RE: Intel x86 family 45 support of btrace

["Metzger, Markus T" <markus dot t dot metzger at intel dot com>]

> -----Original Message-----
> From: Doug Evans [mailto:dje@google.com]
> Sent: Friday, September 18, 2015 9:50 PM
> To: Metzger, Markus T; gdb-patches@sourceware.org
> Subject: Intel x86 family 45 support of btrace

Hi Doug,


> ---snip---
> date: 2013/03/11 08:38:27;  author: mmetzger;  state: Exp;  lines: +186 -1
> LBR, BTM, or BTS records may have incorrect branch "from" information afer
> an
> EIST transition, T-states, C1E, or Adaptive Thermal Throttling (AAJ122).
> 
> This results in sporadic test fails. Disable btrace on those processors.
> ---snip---
> 
> I'm wondering how useful this feature is even if it's not always correct.
> E.g., if it is usually correct, would the user still find the feature useful
> enough to cope with the shortcomings?
> IOW, would it make sense to add the ability to turn on the feature
> anyways (defaulting to "off" of course) ?

Whether the feature is perceived useful or broken depends a lot on the
user, I'd say.

I decided to disable it because we can't always detect that it's broken.  In the
worst case, the trace may appear correct but isn't.  You might get a wrong loop
trip count, for example.  In the best case, you get some bogous trace and
run into an error later.  There is no code that tries to detect this and give a
proper error message at the right place in the trace.

I submitted a kernel patch together with the GDB patch.  Unless someone
patched it out again because he thought that this was still useful, recent
kernels shouldn't support BTS on those processors.

Regards,
Markus.

Intel Deutschland GmbH
Registered Address: Am Campeon 10-12, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de
Managing Directors: Christin Eisenschmid, Prof. Dr. Hermann Eul
Chairperson of the Supervisory Board: Tiffany Doon Silva
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928