From: Mohammad Shuab Siddique <mohammad-shuab.siddique@broadcom.com>
To: dev@dpdk.org
Cc: kishore.padmanabha@broadcom.com,
Mohammad Shuab Siddique <mohammad-shuab.siddique@broadcom.com>,
stable@dpdk.org
Subject: [PATCH v3 2/5] net/bnxt: validate TPA aggregation ID from completions
Date: Mon, 28 Sep 2026 18:24:39 -0600 [thread overview]
Message-ID: <20260929002442.1208481-3-Mohammad-Shuab.Siddique@broadcom.com> (raw)
In-Reply-To: <20260929002442.1208481-1-Mohammad-Shuab.Siddique@broadcom.com>
From: Mohammad Shuab Siddique <mohammad-shuab.siddique@broadcom.com>
TPA start/end and TPA v2 abuf paths indexed rxr->tpa_info[] using
the aggregation ID from the completion without checking
BNXT_TPA_MAX_AGGS. Add bounds checks and schedule a ring reset on
failure. Also fix abuf aggregation ID decode to use
rte_le_to_cpu_16() when reading the device field. On the legacy
(non-Thor) TPA end path, also discard the pending agg bufs before
returning on an invalid agg_id, matching the existing in_reset early
return a few lines above; otherwise those agg buf completions are
left unconsumed and the CQ raw_cons goes out of sync with the ring
even after the scheduled reset.
Fixes: 0958d8b6435d ("net/bnxt: support LRO")
Fixes: b150a7e7ee66 ("net/bnxt: support LRO on Thor adapters")
Cc: stable@dpdk.org
Signed-off-by: Mohammad Shuab Siddique <mohammad-shuab.siddique@broadcom.com>
---
drivers/net/bnxt/bnxt_rxr.c | 48 ++++++++++++++++++++++++++++---------
1 file changed, 37 insertions(+), 11 deletions(-)
diff --git a/drivers/net/bnxt/bnxt_rxr.c b/drivers/net/bnxt/bnxt_rxr.c
index 98bdbc136a..1790ce839f 100644
--- a/drivers/net/bnxt/bnxt_rxr.c
+++ b/drivers/net/bnxt/bnxt_rxr.c
@@ -227,12 +227,19 @@ static void bnxt_tpa_start(struct bnxt_rx_queue *rxq,
struct rx_tpa_start_cmpl_hi *tpa_start1)
{
struct bnxt_rx_ring_info *rxr = rxq->rx_ring;
- uint16_t agg_id;
- uint16_t data_cons;
struct bnxt_tpa_info *tpa_info;
+ uint32_t data_cons, agg_id;
+ struct bnxt *bp = rxq->bp;
struct rte_mbuf *mbuf;
- agg_id = bnxt_tpa_start_agg_id(rxq->bp, tpa_start);
+ agg_id = bnxt_tpa_start_agg_id(bp, tpa_start);
+ if (unlikely(agg_id >= BNXT_TPA_MAX_AGGS(bp))) {
+ PMD_DRV_LOG_LINE(ERR,
+ "TPA start: invalid agg_id %u (max %u)",
+ agg_id, BNXT_TPA_MAX_AGGS(bp));
+ bnxt_sched_ring_reset(rxq);
+ return;
+ }
data_cons = tpa_start->opaque;
tpa_info = &rxr->tpa_info[agg_id];
@@ -259,7 +266,7 @@ static void bnxt_tpa_start(struct bnxt_rx_queue *rxq,
mbuf->port = rxq->port_id;
mbuf->ol_flags = RTE_MBUF_F_RX_LRO;
- bnxt_tpa_get_metadata(rxq->bp, tpa_info, tpa_start, tpa_start1);
+ bnxt_tpa_get_metadata(bp, tpa_info, tpa_start, tpa_start1);
if (likely(tpa_info->hash_valid)) {
mbuf->hash.rss = tpa_info->rss_hash;
@@ -269,7 +276,7 @@ static void bnxt_tpa_start(struct bnxt_rx_queue *rxq,
mbuf->ol_flags |= RTE_MBUF_F_RX_FDIR | RTE_MBUF_F_RX_FDIR_ID;
}
- if (tpa_info->vlan_valid && BNXT_RX_VLAN_STRIP_EN(rxq->bp)) {
+ if (tpa_info->vlan_valid && BNXT_RX_VLAN_STRIP_EN(bp)) {
mbuf->vlan_tci = tpa_info->vlan;
mbuf->ol_flags |= RTE_MBUF_F_RX_VLAN | RTE_MBUF_F_RX_VLAN_STRIPPED;
}
@@ -412,20 +419,21 @@ static inline struct rte_mbuf *bnxt_tpa_end(
{
struct bnxt_cp_ring_info *cpr = rxq->cp_ring;
struct bnxt_rx_ring_info *rxr = rxq->rx_ring;
- uint16_t agg_id;
+ struct bnxt_tpa_info *tpa_info;
+ struct bnxt *bp = rxq->bp;
+ uint8_t payload_offset;
struct rte_mbuf *mbuf;
uint8_t agg_bufs;
- uint8_t payload_offset;
- struct bnxt_tpa_info *tpa_info;
+ uint32_t agg_id;
if (unlikely(rxq->in_reset)) {
PMD_DRV_LOG_LINE(ERR, "rxq->in_reset: raw_cp_cons:%d",
*raw_cp_cons);
- bnxt_discard_rx(rxq->bp, cpr, raw_cp_cons, tpa_end);
+ bnxt_discard_rx(bp, cpr, raw_cp_cons, tpa_end);
