== Progress ==
* GCC upstream validation:
- Reported a few regressions
* GCC
- committed further fix for testcase for PR96770
- sent a few testsuite improvement patches
- resumed work on MVE/auto-vectorization. Added support for vcmp.f16.
Checking fp16 support in previous patches.
* Misc
- scripts patch reviews
== Next ==
* MVE auto-vectorization/intrinsics improvements
* GCC/cortex-M testing improvements & fixes
== Progress ==
* GCC upstream validation:
- Reported a few regressions
- Reduced build frequency on release branches, now same as trunk:
daily bump and arm/aarch64 "interesting" commits
* GCC
- pinged further fix for testcase for PR96770
- preparing cortex-m55 validation setup
- looking at cmse tests vs qemu
- Linaro BZ 5755/upstream 100049
* GDB
- uploaded cleaned up last known-to-work branch to sourceware so that
maintainers can have a look at it
- uploaded latest wip based on 8.2 branch too
* Misc
- scripts patch reviews
== Next ==
* MVE auto-vectorization/intrinsics improvements
* GCC/cortex-M testing improvements & fixes
Progress:
* UM-2 [QEMU upstream maintainership]
- Finished and posted patches implementing remap support for AN524
- Release work for rc3
- Sent a set of refactoring patches that split out some files we
were previously #including into translate.c so they are separate
compilation units.
- Thought a bit more about reset handling (in the course of writing
the AN524 remap patches I again ran into some of the deficiencies
of our current reset implementation; I really must take some time
this release cycle to try to improve things here...)
- Fixed bug where our AN547 board model accidentally disabled the FPU
on the Cortex-M55...
- ...and one where we were accidentally giving it an M33 rather than
an M55 !
- Respun and extended some patches from Paolo which fix our use of
QEMU and system headers in the 2 C++ files in the codebase (which
were broken by a change in new versions of glib)
- started looking at some of the for-6.1 patch review queue, which I
had been postponing in favour of for-6.0 stuff
* QEMU-406 [QEMU support for MVE (M-profile Vector Extension; Helium)]
- re-read the MVE portions of the v8M Arm ARM, sketched out a plan
of where to start with the QEMU implementation
thanks
-- PMM
https://gitlab.com/rth7680/qemu/-/commits/tgt-arm-bf16
It isn't a large extension, so all done. I added aarch64 simd, aarch32 neon,
and sve support all at once. I've tested it via risu vs FVP.
I've based this on my SVE2 branch, since there are some cleanups that made that
easier. I'll post patches to qemu-devel next week.
r~
== Progress ==
* GCC upstream validation:
- No regression to report this week. Issues on gcc-9 and gcc-10
release branches had already been reported by other people.
* GCC
- pinged further fix for testcase for PR96770
- Looking at failures for cortex-M, only found testisms so far
* Misc
- Fixed benchmarking jobs on stm32
== Next ==
* MVE auto-vectorization/intrinsics improvements
* GCC/cortex-M testing improvements & fixes
* Resume GDB/FDPIC work
VirtIO Initiative ([STR-9])
===========================
- had sync-up with Akashi-san/Vincent about approaches for the Zephyr
demo
- need to write-up on Monday
[STR-9] <https://projects.linaro.org/browse/STR-9>
QEMU Support for Xen ([STR-20])
===============================
- continued reviewing [RFC v12 00/65] arm cleanup experiment for
kvm-only build Message-Id: <20210326193701.5981-1-cfontana(a)suse.de>
- this is looking pretty solid now
- cut new version of [arm build clean-ups based on Claudio's series]
- testing reveals some breakage to track down next week
[STR-20] <https://projects.linaro.org/browse/STR-20>
[arm build clean-ups based on Claudio's series]
<https://github.com/stsquad/qemu/tree/xen/arm-build-cleanups-v3>
QEMU Upstream Work ([UM-2])
===========================
- posted [PULL 00/11] rc2 fixes (check-tcg, gitlab, gdbstub)
Message-Id: <20210406150041.28753-1-alex.bennee(a)linaro.org>
- some time trying to debug last weeks PR, eventually merged
[UM-2] <https://projects.linaro.org/browse/UM-2>
GSoC/Intern Mentoring
=====================
GSoC 2021 proposal initiation for AGL
Message-Id: <CAC+yH-Z_KZKuQ0nmAgk-+B+HW88sNjztObtWwazof+cAGJGuuQ(a)mail.gmail.com>
GSoC: cache modelling plugin inquiry
Message-Id: <CAD-LL6hLk1XAB5VWwHgMOWQV=bT1+FCnf8f-q9MVw5e5A4RMqg(a)mail.gmail.com>
GSoC - QEMU Ideas
Message-Id: <CALpwMJz0kb2BDreTWBbt945FR+sLX=sKzHSv8pMb5LkLwXqEJA(a)mail.gmail.com>
Synced up with potential AGL Jailhouse/VirtIO GSoC applicant
Completed Reviews [3/3]
=======================
[PATCH RESEND] docs: clarify absence of set_features in vhost-user
Message-Id: <20210325144846.17520-1-hi(a)alyssa.is>
[PATCH] docs: Add a QEMU Code of Conduct and Conflict Resolution Policy document
Message-Id: <72bc8020-2028-82db-219c-a6ae311e26df(a)redhat.com>
[PATCH v4 00/12] target/arm mte fixes
Message-Id: <20210406174031.64299-4-richard.henderson(a)linaro.org>
Absences
========
Current Review Queue
====================
TODO [RFC v12 00/65] arm cleanup experiment for kvm-only build
Message-Id: <20210326193701.5981-1-cfontana(a)suse.de>
==================================================================================================================
TODO [RFC 0/8] virtio: Improve boot time of virtio-scsi-pci and virtio-blk-pci
Message-Id: <20210325150735.1098387-1-groug(a)kaod.org>
===================================================================================================================================
TODO [PATCH 0/5] virtio: Implement generic vhost-user-i2c backend
Message-Id: <cover.1616570702.git.viresh.kumar(a)linaro.org>
===========================================================================================================================
TODO [PATCH v2 00/29] tcg: Workaround macOS 11.2 mprotect bug
Message-Id: <20210314212724.1917075-1-richard.henderson(a)linaro.org>
================================================================================================================================
--
Alex Bennée
Progress (short week, 2 days):
* UM-2 [QEMU upstream maintainership]
- more release work, bug triage, etc
- started implementing the memory-remapping feature of the
AN524 FPGA image (which allows the BRAM and QSPI to be swapped
either at startup or dynamically under guest control). Some guests
(like ARM TF-M) assume the QSPI mapping.
thanks
-- PMM
== Progress ==
* GCC upstream validation:
- Reported minor testsuite issues (eg failures with -mabi=ilp32 on aarch64)
- re-started looking at validation for cortex-m55, realized that qemu
does not support MVE yet
* GCC
- posted further fix for testcase for PR96770
- fixed PR 99786
- committed fix for PR 99773
* Misc
== Next ==
* MVE auto-vectorization/intrinsics improvements
* GCC/cortex-M testing improvements & fixes
* Resume GDB/FDPIC work
