Re: [fuse-devel] Kernel oops and debug data...

[Valient G. <vg...@po...>]

On Friday 20 May 2005 11:50, Joseph Cohen wrote:
> Hi so I've tried to make my 1st fuse driver but noticed I am able to
> crash my kernel which leads me to the question of how can I do this
> with a userland application?
> But anyways I think the bug is in the fuse driver or library since I
> am only doing a little modification to the example app fusexmp. 

> Call Trace:<ffffffff88040da9>{:unionfs:unionfs_open+31321}
> <ffffffff8018818b>{permission+123}
>        <ffffffff80189e19>{may_open+105} <ffffffff80179ff6>{dentry_open+246}
>        <ffffffff8017a14e>{filp_open+62}
> <ffffffff8017a23b>{get_unused_fd+219} <ffffffff80125a39>{sys32_open+73}
> <ffffffff80123cc1>{ia32_sysret+0}


I'm no expert on the kernel-level part of FUSE, but I'm curious why you think 
this is a fuse problem?  All the kernel traces you included show unionfs as 
the only code involved in the trace.

I was curious about unionfs so I looked it up on google and viewed this 
month's mailing list archives.  I see several threads talking about different 
kinds of OOPS in unionfs.  

If you think the bug is in fuse, then why not duplicate the problem without 
unionfs?

regards,
Valient