Stefan Hajnoczi [Tue, 1 Nov 2022 17:49:33 +0000 (13:49 -0400)]
Merge tag 'block-pull-request' of https://gitlab.com/stefanha/qemu into staging
Pull request
Note that we're still discussing "block/blkio: Make driver nvme-io_uring take a
"path" instead of a "filename"". I have sent the pull request now so everything
is ready for the soft freeze tomorrow if we decide to go ahead with the patch.
* tag 'block-pull-request' of https://gitlab.com/stefanha/qemu:
block/blkio: Make driver nvme-io_uring take a "path" instead of a "filename"
block/blkio: Tolerate device size changes
block/blkio: Add virtio-blk-vfio-pci BlockDriver
Stefan Hajnoczi [Tue, 1 Nov 2022 17:39:05 +0000 (13:39 -0400)]
Merge tag 'pull-testing-for-7.2-011122-3' of https://github.com/stsquad/qemu into staging
testing and plugin updates for 7.2:
- cleanup win32/64 docker files
- update test-mingw test
- add flex/bison to debian-all-test
- handle --enable-static/--disable-pie in config
- extend timeouts on x86_64 avocado tests
- add flex/bison to debian-hexagon-cross
- use regular semihosting for nios2 check-tcg
- fix obscure linker error to nios2 softmmu tests
- various windows portability fixes for tests
- clean-up of MAINTAINERS
- use -machine none when appropriate in avocado
- make raspi2_initrd test detect shutdown
- disable sh4 rd2 tests on gitlab
- re-enable threadcount/linux-test for sh4
- clean-up s390x handling of "ex" instruction
- better handle new CPUs in execlog plugin
- pass CONFIG_DEBUG_TCG to plugin builds
- try and avoid races in test-io-channel-command
- speed up ssh key checking for tests/vm
# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEEZoWumedRZ7yvyN81+9DbCVqeKkQFAmNhI/MACgkQ+9DbCVqe
# KkSFXggAg0HIpBDcNz0V5Mh5p69F14pwbDSygKqGDFBebdOHeL7f+WCvQPUGEWxp
# 814zjvRY3SC4Mo4mtzguRvNu0styaUpemvRw5FDYK48GpEjg2eVxTnAFD4nr7ud0
# dhw3iaHP+RjA6s3EpPUqQ5nlZEgFJ+Tvkckk3wKSpksBYA4tJra6Uey5kpZ27x0T
# KOzB2P6w+9B/B11n/aeSxvRPZdnXt2MyfS/3pwwfoFYioEyaEQ3Ie6ooachtdSL3
# PEvnJVK0VVYbZQwBXJlycNLlK/D++s4AEwmnZ5GmvDFuXlkRO9YMy9Wa5TKJl7gz
# 76Aw1KHsE03SyAPvH4bE7eGkIwhJOQ==
# =6hXE
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 01 Nov 2022 09:49:39 EDT
# gpg: using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <[email protected]>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8 DF35 FBD0 DB09 5A9E 2A44
* tag 'pull-testing-for-7.2-011122-3' of https://github.com/stsquad/qemu: (31 commits)
tests/vm: use -o IdentitiesOnly=yes for ssh
tests/unit: cleanups for test-io-channel-command
contrib/plugins: protect execlog's last_exec expansion
contrib/plugins: enable debug on CONFIG_DEBUG_TCG
tests/tcg: include CONFIG_PLUGIN in config-host.mak
target/s390x: fake instruction loading when handling 'ex'
target/s390x: don't probe next pc for EXecuted insns
target/s390x: don't use ld_code2 to probe next pc
tests/tcg: re-enable threadcount for sh4
tests/tcg: re-enable linux-test for sh4
tests/avocado: disable sh4 rd2 tests on Gitlab
tests/avocado: raspi2_initrd: Wait for guest shutdown message before stopping
tests/avocado: set -machine none for userfwd and vnc tests
MAINTAINERS: fix-up for check-tcg Makefile changes
MAINTAINERS: add features_to_c.sh to gdbstub files
MAINTAINERS: add entries for the key build bits
hw/usb: dev-mtp: Use g_mkdir()
block/vvfat: Unify the mkdir() call
tcg: Avoid using hardcoded /tmp
semihosting/arm-compat-semi: Avoid using hardcoded /tmp
...
