Re: iwlwifi: regression in 3.8-rc4 and 3.7.3

[Seth Forshee]

On Sat, Jan 19, 2013 at 11:40:24PM -0800, Hugh Dickins wrote:
> On Sun, 20 Jan 2013, Grumbach, Emmanuel wrote:
> > > 
> > > On Sun, Jan 20, 2013 at 6:56 AM, Hugh Dickins <hughd@xxxxxxxxxx> wrote:
> > > > After sending the first 2MB, scp over wireless becomes unbearably
> > > > slow, with frequent stalls: on this ThinkPad T420s running 3.8-rc4 or 3.7.3.
> > > > Not always, but often.
> > > >
> > > 
> > > There is one pending iwlwifi-fixes, dunno if it will fix your issue.
> > > [1] http://git.kernel.org/?p=linux/kernel/git/iwlwifi/iwlwifi-
> > > fixes.git;a=commitdiff;h=c3e5d7181afb66657393066bccce0956fab09ab3
> > 
> > Right - so the 2 patches are unrelated, and the one Sedat is pointing at is relevant for 3.8 only. It fixes a bug that has been introduced in 3.8.
> 
> Yes, I have now given that one a try, but it does not affect my issue.
> 
> > Regarding the issue Hugh is suffering from, I have to say that I am a little confused since I am pretty sure the patch is right. Now, it might uncover other pre-existing bugs. All I can say is that I don't think that reverting the patch is *so* problematic since the patch really wants to solve a rare case. If it causes issues, we can simply revert it. It only means that when I will get a little bit of time, I might need to ask you a few logs. Thanks for the report (and the bisection).
> 
> Sure, let me know what to do, once you have time to investigate.
> 
> So far as I know, I'm the first to notice: so we can probably
> leave it in 3.8-rc for now, until/unless it gives wider trouble.
> 
> But reverting from 3.7-stable soon would be a good idea,
> regressions there being more tiresome.

I'm seeing iwlwifi performance problems in 3.8 as well, and have heard
from several other saying the same thing. Testing with iperf shows very
little data getting through most of the time.

Reverting f590dcec944552f9a4a61155810f3abd17d6465d on top of 3.8-rc6
seems to have eliminated the problems for me, so far at least. I'll
request some testing from others as well.

Seth

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/