Progress:
* UM-2 [QEMU upstream maintainership]
+ sent out a pullreq for 6.0, and did a bunch more patch review
+ sent out a respin of the "build one manual, not five" RFC I sent last
month; nobody objected to the RFC so this time around it is a proper
PATCH
+ had a brief scan through the GIC architecture specification to look
at what extra GICv4 and GICv4.1 add (the big thing is that the ITS is
now mandatory; QEMU doesn't emulate that yet)
* VIRT-364 [QEMU support for ARMv8.1-M extensions]
+ SSE-300 timer device model more or less functionally-complete (though
untested). I ran into some issues with QEMU's Clock APIs that I needed
to fix in the process -- sent a patchseries trying to address those.
+ Got most of the outstanding v8.1M patches into master; sent a small
patch series with respins of the last 4 or so patches that needed
more work
thanks
-- PMM
== Progress ==
* GCC upstream validation:
- reported several regressions/new failures
* GCC
- Neon intrinsics: vceqq, vceqz and vceqzq for p64 patch: no feedback
- MVE autovectorization: committed vand and vorr patches. Sent updated
versions of veor, vbic and vmvn. vshr and vshl need some
refactorization with Neon versions, clarified the best way to go with
Arm maintainers.
- PR97875 (suboptimal vectorization on MVE): investigation shows that
movmisalign patterns are missing for MVE.
* benchmarking:
- Scripts to run coremark on stm32 now merged, debugging issues. No
progress this week.
* infra:
- debugging issues with our docker containers
- looking at ABE patches
* misc:
- provided feedback to Arm's partner wrt Intrinsics documentation
== Next ==
* GCC/cortex-M testing improvements & fixes
* cortex-m benchmarking
* MVE auto-vectorization/intrinsics improvements
[UM-59]
More work on softfloat rem; f128 still needs work.
[UM-2]
Patch review, notably v8.1m, mips cleanups, arc port.
Revisions to arm alignment patch set.
r~
== This Week ==
* PR97872 (missed optimization on less-than vector comparison)
- Upstream discussion with Richard, patch close to being approved.
- Causes a regression for pr78102.c.
* PR66791 (replace builtins by vector extensions in arm_neon.h intrinsics)
- Submitted patches for vmvn, vcreate, vneg.
- Kyrill approved patch for vmvn, and asked to remove corresponding builtins
from being created in backend.
== Next Week ==
- Hopefully commit fix for PR97872
- Continue with PR66791
- Start looking at another vectorization PR
VirtIO Initiative ([STR-9])
===========================
- started reviewing latest EPAM IOREQ/VirtIO patches Message-Id:
<1606732298-22107-1-git-send-email-olekstysh(a)gmail.com>
VirtIO RPMB ([STR-5])
=====================
- Continued working on Rust port of vhost-user-rpmb
VirtIO Portability Demo ([STR-13])
==================================
- done and closed
QEMU Support for Xen ([STR-20])
===============================
- re-based and continued to work on Xen development environment
- with my [current branch] I can now boot domU inside dom0
- also can now run qemu-aarch64 as a Dom0 PV backend
[current branch]
<https://github.com/stsquad/qemu/tree/xen/guest-loader-and-arm-build-cleanup…>
QEMU Device and Machine Models ([QEMU-418])
===========================================
- created new initiative for tracking work both in and outside group
- we don't want another watchdog duplication
- worked on refining definition for [CPU model for QEMU]
- wrote to Linaro open discussion list to canvas members
requirements Message-Id: <878saesi0x.fsf(a)linaro.org>
[CPU model for QEMU] <https://projects.linaro.org/browse/QEMU-414>
QEMU Upstream Work ([UM-2])
===========================
- posted [RFC PATCH] configure: add --without-default-features
Message-Id: <20201202153827.17446-1-alex.bennee(a)linaro.org>
Other work
==========
- bootstrapped gitdm metadata for GCC project
Completed Reviews [1/1]
=======================
[PATCH v4 00/11] hvf: Implement Apple Silicon Support
Message-Id: <20201203234857.21051-2-agraf(a)csgraf.de>
--
Alex Bennée
Progress:
* UM-2 [QEMU upstream maintainership]
+ Cut down the patch review queue a bit ready for 6.0 development cycle
* VIRT-364 [QEMU support for ARMv8.1-M extensions]
+ started working on modelling of the SSE-300. Have done some of the easy
"ID registers are different" stuff and started on modelling a new timer
device that isn't in the SSE-200.
thanks
-- PMM
== This Week ==
* PR66791 (Replace __builtins_* in arm_neon.h intrinsics by C vector extensions)
- Submitted patch for vmvn, waiting for feedback.
- Created and tested patches for vcreate and vneg.
* PR97849 (segfault in ifcvt with -march=armv8.2-a+sve)
- Committed fix to trunk.
== Next Week ==
- PR97872
- Continue with PR66791
