Coverity complains of a possible NULL dereference in smc_find_rdma_v2_device_serv().
1782 smc_v2_ext = smc_get_clc_v2_ext(pclc);
CID 121151 (#1 of 1): Dereference null return value (NULL_RETURNS)
5. dereference: Dereferencing a pointer that might be NULL smc_v2_ext when calling smc_clc_match_eid. [show details]
1783 if (!smc_clc_match_eid(ini->negotiated_eid, smc_v2_ext, NULL, NULL))
1784 goto not_found;
Fix this by checking for NULL.
Fixes: e49300a6bf621 ("net/smc: add listen processing for SMC-Rv2")
Cc: Karsten Graul <[email protected]>
Cc: "David S. Miller" <[email protected]>
Cc: Jakub Kicinski <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Signed-off-by: Tim Gardner <[email protected]>
---
net/smc/af_smc.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c
index 5e50e007a7da..ff23d5b40793 100644
--- a/net/smc/af_smc.c
+++ b/net/smc/af_smc.c
@@ -1780,7 +1780,7 @@ static void smc_find_rdma_v2_device_serv(struct smc_sock *new_smc,
goto not_found;
smc_v2_ext = smc_get_clc_v2_ext(pclc);
- if (!smc_clc_match_eid(ini->negotiated_eid, smc_v2_ext, NULL, NULL))
+ if (!smc_v2_ext || !smc_clc_match_eid(ini->negotiated_eid, smc_v2_ext, NULL, NULL))
goto not_found;
/* prepare RDMA check */
--
2.33.1
On 18/10/2021 20:31, Tim Gardner wrote:
> Coverity complains of a possible NULL dereference in smc_find_rdma_v2_device_serv().
>
> 1782 smc_v2_ext = smc_get_clc_v2_ext(pclc);
> CID 121151 (#1 of 1): Dereference null return value (NULL_RETURNS)
> 5. dereference: Dereferencing a pointer that might be NULL smc_v2_ext when calling smc_clc_match_eid. [show details]
> 1783 if (!smc_clc_match_eid(ini->negotiated_eid, smc_v2_ext, NULL, NULL))
> 1784 goto not_found;
>
> Fix this by checking for NULL.
Hmm that's a fundamental question for me: do we want to make the code checkers happy?
While I understand that those warnings give an uneasy feeling I am not sure
if the code should have additional (unneeded) checks only to avoid them.
In this case all NULL checks are initially done in smc_listen_v2_check(),
afterwards no more NULL checks are needed. When we would like to add them
then a lot more checks are needed, e.g. 3 times in smc_find_ism_v2_device_serv()
(not sure why coverity does not complain about them, too).
Thoughts?
On 10/19/21 12:33 AM, Karsten Graul wrote:
> On 18/10/2021 20:31, Tim Gardner wrote:
>> Coverity complains of a possible NULL dereference in smc_find_rdma_v2_device_serv().
>>
>> 1782 smc_v2_ext = smc_get_clc_v2_ext(pclc);
>> CID 121151 (#1 of 1): Dereference null return value (NULL_RETURNS)
>> 5. dereference: Dereferencing a pointer that might be NULL smc_v2_ext when calling smc_clc_match_eid. [show details]
>> 1783 if (!smc_clc_match_eid(ini->negotiated_eid, smc_v2_ext, NULL, NULL))
>> 1784 goto not_found;
>>
>> Fix this by checking for NULL.
>
> Hmm that's a fundamental question for me: do we want to make the code checkers happy?
> While I understand that those warnings give an uneasy feeling I am not sure
> if the code should have additional (unneeded) checks only to avoid them.
>
Coverity produces a lot of false positives. I thought this one might be
legitimate, but if you're comfortable that its not an issue then I'm OK
with that.
> In this case all NULL checks are initially done in smc_listen_v2_check(),
> afterwards no more NULL checks are needed. When we would like to add them
> then a lot more checks are needed, e.g. 3 times in smc_find_ism_v2_device_serv()
> (not sure why coverity does not complain about them, too).
>
> Thoughts?
>
Coverity probably has produced a report from the other call sites if
you've used a similar pattern, I just hadn't gotten to them yet.
I'll just mark them all as false positives.
rtg
--
-----------
Tim Gardner
Canonical, Inc