Mark Cave-Ayland [Wed, 30 Nov 2022 10:04:34 +0000 (10:04 +0000)]
tcg: convert tcg/README to rst
Convert tcg/README to rst and move it to docs/devel as a new "TCG Intermediate
Representation" page. There are a few minor changes to improve the aesthetic
of the final output which are as follows:
- Rename the title from "Tiny Code Generator - Fabrice Bellard" to "TCG
Intermediate Representation"
- Remove the section numbering
- Add the missing parameters to the ssadd_vec operations in the "Host
vector operations" section
- Change the path to the Atomic Operations document to use a proper
reference
- Replace tcg/README in tcg.rst with a proper reference to the new document
Thomas Huth [Wed, 4 Jan 2023 12:35:59 +0000 (13:35 +0100)]
.gitlab-ci.d/windows: Work-around timeout and OpenGL problems of the MSYS2 jobs
The windows jobs (especially the 32-bit job) recently started to
hit the timeout limit. Bump it a little bit to ease the situation
(80 minutes is quite long already - OTOH, these jobs do not have to
wait for a job from the container stage to finish, so this should
still be OK).
Additionally, some update on the container side recently enabled
OpenGL in these jobs - but the corresponding code fails to compile.
Thus disable OpenGL here for the time being until someone figured
out the proper fix in the shader code for this.
Peter Maydell [Wed, 4 Jan 2023 16:47:37 +0000 (16:47 +0000)]
Merge tag 'pull-testing-next-231222-1' of https://gitlab.com/stsquad/qemu into staging
testing updates:
- fix minor shell-ism that can break check-tcg
- turn off verbose logging on custom runners
- make configure echo call in CI
- fix unused variable in linux-test
- add binary compiler docker image for hexagon
- disable doc and gui builds for tci and disable-tcg builds
# gpg: Signature made Fri 23 Dec 2022 15:18:41 GMT
# gpg: using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <[email protected]>" [full]
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8 DF35 FBD0 DB09 5A9E 2A44
* tag 'pull-testing-next-231222-1' of https://gitlab.com/stsquad/qemu:
gitlab-ci: Disable docs and GUIs for the build-tci and build-tcg-disabled jobs
tests/docker: use prebuilt toolchain for debian-hexagon-cross
tests/tcg: fix unused variable in linux-test
configure: repeat ourselves for the benefit of CI
gitlab: turn off verbose logging for make check on custom runners
configure: Fix check-tcg not executing any tests
Peter Maydell [Wed, 4 Jan 2023 14:53:59 +0000 (14:53 +0000)]
Merge tag 'pull-9p-20221223' of https://github.com/cschoenebeck/qemu into staging
9pfs: Windows host prep, cleanup
* Next preparatory patches for upcoming Windows host support.
* Cleanup patches.
# gpg: Signature made Fri 23 Dec 2022 11:04:26 GMT
# gpg: using RSA key 96D8D110CF7AF8084F88590134C2B58765A47395
# gpg: issuer "[email protected]"
# gpg: Good signature from "Christian Schoenebeck <[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: ECAB 1A45 4014 1413 BA38 4926 30DB 47C3 A012 D5F4
# Subkey fingerprint: 96D8 D110 CF7A F808 4F88 5901 34C2 B587 65A4 7395
* tag 'pull-9p-20221223' of https://github.com/cschoenebeck/qemu:
hw/9pfs: Replace the direct call to xxxat() APIs with a wrapper
hw/9pfs: Drop unnecessary *xattr wrapper API declarations
qemu/xattr.h: Exclude <sys/xattr.h> for Windows
MAINTAINERS: Add 9p test client to section "virtio-9p"
9pfs: Fix some return statements in the synth backend
Thomas Huth [Wed, 21 Dec 2022 09:04:11 +0000 (09:04 +0000)]
gitlab-ci: Disable docs and GUIs for the build-tci and build-tcg-disabled jobs
These jobs use their own "script:" section and thus do not profit from
the global "--disable-docs" from the template. While we're at it, disable
also some GUI front ends here since we do not gain any additional test
coverage by compiling those here again.
tests/docker: use prebuilt toolchain for debian-hexagon-cross
The current docker image for cross compiling hexagon guests
is manually built since it takes >2 hours to build from source.