Stefan Hajnoczi [Tue, 1 Nov 2022 13:36:13 +0000 (09:36 -0400)]
Merge tag 'pull-tcg-20221031-2' of https://gitlab.com/rth7680/qemu into staging
Remove sparc32plus support from tcg/sparc.
target/i386: Use cpu_unwind_state_data for tpr access.
target/i386: Expand eflags updates inline
Complete cpu initialization before registration
* tag 'pull-tcg-20221031-2' of https://gitlab.com/rth7680/qemu:
tests/tcg/multiarch: Add munmap-pthread.c
accel/tcg: Complete cpu initialization before registration
target/i386: Expand eflags updates inline
accel/tcg: Remove reset_icount argument from cpu_restore_state_from_tb
accel/tcg: Remove will_exit argument from cpu_restore_state
target/openrisc: Use cpu_unwind_state_data for mfspr
target/openrisc: Always exit after mtspr npc
target/i386: Use cpu_unwind_state_data for tpr access
accel/tcg: Introduce cpu_unwind_state_data
tcg/tci: fix logic error when registering helpers via FFI
tcg/sparc64: Remove sparc32plus constraints
tcg/sparc64: Rename from tcg/sparc
tcg/sparc: Remove support for sparc32plus
* tag 'qemu-macppc-20221031' of https://github.com/mcayland/qemu:
mac_newworld: Turn CORE99_VIA_CONFIG defines into an enum
mac_{old|new}world: Code style fix adding missing braces to if-s
mac_nvram: Use NVRAM_SIZE constant
hw/ppc/mac.h: Rename to include/hw/nvram/mac_nvram.h
hw/ppc/mac.h: Move PROM and KERNEL defines to board code
hw/ppc/mac.h: Move grackle-pcihost type declaration out to a header
hw/ppc/mac.h: Move macio specific parts out from shared header
hw/ppc/mac.h: Move newworld specific parts out from shared header
mac_{old|new}world: Reduce number of QOM casts
mac_newworld: Clean up creation of Uninorth devices
mac_{old|new}world: Avoid else branch by setting default value
mac_{old|new}world: Set tbfreq at declaration
mac_oldworld: Drop some more variables
mac_newworld: Drop some variables
accel/tcg: Complete cpu initialization before registration
Delay cpu_list_add until realize is complete, so that cross-cpu
interaction does not happen with incomplete cpu state. For this,
we must delay plugin initialization out of tcg_exec_realizefn,
because no cpu_index has been assigned.
Fixes a problem with cross-cpu jump cache flushing, when the
jump cache has not yet been allocated.
target/openrisc: Use cpu_unwind_state_data for mfspr
Since we do not plan to exit, use cpu_unwind_state_data
and extract exactly the data requested.
This is a bug fix, in that we no longer clobber dflag.
Consider:
l.j L2 // branch
l.mfspr r1, ppc // delay
L1: boom
L2: l.lwa r3, (r4)
Here, dflag would be set by cpu_restore_state (because that is the current
state of the cpu), but but not cleared by tb_stop on exiting the TB
(because DisasContext has recorded the current value as zero).
The next TB begins at L2 with dflag incorrectly set. If the load has a
tlb miss, then the exception will be delivered as per a delay slot:
with DSX set in the status register and PC decremented (delay slots
restart by re-executing the branch). This will cause the return from
interrupt to go to L1, and boom!
target/i386: Use cpu_unwind_state_data for tpr access
Avoid cpu_restore_state, and modifying env->eip out from
underneath the translator with TARGET_TB_PCREL. There is
some slight duplication from x86_restore_state_to_opc,
but it's just a few lines.
Ilya Leoshkevich [Thu, 27 Oct 2022 18:36:36 +0000 (19:36 +0100)]
tests/vm: use -o IdentitiesOnly=yes for ssh
When one has a lot of keys in ~/.ssh directory, the ssh command will
try all of them before the one specified on the command line, and this
may cause the remote ssh server to reject the connection due to too
many failed authentication attempts.
Fix by adding -o IdentitiesOnly=yes, which makes the ssh client
consider only the keys specified on the command line.
