On Sun, Oct 23, 2016 at 03:37:57PM +0300, Alex Briskin wrote:
Sparse error fix - different lock contexts for basic block.
Acquirement and release of spin lock was dependent on two separate
unprotected variable evaluations. Instead the condition evaluation result
is stored in a local boolean variable to make sure that the same context
that called the spin_lock will evoke spin_unlock.
I kind of feel like this doesn't improve readability.
Sparse kind of sucks at flow analysis so I feel like working around
Sparse limitations is a waste of time.
regards,
dan carpenter