On 2/7/24 02:23, Javier Carrasco wrote:
On 07.02.24 11:16, Dan Carpenter wrote:
On Wed, Feb 07, 2024 at 11:08:18AM +0100, Javier Carrasco wrote:
The ret variable should be initialized to 0 because if no irqs are
defined, the function should not fail i.e. the driver supports that case.
That is probably the reason why I did not notice in my tests.
These days everyone has the GCC extension to automatically zero out
stack variables so it wouldn't have shown up in testing. It's still
technically a bug, but I don't know how many people it would affect in
real life (probably a small number).
regards,
dan carpenter
That is right, but because a small number is greater than zero, I am
glad that you found this issue now before the driver is released.
By the way, Smatch caught this while gcc, coccinelle and the language
server protocol I use did not... great tool.
gcc is notoriously bad with finding uninitialized variable problems.
Maybe that is because of that gcc extension, I don't know. It is
quite annoying, though.
Guenter