Alex Bennée [Thu, 27 Oct 2022 18:36:35 +0000 (19:36 +0100)]
tests/unit: cleanups for test-io-channel-command
This test is hanging under heavy load when the two socats race while
trying to create the socket. I've tried various approaches to avoid
the race but it seems "creat=0" won't stop socat trying to create a
pipe if it executes first. In the end I just use a small sleep which
seems to be reliable enough on the load situations I've tried.
While I was there I also properly created a tmpdir for the socket to
live in which is cleaned up at the end of the test.
We originally naively treated expansion as safe because we expected
each new CPU/thread to appear in order. However the -M raspi2 model
triggered a case where a new high cpu_index thread started executing
just before a smaller one.
Clean this up by converting the GArray into the simpler GPtrArray and
then holding a lock for the expansion.
Alex Bennée [Thu, 27 Oct 2022 18:36:33 +0000 (19:36 +0100)]
contrib/plugins: enable debug on CONFIG_DEBUG_TCG
We used to rely on QEMU_CFLAGS to expose the debug flags but now this
is synthesised by meson and only available to the main build. Add our
own flags if we detect the build has been enabled with
CONFIG_DEBUG_TCG (which is the default for --enable-debug anyway).
Alex Bennée [Thu, 27 Oct 2022 18:36:31 +0000 (19:36 +0100)]
target/s390x: fake instruction loading when handling 'ex'
The s390x EXecute instruction is a bit weird as we synthesis the
executed instruction from what we have stored in memory. This missed
the plugin instrumentation.
Work around this with a special helper to inform the rest of the
translator about the instruction so things stay consistent.
Alex Bennée [Thu, 27 Oct 2022 18:36:28 +0000 (19:36 +0100)]
tests/tcg: re-enable threadcount for sh4
This test was marked as broken due to bug #856 which was fixed by ab419fd8a0 (target/sh4: Fix TB_FLAG_UNALIGN). Local testing shows this
is solid now so lets re-enable the test.
Alex Bennée [Thu, 27 Oct 2022 18:36:27 +0000 (19:36 +0100)]
tests/tcg: re-enable linux-test for sh4
This test was marked as broken due to bug #704 which was fixed by aee14c77f4 (linux-user: Rewrite do_getdents, do_getdents64). Local
testing shows this is solid now so lets re-enable the test.
Alex Bennée [Thu, 27 Oct 2022 18:36:26 +0000 (19:36 +0100)]
tests/avocado: disable sh4 rd2 tests on Gitlab
Running repeated invocations on a number of test boxes show a fairly
high error rate:
$ retry.py -n 100 -c -- ./tests/venv/bin/avocado run tests/avocado/boot_linux_console.py:BootLinuxConsole.test_sh4_r2d
retry.py called with ['./tests/venv/bin/avocado', 'run', 'tests/avocado/boot_linux_console.py:BootLinuxConsole.test_sh4_r2d']
Results:
Run, Ret, Pass/Fail, Time, Total Pass, Total Run
...
Results summary:
0: 94 times (94.00%), avg time 2.254 (0.00 varience/0.04 deviation)
1: 3 times (3.00%), avg time 1.837 (0.02 varience/0.14 deviation)
8: 3 times (3.00%), avg time 91.288 (0.02 varience/0.15 deviation)
Examining the logs they fall into various categories of un-handled
unaligned access by user space and unexpected FPU usage by the kernel
which ultimately lead to the failure to reach the login prompt. This
could be bugs in the translator that only get hit occasionally or just
a flaky kernel - its hard to tell. To avoid these failures gating CI
lets skip on GitLab.
Peter Maydell [Thu, 27 Oct 2022 18:36:25 +0000 (19:36 +0100)]
tests/avocado: raspi2_initrd: Wait for guest shutdown message before stopping
The avocado test
tests/avocado/boot_linux_console.py:BootLinuxConsole.test_arm_raspi2_initrd
finishes with
exec_command(self, 'halt')
# Wait for VM to shut down gracefully
self.vm.wait()
In theory this should be fine. In practice it runs into two bugs:
* when the test calls self.vm.wait() Avocado closes the socket
connection to the guest serial console immediately, so the
avocado logs don't have the last part of the guest output:
https://gitlab.com/qemu-project/qemu/-/issues/1265
* when the socket is closed, a bug in the QEMU socket chardev
means that it loses any data that the guest UART has not
yet consumed. This means that the guest doesn't always read
the full 'halt' command string, so the test intermittently
fails with a timeout:
https://gitlab.com/qemu-project/qemu/-/issues/1264
Work around both of these by waiting for the guest to print the
string that means it has completed the shutdown process. This fixes
a very long standing intermittent failure in this test.
