Opened 11 years ago

Closed 11 years ago

#2578 closed bug (fixed)

[scsi scheduler]: crash in _GetNextUnscheduledRequest

Reported by: tonestone57 Owned by: bonefish
Priority: normal Milestone: R1/alpha1
Component: System/Kernel Version: R1/pre-alpha1
Keywords: Cc:
Blocked By: Blocking:
Has a Patch: no Platform: x86

Description

Ran bonnie++, single process, on Haiku hrev26829. System crashed to KDL.

bonnie++ should have worked without crashing.

Jpeg image of stack trace attached.

Attachments (1)

p1010006.jpg (202.9 KB ) - added by tonestone57 11 years ago.
stack_trace

Download all attachments as: .zip

Change History (9)

by tonestone57, 11 years ago

Attachment: p1010006.jpg added

stack_trace

comment:1 by anevilyak, 11 years ago

Component: Drivers/DiskSystem/Kernel
Owner: changed from marcusoverhagen to axeld

Actually, 26828 just introduced some major (known to not be entirely stable) changes to handling of disk I/O and such, so right now it's not surprising that this doesn't complete.

comment:2 by tonestone57, 11 years ago

Yes, once the new I/O scheduler is working right then this bug ( and some other new ones ) should disappear.

I ran bonnie++ on hrev26804 without any issues or errors - completed successfully.

Once the changes to the disk I/O are complete; I can test again and report back.

comment:3 by emitrax, 11 years ago

Actually, running the test, and reporting bugs with as much as information as possible will help speed things up.

Anyway, if you test it again and encounter the same crash, please also report the stack trace of the running application (bonnie++ in this case). It might help.

comment:4 by emitrax, 11 years ago

Milestone: R1R1/alpha1
Owner: changed from axeld to bonefish
Summary: Running bonnie++ 1.03d on Haiku - crashes system[scsi scheduler]: crash in _GetNextUnscheduledRequest

Ingo, has this been fixed?

in reply to:  4 ; comment:5 by bonefish, 11 years ago

Replying to emitrax:

Ingo, has this been fixed?

You tell me. :-) I've never run bonnie++ (BTW, how about adding an optional test package?). I've never seen such a crash and Trac doesn't seem to want me to show the stack trace. It may already have been fixed; at least some I/O scheduler related problems have been addressed in the meantime.

in reply to:  5 comment:6 by emitrax, 11 years ago

Replying to bonefish:

Replying to emitrax:

Ingo, has this been fixed?

You tell me. :-) I've never run bonnie++ (BTW, how about adding an optional test package?).

I'm not the one working on I/O scheduler. :-) If about "optional test package" you mean bonnie++, it's already included since hrev26920.

I've never seen such a crash and Trac doesn't seem to want me to show the stack trace. It may >already have been fixed; at least some I/O scheduler related problems have been addressed in >the meantime.

Yes, that's why I'm asking. The stack trace shows that it fails in _GetNextUnscheduledRequest with a page fault on 0xdeadbf03, and since some other related ticket have been fixed, I thought this one was fixed too.

comment:7 by tonestone57, 11 years ago

Just finished testing bonnie++ on revision 27189

The test finished without any errors. Ran both single and multi-process tests. Looks like Haiku is stable once again. Well done.

This bug seems to have been fixed with one of the revisions and can now be closed.

comment:8 by bonefish, 11 years ago

Resolution: fixed
Status: newclosed

Thanks for the feedback!

Note: See TracTickets for help on using tickets.