The kmemleak reported that
...
unreferenced object 0xffff888108d6c300 (size 256):
comm "cryptomgr_test", pid 183, jiffies 4294700957
hex dump (first 32 bytes):
00 c1 d6 08 81 88 ff ff 00 cb d6 08 81 88 ff ff ................
50 cd a7 81 ff ff ff ff b8 fb 93 02 00 c9 ff ff P...............
backtrace (crc 29cca632):
__kmalloc_noprof+0x2fa/0x430
test_acomp+0x174/0x960
alg_test_comp+0x6f/0x90
alg_test.part.26+0x105/0x410
cryptomgr_test+0x20/0x40
kthread+0x10c/0x250
ret_from_fork+0x30/0x40
ret_from_fork_asm+0x1a/0x30
unreferenced object 0xffff888108d6c100 (size 256):
comm "cryptomgr_test", pid 183, jiffies 4294700972
hex dump (first 32 bytes):
00 1d da 08 81 88 ff ff 00 c3 d6 08 81 88 ff ff ................
50 cd a7 81 ff ff ff ff b8 fb 93 02 00 c9 ff ff P...............
backtrace (crc 3047d62b):
__kmalloc_noprof+0x2fa/0x430
test_acomp+0x174/0x960
alg_test_comp+0x6f/0x90
alg_test.part.26+0x105/0x410
cryptomgr_test+0x20/0x40
kthread+0x10c/0x250
ret_from_fork+0x30/0x40
ret_from_fork_asm+0x1a/0x30
acomp_request will be chained to req[0], however,
acomp_request_free(), it will not free the whole chain.
Fix it by freeing them one by one.
Fixes: 99585c2192cb ("crypto: testmgr - Add multibuffer acomp testing")
Reported-by: Erhard Furtner <erhard_f@xxxxxxxxxxx>
Closes: https://lore.kernel.org/linux-crypto/20250408002741.089f1e9a@xxxxxxxxxxxxx/
Signed-off-by: Li Zhijian <lizhijian@xxxxxxxxxxx>
---
crypto/testmgr.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/crypto/testmgr.c b/crypto/testmgr.c
index abd609d4c8ef..7f02feee9fb6 100644
--- a/crypto/testmgr.c
+++ b/crypto/testmgr.c
@@ -3498,10 +3498,10 @@ static int test_acomp(struct crypto_acomp *tfm,
ret = 0;
out:
- acomp_request_free(reqs[0]);
for (i = 0; i < MAX_MB_MSGS; i++) {
kfree(output[i]);
kfree(decomp_out[i]);
+ acomp_request_free(reqs[i]);
}
kfree(dst);
kfree(src);
--
2.27.0