Alex Bennée [Thu, 27 Oct 2022 18:36:24 +0000 (19:36 +0100)]
tests/avocado: set -machine none for userfwd and vnc tests
These are exercising core QEMU features and don't actually run code.
Not specifying a machine will fail when avocado chooses the native
arch binary to run. Be explicit.
Alex Bennée [Thu, 27 Oct 2022 18:36:21 +0000 (19:36 +0100)]
MAINTAINERS: add entries for the key build bits
Changes to the build files are a bit special in that they usually go
through other maintainer trees. However considering the build system
is the root of everything a developer is likely to do we should at
least set it out in MAINTAINERS.
I'm going to nominate Paolo for meson stuff given the conversion was
his passion project. I'm happy to cast an eye over configure stuff
considering a lot of the cross compile logic is in there anyway.
Alex Bennée [Thu, 27 Oct 2022 18:36:14 +0000 (19:36 +0100)]
tests/tcg: use regular semihosting for nios2-softmmu
The nios2 code now plugs into the common semihosting code so we can
use the same redirect invocation as the other boards. There is however
a bug raised for the fact the tests don't seem to be completing
properly and silently passing anyway:
Anton Johansson [Thu, 27 Oct 2022 18:36:13 +0000 (19:36 +0100)]
tests/docker: Add flex/bison to `debian-hexagon-cross`
debian-hexagon-cross contains two images, one to build the toolchain
used for building the Hexagon tests themselves, and one image to build
QEMU and run the tests.
This commit adds flex/bison to the final image that builds QEMU so that
it can also build idef-parser.
Note: This container is not built by the CI and needs to be rebuilt and
updated manually.
Anton Johansson [Thu, 27 Oct 2022 18:36:09 +0000 (19:36 +0100)]
tests/docker: Add flex/bison to `debian-all-test`
Adds flex/bison to the debian-all-test-cross container which was missed
in the previous CI patch. These dependencies are required by the
idef-parser patchset for target/hexagon.
Alex Bennée [Thu, 27 Oct 2022 18:36:08 +0000 (19:36 +0100)]
tests/docker: update test-mingw to run single build
While the test-build test happily run for mingw the test-mingw case
runs more of the packaging inline with what our CI does. It however
fails if we don't find both compilers and expects to be run on a
docker image with both.
Remove that distinction and make it work more like the other build
test scripts.
Icenowy Zheng [Fri, 28 Oct 2022 19:23:44 +0000 (06:23 +1100)]
tcg/tci: fix logic error when registering helpers via FFI
When registering helpers via FFI for TCI, the inner loop that iterates
parameters of the helper reuses (and thus pollutes) the same variable
used by the outer loop that iterates all helpers, thus made some helpers
unregistered.
Fix this logic error by using a dedicated temporary variable for the
inner loop.
Since 9b9c37c36439, we have only supported sparc64 cpus.
Debian and Gentoo now only support 64-bit sparc64 userland,
so it is time to drop the 32-bit sparc64 userland: sparc32plus.
BALATON Zoltan [Fri, 28 Oct 2022 11:56:26 +0000 (13:56 +0200)]
hw/ppc/mac.h: Move PROM and KERNEL defines to board code
The PROM_FILENAME and KERNEL_* defines are used by mac_oldworld and
mac_newworld but they don't have to be identical so these could be
moved to the individual boards.
BALATON Zoltan [Fri, 28 Oct 2022 11:56:20 +0000 (13:56 +0200)]
mac_{old|new}world: Avoid else branch by setting default value
Several variables are set in if-else branches where the else branch
can be removed by setting a default value at the variable declaration
which leads to simlpler code that is easier to follow.
BALATON Zoltan [Fri, 28 Oct 2022 11:56:17 +0000 (13:56 +0200)]
mac_newworld: Drop some variables
Values not used frequently enough may not worth putting in a local
variable, especially with names almost as long as the original value
because that does not improve readability, to the contrary it makes it
harder to see what value is used. Drop a few such variables. This is
the same clean up that was done for mac_oldworld in commit b8df32555ce5.