This patch:
1. Solves the above issue by using the prebuilt clang
toolchain hosted on CodeLinaro [1] and maintained by QUIC [2].
2. The dockerfile is also switched from multi-stage to single stage
build to allow the CI docker engine to reuse the layer cache.
3. Re-enables the hexagon-cross-container job to be always run in
CI and makes it a non-optional dependency for the
build-user-hexagon job.
The changes for 1 & 2 together bring down the build time to
~3 minutes in GitLab CI when cache is reused and ~9 minutes
when cache cannot be reused.
Alex Bennée [Wed, 21 Dec 2022 09:04:09 +0000 (09:04 +0000)]
tests/tcg: fix unused variable in linux-test
The latest hexagon compiler picks up that we never consume wcount.
Given the name of the #define that rcount checks against is WCOUNT_MAX
I figured the check just got missed.
Alex Bennée [Wed, 21 Dec 2022 09:04:08 +0000 (09:04 +0000)]
configure: repeat ourselves for the benefit of CI
Our CI system echos the lines it executes but not the expansions. For
the sake of a line of extra verbosity during the configure phase lets
echo the invocation of script to stdout as well as the log when on CI.
Alex Bennée [Wed, 21 Dec 2022 09:04:07 +0000 (09:04 +0000)]
gitlab: turn off verbose logging for make check on custom runners
The verbosity adds a lot of unnecessary output to the CI logs which
end up getting truncated anyway. We can always extract information
from the meson test logs on a failure and for the custom runners its
generally easier to re-create failures anyway.
After configuring with --target-list=hexagon-linux-user
running `make check-tcg` just prints the following:
```
make: Nothing to be done for 'check-tcg'
```
In the probe_target_compiler function, the 'break'
command is used incorrectly. There are no lexically
enclosing loops associated with that break command which
is an unspecfied behaviour in the POSIX standard.
The dash shell implementation aborts the currently executing
loop, in this case, causing the rest of the logic for the loop
in line 2490 to be skipped, which means no Makefiles are
generated for the tcg target tests.
Fixes: c3b570b5a9a24d25 (configure: don't enable
cross compilers unless in target_list)
MAINTAINERS: Add 9p test client to section "virtio-9p"
The 9p test cases use a dedicated, lite-weight 9p client implementation
(using virtio transport) under tests/qtest/libqos/ to communicate with
QEMU's 9p server.
It's already there for a long time. Let's officially assign it to 9p
maintainers.
Greg Kurz [Thu, 24 Nov 2022 15:58:38 +0000 (16:58 +0100)]
9pfs: Fix some return statements in the synth backend
The qemu_v9fs_synth_mkdir() and qemu_v9fs_synth_add_file() functions
currently return a positive errno value on failure. This causes
checkpatch.pl to spit several errors like the one below:
ERROR: return of an errno should typically be -ve (return -EAGAIN)
+ return EAGAIN;
Simply change the sign. This has no consequence since callers
assert() the returned value to be equal to 0.
Peter Maydell [Wed, 21 Dec 2022 18:08:09 +0000 (18:08 +0000)]
Merge tag 'pull-ppc-20221221' of https://gitlab.com/danielhb/qemu into staging
ppc patch queue for 2022-12-21:
This queue contains a MAINTAINERS update, the implementation of the Freescale eSDHC,
the introduction of the DEXCR/HDEXCR instructions and other assorted fixes (most of
them for the e500 board).
# gpg: Signature made Wed 21 Dec 2022 17:18:53 GMT
# gpg: using EDDSA key 17EBFF9923D01800AF2838193CD9CA96DE033164
# gpg: issuer "[email protected]"
# gpg: Good signature from "Daniel Henrique Barboza <[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: 17EB FF99 23D0 1800 AF28 3819 3CD9 CA96 DE03 3164
* tag 'pull-ppc-20221221' of https://gitlab.com/danielhb/qemu:
target/ppc: Check DEXCR on hash{st, chk} instructions
target/ppc: Implement the DEXCR and HDEXCR
hw/ppc/e500: Move comment to more appropriate place
hw/ppc/e500: Resolve variable shadowing
hw/ppc/e500: Prefer local variable over qdev_get_machine()
hw/ppc/virtex_ml507: Prefer local over global variable
target/ppc/mmu_common: Fix table layout of "info tlb" HMP command
target/ppc/mmu_common: Log which effective address had no TLB entry found
hw/ppc/spapr: Reduce "vof.h" inclusion
hw/ppc/vof: Do not include the full "cpu.h"
target/ppc/kvm: Add missing "cpu.h" and "exec/hwaddr.h"
hw/ppc/e500: Add Freescale eSDHC to e500plat
hw/sd/sdhci: Support big endian SD host controller interfaces
MAINTAINERS: downgrade PPC KVM/TCG CPUs and pSeries to 'Odd Fixes'
Define the DEXCR and HDEXCR as special purpose registers.
Each register occupies two SPR indicies, one which can be read in an
unprivileged state and one which can be modified in the appropriate
priviliged state, however both indicies refer to the same underlying
value.
Note that the ISA uses the abbreviation UDEXCR in two different
contexts: the userspace DEXCR, the SPR index which can be read from
userspace (implemented in this patch), and the ultravisor DEXCR, the
equivalent register for the ultravisor state (not implemented).
Currently objects including "hw/ppc/spapr.h" are forced to be
target specific due to the inclusion of "vof.h" in "spapr.h".
"spapr.h" only uses a Vof pointer, so doesn't require the structure
declaration. The only place where Vof structure is accessed is in
spapr.c, so include "vof.h" there, and forward declare the structure
in "spapr.h".
target/ppc/kvm: Add missing "cpu.h" and "exec/hwaddr.h"
kvm_ppc.h is missing various declarations from "cpu.h":
target/ppc/kvm_ppc.h:128:40: error: unknown type name 'CPUPPCState'; did you mean 'CPUState'?
static inline int kvmppc_get_hypercall(CPUPPCState *env,
^~~~~~~~~~~
CPUState
include/qemu/typedefs.h:45:25: note: 'CPUState' declared here
typedef struct CPUState CPUState;
^
target/ppc/kvm_ppc.h:134:40: error: unknown type name 'PowerPCCPU'
static inline int kvmppc_set_interrupt(PowerPCCPU *cpu, int irq, int level)
^
target/ppc/kvm_ppc.h:285:38: error: unknown type name 'hwaddr'
hwaddr ptex, int n)
^
target/ppc/kvm_ppc.h:220:15: error: unknown type name 'target_ulong'
static inline target_ulong kvmppc_configure_v3_mmu(PowerPCCPU *cpu,
^
target/ppc/kvm_ppc.h:286:38: error: unknown type name 'ppc_hash_pte64_t'
static inline void kvmppc_read_hptes(ppc_hash_pte64_t *hptes,
^
hw/sd/sdhci: Support big endian SD host controller interfaces
Some SDHCI IP can be synthetized in various endianness:
https://github.com/u-boot/u-boot/blob/v2021.04/doc/README.fsl-esdhc
- CONFIG_SYS_FSL_ESDHC_BE
ESDHC IP is in big-endian mode. Accessing ESDHC registers can be
determined by ESDHC IP's endian mode or processor's endian mode.
Our current implementation is little-endian. In order to support
big endianness:
- Rename current MemoryRegionOps as sdhci_mmio_le_ops ('le')
- Add an 'endianness' property to SDHCIState (default little endian)
- Set the 'io_ops' field in realize() after checking the property
- Add the sdhci_mmio_be_ops (big-endian) MemoryRegionOps.
