> -----Original Message-----
> From: Richard Henderson <richard.henderson@linaro.org>
> Sent: Sunday, February 14, 2021 7:23 PM
> To: Taylor Simpson <tsimpson@quicinc.com>; qemu-devel@nongnu.org
> Cc: philmd@redhat.com; alex.bennee@linaro.org; laurent@vivier.eu;
> ale@rev.ng; Brian Cain <bcain@quicinc.com>; Peter Maydell
> <peter.maydell@linaro.org>
> Subject: Re: [PATCH v8 00/35] Hexagon patch series
>
> On 2/7/21 9:45 PM, Taylor Simpson wrote:
> > This series adds support for the Hexagon processor with Linux user support
> >
> > See patch 02 Hexagon README for detailed information.
> >
> > This series assumes int128_or() is implemented.
> > https://lists.nongnu.org/archive/html/qemu-devel/2020-10/msg06004.html
> >

What's the status of this patch?  Could you go ahead and merge it ahead of the rest of the series that it's in?

> >
> > Once the series is applied, the Hexagon port will pass "make check-tcg".
> > The series also includes Hexagon-specific tests in tcg/tests/hexagon.
> >
> > The final patch in the series add docker support.  Thanks to Alessandro
> > Di Federico <ale@rev.ng> and Brian Cain <bcain@quicinc.com> for making
> this
> > happen.  The default container (debian-hexagon-cross) uses a toolchain
> built
> > by rev.ng.  Alternatively, there is a container that will build the toolchain
> > locally (debian-hexagon-cross-build-local).
>
> Right.  This is in really good shape.
>
> I've completed review of this round, and there are some nits.  But they're
> minor enough that I wouldn't even mind them being addressed via the
> normal
> development process.  I.e. I'd be keen to not look through that diffstat again.
>  ;-)

Thanks Richard!!  We're having unprecedented winter weather in Austin, so I'll address your latest feedback once our datacenter recovers from a power outage.

>
> Any objections from anyone else on that?
>
> I don't suppose you and Peter Maydell signed gpg keys when we all met in
> Lyon?

Nope.  Peter, please advise

On Tue, 16 Feb 2021 at 20:59, Taylor Simpson <tsimpson@quicinc.com> wrote:
> > -----Original Message-----
> > From: Richard Henderson <richard.henderson@linaro.org>
> > I've completed review of this round, and there are some nits.  But they're
> > minor enough that I wouldn't even mind them being addressed via the
> > normal
> > development process.  I.e. I'd be keen to not look through that diffstat again.
> >  ;-)

> > Any objections from anyone else on that?
> >
> > I don't suppose you and Peter Maydell signed gpg keys when we all met in
> > Lyon?
>
> Nope.  Peter, please advise

We effectively are operating a TOFU policy for gpg keys,
ie put them on a public keyserver, to the extent that you can arrange
to get them signed by other people who also have gpg keys please do,
and at some point we may be able to meet up and get a shorter
trust path.

For this patchset, I would prefer it if Richard collected the patches
and sent me a pullrequest. First pullrequests from new submaintainers
are higher-effort for me, because I need to look them through carefully
to be sure that they're the right format and so on; so I'd rather
not do that with an enormous patchset. It's easier for me if that
work is postponed and done with something smaller later.

thanks
-- PMM

On 2/16/21 10:17 PM, Peter Maydell wrote:
> On Tue, 16 Feb 2021 at 20:59, Taylor Simpson <tsimpson@quicinc.com> wrote:
>>> -----Original Message-----
>>> From: Richard Henderson <richard.henderson@linaro.org>
>>> I've completed review of this round, and there are some nits.  But they're
>>> minor enough that I wouldn't even mind them being addressed via the
>>> normal
>>> development process.  I.e. I'd be keen to not look through that diffstat again.
>>>  ;-)
> 
>>> Any objections from anyone else on that?

I'm very pleased with how this series evolved over the time,
and Taylor interaction with the community during the long
review process.

Patches 1-34 are OK but I don't think patch 35 (Dockerfile hexagon)
should enter mainstream that way. It probably makes sense to add
it along with the CI testing job (missing so far).

Patch 30 (Linux user emulation) could have an Acked-by from Laurent.

Also, while having TCG unit tests is great, it doesn't replace
integration tests using real world binaries. So personally I'd
like to be able to test an Hexagon BusyBox, ideally one coming
from the official Hexagon distribution (extracted from the SDK).
We talked about this and Brian will probably share it.

Whole series tested it on x86, so:
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
(I still plan to test it on big-endian host)

>>> I don't suppose you and Peter Maydell signed gpg keys when we all met in
>>> Lyon?
>>
>> Nope.  Peter, please advise
> 
> We effectively are operating a TOFU policy for gpg keys,
> ie put them on a public keyserver, to the extent that you can arrange
> to get them signed by other people who also have gpg keys please do,
> and at some point we may be able to meet up and get a shorter
> trust path.
> 
> For this patchset, I would prefer it if Richard collected the patches
> and sent me a pullrequest. First pullrequests from new submaintainers
> are higher-effort for me, because I need to look them through carefully
> to be sure that they're the right format and so on; so I'd rather
> not do that with an enormous patchset. It's easier for me if that
> work is postponed and done with something smaller later.

Ahah I just suggested to Taylor to look at your previous explanation
before reading the same explanation here :)
https://www.mail-archive.com/qemu-devel@nongnu.org/msg765788.html

Regards,

Phil.

On 2/17/21 6:23 AM, Philippe Mathieu-Daudé wrote:
> I'm very pleased with how this series evolved over the time,
> and Taylor interaction with the community during the long
> review process.
> 
> Patches 1-34 are OK but I don't think patch 35 (Dockerfile hexagon)
> should enter mainstream that way. It probably makes sense to add
> it along with the CI testing job (missing so far).

Ok, I'll drop 35 from the initial pull,

> Patch 30 (Linux user emulation) could have an Acked-by from Laurent.

and wait a couple of days more for this.  But without, I'll send the PR on the
weekend.

> Whole series tested it on x86, so:
> Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
> (I still plan to test it on big-endian host)

Thanks.


r~

On 2/16/21 12:59 PM, Taylor Simpson wrote:
>>> This series assumes int128_or() is implemented.
>>> https://lists.nongnu.org/archive/html/qemu-devel/2020-10/msg06004.html
>>>
> 
> What's the status of this patch?  Could you go ahead and merge it ahead of the rest of the series that it's in?

The fp patch set that created this patch has been rewritten and no longer uses
it.  I'll queue this to the branch that will include this pull.


r~