Skip to contentSkip to conversation

selftests/bpf: Fix csum_partial() dropping trailing byte on odd length

PATCH Superseded Madhav Khosla 3 participants 4 messages Sep 16 – Sep 18

Subsystems (5)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 selftests/bpf: Fix csum_partial() dropping trailing byte on odd length

csum_partial() computes num_u16 = len >> 1 and only sums that many 16-bit words, so the last byte of an odd-length buffer never gets added to the checksum. RFC 1071 says it should be padded with a zero byte and summed as one more word, not dropped.

This backs build_ip_csum(), build_udp_v4_csum() and build_udp_v6_csum(), used by flow_dissector_classification.c and xdp_metadata.c to hand-build packets. No current caller builds an odd-length payload, so nothing fails today, but a future one would get a silently wrong checksum.

Also bump flow_dissector_classification's TEST_PACKET_LEN from 100 to 99 so this actually gets exercised instead of staying latent.

Without the fix and with TEST_PACKET_LEN=99, flow_dissector_classification fails under vmtest.sh:

test_flow_dissector_classification:FAIL:test third port unexpected test third port: actual 0 != expected 10
#137/6 flow_dissector_classification/ipv6:FAIL
#137 flow_dissector_classification:FAIL
Summary: 1/0 PASSED, 0 SKIPPED, 1/6 FAILED

Kernel just drops the packet over a bad checksum. With the fix, both flow_dissector_classification and xdp_metadata pass.

v1 -> v2:
- comment style: opening /* on its own line, per BPF selftests style
- Retarget the Fixes tag to bcc00987bc56. commit f4504af68575 ("selftests/bpf: move ip checksum helper to network helpers") moved the helper, but sizeof(iphdr) is always a multiple of 32 bit words / 4 Bytes (iph->ihl counts in 4-byte words), so the odd-length path was never reachable through build_ip_csum(). csum_partial() first gets called with a length that isn't guaranteed even in bcc00987bc56, via build_udp_v4_csum()/build_udp_v6_csum().
- TEST_PACKET_LEN 100 -> 99 so an existing test catches this instead of the bug staying unexercised

Fixes bcc00987bc56 ("selftests/bpf: add network helpers to generate udp checksums")
---
 tools/testing/selftests/bpf/network_helpers.h     | 15 +++++++++++++--
 .../prog_tests/flow_dissector_classification.c    |  2 +-
 2 files changed, 14 insertions(+), 3 deletions(-)
tools/testing/selftests/bpf/network_helpers.h +13 -2
index 75133119c04a..878c9fc5c37c 100644
@@ -129,12 +129,23 @@ static __u16 csum_fold(__u32 csum)
 
 static __wsum csum_partial(const void *buf, int len, __wsum sum)
 {
-	__u16 *p = (__u16 *)buf;
+	const __u8 *p = buf;
 	int num_u16 = len >> 1;
 	int i;
 
 	for (i = 0; i < num_u16; i++)
-		sum += p[i];
+		sum += ((const __u16 *)p)[i];
+
+	/*
+	 * RFC 1071: an odd-length buffer's trailing byte is paired with
+	 * a zero pad byte to form the final 16-bit word.
+	 */
+	if (len & 1) {
+		__u16 tail = 0;
+
+		__builtin_memcpy(&tail, p + len - 1, 1);
+		sum += tail;
+	}
 
 	return sum;
 }
tools/testing/selftests/bpf/prog_tests/flow_dissector_classification.c +1 -1
index 80b153d3ddec..421dfa6c4ea3 100644
@@ -27,7 +27,7 @@
 #define TEST_NAME_MAX_LEN (32 + SUBTEST_NAME_MAX_LEN)
 #define MAX_SOURCE_PORTS 3
 #define TEST_PACKETS_COUNT 10
-#define TEST_PACKET_LEN 100
+#define TEST_PACKET_LEN 99
 #define TEST_PACKET_PATTERN 'a'
 #define TEST_IPV4 "192.168.0.1/32"
 #define TEST_IPV6 "100::a/128"
Signature · 1 line
2.55.0
#1 lore raw
The commit message body includes:
Quoted 35 lines
tools/testing/selftests/bpf/network_helpers.h +13 -2
index 75133119c04a3..878c9fc5c37ca 100644
@@ -129,12 +129,23 @@ static __u16 csum_fold(__u32 csum)
 
 static __wsum csum_partial(const void *buf, int len, __wsum sum)
 {
-	__u16 *p = (__u16 *)buf;
+	const __u8 *p = buf;
 	int num_u16 = len >> 1;
 	int i;
 
 	for (i = 0; i < num_u16; i++)
-		sum += p[i];
+		sum += ((const __u16 *)p)[i];
+
+	/*
+	 * RFC 1071: an odd-length buffer's trailing byte is paired with
+	 * a zero pad byte to form the final 16-bit word.
+	 */
+	if (len & 1) {
+		__u16 tail = 0;
+
+		__builtin_memcpy(&tail, p + len - 1, 1);
+		sum += tail;
+	}
 
