Media checked KDL on get_cache_blocked

[emitrax]

I run a git checkout on a unfinished webkit clone. Git crashed (bt attached). Then continue, and media checker KDLed. (bt attached).

Vmware, rev of 4th Nov.

[emitrax]

Happened again this morning.

Revision of 10-11-09.

This time happened from the first checkout try.

Stack trace is different, and after continuing block notifier KDL'ed. Probably though it's related.

I'm not sure if I should open a different bug since the stack trace is different, although the scenario is the same (checking out webkit git repository from vmware directly on USB BFS formatted stick).

I'm sorry but I'll keep doing it until I get it! :-)

[axeld]

From what I can tell this seem to be 3 different bugs:

1) the first bug is the originally reported one; this seems to be a BFS bug. Please use the "bfs_allocator", and "bfs" command next time you stumble across that one (you get the pointer-to-volume from the "mount" KDL command).

2) Writing back block 5 failed - this seems to be a problem of the underlying device, ie. probably a USB problem (you may want to have a look into your syslog for hints). I guess it's something similar to #4901.

3) The block/notifier crash - there already seems to be a ticket for that one (see #2690), also reported by you. Are you sure you don't have local changes on that file? ;-))

[emitrax]

Replying to axeld:

From what I can tell this seem to be 3 different bugs: 1) the first bug is the originally reported one; this seems to be a BFS bug. Please use the "bfs_allocator", and "bfs" command next time you stumble across that one (you get the pointer-to-volume from the "mount" KDL command).

Ok. Thanks.

2) Writing back block 5 failed - this seems to be a problem of the underlying device, ie. probably a USB problem (you may want to have a look into your syslog for hints). I guess it's something similar to #4901.

I just remembered I had serial output enabled and found an hint on this one.

42 usb error control pipe 16: timeout waiting for queued request to complete 43 usb error hub 15: error updating port status

But this is probably related to the fact that I was on KDL.

3) The block/notifier crash - there already seems to be a ticket for that one (see #2690), also reported by you. Are you sure you don't have local changes on that file? ;-))

100% sure. The image I'm using at work is basically the same as the trunk one. Not local changes on that area. BTW: I didn't remember about that bug at all. :-)

[emitrax]

Ok.

What I'm trying to do seem to trigger a few bugs. :-)

This definitely seems to be USB related from the serial log, but I'm attaching it here for now.