[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#664767: Brcmsmac driver woes, possible regression?

El 2012-06-23 a las 12:23 -0500, Jonathan Nieder escribió:

> Camaleón wrote:

> > I noticed there was a new kernel update but as I was centered in this 
> > bug report I had configured GRUB to directly boot the older kernel. I'm 
> > going to leave the system for a while with the new kernel loaded to see 
> > what problem hits (if any)... 
> Ah, so the reconects do not happen often enough to tell immediately
> whether switching kernels caused relief?
> We need a program to automatically create histograms with number of
> occurences per day, per hour, and per minute for intermittent bugs like
> this one. :)

I wish there's such a tool :-P

Well, the latest kernel 3.2.20-1 runs more stable (only one reconnect 
since it was loaded a few hours ago), I mean, now more crazy crashes 
like the above.
> The new data from this last observation is that 3.2.2-1 is capable of
> producing random reconnects and instability, just like the newer
> kernels.  So the mystery of the 3.2.2 -> 3.2.9 regression has been
> dispelled.

The kernel 3.2.2-1 craziness started a day ago, it was running nicely, 
but well, you're right, whatever happened in the meantime reproduced 
-very agressively- the issue again which translates into "the problem 
was still there" :-(
> I don't expect 3.2.21-1 to be any different from, say, 3.2.18-1, so I
> don't think it's worth spending much time with it loaded unless
> there's no better test to run.  I can suggest a kernel if you'd like,
> or if you want to use whichever has worked best in the past (3.4.y?)
> until we haved spent a little more time mulling over the data you've
> already provided, that's fine with me, too.

I'll keep testing brcmsmac with the upstream branch (mainline and 
unstable) so yes, if you have a kernel in your "radar" you think I 
could try just tell and I'll go with it.

Anyway, it's a good signal that there are no more users added to this 
bug report, mine has to be a very corner (though annoying) case.



Reply to: