Re: [PING, ARM] Handle DWARF2_UNWIND_INFO in arm_except_unwind_info

[Olivier Hainque <hainque at adacore dot com>]

Hello Ramana,

> On 20 Sep 2017, at 15:06, Ramana Radhakrishnan <ramana.gcc@googlemail.com> wrote:

>>       * common/config/arm/arm-common.c (arm_except_unwind_info):
>>      Handle DWARF2_UNWIND_INFO.

>> As I mentioned in the course of another discussion, the dwarf2 unwind support
>> could most likely go away when support for the old ABI gets removed.
> 
> This is OK

Great, thanks!

> and thanks for thinking about this - unrelated to this
> specific patch, do you have some thoughts about what the timeline for
> removing the old ABI is for vxworks.

This is a point we discussed some time ago with Richard, hmm, here:

  https://gcc.gnu.org/ml/gcc-patches/2017-07/msg02002.html

VxWorks is based on the old ABI up to VxWorks 6. The move to
the new ABI occurred recently with VxWorks 7, latest version to date.

Richard suggested you would deprecate support for the old ABI
in GCC 8, and remove it in GCC 9. This seemed like a good compromise
to me.

> Sorry about the delay in reviews - I've been buried with tons of other
> things recently.

No worries at all. Thanks for having looked at this one!

It turns out there's a related patch still pending review at:

  https://gcc.gnu.org/ml/gcc-patches/2017-08/msg00271.html

2017-08-03  Olivier Hainque  <hainque@adacore.com>

        * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
        on the target mem instead of RTX_FRAME_RELATED_P on the insn to
        prevent DSE.
	(thumb_set_return_address): Likewise.

This one fixes ICEs during the build of libgcc with dwarf2 unwind info
activated. The explanation message is longish but the patch is short.

Would you have an opinion on this other one as well ?

Thanks much in advance for your feedback,

With Kind Regards,

Olivier