Since second call of ssb_get_devtypedata() may fail as well as the
first one, the NULL return value in 'wl' will be later dereferenced in
calls to b43legacy_one_core_attach() and schedule_work().
Instead of merely warning about this failure with
B43legacy_WARN_ON(), properly return with error to avoid any further
NULL pointer dereferences.
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
Fixes: 75388acd0cd8 ("[B43LEGACY]: add mac80211-based driver for legacy BCM43xx devices")
Co-developed-by: Natalia Petrova <[email protected]>
Signed-off-by: Nikita Zhandarovich <[email protected]>
---
v2: fix issues with overlooked null-ptr-dereferences pointed out by
Simon Horman <[email protected]>
Link: https://lore.kernel.org/all/[email protected]/
drivers/net/wireless/broadcom/b43legacy/main.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/net/wireless/broadcom/b43legacy/main.c b/drivers/net/wireless/broadcom/b43legacy/main.c
index 760136638a95..5a706dd0b1a4 100644
--- a/drivers/net/wireless/broadcom/b43legacy/main.c
+++ b/drivers/net/wireless/broadcom/b43legacy/main.c
@@ -3857,7 +3857,11 @@ static int b43legacy_probe(struct ssb_device *dev,
if (err)
goto out;
wl = ssb_get_devtypedata(dev);
- B43legacy_WARN_ON(!wl);
+ if (!wl) {
+ B43legacy_WARN_ON(!wl);
+ err = -ENODEV;
+ goto out;
+ }
}
err = b43legacy_one_core_attach(dev, wl);
if (err)
--
2.25.1
On Fri, 21 Apr 2023 17:14:18 -0500
Larry Finger <[email protected]> wrote:
> > (err) goto out;
> > wl = ssb_get_devtypedata(dev);
> > - B43legacy_WARN_ON(!wl);
> > + if (!wl) {
> > + B43legacy_WARN_ON(!wl);
> > + err = -ENODEV;
> > + goto out;
> > + }
> > }
> > err = b43legacy_one_core_attach(dev, wl);
> > if (err)
>
> I do not recall seeing v1. One additional nitpick: The latest
> convention would have the subject as "wifi: b43legacy:...". Kalle may
> be able to fix this on merging, but it not, a v3 might be required.
> Otherwise, the patch is good.
>
> Reviewed-by: Larry Finger <[email protected]>
No, it's not good. It's wrong. I already replied to it.
wl can never be NULL here and the goto-out path is wrong (if there
was a chance for it to trigger).
Please drop this patch, Kalle.
--
Michael Büsch
https://bues.ch/