Re: [PATCH] Revert 9fc2105aeaaf56b0cf75296a84702d0f9e64437b to fix pyaudio (and probably more)

[Catalin Marinas <catalin.marinas@arm.com>]

On Thu, Jan 08, 2015 at 10:46:14PM +0000, Pavel Machek wrote:
> On Thu 2015-01-08 16:57:05, Russell King - ARM Linux wrote:
> > On Thu, Jan 08, 2015 at 04:41:12PM +0000, Catalin Marinas wrote:
> > > On Thu, Jan 08, 2015 at 04:34:50PM +0000, Russell King - ARM Linux wrote:
> > > > On Thu, Jan 08, 2015 at 04:19:08PM +0000, Catalin Marinas wrote:
> > > > > I really think we should stop this thread. User ABI breakage fixed now.
> > > > 
> > > > No.  Vince's email shows a more serious problem.
> > > > 
> > > > The ABI breakage issue was reported before the patches were apparently
> > > > merged, but that information seems to have been lost.  We need to
> > > > understand how that happened so similar instances don't happen in the
> > > > future.
> > > 
> > > FYI, searching for "Vince Weaver" and "bogomips" found this:
> > > 
> > > https://lkml.org/lkml/2013/7/11/454
> > 
> > Gah.
> > 
> > It looks like the report was made by hanging it into a totally different
> > thread to the patches which caused the problem, and which went nowhere
> > near the ARM kernel mailing lists.
> 
> This went at least to Will Deacon and Nicolas Pitre.
> 
> https://lkml.org/lkml/2013/5/15/217
> 
> So AFAICT yes, we knowingly broke at least
> com.securiteinfo.android.bogomips application.

Hmm, "BogoMIPS is a tool used to create a web page about Android
Benchmarks". And a stupid table here:
https://www.securiteinfo.com/divers/android_bogomips_benchmark.shtml.
Even though it's against the user ABI kernel policy, I must say I'm glad
we broke this one ;).

pyaudio is a valid complaint though (and not about the bogomips value
but about the presence of the "bogomips" string).

-- 
Catalin