[RFC AR6000 hif patch] Clean up code and do_irq on a deferrable worker

Michael Trimarchi trimarchi at gandalf.sssup.it
Sat Mar 21 10:03:06 CET 2009


Werner Almesberger wrote:
> Michael Trimarchi wrote:
>> Little clean up and move the do_irq on a deferrable work
> Hmm, did the current solution cause any problems ? 
I forgot to answer to this question. I find panics in the wifi and
I start to understand the code. So first of all I clean up the code
of the hif2 and move to a deferrable lock to avoid the release of the
lock. The bug remain and decide to take a look to the wext_ioctl.c and
I don't find any locking during remove of the device, so I change it too.
I don't mesaure the performance now. For reproduce the panic in the current
version of the kernel is simple:

request a scan from the user space and in the same time remove the device
with rfkill and the ioctl saw the priv data removed. I do it by hand, so
in android it can happen frequently just send the device to suspend when
the system in background scan for the network.

I post all the patches because we can discuss each modification and maybe
revert somenthing and provide better patches.


More information about the openmoko-kernel mailing list