Re: [PATCH] Handle MIPS Linux SIGTRAP siginfo.si_code values

[Luis Machado <lgustavo at codesourcery dot com>]

On 02/24/2016 05:49 PM, Maciej W. Rozycki wrote:
> On Wed, 24 Feb 2016, Luis Machado wrote:
>
> > I think we should converge to a standard solution across all architectures in
> > the future rather than potentially perpetuate old non-standard ways. So the
> > movement towards returning well defined si_code values in the MIPS Linux
> > Kernel is a plus, even though we might not benefit from it right now. I'm in
> > favor of having the change to the kernel made.
>
>   Not right now as in: "not earlier than in 2-3 months' time needed for a
> proper kernel release", which is when 4.6.0 is likely to happen?  Well, I
> think it's not much waiting really unless you don't care, in which case
> you'll get it with your next regular kernel upgrade scheduled.
>
>    Maciej

The benefit (long term one) i'm looking at is mostly from GDB's 
perspective, not having to special-case things to accomodate old 
mechanisms that tend to grow code that wouldn't be needed otherwise. 
We'll still have to support the old mechanism for years, but having this 
change sooner (be it 3 months) rather than later will get us on the 
right path for this matter.

Looking at gdb/nat/linux-ptrace.h, we already have a few variations on 
the theme, with architectures doing different things. :-)