mbox series

[pull,request] arch/arm: add some new armv8a cores

Message ID cover.1504444617.git.yann.morin.1998@free.fr
State Not Applicable
Headers show
Series [1/9] arch/arm: re-order cores choice | expand

Pull-request

git://git.buildroot.org/~ymorin/git/buildroot.git

Message

Yann E. MORIN Sept. 3, 2017, 1:17 p.m. UTC
Hello All!

This series adds all the armv8a cores currently known to any released
version of gcc.

First, we do a bit of cleanups in the existing menu: re-order the cores,
simplify hiding non-64-bit cores, rename armv8 to armv8a.

Second, as all armv8a cores are guaranteed by the ISA to have a VFPv4,
we disallow building a soft-float toolchain on those cores when in
64-bit mode.

Now we add a bunch of armv8a cores, cortex and non-cortex alike, with a
few armv8.1a variants as well.

It is to be noted that we do have at least one armv8a core that is in
fact 32-bit only, while we also do have a few armv8a cores that are
64-bit only (or at least gcc does not recognises them in arm mode).

Eventually, we add a build-only test for all those toolchains, to check
that at least the toolchains do build. This final commit is not meant to
be applied, it adds way too many tests.

This series relies on the previous architecture series I've recently
sent:
    http://lists.busybox.net/pipermail/buildroot/2017-September/201403.html
    http://lists.busybox.net/pipermail/buildroot/2017-September/201409.html

Regards,
Yann E. MORIN.

PS. I've sent this series as a real pull-request, just to see how
the new patchwork handles it: currently, series are sorted by the
first patch, and patchwork misses the cover-letter. Let's see if
a cover-letter that is a pull-request gets noticed.


The following changes since commit 41eeac74f986790fe26f03ebf05322120d2c04da

  package/gcc: slight cleanup and reorg in remaining arch depends (2017-09-03 11:14:53 +0200)


are available in the git repository at:

  git://git.buildroot.org/~ymorin/git/buildroot.git

for you to fetch changes up to a876f6b64754ba3b96492c57b18654f8ec465ca0

  [DON'T COMMIT] tests for all new arm cores (2017-09-03 15:02:47 +0200)


----------------------------------------------------------------
Yann E. MORIN (9):
      arch/arm: re-order cores choice
      arch/arm: simplify hiding non 64-bit cores
      arch/arm: armv8 is really armv8a
      arch/arm: do not allow soft-float for armv8a
      arch/arm: add cortex-A32
      arch/arm: add some armv8a cortex variants
      arch/arm: add some non-cortex armv8a cores
      arch/arm: add armv8.1a cores
      [DON'T COMMIT] tests for all new arm cores

 arch/Config.in.arm                                 | 297 ++++++++++++++++-----
 package/pkg-cmake.mk                               |   2 +-
 .../tests/toolchain/test_internal_aarch64_all.py   | 162 +++++++++++
 .../tests/toolchain/test_internal_arm_all.py       |  95 +++++++
 .../toolchain-external-codesourcery-arm/Config.in  |   2 +-
 .../toolchain-external-linaro-arm/Config.in        |   2 +-
 .../toolchain-external-linaro-armeb/Config.in      |   2 +-
 7 files changed, 489 insertions(+), 73 deletions(-)
 create mode 100644 support/testing/tests/toolchain/test_internal_aarch64_all.py
 create mode 100644 support/testing/tests/toolchain/test_internal_arm_all.py

Comments

Yann E. MORIN Sept. 3, 2017, 1:24 p.m. UTC | #1
All,

On 2017-09-03 15:17 +0200, Yann E. MORIN spake thusly:
> PS. I've sent this series as a real pull-request, just to see how
> the new patchwork handles it: currently, series are sorted by the
> first patch, and patchwork misses the cover-letter. Let's see if
> a cover-letter that is a pull-request gets noticed.

Definitely not working, and even breaks the series-detection of
patchowrk:

  - it does not classifies the first patch in the series as being
    part of the series;

  - it treats the cover-letter as bieing on its own, totally
    decorelated from the rest of the series.

Too bad...

Sorry for the noise... :-(

Regards,
Yann E. MORIN.
Thomas Petazzoni Nov. 24, 2017, 10:32 p.m. UTC | #2
Hello,

On Sun,  3 Sep 2017 15:17:44 +0200, Yann E. MORIN wrote:

> Yann E. MORIN (9):
>       arch/arm: re-order cores choice
>       arch/arm: simplify hiding non 64-bit cores
>       arch/arm: armv8 is really armv8a

Those patches applied to next.

>       arch/arm: do not allow soft-float for armv8a

This one not applied, see my reply to it. I've kept it as "New" in
patchwork for now.

>       arch/arm: add cortex-A32
>       arch/arm: add some armv8a cortex variants
>       arch/arm: add some non-cortex armv8a cores
>       arch/arm: add armv8.1a cores

I've applied those four patches to next as well. As I replied to one of
them, I'm not super convinced exhaustively adding all cores is really
useful, but since you did it, and build tested all of them, I applied
the patches anyway.

Thanks!

Thomas Petazzoni