Alberto Faria [Fri, 28 Oct 2022 23:38:54 +0000 (00:38 +0100)]
block/blkio: Make driver nvme-io_uring take a "path" instead of a "filename"
The nvme-io_uring driver expects a character special file such as
/dev/ng0n1. Follow the convention of having a "filename" option when a
regular file is expected, and a "path" option otherwise.
This makes io_uring the only libblkio-based driver with a "filename"
option, as it accepts a regular file (even though it can also take a
block special file).
Alberto Faria [Sat, 29 Oct 2022 12:20:31 +0000 (13:20 +0100)]
block/blkio: Tolerate device size changes
Some libblkio drivers may be able to work with regular files (e.g.,
io_uring) or otherwise resizable devices. Conservatively set
BlockDriver::has_variable_length to true to ensure bdrv_nb_sectors()
always gives up-to-date results.
Also implement BlockDriver::bdrv_co_truncate for the case where no
preallocation is needed and the device already has a size compatible
with what was requested.
Peter Maydell [Mon, 31 Oct 2022 13:29:01 +0000 (13:29 +0000)]
net/vhost-vdpa.c: Fix clang compilation failure
Commit 8801ccd0500437 introduced a compilation failure with clang
version 10.0.0-4ubuntu1:
../../net/vhost-vdpa.c:654:16: error: variable 'vdpa_device_fd' is
used uninitialized whenever 'if' condition is false
[-Werror,-Wsometimes-uninitialized]
} else if (opts->has_vhostfd) {
^~~~~~~~~~~~~~~~~
../../net/vhost-vdpa.c:662:33: note: uninitialized use occurs here
r = vhost_vdpa_get_features(vdpa_device_fd, &features, errp);
^~~~~~~~~~~~~~
../../net/vhost-vdpa.c:654:12: note: remove the 'if' if its condition
is always true
} else if (opts->has_vhostfd) {
^~~~~~~~~~~~~~~~~~~~~~~
../../net/vhost-vdpa.c:629:23: note: initialize the variable
'vdpa_device_fd' to silence this warning
int vdpa_device_fd;
^
= 0
1 error generated.
It's a false positive -- the compiler doesn't manage to figure out
that the error checks further up mean that there's no code path where
vdpa_device_fd isn't initialized. Put another way, the problem is
that we check "if (opts->has_vhostfd)" when in fact that condition
must always be true. A cleverer static analyser would probably warn
that we were checking an always-true condition.
Fix the compilation failure by removing the unnecessary if().
Stefan Hajnoczi [Mon, 31 Oct 2022 17:01:12 +0000 (13:01 -0400)]
Merge tag 'pull-qemu-20221031' of https://gitlab.com/stweil/qemu into staging
Patches for Windows
# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEESSNv6nXJXWmOwreK4Iwh1Wd0UK0FAmNfkC0ACgkQ4Iwh1Wd0
# UK2tEw//QQapqOnJQmjiGVWJ3xUEVoDUmZbhXlPaLKOzSkAnrnIdIO2p4jsXrITi
# LfL8PF7wCg1+oldPsKmhh+ZD6XiUyNBpTt61atXRJS/TybKAGjI33XC+/Hliwity
# 4A0+WxJ960ExLwZzaE/ANTYM86Jo5SqhzACosGh8txRbL1rkmqlrCU1DwnhU6vq1
# 5ph1HFgpqkii43Eiq+v1nmkbh5MWVuMap6MOoQzgwVTkaUU0cTmR8/KqhSHrIryj
# xFPH2wY8pA3vFgMv5OSOjq5Lg197kiWUyhJa6eBBsi4MKnQgRzAxE6yHhpyYZ5EA
# dMW9iLhPVFRDAoQOiSRLj/NA1nl2gwDdjs5WhKqF6AtxMck5IDqltLKFvLlXIxiK
# BYi3ghVeA5LUarcxuAOHse8rCXxBaIJI3aSolO5fDe0mcpNIb7CgCleBKlnBWEsP
# GtRhr1AkoKHcetO5iEfg1QG71/XWdWWy3hfW39GJeBl9C7/AxzoLC7yStI7Iv3b4
# tv/Tylt+Js1KadA9z/tof4wm4NkGf2Q9aFoSbm4pSZH+7b4ZI5LVLlDKYCnjT37v
# LekyJgkU3wRjKdLkM1n6qhsa5Ey2D7STw9ANWQwqOImoj5Dkix2FIqfaydctgrxq
# zmdQpJhOzIO8b9vVSRLn2xYtae5LNlxiAx85r5l11jwfqDOWTts=
# =ZJqu
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 31 Oct 2022 05:06:53 EDT
