Skip to contentSkip to conversation

selftests/bpf: fix test_cgroup_storage on s390

PATCH Applied Ilya Leoshkevich 2 participants 2 messages Aug 20, 2019 – Aug 21, 2019

Subsystems (3)
Kselftest BPF BPF selftests
lore
Share & tools
lore

Ilya Leoshkevich

author
#0 lore raw

PATCH 1/1 selftests/bpf: fix test_cgroup_storage on s390

test_cgroup_storage fails on s390 with an assertion failure: packets are dropped when they shouldn't. The problem is that BPF_DW packet count is accessed as BPF_W with an offset of 0, which is not correct on big-endian machines.

Since the point of this test is not to verify narrow loads/stores, simply use BPF_DW when working with packet counts.

Fixes 68cfa3ac6b8d ("selftests/bpf: add a cgroup storage test")
Fixes 919646d2a3a9 ("selftests/bpf: extend the storage test to test per-cpu cgroup storage")
Signed-off-by Ilya Leoshkevich <[email protected]>
---
 tools/testing/selftests/bpf/test_cgroup_storage.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
tools/testing/selftests/bpf/test_cgroup_storage.c +3 -3
index 2fc4625c1a15..655729004391 100644
@@ -20,9 +20,9 @@ int main(int argc, char **argv)
 		BPF_MOV64_IMM(BPF_REG_2, 0), /* flags, not used */
 		BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
 			     BPF_FUNC_get_local_storage),
-		BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_0, 0),
+		BPF_LDX_MEM(BPF_DW, BPF_REG_3, BPF_REG_0, 0),
 		BPF_ALU64_IMM(BPF_ADD, BPF_REG_3, 0x1),
-		BPF_STX_MEM(BPF_W, BPF_REG_0, BPF_REG_3, 0),
+		BPF_STX_MEM(BPF_DW, BPF_REG_0, BPF_REG_3, 0),
 
 		BPF_LD_MAP_FD(BPF_REG_1, 0), /* map fd */
 		BPF_MOV64_IMM(BPF_REG_2, 0), /* flags, not used */

@@ -30,7 +30,7 @@ int main(int argc, char **argv)
 			     BPF_FUNC_get_local_storage),
 		BPF_MOV64_IMM(BPF_REG_1, 1),
 		BPF_STX_XADD(BPF_DW, BPF_REG_0, BPF_REG_1, 0),
-		BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, 0),
+		BPF_LDX_MEM(BPF_DW, BPF_REG_1, BPF_REG_0, 0),
 		BPF_ALU64_IMM(BPF_AND, BPF_REG_1, 0x1),
 		BPF_MOV64_REG(BPF_REG_0, BPF_REG_1),
 		BPF_EXIT_INSN(),
Signature · 1 line
2.21.0
maintainer Applied
#1 lore raw
Quoted 11 lines · Ilya Leoshkevich

test_cgroup_storage fails on s390 with an assertion failure: packets are dropped when they shouldn't. The problem is that BPF_DW packet count is accessed as BPF_W with an offset of 0, which is not correct on big-endian machines.

Since the point of this test is not to verify narrow loads/stores, simply use BPF_DW when working with packet counts.

Fixes 68cfa3ac6b8d ("selftests/bpf: add a cgroup storage test")
Fixes 919646d2a3a9 ("selftests/bpf: extend the storage test to test per-cpu cgroup storage")
Signed-off-by Ilya Leoshkevich <[email protected]>

Applied, thanks!