Bug #186

Recent commits to -HEAD broke the linux emulator

Added by rumcic over 8 years ago. Updated about 8 years ago.

Status:ClosedStart date:
Priority:NormalDue date:
Assignee:-% Done:

0%

Category:-
Target version:-

Description

The commits that occured recently in -HEAD (todays and maybe also yesterdays)
broke the linux emulator. While booting up with SMP or not, it hangs while
loading the linux emulator.

History

#1 Updated by qhwt+dfly over 8 years ago

On Sat, May 27, 2006 at 07:26:09PM +0200, Rumko wrote:
> The commits that occured recently in -HEAD (todays and maybe also yesterdays)
> broke the linux emulator. While booting up with SMP or not, it hangs while
> loading the linux emulator.

Me too, and when it's compiled in the kernel, it boots, but it hangs
when I start acroread7. I think remember that the kernel built from
the source as of May 24 was OK.

#2 Updated by dillon over 8 years ago

:
:On Sat, May 27, 2006 at 07:26:09PM +0200, Rumko wrote:
:> The commits that occured recently in -HEAD (todays and maybe also yesterdays)
:> broke the linux emulator. While booting up with SMP or not, it hangs while
:> loading the linux emulator.
:
:Me too, and when it's compiled in the kernel, it boots, but it hangs
:when I start acroread7. I think remember that the kernel built from
:the source as of May 24 was OK.

I need something I can duplicate from a freshly installed machine.
I need to know which packages I need to install and a sequence of
commands that reproduces the problem.

-Matt
Matthew Dillon
<>

#3 Updated by qhwt+dfly over 8 years ago

On Fri, Jun 02, 2006 at 10:04:14AM -0700, Matthew Dillon wrote:
>
> :
> :On Sat, May 27, 2006 at 07:26:09PM +0200, Rumko wrote:
> :> The commits that occured recently in -HEAD (todays and maybe also yesterdays)
> :> broke the linux emulator. While booting up with SMP or not, it hangs while
> :> loading the linux emulator.
> :
> :Me too, and when it's compiled in the kernel, it boots, but it hangs
> :when I start acroread7. I think remember that the kernel built from
> :the source as of May 24 was OK.
>
> I need something I can duplicate from a freshly installed machine.
> I need to know which packages I need to install and a sequence of
> commands that reproduces the problem.

If linux emulator is compiled in, just install pkgsrc/emulators/suse_base
and run
$ /usr/pkg/emul/bin/ls -l
and it never returns. You may have to run `sysctl kern.sync_on_panic=1'
in advance, as panic command from DDB may be stuck.
If linux emulator is not compiled in the kernel, loading it via
/etc/rc.d/abi will trigger the lock up(probably because it runs
/compat/linux/sbin/ldconfig).

I couldn't drop in to DDB on my SMP machine, but I could on my laptop.
I'm uploading a kernel dump on my leaf account.

#4 Updated by qhwt+dfly over 8 years ago

On Sat, Jun 03, 2006 at 02:55:38PM +0900, YONETANI Tomokazu wrote:
> I'm uploading a kernel dump on my leaf account.

Done.

#5 Updated by dillon over 8 years ago

:
:On Sat, Jun 03, 2006 at 02:55:38PM +0900, YONETANI Tomokazu wrote:
:> I'm uploading a kernel dump on my leaf account.
:
:Done.

Ok, I've just committed a fix for that issue.

-Matt
Matthew Dillon
<>

#6 Updated by qhwt+dfly over 8 years ago

On Sat, Jun 03, 2006 at 01:09:57AM -0700, Matthew Dillon wrote:
> :On Sat, Jun 03, 2006 at 02:55:38PM +0900, YONETANI Tomokazu wrote:
> :> I'm uploading a kernel dump on my leaf account.
> :Done.
> Ok, I've just committed a fix for that issue.

Thanks, so far it doesn't lock up. But I don't understand why it
worked without looking at eofflag.

#7 Updated by dillon over 8 years ago

:
:On Sat, Jun 03, 2006 at 01:09:57AM -0700, Matthew Dillon wrote:
:> :On Sat, Jun 03, 2006 at 02:55:38PM +0900, YONETANI Tomokazu wrote:
:> :> I'm uploading a kernel dump on my leaf account.
:> :Done.
:> Ok, I've just committed a fix for that issue.
:
:Thanks, so far it doesn't lock up. But I don't understand why it
:worked without looking at eofflag.

I'm not sure why it worked before either. The algorithm in the linux
readir code was clearly broken. Any situation where a readdir returns
just one cookie which happens to have an offset before the requested
offset would have triggered the bug.

-Matt
Matthew Dillon
<>

Also available in: Atom PDF