Hello Sergey,
On Wed, Apr 29, 2015 at 04:23:28PM +0900, Sergey Senozhatsky wrote:
> On (04/29/15 16:02), Sergey Senozhatsky wrote:
> > sure. I was talking about this one:
> >
> > CPU0 CPU1
> > umount
> > zram_remove()
> > lock ->bd_mutex
> > zram_reset_device()
> > unlock ->bd_mutex
> > disksize_store
> > mount
> > echo 'test' > /mnt/test
> > kfree zram
> > zram write
> >
>
> I'll take a look later today. currently I think of something like:
>
>
> sysfs_remove_group()
> lock ->bd_mutex
> ... check ->bd_openers
>
> zram_reset_device()
> blk_cleanup_queue()
> del_gendisk()
> put_disk()
>
> unlock ->bd_mutex
> bdput bdev
>
> idr_remove()
> kfree(zram)
>
>
> iow, idr_remove() and kfree() are done outside of ->bd_mutex lock.
> but I may be wrong. haven't tested yet. but seems reasonable: we
> invalidate ->bdev, delete partitions, etc., holding ->bd_mutex and
> then release ->bdev, which does final put. need to check that in
> detail.
>
> -ss
Isn't it related to bd_mutex?
I think the problem of deadlock is that you are trying to remove sysfs file
in sysfs handler.
#> echo 1 > /sys/xxx/zram_remove
kernfs_fop_write - hold s_active
-> zram_remove_store
-> zram_remove
-> sysfs_remove_group - hold s_active *again*
Right?
--
Kind regards,
Minchan Kim
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
