-
for-8.2/final-fixes-v1
final fixes for 8.2 8.2 is looking fairly stable but I do have one bug fix for gdbstub which I came across while debugging something else. The changes for avocado rationalise all flaky tests under the QEMU_TEST_FLAKY_TESTS environment variable. The final patch re-adds the flaky tests to the CI as a manually run allow_fail job so we can still attempt to debug their failure in the place they tend to fall over. Alex.
-
pull-for-8.2-fixes-231123-1
various random fixes for 8.2 - replace fedora-i386 cross compiler with debian - update cirrus MacOS image to Ventura - merge debian-native and debian-amd64 docker images - fix compile of plugins on Windows mingw cross - add some doc notes on semihosting READC - add some doc notes on gdbstub - skip loading debug symbols if we have failed - enable arm-softmmu TCG tests - don't attempt to use native cross builds for linux-user - clean up registers gdb test case (ppc64/s390x)
-
for-8.2/random-fixes-v2
random fixes for 8.2 pre-PR (tests, plugins, docs, semihosting) I've just been trying to clear some of the bugs against various subsystems. These are mostly testing changes and a minor tweak I found while checking up on semihosting behaviour. Enabling arm-softmmu check-tcg tests expands our testing abilities slightly for the 32 bit side of the family. For v2 ------ - fixed up register test for ppc64/s390x - added gdbstub doc patch - review comments for arm-softmmu testing - changed handling of "native" cross for linux-user I intend to roll the PR on Wednesday. The following need still review: tests/tcg: finesse the registers check for "hidden" regs configure: don't try a "native" cross for linux-user tests/tcg: enable semiconsole test for Arm tests/tcg: enable arm softmmu tests docs/system: clarify limits of using gdbstub in system emulation docs/emulation: expand warning about semihosting tests/tcg: fixup Aarch64 semiconsole test
-
for-8.2/random-fixes-v1
random fixes for 8.2 (tests, plugins, docs, semihosting) I've just been trying to clear some of the bugs against various subsystems. These are mostly testing changes and a minor tweak I found while checking up on semihosting behaviour. Enabling arm-softmmu check-tcg tests expands our testing abilities slightly for the 32 bit side of the family. Let me know if there are any other patches worth picking up. The following need review: tests/tcg: enable semiconsole test for Arm tests/tcg: enable arm softmmu tests testing: move arm system tests into their own folder hw/core: skip loading debug on all failures docs/emulation: expand warning about semihosting tests/tcg: fixup Aarch64 semiconsole test
-
pull-halloween-omnibus-081123-1
Final test, gdbstub, plugin and gitdm updates for 8.2 - fix duplicate register in arm xml - hide various duplicate system registers from gdbstub - add new gdb register test to the CI (skipping s390x/ppc64 for now) - introduce GDBFeatureBuilder - move plugin initialisation to after vCPU init completes - enable building TCG plugins on Windows platform - various gitdm updates - some mailmap fixes - disable testing for nios2 signals which have regressed
-
pull-halloween-omnibus-071123-1
Final test, gdbstub, plugin and gitdm updates for 8.2 - fix duplicate register in arm xml - hide various duplicate system registers from gdbstub - add new gdb register test to the CI (skipping s390x/ppc64 for now) - introduce GDBFeatureBuilder - move plugin initialisation to after vCPU init completes - enable building TCG plugins on Windows platform - various gitdm updates - some mailmap fixes - disable testing for nios2 signals which have regressed
-
maintainer/november-for-8.2-v1
Maintainer updates for 8.2 (gdbstub, tests, plugins) pre-PR I ran into various issues getting the register API ready in time so those have been dropped although I've included a bunch of the pre-requisites. There has been more tweaking of the registers test which keeps finding kinks in our gdbstub report. I've also rolled up the acked gitdm changes. There has been a bit of juggling to the configure and meson bits for the Windows plugin support. Finally I noticed nios2 signal handling is broken again so I've disabled testing again and we shouldn't re-enable it until we can defend in CI (if we ever want to, the architecture is currently orphaned). I'll be rolling the PR tomorrow for soft-freeze so this is your last chance to complain ;-) The following still need review: Revert "tests/tcg/nios2: Re-enable linux-user tests" configure: tell meson and contrib_plugins about DLLTOOL tests/avocado: update the tcg_plugins test tests/tcg: add an explicit gdbstub register tester target/arm: hide aliased MIDR from gdbstub target/arm: hide all versions of DBGD[RS]AR from gdbstub target/arm: hide the 32bit version of PAR from gdbstub
-
maintainer/gdb-and-plugins-for-8.2-v1
