RFC: FUSE kernel module for the kernel...

[Gustau Pérez]

On 13/03/2012 18:56, gnn at freebsd.org wrote:
> At Sun, 11 Mar 2012 11:43:06 +0100,
> Gustau Pérez wrote:
>> On 08/03/2012 22:20, George Neville-Neil wrote:
>>> Howdy,
>>>
>>> I've taken the GSoC work done with the FUSE kernel module, and created a patch against HEAD
>>> which I have now subjected to testing using tools/regression/fsx.
>>>
>>> The patch is here: http://people.freebsd.org/~gnn/head-fuse-1.diff
>>>
>>> I would like to commit this patch in the next few days, so, please, if you care
>>> about this take a look and get back to me.
>>>
>>> Thanks,
>>> George
>>      Hi,
>>
>>     I'm running HEAD r232383 (as of 2 March) + head-fuse-2.diff in AMD64.
>>
>>     I've been able to use some fuse fs. I run fsx for a while without
>> problems with some of them (ext4fuse is readonly). Then ones working were:
>>
>>           sshfs
>>           ntfs-3g
>>           ext4fuse
>>
>>     others like:
>>
>>           truecrypt
>>           gvfs (gnome fuse daemon)
>>
>>     do fail. I tried fsx with gvfs, that's what I got:
>>
>>           [gus at portgus ~]$ /root/deviant2/tools/regression/fsx/fsx
>> .gvfs/multimedia\ a\ harkserver/prova
>>           no extend on truncate! not posix!
>>
>>     They (truecrypt and gvfs) fail when doing setattr/getattr syscalls.
>> truecrypt complains about not being able to find the recently created
>> encrypted volume (a simple one like $HOME/Desktop/prova).
>>
>>      With gvfs, the nautilus (or the application trying to use the file)
>> tries to setattr the file causing gvfs to get an I/O. It happens with
>> nearly all kind of files opened with gvfs, although there are some that
>> are useable. With those files useable with gvfs, when the application
>> closes them causes gvfs to block somewhere, rendering gvfs unuseable.
>>
>>     Those two filesystems can be very useful in the desktop, I guess
>> PCBSD could benefit from them.
>>
>>     I would say there is something blocking in
>> fuse_vnop_setattr/fuse_vnop_getattr, but I'm not sure how to debug it.
>>
>>     Thanks for your help.
>>
> Thanks for the detailed report.  I'll look into this in a bit, I'm
> traveling for two weeks.
>
> Best,
> George

    Hi,

    testing ntfs-3g, after doing a bit large transfer with rsync, I 
found I couldn't unmount the filesystem. After some tries and before 
checking that no process was accessing the filesystem I tried to force 
the unmont. After that the system paniced instantly.

    I'm running HEAD/AMD64 r232862+head-fuse-2.diff.

    I have a dump of it, but it would seem that fuse is missing debug 
symbols (I don't know why), so the backtrace is incomplete. I compiled 
fuse just by doing make on $SRCDIR/sys/modules/fuse. I'll try to 
reproduce the panic and figure out what happens. Any help would be also 
appreciated on this other issue.

    Gustau