wait_for_completion_timeout() returns unsigned long (0 on timeout or
remaining jiffies) not int - so this type error allows for a
theoretically int overflow - though not in this case where TIMEOUT is
only HZ*2). To fix this type inconsistency the completion is wrapped
into the if() rather than introducing an additional unsigned long
variable.
Along with fixing this type inconsistency the fall-through if is
consolidated to a single if-block.
Signed-off-by: Nicholas Mc Guire <hof...@osadl.org>
---
Problem located with experimental API conformance checking cocci script
V3: As requested by Sven Van Asbroeck <thesve...@gmail.com> cleanup
the commit message to make it more clear what the impact of the
proposed change is....lets see if this is any better.
V2: The original patch's logic was wrong as it was skipping the
fall-through if so using the fix proposed by Sven Van Asbroeck
<thesve...@gmail.com> - This solution also eliminates the need
to introduce an additional variable - Thanks !
Patch was compile-tested with. x86_64_defconfig + OF=y, FIELDBUS_DEV=m,
HMS_ANYBUSS_BUS=m
(with an unrelated sparse warnings (cast to restricted __be16))
Patch is against 5.1-rc6 (localversion-next is next-20190426)
drivers/staging/fieldbus/anybuss/host.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/staging/fieldbus/anybuss/host.c
b/drivers/staging/fieldbus/anybuss/host.c
index e34d424..6227daf 100644
--- a/drivers/staging/fieldbus/anybuss/host.c
+++ b/drivers/staging/fieldbus/anybuss/host.c
@@ -1325,11 +1325,11 @@ anybuss_host_common_probe(struct device *dev,
* interrupt came in: ready to go !
*/
reset_deassert(cd);
- ret = wait_for_completion_timeout(&cd->card_boot, TIMEOUT);
- if (ret == 0)
+ if (!wait_for_completion_timeout(&cd->card_boot, TIMEOUT)) {
ret = -ETIMEDOUT;
- if (ret < 0)
goto err_reset;
+ }
+
/*
* according to the anybus docs, we're allowed to read these
* without handshaking / reserving the area
--
2.1.4
