Hi Dan,
The mempool_free() function frees "request". Don't free the request
until after smbd_disconnect_rdma_connection() to avoid a use after free
bug.
Fixes: 5e65668c75c0 ("smb: client: let send_done() cleanup before calling smbd_disconnect_rdma_connection()")
Signed-off-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
---
fs/smb/client/smbdirect.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/smb/client/smbdirect.c b/fs/smb/client/smbdirect.c
index 58321e483a1a..162f8d1c548a 100644
--- a/fs/smb/client/smbdirect.c
+++ b/fs/smb/client/smbdirect.c
@@ -286,8 +286,8 @@ static void send_done(struct ib_cq *cq, struct ib_wc *wc)
if (wc->status != IB_WC_SUCCESS || wc->opcode != IB_WC_SEND) {
log_rdma_send(ERR, "wc->status=%d wc->opcode=%d\n",
wc->status, wc->opcode);
- mempool_free(request, request->info->request_mempool);
smbd_disconnect_rdma_connection(request->info);
+ mempool_free(request, request->info->request_mempool);
The correct fix is to use 'info' instead of 'request->info'
other than that the order needs to stay that way.
I already asked Steve to squash such a change into the
original commit (which is not yet upstream).
See:
https://lore.kernel.org/linux-cifs/cover.1754308712.git.metze@xxxxxxxxx/T/#m98a8607d7b83a11fd78547306836a872a2a27192
What was the test that triggered the problem?
Or did you only noticed it by looking at the code?
Thanks!
metze
