Thread (23 messages) 23 messages, 3 authors, 2026-01-29

Re: [net-next,v6,01/13] octeontx2-af: npc: cn20k: Index management

From: Jakub Kicinski <kuba@kernel.org>
Date: 2026-01-29 05:13:03
Also in: lkml

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.

For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
octeontx2-af: npc: cn20k: Index management

This patch adds MCAM index management for CN20K silicon, implementing
contiguous and non-contiguous allocation schemes across the new subbank
architecture with 32 subbanks of 256 entries each.
quoted hunk ↗ jump to hunk
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/cn20k/npc.c b/drivers/net/ethernet/marvell/octeontx2/af/cn20k/npc.c
new file mode 100644
index 000000000000..30cc05c64487
--- /dev/null
+++ b/drivers/net/ethernet/marvell/octeontx2/af/cn20k/npc.c
[ ... ]
+static int npc_create_srch_order(int cnt)
+{
+	int val = 0;
+
+	subbank_srch_order = kcalloc(cnt, sizeof(int),
+				     GFP_KERNEL);
+	if (!subbank_srch_order)
+		return -ENOMEM;
+
+	for (int i = 0; i < cnt; i += 2) {
+		subbank_srch_order[i] = cnt / 2 - val - 1;
+		subbank_srch_order[i + 1] = cnt / 2 + 1 + val;
                                   ^^^^^

When cnt is odd, the final loop iteration has i = cnt - 1, which means
subbank_srch_order[i + 1] writes to index cnt, which is out of bounds.

For example with cnt = 5, the loop iterates with i = 0, 2, 4. At i = 4,
the array access subbank_srch_order[5] overflows the allocated array of
5 elements.

While CN20K hardware typically has 32 subbanks (even), the value is read
from a hardware register at npc_priv_init() without validation beyond
checking for zero. Could this lead to memory corruption if the hardware
reports an odd subbank count?
+		val++;
+	}
+
+	subbank_srch_order[cnt - 1] = cnt / 2;
+	return 0;
+}
[ ... ]
-- 
pw-bot: cr
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help