On Thu, 2014-01-30 at 14:19 -0600, Larry Finger wrote:

On 01/30/2014 12:22 AM, Olivier Langlois wrote:

quoted

Signed-off-by: Olivier Langlois <redacted>
---
  drivers/net/wireless/rtlwifi/ps.c           |  2 +-
  drivers/net/wireless/rtlwifi/rtl8192ce/hw.c | 18 ++++++++++++++++--
  2 files changed, 17 insertions(+), 3 deletions(-)

Olivier,

I like the fact that you have proposed a solution that has minimal effect on the 
other PCI drivers under rtlwifi. That certainly decreases the amount of testing 
needed. Of course, all of them may need the same kind of fix.

This has been done on purpose. I knew that you would appreciate. The
problem is probably present with the other cards but since I could not
validate or test the patch with them, I have left them unchanged. If you
were to determine that it apply to all of them then maybe the patch
could be moved to ps.c to wrap hw_init() call.

The changes are certainly minimal; however, I do need to test them before giving 
my Ack.

As the problem(s) fixed by this patch will affect stable kernels, you should add 
a "Cc: Stable [off-list ref]" patch.

Originally, I was going to have you add a comment to the commit message on why 
you were changing the return value in rtl_ps_enable_nic(), but now I am thinking 
that this should be a separate commit as it fixes a totally different bug. Yes 
it is involved with the callback to hw_init, but the bug is independent.

done. I am about to resend the patch splitted into 2 commits.

Good work on finding this problem.

my pleasure. it has been fun and gratifying.