diff mbox series

[3/3] net/tap: compute TC handle correctly

Message ID 20240229173154.121491-4-stephen@networkplumber.org (mailing list archive)
State Accepted, archived
Delegated to: Ferruh Yigit
Headers
Series TAP device fixes |

Checks

Context Check Description
ci/checkpatch warning coding style issues
ci/loongarch-compilation success Compilation OK
ci/loongarch-unit-testing success Unit Testing PASS
ci/Intel-compilation success Compilation OK
ci/intel-Testing success Testing PASS
ci/intel-Functional success Functional PASS
ci/github-robot: build success github build: passed
ci/iol-intel-Performance success Performance Testing PASS
ci/iol-mellanox-Performance success Performance Testing PASS
ci/iol-abi-testing success Testing PASS
ci/iol-unit-arm64-testing success Testing PASS
ci/iol-compile-amd64-testing success Testing PASS
ci/iol-compile-arm64-testing success Testing PASS
ci/iol-unit-amd64-testing success Testing PASS
ci/iol-intel-Functional success Functional Testing PASS
ci/iol-broadcom-Functional success Functional Testing PASS
ci/iol-broadcom-Performance success Performance Testing PASS
ci/iol-sample-apps-testing success Testing PASS

Commit Message

Stephen Hemminger Feb. 29, 2024, 5:31 p.m. UTC 
  The code to take a flow pointer and make a TC handle was incorrect
and would always generate the same handle. This is because it was
hashing the address of the union on the stack (which is invariant)
rather than the contents of the union.

The following testpmd case would cause an error:
testpmd> flow create 0 ingress pattern eth src is 06:05:04:03:02:01 \
  / end actions queue index 2 / end
Flow rule #0 created
testpmd> flow create 0 ingress pattern eth src is 06:05:04:03:02:02 \
  / end actions queue index 3 / end
tap_nl_dump_ext_ack(): Filter already exists
tap_flow_create(): Kernel refused TC filter rule creation (17): File exists
port_flow_complain(): Caught PMD error type 2 (flow rule (handle)):
  overlapping rules or Kernel too old for flower support: File exists

This fix does it in a more robust manner using size independent
code. It also initializes the hash seed so the same hash won't
show up every time and risk potential leakage of address to
other places.

Bugzilla ID: 1382
Fixes: de96fe68ae95 ("net/tap: add basic flow API patterns and actions")
Fixes: a625ab89df11 ("net/tap: fix build with GCC 11")
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
---
 drivers/net/tap/tap_flow.c | 23 ++++++++++++-----------
 1 file changed, 12 insertions(+), 11 deletions(-)
diff mbox series

Patch

diff --git a/drivers/net/tap/tap_flow.c b/drivers/net/tap/tap_flow.c
index 5b0fee906493..94f9b99cdaa6 100644
--- a/drivers/net/tap/tap_flow.c
+++ b/drivers/net/tap/tap_flow.c
@@ -11,6 +11,7 @@ 
 
 #include <rte_byteorder.h>
 #include <rte_jhash.h>
+#include <rte_random.h>
 #include <rte_malloc.h>
 #include <rte_eth_tap.h>
 #include <tap_flow.h>
@@ -1297,9 +1298,7 @@  tap_flow_validate(struct rte_eth_dev *dev,
  * In those rules, the handle (uint32_t) is the part that would identify
  * specifically each rule.
  *
- * On 32-bit architectures, the handle can simply be the flow's pointer address.
- * On 64-bit architectures, we rely on jhash(flow) to find a (sufficiently)
- * unique handle.
+ * Use jhash of the flow poitner to make a unique handle.
  *
  * @param[in, out] flow
  *   The flow that needs its handle set.
@@ -1309,16 +1308,18 @@  tap_flow_set_handle(struct rte_flow *flow)
 {
 	union {
 		struct rte_flow *flow;
-		const void *key;
-	} tmp;
-	uint32_t handle = 0;
+		uint32_t words[sizeof(flow) / sizeof(uint32_t)];
+	} tmp = {
+		.flow = flow,
+	};
+	uint32_t handle;
+	static uint64_t hash_seed;
 
-	tmp.flow = flow;
+	if (hash_seed == 0)
+		hash_seed = rte_rand();
+
+	handle = rte_jhash_32b(tmp.words, sizeof(flow) / sizeof(uint32_t), hash_seed);
 
-	if (sizeof(flow) > 4)
-		handle = rte_jhash(tmp.key, sizeof(flow), 1);
-	else
-		handle = (uintptr_t)flow;
 	/* must be at least 1 to avoid letting the kernel choose one for us */
 	if (!handle)
 		handle = 1;