Skip to contentSkip to conversation

net/bnxt/tf_core: fix ignored return of EM delete

PATCH In review Denis Sergeev 2 participants 5 messages May 19 – Jun 3 Network drivers

lore
Share & tools
lore
#0 lore raw

cover net/bnxt/tf_core: fix ignored return of EM delete

This series fixes a missing error check in the bnxt TF core EM delete path that can lead to HW/SW state inconsistency.

The return value of tfc_em_delete_raw() in tfc_em_delete() was silently discarded, so if the HW EM entry removal fails, the function continues to free the corresponding SW pool entry, creating stale flow matches or incorrect pool slot reuse.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Patches:
- net/bnxt/tf_core: fix ignored return of EM delete
- mailmap: add Denis Sergeev

Denis Sergeev (2):
net/bnxt/tf_core: fix ignored return of EM delete
mailmap: add Denis Sergeev

 .mailmap                             | 1 +
 drivers/net/bnxt/tf_core/v3/tfc_em.c | 5 +++++
 2 files changed, 6 insertions(+)
Signature · 1 line
2.50.1
#1 lore raw

PATCH 1/2 net/bnxt/tf_core: fix ignored return of EM delete

The return value of tfc_em_delete_raw() in tfc_em_delete() was silently discarded: rc was unconditionally overwritten by the subsequent tfc_cpm_get_cmm_inst() call without any error check.

If tfc_em_delete_raw() fails, the HW EM entry is not removed but the function continues to free the corresponding SW pool entry, creating a HW/SW state inconsistency that can lead to stale flow matches or incorrect pool slot reuse.

Add an error check after the call and return -EINVAL on failure.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes 80317ff6adfd ("net/bnxt/tf_core: support Thor2")
---
 drivers/net/bnxt/tf_core/v3/tfc_em.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/bnxt/tf_core/v3/tfc_em.c +5 -0
index 3fe4dbe3fe..4c126dc2f4 100644
@@ -661,6 +661,11 @@ int tfc_em_delete(struct tfc *tfcp, struct tfc_em_delete_parms *parms)
 			       &db_offset
 #endif
 			       );
+	if (rc != 0) {
+		PMD_DRV_LOG_LINE(ERR, "tfc_em_delete_raw() failed: %s",
+				 strerror(-rc));
+		return -EINVAL;
+	}
 
 	record_offset = REMOVE_POOL_FROM_OFFSET(pi.lkup_pool_sz_exp,
 						record_offset);
Signature · 1 line
2.50.1
#2 lore raw
Quoted 38 lines · Denis Sergeev

The return value of tfc_em_delete_raw() in tfc_em_delete() was silently discarded: rc was unconditionally overwritten by the subsequent tfc_cpm_get_cmm_inst() call without any error check.

If tfc_em_delete_raw() fails, the HW EM entry is not removed but the function continues to free the corresponding SW pool entry, creating a HW/SW state inconsistency that can lead to stale flow matches or incorrect pool slot reuse.

Add an error check after the call and return -EINVAL on failure.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes 80317ff6adfd ("net/bnxt/tf_core: support Thor2")
---
 drivers/net/bnxt/tf_core/v3/tfc_em.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/bnxt/tf_core/v3/tfc_em.c +5 -0
index 3fe4dbe3fe..4c126dc2f4 100644
@@ -661,6 +661,11 @@ int tfc_em_delete(struct tfc *tfcp, struct
 tfc_em_delete_parms *parms)
                               &db_offset
 #endif
                               );
+       if (rc != 0) {
+               PMD_DRV_LOG_LINE(ERR, "tfc_em_delete_raw() failed: %s",
+                                strerror(-rc));
+               return -EINVAL;
+       }
 
        record_offset = REMOVE_POOL_FROM_OFFSET(pi.lkup_pool_sz_exp,
Quoted 2 lines
                                                record_offset);

This change is not required, even if deletion of the HW entry fails, it should continue to delete the SW state, since at the end all the HW entries are invalidated. Have you been able to reproduce a scenario where this failure is seen.

Quoted 4 lines

--
2.50.1

#3 lore raw
Quoted 45 lines · Kishore Padmanabha

On Tue, May 19, 2026 at 12:44 AM Denis Sergeev <[email protected]> wrote:

    The return value of tfc_em_delete_raw() in tfc_em_delete() was
    silently discarded: rc was unconditionally overwritten by the
    subsequent tfc_cpm_get_cmm_inst() call without any error check.

    If tfc_em_delete_raw() fails, the HW EM entry is not removed but
    the function continues to free the corresponding SW pool entry,
    creating a HW/SW state inconsistency that can lead to stale flow
    matches or incorrect pool slot reuse.

    Add an error check after the call and return -EINVAL on failure.

    Found by Linux Verification Center (linuxtesting.org
    <http://linuxtesting.org>) with SVACE.

    Fixes: 80317ff6adfd ("net/bnxt/tf_core: support Thor2")
    Cc: [email protected]

    Signed-off-by: Denis Sergeev <[email protected]>
    ---
     drivers/net/bnxt/tf_core/v3/tfc_em.c | 5 +++++
     1 file changed, 5 insertions(+)

    diff --git a/drivers/net/bnxt/tf_core/v3/tfc_em.c
    b/drivers/net/bnxt/tf_core/v3/tfc_em.c
    index 3fe4dbe3fe..4c126dc2f4 100644
    --- a/drivers/net/bnxt/tf_core/v3/tfc_em.c
    +++ b/drivers/net/bnxt/tf_core/v3/tfc_em.c
    @@ -661,6 +661,11 @@ int tfc_em_delete(struct tfc *tfcp, struct
    tfc_em_delete_parms *parms)
                                   &db_offset
     #endif
                                   );
    +       if (rc != 0) {
    +               PMD_DRV_LOG_LINE(ERR, "tfc_em_delete_raw() failed:
    %s",
    +                                strerror(-rc));
    +               return -EINVAL;
    +       }

            record_offset = REMOVE_POOL_FROM_OFFSET(pi.lkup_pool_sz_exp,
    record_offset);
Quoted 5 lines

This change is not required, even if deletion of the HW entry fails, it should continue to delete the SW state, since at the end all the HW entries are invalidated. Have you been able to reproduce a scenario where this failure is seen.

    -- 
    2.50.1

Thanks for the review. I'll drop this patch.

The issue was flagged by a static analyzer that caught the discarded return value of tfc_em_delete_raw(). Over the past week I tried to reproduce an actual failure at runtime, but I couldn't. So it seems this failure doesn't occur in real runtime.

Signature · 2 lines
Best regards,
Denis Sergeev
#4 lore raw

PATCH 2/2 mailmap: add Denis Sergeev

In reply to Denis Sergeev

Add my name and email mapping to .mailmap

---
 .mailmap | 1 +
 1 file changed, 1 insertion(+)
.mailmap +1 -0
index 4d26d9c286..3d0fc8c7da 100644
@@ -373,6 +373,7 @@ Deirdre O'Connor <[email protected]>
 Dekel Peled <[email protected]> <[email protected]>
 Dengdui Huang <[email protected]>
 Denis Pryazhennikov <[email protected]>
+Denis Sergeev <[email protected]>
 Dennis Marinus <[email protected]>
 Derek Chickles <[email protected]>
 Des O Dea <[email protected]>
Signature · 1 line
2.50.1