RE: [PATCH 0/2] Split tdesc_(amd64|i386)_mpx into tdesc(amd64|i386)_mpx_* and tdesc(amd64|i386)_avx_mpx_*

["Tedeschi, Walfred" <walfred.tedeschi@intel.com>]

-----Original Message-----
From: Marcin Kościelnicki [mailto:koriakin@0x04.net] 
Sent: Thursday, February 25, 2016 8:21 PM
To: Tedeschi, Walfred; palves@redhat.com; eliz@gnu.org; brobecker@adacore.com
Cc: gdb-patches@sourceware.org
Subject: Re: [PATCH 0/2] Split tdesc_(amd64|i386)_mpx into tdesc(amd64|i386)_mpx_* and tdesc(amd64|i386)_avx_mpx_*

On 22/02/16 09:49, Walfred Tedeschi wrote:
> CPU features can occur in any combination. The current assumption that 
> feature "A" implies in feature "B" does not necessarily hold.
>
> This patch series construct an additional combination of the Intel(R) 
> Memory Protection Extensions (MPX) with Intel(R) Advanced Vector 
> Extensions (AVX).
>
> Starting from the current implementation that has MPX combined with AVX.
> A new target description having only MPX will be created using a two 
> step approach as described below:
>
> First step:
> Mirror (i386|amd64)mpx target descriptors onto (i386|amd64)-avx-mpx 
> ones.
>
> Add a redundant target description for the MPX and AVX case using a 
> combined feature name to reflect that, i.e. avx-mpx.  A new flag 2is 
> also added to address MPX case without AVX.
>
>
> Second step:
> Refactor (i386|amd64)-mpx target descriptors.
>
> AVX feature is removed from the set of files that described MPX alone 
> feature.  Present on GDB code previous to this patch.
>
>
> Tests were done with hardware having MPX and AVX as is.
> For MPX standing alone tests were done forcing the XCR0 bits.
>
>

FYI, I've just refactored a few things in x86 tdescs, to properly pass them into the IPA - you will have a conflict with my commit ae91f6253926e4dadebcae90772f4f5a5bd06056.  You will have to add an X86_TDESC_AVX_MPX enum value, emit it in gdbserver when you're using one of the new tdescs, and recognize it in the IPAs.  Sorry for the inconvenience.


Marcin,

Thanks a lot for the heads up!

I looks like it is time to refactor the tdesc_selection, or? I suppose I have seen this code now 5 times.

May I do it and send as a separated patch?

Thanks and regards,
-Fred
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, Christian Lamprechter
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928