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
Progress:
* UM-2 [QEMU upstream maintainership]
- tagged rc0
- various bug investigation/patches for 6.0-ish stuff:
+ machines with a 'platform bus' incorrectly treated all sysbus
devices as hotpluggable: wrote and sent patchseries
+ respin of gpex pci-host "don't fault for unmapped MMIO/PIO window
accesses" patch
+ document how to use gdbstub/gdb for a multi-cluster machine where
the clusters are different gdb inferiors
+ looked for workarounds for macos Appkit bug where Apple broke menubars
for apps that start off as console apps and programmatically switch
themselves to being GUI apps; various online suggestions don't seem
to help, except for "when app starts, force the Dock to become active
and then grab back focus 200ms later". Maybe we'll do that, but it's
pretty ugly...
thanks
-- PMM
== Progress ==
* GCC upstream validation:
- Small improvement to pre-commit testing scripts to allow running a
subset of the tests (and thus save a lot of time)
* GCC
- MVE autovectorization:
- vcmp support of FP types OK.
- testsuite cleanup: looking at current failures, only found issues
with tests so far. Testing several small testsuite patches is taking a
long time due to the number of configurations
* Misc
== Next ==
* MVE auto-vectorization/intrinsics improvements
* GCC/cortex-M testing improvements & fixes
* Resume GDB/FDPIC work
Progress:
* UM-2 [QEMU upstream maintainership]
- softfreeze was this week, lots of pullreq processing
- spun v2 of "fix M-profile load of PC/SP from ELF images via
memory aliases"
thanks
-- PMM
Hello Linaro Toolchain Working Group,
Please connect your new bots to the staging first and make them reliably
green there:
linaro-aarch64-flang-debug
linaro-aarch64-flang-latest-clang
linaro-aarch64-flang-latest-gcc
clang-cmake-aarch64-full is red for a month.
https://lab.llvm.org/buildbot/#/builders/7?numbuilds=700
Could you move linaro-aarch64-full to the staging and make it green again,
please?
Please let me know if I could help or you have questions.
Thanks
Galina