Commits
Commits on Nov 21, 2023
-
ppc/pnv: Fix PNV I2C invalid status after reset
The PNV I2C Controller was clearing the status register after a reset without repopulating the "upper threshold for I2C ports", "Command Complete" and the SCL/SDA input level fields. Fixed this for resets caused by a system reset as well as from writing to the "Immediate Reset" register. Fixes: 263b81e ("ppc/pnv: Add an I2C controller model") Signed-off-by: Glenn Miles <milesg@linux.vnet.ibm.com> Signed-off-by: Cédric Le Goater <clg@kaod.org>
-
ppc/pnv: PNV I2C engines assigned incorrect XSCOM addresses
The PNV I2C engines for power9 and power10 were being assigned a base XSCOM address that was off by one I2C engine's address range such that engine 0 had engine 1's address and so on. The xscom address assignment was being based on the device tree engine numbering, which starts at 1. Rather than changing the device tree numbering to start with 0, the addressing was changed to be based on the existing device tree numbers minus one. Fixes: 1ceda19 ("ppc/pnv: Connect PNV I2C controller to powernv10) Signed-off-by: Glenn Miles <milesg@linux.vnet.ibm.com> Signed-off-by: Cédric Le Goater <clg@kaod.org>
-
target/ppc: Fix bugs in VSX_CVT_FP_TO_INT and VSX_CVT_FP_TO_INT2 macros
The patch below fixes a bug in the VSX_CVT_FP_TO_INT and VSX_CVT_FP_TO_INT2 macros in target/ppc/fpu_helper.c where a non-NaN floating point value from the source vector is incorrectly converted to 0, 0x80000000, or 0x8000000000000000 instead of the expected value if a preceding source floating point value from the same source vector was a NaN. The bug in the VSX_CVT_FP_TO_INT and VSX_CVT_FP_TO_INT2 macros in target/ppc/fpu_helper.c was introduced with commit c3f2425. This patch also adds a new vsx_f2i_nan test in tests/tcg/ppc64 that checks that the VSX xvcvspsxws, xvcvspuxws, xvcvspsxds, xvcvspuxds, xvcvdpsxws, xvcvdpuxws, xvcvdpsxds, and xvcvdpuxds instructions correctly convert non-NaN floating point values to integer values if the source vector contains NaN floating point values. Fixes: c3f2425 ("target/ppc: Clear fpstatus flags on helpers missing it") Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1941 Signed-off-by: John Platts <john_platts@hotmail.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Cédric Le Goater <clg@kaod.org>
-
ppc/pnv: Fix potential overflow in I2C model
Coverity warns that "i2c_bus_busy(i2c->busses[i]) << i" might overflow because the expression is evaluated using 32-bit arithmetic and then used in a context expecting a uint64_t. While we are at it, introduce a PNV_I2C_MAX_BUSSES constant and check the number of busses at realize time. Fixes: Coverity CID 1523918 Cc: Glenn Miles <milesg@linux.vnet.ibm.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Glenn Miles <milesg@linux.vnet.ibm.com> Signed-off-by: Cédric Le Goater <clg@kaod.org>
Commits on Nov 20, 2023
Commits on Nov 19, 2023
-
linux-user: xtensa: fix signal delivery in FDPIC
In FDPIC signal handlers are passed around as FD pointers. Actual code address and GOT pointer must be fetched from memory by the QEMU code that implements kernel signal delivery functionality. This change is equivalent to the following kernel change: 9c2cc74fb31e ("xtensa: fix signal delivery to FDPIC process") Cc: qemu-stable@nongnu.org Fixes: d2796be ("linux-user: add support for xtensa FDPIC") Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Commits on Nov 17, 2023
-
disas/hppa: Show hexcode of instruction along with disassembly
On hppa many instructions can be expressed by different bytecodes. To be able to debug qemu translation bugs it's therefore necessary to see the currently executed byte codes without the need to lookup the sequence without the full executable. With this patch the instruction byte code is shown beside the disassembly. Signed-off-by: Helge Deller <deller@gmx.de> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
-
target/hppa: Fix 64-bit SHRPD instruction
When shifting the two joined 64-bit registers right, shift the upper 64-bit register to the left and the lower 64-bit register to the right before merging them with OR. Signed-off-by: Helge Deller <deller@gmx.de> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Commits on Nov 16, 2023
-
tests/avocado: Enable reverse_debugging.py tests in gitlab CI
Let's try enable reverse_debugging.py in gitlab CI. Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Message-ID: <20231116115354.228678-3-npiggin@gmail.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
-
tests/avocado: reverse_debugging drain console to prevent hang
