[PATCH v2] ceph: avoid putting the realm twice when decoding snaps fails

[PATCH v2] ceph: avoid putting the realm twice when decoding snaps fails

[xiubli]

From: Xiubo Li <xiubli@redhat.com>

When decoding the snaps fails it maybe leaving the 'first_realm'
and 'realm' pointing to the same snaprealm memory. And then it'll
put it twice and could cause random use-after-free, BUG_ON, etc
issues.

Cc: stable@vger.kernel.org
URL: https://tracker.ceph.com/issues/57686
Reviewed-by: Luís Henriques <lhenriques@suse.de>
Signed-off-by: Xiubo Li <xiubli@redhat.com>
---
 fs/ceph/snap.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c
index 9bceed2ebda3..77b948846d4d 100644
--- a/fs/ceph/snap.c
+++ b/fs/ceph/snap.c
@@ -854,6 +854,8 @@ int ceph_update_snap_trace(struct ceph_mds_client *mdsc,
 	else
 		ceph_put_snap_realm(mdsc, realm);
 
+	realm = NULL;
+
 	if (p < e)
 		goto more;
 
-- 
2.31.1

Re: [PATCH v2] ceph: avoid putting the realm twice when decoding snaps fails

[Ilya Dryomov]

On Tue, Nov 8, 2022 at 3:32 AM <xiubli@redhat.com> wrote:
>
> From: Xiubo Li <xiubli@redhat.com>
>
> When decoding the snaps fails it maybe leaving the 'first_realm'
> and 'realm' pointing to the same snaprealm memory. And then it'll
> put it twice and could cause random use-after-free, BUG_ON, etc
> issues.
>
> Cc: stable@vger.kernel.org
> URL: https://tracker.ceph.com/issues/57686
> Reviewed-by: Luís Henriques <lhenriques@suse.de>
> Signed-off-by: Xiubo Li <xiubli@redhat.com>
> ---
>  fs/ceph/snap.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c
> index 9bceed2ebda3..77b948846d4d 100644
> --- a/fs/ceph/snap.c
> +++ b/fs/ceph/snap.c
> @@ -854,6 +854,8 @@ int ceph_update_snap_trace(struct ceph_mds_client *mdsc,
>         else
>                 ceph_put_snap_realm(mdsc, realm);
>
> +       realm = NULL;

Hi Xiubo,

Nit: I think it would be better to clear realm in the part of
the function that actually needs it to be cleared -- otherwise
this assignment can be easily lost in refactoring.

        dout("%s deletion=%d\n", __func__, deletion);
more:
        realm = NULL;
        rebuild_snapcs = 0;

And then realm wouldn't need to be initialized.

Thanks,

                Ilya

Re: [PATCH v2] ceph: avoid putting the realm twice when decoding snaps fails

[Xiubo Li]

On 08/11/2022 17:56, Ilya Dryomov wrote:
> On Tue, Nov 8, 2022 at 3:32 AM <xiubli@redhat.com> wrote:
>> From: Xiubo Li <xiubli@redhat.com>
>>
>> When decoding the snaps fails it maybe leaving the 'first_realm'
>> and 'realm' pointing to the same snaprealm memory. And then it'll
>> put it twice and could cause random use-after-free, BUG_ON, etc
>> issues.
>>
>> Cc: stable@vger.kernel.org
>> URL: https://tracker.ceph.com/issues/57686
>> Reviewed-by: Luís Henriques <lhenriques@suse.de>
>> Signed-off-by: Xiubo Li <xiubli@redhat.com>
>> ---
>>   fs/ceph/snap.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c
>> index 9bceed2ebda3..77b948846d4d 100644
>> --- a/fs/ceph/snap.c
>> +++ b/fs/ceph/snap.c
>> @@ -854,6 +854,8 @@ int ceph_update_snap_trace(struct ceph_mds_client *mdsc,
>>          else
>>                  ceph_put_snap_realm(mdsc, realm);
>>
>> +       realm = NULL;
> Hi Xiubo,
>
> Nit: I think it would be better to clear realm in the part of
> the function that actually needs it to be cleared -- otherwise
> this assignment can be easily lost in refactoring.
>
>          dout("%s deletion=%d\n", __func__, deletion);
> more:
>          realm = NULL;
>          rebuild_snapcs = 0;
>
> And then realm wouldn't need to be initialized.

Okay, will fix it.

Thanks!

- Xiubo


>
> Thanks,
>
>                  Ilya
>