Re: [PATCH] wireless: convert reg_regdb_search_lock to mutex

[Johannes Berg <johannes@sipsolutions.net>]

On Tue, 2010-03-16 at 22:56 -0400, John W. Linville wrote:
> On Tue, Mar 16, 2010 at 05:54:50PM +0530, Senthil Balasubramanian wrote:
> > On Wed, Mar 17, 2010 at 05:54:16AM +0530, John W. Linville wrote:
> > > On Tue, Mar 16, 2010 at 03:33:38PM +0530, Senthil Balasubramanian wrote:
> > > > On Wed, Mar 17, 2010 at 01:19:00AM +0530, John W. Linville wrote:
> > > > > Stanse discovered that kmalloc can be called with GFP_KERNEL while
> > > > This commit log is confusing. It Should be "Stanse discovered kmalloc
> > > > was called with GFP_KERNEL". Obviously kmalloc with GFP_KERNEL shouldn't
> > > > be used while holding a spinlock.
> > > > > holding this spinlock.  It can be a mutex instead.
> > > 
> > > Not half so confusing as your criticism... :-)
> > sorry! if my mail wasn't proper. I didn't meant to blame/criticize.
> > I was confused when I read the commit log and so I replied.
> > > 
> > > Are you objecting to "can be" instead of "was"?
> > I am not objecting anything...
> 
> Criticism is fine -- I just don't understand what you were saying or
> what you think I should have said in the commit log.

English at work ... let me guess:

For John: "can" == "it could happen that"
For Senthil: "can" == "may"

(when really you may NOT do GFP_KERNEL allocations in atomic context)

johannes