 	return sum;
 }
tools/testing/selftests/bpf/prog_tests/flow_dissector_classification.c +0 -0
index 80b153d3ddecf..421dfa6c4ea36 100644

tools/testing/selftests/bpf/prog_tests/flow_dissector_classification.c line 33

Quoted 8 lines
tools/testing/selftests/bpf/prog_tests/flow_dissector_classification.c +1 -1
@@ -27,7 +27,7 @@
 #define TEST_NAME_MAX_LEN (32 + SUBTEST_NAME_MAX_LEN)
 #define MAX_SOURCE_PORTS 3
 #define TEST_PACKETS_COUNT 10
-#define TEST_PACKET_LEN 100
+#define TEST_PACKET_LEN 99
 #define TEST_PACKET_PATTERN 'a'
 #define TEST_IPV4 "192.168.0.1/32"
 #define TEST_IPV6 "100::a/128"

The commit message body includes:

Quoted 11 lines

v1 -> v2:
- comment style: opening /* on its own line, per BPF selftests style
- Retarget the Fixes tag to bcc00987bc56. commit f4504af68575 ("selftests/bpf: move ip checksum helper to network helpers") moved the helper, but sizeof(iphdr) is always a multiple of 32 bit words / 4 Bytes (iph->ihl counts in 4-byte words), so the odd-length path was never reachable through build_ip_csum(). csum_partial() first gets called with a length that isn't guaranteed even in bcc00987bc56, via build_udp_v4_csum()/build_udp_v6_csum().
- TEST_PACKET_LEN 100 -> 99 so an existing test catches this instead of the bug staying unexercised

This isn't a bug, but could the 'v1 -> v2:' block move below the '---' line so it stays out of the committed changelog? Documentation/process/ submitting-patches.rst asks for inter-version changelogs to go below the '---' separator so they are dropped when the patch is applied. The content itself is useful review material, especially the reasoning for retargeting the Fixes tag.

The commit message also includes this test failure excerpt:

Quoted 8 lines

Without the fix and with TEST_PACKET_LEN=99, flow_dissector_classification fails under vmtest.sh:

test_flow_dissector_classification:FAIL:test third port unexpected test third port: actual 0 != expected 10
#137/6 flow_dissector_classification/ipv6:FAIL
#137 flow_dissector_classification:FAIL
Summary: 1/0 PASSED, 0 SKIPPED, 1/6 FAILED

This isn't a bug, but would it be clearer to paste the full failure output or mark the elision? The summary line says six subtests failed (1/6 FAILED) while only the ipv6 one is shown, and test_progs.c runs three ASSERT_EQ calls per subtest ('test first port', 'test second port', 'test third port'), so 'test first port' would have failed too. And '1/0 PASSED' reports one passing test, which cannot come from a run whose only named test failed.

---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

CI run summary: https://github.com/kernel-patches/bpf/actions/runs/35091050280

#2 lore raw
Quoted 7 lines

This isn't a bug, but would it be clearer to paste the full failure output or mark the elision? The summary line says six subtests failed (1/6 FAILED) while only the ipv6 one is shown, and test_progs.c runs three ASSERT_EQ calls per subtest ('test first port', 'test second port', 'test third port'), so 'test first port' would have failed too. And '1/0 PASSED' reports one passing test, which cannot come from a run whose only named test failed.

Not a bug. That log is the test failing on the TEST_PACKET_LEN=99 case against the old (pre-fix) csum_partial(), not my patch. My change is what makes the test catch it; it passes with the fix applied.

Madhav

Quoted 91 lines · On Wed, Sep 16, 2026 at 12:13 PM <[email protected]> wrote:
Quoted 36 lines
tools/testing/selftests/bpf/network_helpers.h +13 -2
index 75133119c04a3..878c9fc5c37ca 100644
@@ -129,12 +129,23 @@ static __u16 csum_fold(__u32 csum)
 
 static __wsum csum_partial(const void *buf, int len, __wsum sum)
 {
-     __u16 *p = (__u16 *)buf;
+     const __u8 *p = buf;
      int num_u16 = len >> 1;
      int i;
 
      for (i = 0; i < num_u16; i++)
-             sum += p[i];
+             sum += ((const __u16 *)p)[i];
+
+     /*
+      * RFC 1071: an odd-length buffer's trailing byte is paired with
+      * a zero pad byte to form the final 16-bit word.
+      */
+     if (len & 1) {
+             __u16 tail = 0;
+
+             __builtin_memcpy(&tail, p + len - 1, 1);
+             sum += tail;
+     }
 
      return sum;
 }
tools/testing/selftests/bpf/prog_tests/flow_dissector_classification.c +0 -0
index 80b153d3ddecf..421dfa6c4ea36 100644

tools/testing/selftests/bpf/prog_tests/flow_dissector_classification.c line 33

Quoted 8 lines
tools/testing/selftests/bpf/prog_tests/flow_dissector_classification.c +1 -1
@@ -27,7 +27,7 @@
 #define TEST_NAME_MAX_LEN (32 + SUBTEST_NAME_MAX_LEN)
 #define MAX_SOURCE_PORTS 3
 #define TEST_PACKETS_COUNT 10
-#define TEST_PACKET_LEN 100
+#define TEST_PACKET_LEN 99
 #define TEST_PACKET_PATTERN 'a'
 #define TEST_IPV4 "192.168.0.1/32"
 #define TEST_IPV6 "100::a/128"