gdbstub and plugin read register and windows support Here are my final updates for the 8.2 cycle which I can hopefully merge if we get enough review. Aside from the usual tweaks and fixes there are two new features: TCG Plugin Register Access This is based on Akihiko's previously posted series with some changes by myself. I wasn't keen on the plugin facing API so I've re-written it to use an opaque handle and hide the gdb details from the plugin. I think this allows for potential future improvements as well as being ready for up-coming heterogeneous support. The new API allowed for making the execlog register tracking a bit more flexible and able to track multiple registers. Windows Support This fairly late breaking patch finally adds support for Windows to the TCG plugins subsystem. I'm pretty pleased with Greg's approach which improves on previous attempts by avoiding re-implementing a linker for POSIX targets. I don't have access to Windows though so I'm calling on Windows users to test the solution. The following patches still need review: contrib/plugins: extend execlog to track register changes plugins: add an API to read registers gdbstub: expose api to find registers tests/avocado: update the tcg_plugins test tests/tcg: add an explicit gdbstub register tester target/arm: hide aliased MIDR from gdbstub target/arm: hide all versions of DBGD[RS]AR from gdbstub target/arm: hide the 32bit version of PAR from gdbstub gdb-xml: fix duplicate register in arm-neon.xml
-
pull-halloween-omnibus-311023-2
Maintainer updates for testing, gitlab, gdbstub and plugins: - add dtc package to openbsd VMs - use -fno-stack-protector for non-stdlib tests - split alpha and sh4 compilers into legacy image - harmonise other compilers into debian-all-test-cross - fix NULL check in gdb_regs - fix memleak in semihosting - remove unused parameter in plugin code - fix fd leak in lockstep plugin
-
pull-halloween-omnibus-311023-1
Maintainer updates for testing, gitlab, gdbstub and plugins: - add dtc package to openbsd VMs - use -fno-stack-protector for non-stdlib tests - split alpha and sh4 compilers into legacy image - harmonise other compilers into debian-all-test-cross - fix NULL check in gdb_regs - fix memleak in semihosting - remove unused parameter in plugin code - fix fd leak in lockstep plugin
-
maintainer/halloween-branch-v2
Maintainer updates for testing, gdb, semihosting and plugins (pre-PR) This collects up a number of fixes across my various maintainer trees. There will be a longer plugins related one coming later but I wanted to turn these around first. The update to bookworm was delayed from the last release but I've now "solved" that by moving sh4 and alpha into a legacy container stuck on bullseye. v2 - further split the first alpha patch - re-base and merge conflicts - address comments and tags The following still need review: tests/docker: upgrade debian-all-test-cross to bookworm (1 acks, 1 sobs, 0 tbs) tests/docker: use debian-all-test-cross for sparc64 tests/docker: move sh4 to use debian-legacy-test-cross gitlab: add build-loongarch to matrix gitlab: clean-up build-soft-softmmu job gitlab: split alpha testing into a legacy container
-
maintainer/halloween-branch-v1
Maintainer updates for testing, gdb, semihosting and plugins This collects up a number of fixes across my various maintainer trees. There will be a longer plugins related one coming later but I wanted to turn these around first. The update to bookworm was delayed from the last release but I've now "solved" that by moving sh4 and alpha into a legacy container stuck on bullseye. I've also taken the opportunity to rationalise our zoo of containers to better sync up with what we actually use on gitlab. The following still need review: plugins: Remove an extra parameter tests/docker: upgrade debian-all-test-cross to bookworm tests/docker: use debian-all-test-cross for sparc64 tests/docker: use debian-all-test-cross for riscv64 tests/docker: use debian-all-test-cross for mips tests/docker: use debian-all-test-cross for mips64 tests/docker: use debian-all-test-cross for m68k tests/docker: use debian-all-test-cross for hppa tests/docker: use debian-all-test-cross for power tests/docker: move sh4 to use debian-legacy-test-cross tests/docker: use debian-legacy-test-cross for alpha gitlab: split alpha testing into a legacy container
-
gitdm/next-v1
gitdm metadata updates for Oct '23 Hi, I was looking at the stats today so I've done my usual run of gitdm and then attempted to fix up the missing bits. I've had to patch gitdm to deal with gitlab URLS in place of email addresses and will have to work out if that is worth upstreaming at some point. The Rivos Inc updates actually got acked a while back but have been sitting on the tree waiting for me to bundle it in a PR. I've added the HiSilicon, Daynix and mailmap patches. The others never got acked so the come around one more time. Thanks, Alex.
