Re: [PATCH] vhost-vdpa: fix potential memory leak during the release

[Stefano Garzarella]

On Wed, Nov 09, 2022 at 12:47:19PM -0500, Michael S. Tsirkin wrote:
> On Wed, Nov 09, 2022 at 04:42:13PM +0100, Stefano Garzarella wrote:
> > Before commit 3d5698793897 ("vhost-vdpa: introduce asid based IOTLB")
> > we call vhost_vdpa_iotlb_unmap(v, iotlb, 0ULL, 0ULL - 1) during the
> > release to free all the resources allocated when processing user IOTLB
> > messages through vhost_vdpa_process_iotlb_update().
> > That commit changed the handling of IOTLB a bit, and we accidentally
> > removed some code called during the release.
> >
> > We partially fixed with commit 037d4305569a ("vhost-vdpa: call
> > vhost_vdpa_cleanup during the release") but a potential memory leak is
> > still there as showed by kmemleak if the application does not send
> > VHOST_IOTLB_INVALIDATE or crashes:
> >
> >   unreferenced object 0xffff888007fbaa30 (size 16):
> >     comm "blkio-bench", pid 914, jiffies 4294993521 (age 885.500s)
> >     hex dump (first 16 bytes):
> >       40 73 41 07 80 88 ff ff 00 00 00 00 00 00 00 00  @sA.............
> >     backtrace:
> >       [<0000000087736d2a>] kmem_cache_alloc_trace+0x142/0x1c0
> >       [<0000000060740f50>] vhost_vdpa_process_iotlb_msg+0x68c/0x901 [vhost_vdpa]
> >       [<0000000083e8e205>] vhost_chr_write_iter+0xc0/0x4a0 [vhost]
> >       [<000000008f2f414a>] vhost_vdpa_chr_write_iter+0x18/0x20 [vhost_vdpa]
> >       [<00000000de1cd4a0>] vfs_write+0x216/0x4b0
> >       [<00000000a2850200>] ksys_write+0x71/0xf0
> >       [<00000000de8e720b>] __x64_sys_write+0x19/0x20
> >       [<0000000018b12cbb>] do_syscall_64+0x3f/0x90
> >       [<00000000986ec465>] entry_SYSCALL_64_after_hwframe+0x63/0xcd
> >
> > Let's fix calling vhost_vdpa_iotlb_unmap() on the whole range in
> > vhost_vdpa_remove_as(). We move that call before vhost_dev_cleanup()
> > since we need a valid v->vdev.mm in vhost_vdpa_pa_unmap().
> > vhost_iotlb_reset() call can be removed, since vhost_vdpa_iotlb_unmap()
> > on the whole range removes all the entries.
> >
> > The kmemleak log reported was observed with a vDPA device that has `use_va`
> > set to true (e.g. VDUSE). This patch has been tested with both types of
> > devices.
> >
> > Fixes: 037d4305569a ("vhost-vdpa: call vhost_vdpa_cleanup during the release")
> > Fixes: 3d5698793897 ("vhost-vdpa: introduce asid based IOTLB")
> > Signed-off-by: Stefano Garzarella <sgarzare@xxxxxxxxxx>
>
> It's fine, just pls don't say "potential" here in the subject, let's
> avoid pleonasms

Got it, I'll avoid in the future.

> - it's a memory leak, yes it triggers under some coditions
> but little is unconditional in this world :)

Nothing could be more true :-)

> No need to repost.

Thanks,
Stefano