The commit message body includes:

Quoted 11 lines

v1 -> v2:
- comment style: opening /* on its own line, per BPF selftests style
- Retarget the Fixes tag to bcc00987bc56. commit f4504af68575 ("selftests/bpf: move ip checksum helper to network helpers") moved the helper, but sizeof(iphdr) is always a multiple of 32 bit words / 4 Bytes (iph->ihl counts in 4-byte words), so the odd-length path was never reachable through build_ip_csum(). csum_partial() first gets called with a length that isn't guaranteed even in bcc00987bc56, via build_udp_v4_csum()/build_udp_v6_csum().
- TEST_PACKET_LEN 100 -> 99 so an existing test catches this instead of the bug staying unexercised

This isn't a bug, but could the 'v1 -> v2:' block move below the '---' line so it stays out of the committed changelog? Documentation/process/ submitting-patches.rst asks for inter-version changelogs to go below the '---' separator so they are dropped when the patch is applied. The content itself is useful review material, especially the reasoning for retargeting the Fixes tag.

The commit message also includes this test failure excerpt:

Quoted 8 lines

Without the fix and with TEST_PACKET_LEN=99, flow_dissector_classification fails under vmtest.sh:

test_flow_dissector_classification:FAIL:test third port unexpected test third port: actual 0 != expected 10
#137/6 flow_dissector_classification/ipv6:FAIL
#137 flow_dissector_classification:FAIL
Summary: 1/0 PASSED, 0 SKIPPED, 1/6 FAILED

This isn't a bug, but would it be clearer to paste the full failure output or mark the elision? The summary line says six subtests failed (1/6 FAILED) while only the ipv6 one is shown, and test_progs.c runs three ASSERT_EQ calls per subtest ('test first port', 'test second port', 'test third port'), so 'test first port' would have failed too. And '1/0 PASSED' reports one passing test, which cannot come from a run whose only named test failed.

---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

CI run summary: https://github.com/kernel-patches/bpf/actions/runs/35091050280

#3 lore raw
In reply to Madhav Khosla

Hi Madhav,

Quoted 5 lines · Madhav Khosla

csum_partial() computes num_u16 = len >> 1 and only sums that many 16-bit words, so the last byte of an odd-length buffer never gets added to the checksum. RFC 1071 says it should be padded with a zero byte and summed as one more word, not dropped.

Quoted 8 lines

This backs build_ip_csum(), build_udp_v4_csum() and build_udp_v6_csum(), used by flow_dissector_classification.c and xdp_metadata.c to hand-build packets. No current caller builds an odd-length payload, so nothing fails today, but a future one would get a silently wrong checksum.

Also bump flow_dissector_classification's TEST_PACKET_LEN from 100 to 99 so this actually gets exercised instead of staying latent.

Great, thanks for the update.

Quoted 12 lines

Without the fix and with TEST_PACKET_LEN=99, flow_dissector_classification fails under vmtest.sh:

test_flow_dissector_classification:FAIL:test third port unexpected test third port: actual 0 != expected 10
#137/6 flow_dissector_classification/ipv6:FAIL
#137 flow_dissector_classification:FAIL
Summary: 1/0 PASSED, 0 SKIPPED, 1/6 FAILED

Kernel just drops the packet over a bad checksum. With the fix, both flow_dissector_classification and xdp_metadata pass.

Quoted 11 lines

v1 -> v2:
- comment style: opening /* on its own line, per BPF selftests style
- Retarget the Fixes tag to bcc00987bc56. commit f4504af68575 ("selftests/bpf: move ip checksum helper to network helpers") moved the helper, but sizeof(iphdr) is always a multiple of 32 bit words / 4 Bytes (iph->ihl counts in 4-byte words), so the odd-length path was never reachable through build_ip_csum(). csum_partial() first gets called with a length that isn't guaranteed even in bcc00987bc56, via build_udp_v4_csum()/build_udp_v6_csum().
- TEST_PACKET_LEN 100 -> 99 so an existing test catches this instead of the bug staying unexercised

Could you please move the patch changelog under --- so it does not end up in the commit once it is merged ? You can also drop the Link tag, one will automatically be added on the latest revision once it is merged.

Up to some extent, you could also move the ./test_progs sample output in the comments under --- as well.

Quoted 1 line
Fixes bcc00987bc56 ("selftests/bpf: add network helpers to generate udp checksums")

You chose to use bcc00987bc56 rather than e2a46d54d7a1, but I guess it is ok.

Quoted 2 lines

Thanks,

Alexis

Signature · 3 lines
Alexis Lothoré, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com