return NULL;
}
- if (BNXT_CHIP_P5_P7(rxq->bp)) {
+ if (BNXT_CHIP_P5_P7(bp)) {
struct rx_tpa_v2_end_cmpl *th_tpa_end;
struct rx_tpa_v2_end_cmpl_hi *th_tpa_end1;
@@ -442,6 +450,15 @@ static inline struct rte_mbuf *bnxt_tpa_end(
payload_offset = tpa_end->payload_offset;
}
+ if (unlikely(agg_id >= BNXT_TPA_MAX_AGGS(bp))) {
+ PMD_DRV_LOG_LINE(ERR,
+ "TPA end: invalid agg_id %u (max %u)",
+ agg_id, BNXT_TPA_MAX_AGGS(bp));
+ bnxt_discard_rx(bp, cpr, raw_cp_cons, tpa_end);
+ bnxt_sched_ring_reset(rxq);
+ return NULL;
+ }
+
tpa_info = &rxr->tpa_info[agg_id];
mbuf = tpa_info->mbuf;
RTE_ASSERT(mbuf != NULL);
@@ -1126,9 +1143,18 @@ static int bnxt_rx_pkt(struct rte_mbuf **rx_pkt,
if (cmp_type == RX_TPA_V2_ABUF_CMPL_TYPE_RX_TPA_AGG) {
struct rx_tpa_v2_abuf_cmpl *rx_agg = (void *)rxcmp;
- uint16_t agg_id = rte_cpu_to_le_16(rx_agg->agg_id);
+ uint32_t agg_id = rte_le_to_cpu_16(rx_agg->agg_id);
struct bnxt_tpa_info *tpa_info;
+ if (unlikely(agg_id >= BNXT_TPA_MAX_AGGS(bp))) {
+ PMD_DRV_LOG_LINE(ERR,
+ "TPA abuf: invalid agg_id %u (max %u)",
+ agg_id, BNXT_TPA_MAX_AGGS(bp));
+ bnxt_sched_ring_reset(rxq);
+ rc = -EINVAL;
+ goto next_rx;
+ }
+
tpa_info = &rxr->tpa_info[agg_id];
RTE_ASSERT(tpa_info->agg_count < 16);
tpa_info->agg_arr[tpa_info->agg_count++] = *rx_agg;
--
2.47.3
next prev parent reply other threads:[~2026-09-29 0:22 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-18 3:27 [PATCH 0/5] net/bnxt: fix flow, Rx and naming bounds issues Mohammad Shuab Siddique
2026-09-18 3:27 ` [PATCH 1/5] net/bnxt: fix stack exhaustion in flow stats Mohammad Shuab Siddique
2026-09-18 3:27 ` [PATCH 2/5] net/bnxt: fix bounds on TPA aggregation ID from completions Mohammad Shuab Siddique
2026-09-18 3:27 ` [PATCH 3/5] net/bnxt: harden sprintf bounds for device memory names Mohammad Shuab Siddique
2026-09-18 3:27 ` [PATCH 4/5] net/bnxt: fix bounds in MAC pool index and flow parsing Mohammad Shuab Siddique
2026-09-18 3:27 ` [PATCH 5/5] net/bnxt: fix TPA agg Rx descriptor and VNIC query bounds Mohammad Shuab Siddique
2026-09-21 2:24 ` [PATCH v2 0/5] net/bnxt: fix flow, Rx and naming bounds issues Mohammad Shuab Siddique
2026-09-21 2:24 ` [PATCH v2 1/5] net/bnxt: fix stack exhaustion in flow stats Mohammad Shuab Siddique
2026-09-21 2:24 ` [PATCH v2 2/5] net/bnxt: fix bounds on TPA aggregation ID from completions Mohammad Shuab Siddique
2026-09-21 2:24 ` [PATCH v2 3/5] net/bnxt: harden sprintf bounds for device memory names Mohammad Shuab Siddique
2026-09-21 15:49 ` Stephen Hemminger
2026-09-21 2:24 ` [PATCH v2 4/5] net/bnxt: fix bounds in MAC pool index and flow parsing Mohammad Shuab Siddique
2026-09-21 15:50 ` Stephen Hemminger
2026-09-21 2:24 ` [PATCH v2 5/5] net/bnxt: fix TPA agg Rx descriptor and VNIC query bounds Mohammad Shuab Siddique
2026-09-29 0:24 ` [PATCH v3 0/5] net/bnxt: fix flow, Rx and naming bounds issues Mohammad Shuab Siddique
2026-09-29 0:24 ` [PATCH v3 1/5] net/bnxt: fix stack exhaustion in flow stats Mohammad Shuab Siddique
2026-09-29 0:24 ` Mohammad Shuab Siddique [this message]
2026-09-29 0:24 ` [PATCH v3 3/5] net/bnxt: harden sprintf bounds for device memory names Mohammad Shuab Siddique
2026-09-29 0:24 ` [PATCH v3 4/5] net/bnxt: fix bounds in MAC address pool index Mohammad Shuab Siddique
2026-09-29 0:24 ` [PATCH v3 5/5] net/bnxt: fix TPA agg Rx descriptor and VNIC query bounds Mohammad Shuab Siddique
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260929002442.1208481-3-Mohammad-Shuab.Siddique@broadcom.com \
--to=mohammad-shuab.siddique@broadcom.com \
--cc=dev@dpdk.org \
--cc=kishore.padmanabha@broadcom.com \
--cc=stable@dpdk.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox