Re: hexagon container update

[Alex Bennée]

Brian Cain <bcain@quicinc.com> writes:

>> -----Original Message-----
>> From: Brian Cain
>> Sent: Friday, October 1, 2021 7:23 PM
>> To: Richard Henderson <richard.henderson@linaro.org>; Alex Bennée
>> <alex.bennee@linaro.org>; qemu-devel@nongnu.org
>> Cc: Taylor Simpson <tsimpson@quicinc.com>
>> Subject: RE: hexagon container update
>> 
>> > -----Original Message-----
>> > From: Brian Cain
>> ...
>> > > -----Original Message-----
>> > > From: Richard Henderson <richard.henderson@linaro.org>
>> > ...
>> > > On 10/1/21 12:59 PM, Brian Cain wrote:
>> > > > Alex,
>> > > >
>> > > > We need to update the docker container used for hexagon for new test
>> > cases
>> > > proposed in Taylor's recent patch series under review.  Thankfully,
>> > CodeLinaro
>> > > has provided a binary of the hexagon cross toolchain so now I think we 
>> > > can
>> > > simplify the hexagon docker file to something like the below.  I hope 
>> > > this
>> also
>> > > means that we can remove the exceptional handling for the hexagon
>> > container.
>
> We had some issues with the previous attempt to update the container.
> The linux-user "signals" test failed. Richard pointed out that the
> archive of the C library had what looks like a defect that would cause
> the test to fail.
>
> https://www.mail-archive.com/qemu-devel@nongnu.org/msg849635.html
>
> I'm following up now with a workaround - the attached patch references
> a toolchain which avoids the problem and passes the signals test. This
> toolchain is based on llvm+clang 13.0.1-rc2. BTW, the release page has
> a signature provided in case anyone wants to verify the download:
> https://github.com/quic/toolchain_for_hexagon/releases/tag/13.0.1-rc2_
>
> Can we try again with this new container definition?
>
> -Brian
>
> [2. 0001-Update-Hexagon-toolchain-to-13.0.1-rc2.patch --- text/x-diff; 
> 0001-Update-Hexagon-toolchain-to-13.0.1-rc2.patch]...

Could you send the updated patch against the current master in the usual
way please?

That way I know it's been tested and there is less risk of me botching
up a franken patch by applying to the wrong base.

-- 
Alex Bennée