* [PATCH net] octeontx2-af: Fix uninitialized variables and bounds in MCS alloc handlers
@ 2026-11-19 17:05 Ian Rogers
0 siblings, 0 replies; only message in thread
From: Ian Rogers @ 2026-11-19 17:05 UTC (permalink / raw)
To: Sunil Goutham, Ratheesh Kannoth, Geetha sowjanya,
Subbaraya Sundeep, Andrew Lunn, David S. Miller, Eric Dumazet,
Jakub Kicinski, Paolo Abeni
Cc: netdev, linux-kernel, llvm, Elkin Cruz, Ian Rogers
In rvu_mbox_handler_mcs_alloc_ctrl_pkt_rule(), switch (req->rule_type) has
no default case; an unrecognized rule_type in a PF/VF mailbox request
leaves 'offset' uninitialized when passed to mcs_alloc_ctrlpktrule(),
allocating rules from the wrong hardware partition or failing spuriously.
In rvu_mbox_handler_mcs_alloc_resources(), 'rsrc_id' is left uninitialized
when !req->all and either req->rsrc_cnt == 0 or req->rsrc_type is not one
of the four handled types. Moreover, req->rsrc_cnt (u8, 0..255) is not
checked against the 128-element rsp->flow_ids, rsp->secy_ids, and
rsp->sc_ids response arrays before writing into them.
Reject invalid rule_type, unrecognized rsrc_type, zero rsrc_cnt when
!req->all, and rsrc_cnt values exceeding the corresponding response array
bounds.
Found via an x86_64 LLVM=1 (Clang 21.1.8) allyesconfig audit with
-Wconditional-uninitialized enabled; fix and changelog drafted with an AI
coding assistant. Build-tested under LLVM=1 W=1 and
-Wconditional-uninitialized; not tested on Marvell CN10K hardware.
Fixes: cfc14181d497 ("octeontx2-af: cn10k: mcs: Manage the MCS block hardware resources")
Signed-off-by: Ian Rogers <irogers@google.com>
---
.../ethernet/marvell/octeontx2/af/mcs_rvu_if.c | 16 +++++++++++++++-
1 file changed, 15 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c b/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
index cd9cca5a9cf5..54c1d9809ae6 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
@@ -682,12 +682,15 @@ int rvu_mbox_handler_mcs_alloc_resources(struct rvu *rvu,
{
u16 pcifunc = req->hdr.pcifunc;
struct mcs_rsrc_map *map;
+ int rsrc_id = -EINVAL, i;
struct mcs *mcs;
- int rsrc_id, i;
if (req->mcs_id >= rvu->mcs_blk_cnt)
return MCS_AF_ERR_INVALID_MCSID;
+ if (!req->all && !req->rsrc_cnt)
+ return -EINVAL;
+
mcs = mcs_get_pdata(req->mcs_id);
if (req->dir == MCS_RX)
@@ -709,6 +712,8 @@ int rvu_mbox_handler_mcs_alloc_resources(struct rvu *rvu,
switch (req->rsrc_type) {
case MCS_RSRC_TYPE_FLOWID:
+ if (req->rsrc_cnt > ARRAY_SIZE(rsp->flow_ids))
+ goto exit;
for (i = 0; i < req->rsrc_cnt; i++) {
rsrc_id = mcs_alloc_rsrc(&map->flow_ids, map->flowid2pf_map, pcifunc);
if (rsrc_id < 0)
@@ -718,6 +723,8 @@ int rvu_mbox_handler_mcs_alloc_resources(struct rvu *rvu,
}
break;
case MCS_RSRC_TYPE_SECY:
+ if (req->rsrc_cnt > ARRAY_SIZE(rsp->secy_ids))
+ goto exit;
for (i = 0; i < req->rsrc_cnt; i++) {
rsrc_id = mcs_alloc_rsrc(&map->secy, map->secy2pf_map, pcifunc);
if (rsrc_id < 0)
@@ -727,6 +734,8 @@ int rvu_mbox_handler_mcs_alloc_resources(struct rvu *rvu,
}
break;
case MCS_RSRC_TYPE_SC:
+ if (req->rsrc_cnt > ARRAY_SIZE(rsp->sc_ids))
+ goto exit;
for (i = 0; i < req->rsrc_cnt; i++) {
rsrc_id = mcs_alloc_rsrc(&map->sc, map->sc2pf_map, pcifunc);
if (rsrc_id < 0)
@@ -744,6 +753,8 @@ int rvu_mbox_handler_mcs_alloc_resources(struct rvu *rvu,
rsp->rsrc_cnt++;
}
break;
+ default:
+ goto exit;
}
rsp->rsrc_type = req->rsrc_type;
@@ -793,6 +804,9 @@ int rvu_mbox_handler_mcs_alloc_ctrl_pkt_rule(struct rvu *rvu,
case MCS_CTRL_PKT_RULE_TYPE_MAC:
offset = MCS_CTRLPKT_MAC_EN_RULE_OFFSET;
break;
+ default:
+ rsrc_id = -EINVAL;
+ goto exit;
}
rsrc_id = mcs_alloc_ctrlpktrule(&map->ctrlpktrule, map->ctrlpktrule2pf_map, offset,
--
2.56.1.708.g77570a6a1d-goog
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2026-11-19 17:05 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-19 17:05 [PATCH net] octeontx2-af: Fix uninitialized variables and bounds in MCS alloc handlers Ian Rogers
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®