years ago.
shutdown-error.png (34.1 KB ) - added by Karvjorm 9 years ago.
Another picture about r1alpha2 shutdown crash in WMware player.
shutdown-error2.png (50.3 KB ) - added by Karvjorm 9 years ago.
The end of picture about r1alpha2 shutdown crash (AMD Turion X2)

Download all attachments as: .zip

Change History (15)

by tonestone57, 9 years ago

Attachment: vm_page_fault.jpg added

comment:1 by bonefish, 9 years ago

Owner: changed from axeld to mmlr
Status: newassigned

comment:2 by mmlr, 9 years ago

Curios one really. From the output it looks like the gPCI pointer is NULL, but the PCI::FindDevice() call really is already the second use of gPCI in this function. It is already used to call PCI::ResolveVirtualBus(), so it should've crashed there already if not available.

comment:3 by mmlr, 9 years ago

Might be fixed in hrev36745. Can you please retry?

in reply to:  2 comment:4 by bonefish, 9 years ago

Replying to mmlr:

Curios one really. From the output it looks like the gPCI pointer is NULL, but the PCI::FindDevice() call really is already the second use of gPCI in this function. It is already used to call PCI::ResolveVirtualBus(), so it should've crashed there already if not available.

The call itself doesn't dereference the object pointer (non-virtual method) and the method implementation doesn't dereference the this pointer either.

comment:5 by tonestone57, 9 years ago

I retested with hrev36747 gcc2 hybrid and still same issue. Not fixed. I will attach new screenshot.

off-topic: hrev36740 (usb_webcam) broke build for me, please fix

by tonestone57, 9 years ago

Attachment: r36747-bt.jpg added

comment:6 by tonestone57, 9 years ago

Tested with Alpha2 hrev36737 which works without any issues and very well in Live-CD mode. System only crashes with this error when using Haiku's trunk build.

I would say that crashing started with Ingo's vm changes which somehow affected my laptop.

comment:7 by tonestone57, 9 years ago

I looked and saw Ingo's vm changes were all fully merged into Alpha2 before hrev36737 release. So, must be other code committed to trunk causing the crash. Sorry for the mistake.

comment:8 by tonestone57, 9 years ago

Please hold off on fixing this ticket! I believe my haiku source files have become corrupted somehow (were not updated for 10 months) and I would like to test with nightly from haiku files when they become available again for download.

I will update this ticket when I can test with nightly from haiku files too. I will rebuild haiku from clean and if that does not work then I will check out clean copy of source too. Very sorry about that guys!

by Karvjorm, 9 years ago

Attachment: shutdown-error.png added

Another picture about r1alpha2 shutdown crash in WMware player.

by Karvjorm, 9 years ago

Attachment: shutdown-error2.png added

The end of picture about r1alpha2 shutdown crash (AMD Turion X2)

comment:9 by bonefish, 9 years ago

Karvjorm: This crash isn't related. It is already tracked in #4473.

comment:10 by tonestone57, 9 years ago

This ticket can be closed for now and I may reopen at later date.

Images from haiku-files seem to work without crashing on my laptop. Images I build crash on my laptop ever since the MSI changeset happened.

I will look into how my images differ from those off haiku-files later on when i have more time. Might be awhile before I check into this.

comment:11 by mmlr, 8 years ago

Resolution: invalid
Status: assignedclosed

Closing as it looks like the issue was stale object files. "Official" images didn't experience this issue.

Note: See TracTickets for help on using tickets.