Like replay_linux.py, reverse_debugging.py starts the vm with console set but does not interact with it (e.g., with wait_for_console_pattern). In this situation, the console should have a drainer attached so the socket does not fill. replay_linux.py has a drainer, but it is missing from reverse_debugging.py. Per analysis in Link: this can cause the console socket/pipe to fill and QEMU get stuck in qemu_chr_write_buffer, leading to strange test case failures (ppc64 fails because it prints a lot to console in early bios). Attaching a drainer prevents this. Note, this commit does not fix bugs introduced by the commits referenced in the first two Fixes: tags, but together those commits conspire to irritate the problem and cause test case failure, which this commit fixes. Link: https://lore.kernel.org/qemu-devel/ZVT-bY9YOr69QTPX@redhat.com/ Fixes: 1d4796c ("python/machine: use socketpair() for console connections") Fixes: 761a13b ("tests/avocado: ppc64 reverse debugging tests for pseries and powernv") Fixes: be52eca ("tests/acceptance: add reverse debugging test") Tested-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Nicholas Piggin <npiggin@gmail.com> Message-ID: <20231116115354.228678-1-npiggin@gmail.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
-
tests/avocado/mem-addr-space-check: Replace assertEquals() for Python…
… 3.12 assertEquals() has been removed in Python 3.12 and should be replaced by assertEqual(). See: https://docs.python.org/3.12/whatsnew/3.12.html#id3 Message-ID: <20231116061956.14676-1-thuth@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Acked-by: Ani Sinha <anisinha@redhat.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
-
tests/avocado/replay_kernel: Mark the test_x86_64_pc as flaky
It's failing very often, so don't run this by default anymore until it gets fixed. Message-ID: <20231114153019.295131-1-thuth@redhat.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Thomas Huth <thuth@redhat.com>
-
tests/avocado: Make fetch_asset() unconditionally require a crypto hash
In a perfect world we'd have reproducible tests, but then we'd be sure we run the same binaries. If a binary artifact isn't hashed, we have no idea what we are running. Therefore enforce hashing for all our artifacts. With this change, unhashed artifacts produce: $ avocado run tests/avocado/multiprocess.py (1/2) tests/avocado/multiprocess.py:Multiprocess.test_multiprocess_x86_64: ERROR: QemuBaseTest.fetch_asset() missing 1 required positional argument: 'asset_hash' (0.19 s) Inspired-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Thomas Huth <thuth@redhat.com> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Message-ID: <20231115205149.90765-1-philmd@linaro.org> Signed-off-by: Thomas Huth <thuth@redhat.com>
-
tests/avocado/multiprocess: Add asset hashes to silence warnings
The multiprocess test is currently succeeding with an annoying warning: (1/2) tests/avocado/multiprocess.py:Multiprocess.test_multiprocess_x86_64: WARN: Test passed but there were warnings during execution. Check the log for details In the log, you can find an entry like: WARNI| No hash provided. Cannot check the asset file integrity. Add the proper asset hashes to avoid those warnings. Message-ID: <20231115145852.494052-1-thuth@redhat.com> Signed-off-by: Thomas Huth <thuth@redhat.com> -
tests/avocado/intel_iommu: Add asset hashes to avoid warnings
The intel_iommu test is currently succeeding with annoying warnings. Add the proper asset hashes to avoid those. Message-ID: <20231114143531.291820-1-thuth@redhat.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
-
tests/avocado/virtio-gpu: Fix test_vhost_user_vga_virgl for edid support
The "edid" feature has been added to vhost-user-gpu in commit c064442 ("contrib/vhost-user-gpu: implement get_edid feature"), so waiting for "features: +virgl -edid" in the test does not work anymore, it's "+edid" instead of "-edid" now! While we're at it, move the expected string to the preceeding exec_command_and_wait_for_pattern() instead (since waiting for empty string here does not make too much sense). Message-ID: <20231114203456.319093-1-thuth@redhat.com> Reviewed-by: Antonio Caggiano <quic_acaggian@quicinc.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
-
tests/avocado: Replace assertRegexpMatches() for Python 3.12 compatib…
…ility assertRegexpMatches() has been removed in Python 3.12 and should be replaced by assertRegex(). See: https://docs.python.org/3.12/whatsnew/3.12.html#id3 Inspired-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Message-ID: <20231114144832.71612-1-philmd@linaro.org> Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
-
tests/avocado: Replace assertEquals() for Python 3.12 compatibility
assertEquals() has been removed in Python 3.12 and should be replaced by assertEqual(). See: https://docs.python.org/3.12/whatsnew/3.12.html#id3 Message-ID: <20231114134326.287242-1-thuth@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Thomas Huth <thuth@redhat.com>
