Re: [PATCH] xen-blkback: prevent premature module unload

From: JÃrgen GroÃ
Date: Wed Dec 11 2019 - 08:57:00 EST


On 11.12.19 14:55, Roger Pau Monnà wrote:
On Wed, Dec 11, 2019 at 01:27:42PM +0000, Durrant, Paul wrote:
-----Original Message-----
From: Roger Pau Monnà <roger.pau@xxxxxxxxxx>
Sent: 11 December 2019 11:29
To: Durrant, Paul <pdurrant@xxxxxxxxxx>
Cc: xen-devel@xxxxxxxxxxxxxxxxxxxx; linux-block@xxxxxxxxxxxxxxx; linux-
kernel@xxxxxxxxxxxxxxx; Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>;
Jens Axboe <axboe@xxxxxxxxx>
Subject: Re: [PATCH] xen-blkback: prevent premature module unload

On Tue, Dec 10, 2019 at 02:53:05PM +0000, Paul Durrant wrote:
Objects allocated by xen_blkif_alloc come from the 'blkif_cache' kmem
cache. This cache is destoyed when xen-blkif is unloaded so it is
necessary to wait for the deferred free routine used for such objects to
complete. This necessity was missed in commit 14855954f636 "xen-blkback:
allow module to be cleanly unloaded". This patch fixes the problem by
taking/releasing extra module references in xen_blkif_alloc/free()
respectively.

Signed-off-by: Paul Durrant <pdurrant@xxxxxxxxxx>

Reviewed-by: Roger Pau Monnà <roger.pau@xxxxxxxxxx>

One nit below.

---
Cc: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
Cc: "Roger Pau MonnÃ" <roger.pau@xxxxxxxxxx>
Cc: Jens Axboe <axboe@xxxxxxxxx>
---
drivers/block/xen-blkback/xenbus.c | 10 ++++++++++
1 file changed, 10 insertions(+)

diff --git a/drivers/block/xen-blkback/xenbus.c b/drivers/block/xen-
blkback/xenbus.c
index e8c5c54e1d26..59d576d27ca7 100644
--- a/drivers/block/xen-blkback/xenbus.c
+++ b/drivers/block/xen-blkback/xenbus.c
@@ -171,6 +171,15 @@ static struct xen_blkif *xen_blkif_alloc(domid_t
domid)
blkif->domid = domid;
atomic_set(&blkif->refcnt, 1);
init_completion(&blkif->drain_complete);
+
+ /*
+ * Because freeing back to the cache may be deferred, it is not
+ * safe to unload the module (and hence destroy the cache) until
+ * this has completed. To prevent premature unloading, take an
+ * extra module reference here and release only when the object
+ * has been free back to the cache.
^ freed

Oh yes. Can this be done on commit, or would you like me to send a v2?

Adjusting on commit would be fine for me, but it's up to Juergen since
he is the one that will pick this up. IIRC the module unload patches
didn't go through the block subsystem.

Oh, right. Yes, will fix this when committing.


Juergen