# gpg: using RSA key 49236FEA75C95D698EC2B78AE08C21D5677450AD
# gpg: Good signature from "Stefan Weil <[email protected]>" [unknown]
# gpg: aka "Stefan Weil (Universitätsbibliothek Mannheim) <[email protected]>" [unknown]
# gpg: aka "Stefan Weil <[email protected]>" [unknown]
# gpg: aka "Stefan Weil <[email protected]>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 4923 6FEA 75C9 5D69 8EC2 B78A E08C 21D5 6774 50AD
* tag 'pull-qemu-20221031' of https://gitlab.com/stweil/qemu:
block/nfs: Fix 32-bit Windows build
scripts/nsis.py: Automatically package required DLLs of QEMU executables
scripts/nsis.py: Fix destination directory name when invoked on Windows
scripts/nsis.py: Drop the unnecessary path separator
Stefan Hajnoczi [Mon, 31 Oct 2022 10:36:15 +0000 (06:36 -0400)]
Merge tag 'mips-20221030' of https://github.com/philmd/qemu into staging
MIPS patches queue
- Convert nanoMIPS disassembler from C++ to C (Milica Lazarevic)
- Consolidate VT82xx/PIIX south bridges (Bernhard Beschow)
- Remove unused MAX_IDE_BUS definition (Zoltan Balaton)
- Fix branch displacement for BEQZC/BNEZC (David Daney)
- Don't set link_up for Boston's xilinx-pcie (Jiaxun Yang)
- Use bootloader API to set BAR registers in Malta (Jiaxun Yang)
* tag 'mips-20221030' of https://github.com/philmd/qemu: (55 commits)
hw/mips/malta: Use bootloader helper to set BAR registers
hw/mips: Use bl_gen_kernel_jump to generate bootloaders
hw/mips/bootloader: Allow bl_gen_jump_kernel to optionally set register
hw/mips/boston: Don't set link_up for xilinx-pcie
hw/isa/piix4: Move pci_ide_create_devs() call to board code
hw/isa/piix4: Add missing initialization
hw/isa/Kconfig: Fix dependencies of piix4 southbridge
hw/mips/malta: Reuse dev variable
hw/isa/piix3: Remove unused include
hw/ide/piix: Introduce TYPE_ macros for PIIX IDE controllers
hw/isa/piix4: Rename wrongly named method
hw/isa/piix3: Prefer pci_address_space() over get_system_memory()
hw/isa/piix3: Modernize reset handling
hw/isa/piix3: Add size constraints to rcr_ops
hw/isa/piix3: Remove extra ';' outside of functions
hw/i386/pc: Create DMA controllers in south bridges
disas/mips: Fix branch displacement for BEQZC and BNEZC
disas/nanomips: Rename nanomips.cpp to nanomips.c
disas/nanomips: Remove argument passing by ref
disas/nanomips: Replace Cpp enums for C enums
...
hw/mips/bootloader: Allow bl_gen_jump_kernel to optionally set register
When one of the $sp/$a[0..3] register is already set, we might
want bl_gen_jump_kernel() to NOT set it again. Pass a boolean
argument for each register, to allow to optionally set them.
Jiaxun Yang [Mon, 24 Oct 2022 14:35:40 +0000 (15:35 +0100)]
hw/mips/boston: Don't set link_up for xilinx-pcie
PCIe port 0 and 1 had link_up set as false previously,
that makes those two ports effectively useless. It can
be annoying for users to find that the device they plug
on those buses won't work at all.
As link_up is true by default, just don't set it again in
boston platform code.
Bernhard Beschow [Sat, 22 Oct 2022 15:04:31 +0000 (17:04 +0200)]
hw/isa/piix3: Prefer pci_address_space() over get_system_memory()
get_system_memory() accesses global state while pci_address_space() uses
whatever has been passed to the device instance, so avoid the global.
Moreover, PIIX4 uses pci_address_space() here as well.
