Progress: * UM-2 [QEMU upstream maintainership] - further diagnosed the cause of the reported regression in virtio-pci running Linux on 32-bit arm emulation: the kernel the user was using was a non-LPAE one. So the fix is "use an LPAE kernel, or else manually set highmem=off". It would be nice if the kernel produced a more helpful error message for the case where it was built without CONFIG_LPAE and the PCI controller ECAM window is above the 4GB mark... - lots of code review this week, including: + a patch trying to add support to the memory system core for unaligned accesses to a device whose implementation doesn't directly handle them itself + another round of the Nuvoton GMAC and Mailbox device patchset + various parts of the new STM32L4x5 emulation + other smaller patches - patch: remove some now-unneeded code from arm_pamax() - patch: improve docs for the virt board 'highmem' option - patch: work around a bug in 'htags' that was causing intermittent failures of one of our CI jobs - patch: fix 'raven' PCI controller handling of unaligned accesses - handling merging of pull requests; spent some time making sure we have reported issues in our bug tracker (new tag "flaky-ci") for any intermittent CI failures, and asking people to investigate some of them
-- PMM