[PATCH 4.19 20/49] rapidio: handle create_workqueue() failure

From: Greg Kroah-Hartman
Date: Mon May 24 2021 - 11:49:27 EST


From: Anirudh Rayabharam <mail@xxxxxxxxxxxxx>

commit 69ce3ae36dcb03cdf416b0862a45369ddbf50fdf upstream.

In case create_workqueue() fails, release all resources and return -ENOMEM
to caller to avoid potential NULL pointer deref later. Move up the
create_workequeue() call to return early and avoid unwinding the call to
riocm_rx_fill().

Cc: Alexandre Bounine <alex.bou9@xxxxxxxxx>
Cc: Matt Porter <mporter@xxxxxxxxxxxxxxxxxxx>
Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: stable <stable@xxxxxxxxxxxxxxx>
Signed-off-by: Anirudh Rayabharam <mail@xxxxxxxxxxxxx>
Link: https://lore.kernel.org/r/20210503115736.2104747-46-gregkh@xxxxxxxxxxxxxxxxxxx
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
drivers/rapidio/rio_cm.c | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)

--- a/drivers/rapidio/rio_cm.c
+++ b/drivers/rapidio/rio_cm.c
@@ -2136,6 +2136,14 @@ static int riocm_add_mport(struct device
return -ENODEV;
}

+ cm->rx_wq = create_workqueue(DRV_NAME "/rxq");
+ if (!cm->rx_wq) {
+ rio_release_inb_mbox(mport, cmbox);
+ rio_release_outb_mbox(mport, cmbox);
+ kfree(cm);
+ return -ENOMEM;
+ }
+
/*
* Allocate and register inbound messaging buffers to be ready
* to receive channel and system management requests
@@ -2146,7 +2154,6 @@ static int riocm_add_mport(struct device
cm->rx_slots = RIOCM_RX_RING_SIZE;
mutex_init(&cm->rx_lock);
riocm_rx_fill(cm, RIOCM_RX_RING_SIZE);
- cm->rx_wq = create_workqueue(DRV_NAME "/rxq");
INIT_WORK(&cm->rx_work, rio_ibmsg_handler);

cm->tx_slot = 0;