-
pull-omnibus-111023-1
testing, gdbstub and plugin updates - enable more sbsa-ref tests in avocado - add swtpm to the package lists - reduce avocado noise in gitlab by limiting tests - make docker engine choice driven by configure and enable override - remove unneeded gcc suffix on some cross compilers - fix some NULL returns in gdbstub - improve locking in execlog plugin - introduce the GDBFeature structure - consistently set gdb_core_xml_file - use cleaner escaping for gdb xml - drop ancient gdb_has_xml() test - disable multi-instruction GUSA emulation when plugins enabled - fix some coverity issues in plugins
-
maintainer/october-omnibus-v1
October maintainer omnibus pre-PR (tests, gdbstub, plugins) Hi, This wraps up my current testing, gdbstub and plugin trees in an attempt to do my part to reduce the qemu-devel fire hose. It is pretty much the same as the September tree except: - dropped RFC patches - dropped CRIS deprecation patches - dropped makefile dependency tweaks - tweaked RUNC as per Paolo's suggestion - used later version of Marcin's sbsa-ref changes - added a FLAKY_TEST removal - added tags as appropriate Now I'm using the pre-PR subject to give people a last chance to object before I roll a proper pull request later this week. The following still could do with review: contrib/plugins: fix coverity warning in hotblocks contrib/plugins: fix coverity warning in lockstep contrib/plugins: fix coverity warning in cache configure: remove gcc version suffixes configure: allow user to override docker engine tests/docker: make docker engine choice entirely configure driven tests/avocado: remove flaky test marking for test_sbsaref_edk2_firmware
-
maintainer/september-2023-omnibus-v1
September maintainer omnibus (tests, gdbstub, plugins) Hi, This wraps up my current testing, gdbstub and plugin trees in an attempt to do my part to reduce the qemu-devel fire hose. For testing we have a number of cleanups to configure to better handle selecting the container engine (removing the ability to dynamically switch). I had to do this as one of my build boxes has recently acquired a broken podman install. There are also some updates to avocado sbsa-ref as well as tweaks to the gitlab setup to minimise the amount of wheel spinning we do. The deprecation of CRIS is a little provocative but if we do want to keep it we need to stop relying on a rapidly dated fedora image to do it. Finally we include the swtpm package to widen the testing we do through avocado. The gdbstub updates are from Akihiko and include a bunch of clean-ups that will hopefully pave the way for another series which allows the plugins to access register values using the gdb backend to source the data. Finally the plugins has a few fixes which includes a tweak to make SH4 atomic modelling more plugin friendly. This allows for Matt's fix to deal with non-regular instruction encoding spanning pages. I also fix a number of coverity warnings. The final time control patches are still RFC and not ready for merging but I include them for completeness. The following still need review: contrib/plugins: add iops plugin example for cost modelling plugins: add time control API sysemu: generalise qtest_warp_clock as qemu_clock_advance_virtual_time qtest: use cpu interface in qtest_clock_warp (1 acks, 1 sobs, 0 tbs) contrib/plugins: fix coverity warning in hotblocks contrib/plugins: fix coverity warning in lockstep contrib/plugins: fix coverity warning in cache configure: ensure dependency for cross-compile setup configure: remove gcc version suffixes configure: allow user to override docker engine tests/docker: make docker engine choice entirely configure driven docs: mark CRIS support as deprecated tests/lcitool: add swtpm to the package list
-
pull-testing-200923-1
testing updates: - update most Debian to bookworm - fix some typos - update loongarch toolchain - fix microbit test - handle GitLab/Cirrus timeout discrepancy - improve avocado console handling - disable mips avocado images pending bugfix
-
pull-maintainer-ominbus-190923-1
testing updates: - update most Debian to bookworm - fix some typos - update loongarch toolchain - fix microbit test - handle GitLab/Cirrus timeout discrepancy - improve avocado console handling - disable mips avocado images pending bugfix
-
pull-maintainer-ominbus-150923-1
testing updates: - update most Debian to bookworm - fix some typos - update loongarch toolchain - fix microbit test - handle GitLab/Cirrus timeout discrepancy - improve avocado console handling - disable mips avocado images pending bugfix
-
pull-maintainer-ominbus-300823-1
testing and gdbstub updates: - enable ccache for gitlab builds - fix various test info leakages for non V=1 - update style to allow loop vars - bump FreeBSD to v13.2 - clean-up gdbstub tests - various gdbstub doc and refactorings