`efa_alloc_pd()` assumes that `udata` is not null but it seems like
`__ib_alloc_pd()` in drivers/infiniband/core/verbs.c:279 can call it
through the `ib_device` interface.  This checks if `udata` is null
before dereferencing it.

Fixes: 40909f664d27 ("RDMA/efa: Add EFA verbs implementation")
Signed-off-by: Thomas Fourier <fourier.thomas@gmail.com>
---
 drivers/infiniband/hw/efa/efa_verbs.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/efa/efa_verbs.c b/drivers/infiniband/hw/efa/efa_verbs.c
index a8645a40730f..fea634b7d825 100644
--- a/drivers/infiniband/hw/efa/efa_verbs.c
+++ b/drivers/infiniband/hw/efa/efa_verbs.c
@@ -427,7 +427,7 @@ int efa_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
 	struct efa_pd *pd = to_epd(ibpd);
 	int err;
 
-	if (udata->inlen &&
+	if (udata && udata->inlen &&
 	    !ib_is_udata_cleared(udata, 0, udata->inlen)) {
 		ibdev_dbg(&dev->ibdev,
 			  "Incompatible ABI params, udata not cleared\n");
@@ -442,7 +442,7 @@ int efa_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
 	pd->pdn = result.pdn;
 	resp.pdn = result.pdn;
 
-	if (udata->outlen) {
+	if (udata && udata->outlen) {
 		err = ib_copy_to_udata(udata, &resp,
 				       min(sizeof(resp), udata->outlen));
 		if (err) {
-- 
2.43.0

On 13/05/2025 16:22, Thomas Fourier wrote:
> `efa_alloc_pd()` assumes that `udata` is not null but it seems like
> `__ib_alloc_pd()` in drivers/infiniband/core/verbs.c:279 can call it
> through the `ib_device` interface.  This checks if `udata` is null
> before dereferencing it.
> 
> Fixes: 40909f664d27 ("RDMA/efa: Add EFA verbs implementation")
> Signed-off-by: Thomas Fourier <fourier.thomas@gmail.com>

udata will never be NULL since EFA does not support kverbs, and
userspace will always pass udata.

Look at ib_device_check_mandatory().