Bernhard Beschow [Sat, 22 Oct 2022 15:04:30 +0000 (17:04 +0200)]
hw/isa/piix3: Modernize reset handling
Rather than registering the reset handler via a function which
appends the handler to a global list, prefer to implement it as
a virtual method - PIIX4 does the same already.
Note that this means that piix3_reset can now also be called writing to
the relevant configuration space register on a PCI bridge.
David Daney [Fri, 14 Oct 2022 11:23:22 +0000 (13:23 +0200)]
disas/mips: Fix branch displacement for BEQZC and BNEZC
disas/mips.c got added in commit 6643d27ea0 ("MIPS disas support")
apparently based on binutils tag 'gdb_6_1-branchpoint' [1].
Back then, MIPSr6 was not supported (added in binutils commit 7361da2c952 during 2014 [2]).
Binutils codebase diverged so much over the last 18 years, it is
not possible to simply cherry-pick their changes, so fix it BEQZC /
BNEZC 21-bit signed branch displacement locally.
Since there's no support for exception handling in C, the try-catch
blocks have been deleted, and throw clauses are replaced. When a runtime
error happens, we're printing out the error message. Disassembling of
the current instruction interrupts. This behavior is achieved by adding
sigsetjmp() to discard further disassembling after the error message
prints and by adding the siglongjmp() function to imitate throwing an
error. The goal was to maintain the same output as it was.
This patch expands the Dis_info struct, which should hold the
necessary data for handling runtime errors. Fields fprintf_func and
stream are in charge of error printing. Field buf enables the use of
sigsetjmp() and siglongjmp() functions. Support for runtime error
handling will be added later.
We're filling Dis_info at the entrance of the nanoMIPS disassembler,
i.e. print_insn_nanomips. Next, we're adding that information as an
argument wherever we need to.
Before this patch, we'd been calling img_format twice, the first time
through the CPR function to get an appropriate string and the second
time to print that formatted string. There's no more need for that.
Therefore, calls to CPR are removed, and now we're directly printing
"CP" and integer value instead.
Both versions of IMMEDIATE functions have been removed.
Before this patch, we'd been calling img_format twice, the first time
through the IMMEDIATE to get an appropriate string and the second time
to print that string. There's no more need for that. Therefore, calls to
IMMEDIATE are removed, and now we're directly printing the integer
values instead.
The return type of typedef disassembly_function is changed to char *
instead of std::string. Therefore, for every particular
disassembly_function function signature is changed.
For example:
- static std::string ABS_D(uint64 instruction, img_address m_pc) {...}
is replaced with
- static char *ABS_D(uint64 instruction, img_address m_pc) {...}
Every helper function used to return std::string is changed to return
const char * or char *. Where the return value points to a static string
that the caller must not free, the return type is const char *. If a
function allocates memory and the caller is required to free it, the
return type is a char *. This applies to the following functions:
img_format, to_string, GPR, save_restore_list, FPR, etc.
Now that we replaced every std::string for const char * or char *, it is
possible to delete multiple versions of the img_format function. The
general version:
- static char *img_format(const char *format, ...) {...}
can handle all string formatting, so others have been deleted.
Where necessary, strings are dynamically allocated with g_strjoinv,
g_strdup, g_strdup_vprintf, and g_strdup_printf. Memory leaking will be
prevented later.
String concatenation in the save_restore_list() function is handled
using g_strjoinv() function instead of += operator.
The type of the "dis" parameter in the Disassemble function is changed
- from std::string &
- to char **
Without applying all of these changes, the nanomips disassembler may be
buildable but can't produce the appropriate output, so all of them are
made together.
Following functions just wrap the decode_gpr_gpr3() function:
- encode_rs3_and_check_rs3_ge_rt3()
- encode_rs3_and_check_rs3_lt_rt3()
Therefore those have been deleted. Calls to these two functions have
been replaced with calls to decode_gpr_gpr3.
Header file nanomips.h has been deleted for the nanomips disassembler to
stay consistent with the rest of the disassemblers which don't include
extra header files.
The following is moved from the nanomips.h to nanomips.cpp file:
- #include line
- typedefs
- enums
- definition of the Pool struct.
Header file nanomips.h will be deleted to be consistent with the rest of
the disas/ code.
projects / qemu.git / log