Re: time to create the GDB 8.1.x branch?

[Joel Brobecker <brobecker at adacore dot com>]

[question for everyone at the bottom]

Hi Yao,

> Arm still has pending patch series about new ARMv8.3-A Pointer
> Authentication feature support in GDB.  They were posted in Aug, and
> reviewed,
> 
> https://sourceware.org/ml/gdb-patches/2017-08/msg00174.html
> https://sourceware.org/ml/gdb-patches/2017-08/msg00175.html
> https://sourceware.org/ml/gdb-patches/2017-08/msg00171.html
> https://sourceware.org/ml/gdb-patches/2017-08/msg00172.html
> https://sourceware.org/ml/gdb-patches/2017-08/msg00173.html
> https://sourceware.org/ml/gdb-patches/2017-08/msg00176.html
> 
> The GDB support needs the new regset provided by Linux kernel, so
> these GDB patches can't go in until the kernel patches are merged.
> I think kernel patches will be merged in several (5?) weeks, is it OK to
> get these GDB patches into 8.1 release (mainline first and backport to
> 8.1 branch)?  To be clear, I don't expect GDB 8.1 release should be
> delayed in case that kernel patches merge won't happen in several weeks.
> The reason I ask this is that I want to make sure we are OK to merge
> such new feature support to release branch.

Considering that the new feature only seems to be activated when
the necessary hwcap flag is set, it seems fine for me to backport it.
The one concern I have, however, is that 5 weeks is a bit long for
us to wait. Typically, what I would expect is for us to cut the
branch, then take a two/three weeks for people to field-test and
stabilize, before we actually create the release.

I don't have a problem with delaying the release after the branch
get created, but 5 weeks takes us roughly to start of 2018. I would
add an extra week or two after your backport before we actually
release, just in case there are unexpected complications with
the patch. So, we're talking mid-January.

Any objections to a release around that time?

-- 
Joel