another fuse panic

[John Baldwin]

On Monday, November 08, 2010 6:35:55 am Ivan Voras wrote:
> On 11/08/10 10:55, Andriy Gapon wrote:
> > 
> > JFYI.
> > Fatal trap 12: page fault while in kernel mode
> 
> Can you find any set of circumstances which make this repeatable?
> 
> This panic apparently goes like this:
> 
> 1) used by devfs_open():
>  47 static struct cdevsw fuse_cdevsw = {
>  48         .d_open = fusedev_open,
> 
> 2) in fusedev_open():
> 119         fdata = fdata_alloc(dev, td->td_ucred);
> 
> 3) in fdata_alloc():
> 297         data->daemoncred = crhold(cred);
> 
> in other words, td->td_ucred from td passed to fusedev_open (presumably
> when the device is opened from the userland) appears to be NULL.
> 
> I don't know if there is any normal set of circumstances under which
> this is expected.

No, td_ucred should never be NULL.

-- 
John Baldwin