MAINTAINERS: downgrade PPC KVM/TCG CPUs and pSeries to 'Odd Fixes'
The maintainer is no longer being paid to maintain these components. All
maintainership work is being done in his personal time since the middle
of the 7.2 development cycle.
Change the status of PPC KVM CPUs, PPC TCG CPUs and the pSeries machine
to 'Odd Fixes', reflecting that the maintainer no longer has exclusive
time to dedicate to them. It'll also (hopefully) keep expectations under
check when/if these components are used in a customer product.
Peter Maydell [Wed, 21 Dec 2022 15:44:08 +0000 (15:44 +0000)]
Merge tag 'for_upstream' of https://git.kernel.org/pub/scm/virt/kvm/mst/qemu into staging
virtio,pc,pci: features, cleanups, fixes
make TCO watchdog work by default
part of generic vdpa support
asid interrupt for vhost-vdpa
added flex bus port DVSEC for cxl
misc fixes, cleanups, documentation
Signed-off-by: Michael S. Tsirkin <[email protected]>
# gpg: Signature made Wed 21 Dec 2022 12:32:36 GMT
# gpg: using RSA key 5D09FD0871C8F85B94CA8A0D281F0DB8D28D5469
# gpg: issuer "[email protected]"
# gpg: Good signature from "Michael S. Tsirkin <[email protected]>" [full]
# gpg: aka "Michael S. Tsirkin <[email protected]>" [full]
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17 0970 C350 3912 AFBE 8E67
# Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA 8A0D 281F 0DB8 D28D 5469
* tag 'for_upstream' of https://git.kernel.org/pub/scm/virt/kvm/mst/qemu: (41 commits)
contrib/vhost-user-blk: Replace lseek64 with lseek
libvhost-user: Switch to unsigned int for inuse field in struct VuVirtq
hw/virtio: Extract QMP related code virtio-qmp.c
hw/virtio: Extract config read/write accessors to virtio-config-io.c
hw/virtio: Constify qmp_virtio_feature_map_t[]
hw/virtio: Guard and restrict scope of qmp_virtio_feature_map_t[]
hw/virtio: Rename virtio_ss[] -> specific_virtio_ss[]
hw/virtio: Add missing "hw/core/cpu.h" include
hw/cxl/device: Add Flex Bus Port DVSEC
hw/acpi: Rename tco.c -> ich9_tco.c
acpi/tests/avocado/bits: add mformat as one of the dependencies
docs/acpi/bits: document BITS_DEBUG environment variable
pci: drop redundant PCIDeviceClass::is_bridge field
remove DEC 21154 PCI bridge
vhost: fix vq dirty bitmap syncing when vIOMMU is enabled
acpi/tests/avocado/bits: add SPDX license identifiers for bios bits tests
include/hw: attempt to document VirtIO feature variables
vhost-user: send set log base message only once
vdpa: always start CVQ in SVQ mode if possible
vdpa: add shadow_data to vhost_vdpa
...
Peter Maydell [Wed, 21 Dec 2022 14:15:18 +0000 (14:15 +0000)]
Merge tag 'pull-tcg-20221220' of https://gitlab.com/rth7680/qemu into staging
Use interval trees for user-only vma mappings.
Assorted cleanups to page locking.
# gpg: Signature made Wed 21 Dec 2022 05:00:30 GMT
# gpg: using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg: issuer "[email protected]"
# gpg: Good signature from "Richard Henderson <[email protected]>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A 05C0 64DF 38E8 AF7E 215F
* tag 'pull-tcg-20221220' of https://gitlab.com/rth7680/qemu:
accel/tcg: Restrict page_collection structure to system TB maintainance
accel/tcg: Factor tb_invalidate_phys_range_fast() out
accel/tcg: Rename tb_invalidate_phys_page_fast{,__locked}()
accel/tcg: Remove trace events from trace-root.h
accel/tcg: Restrict cpu_io_recompile() to system emulation
accel/tcg: Move remainder of page locking to tb-maint.c
accel/tcg: Move PageDesc tree into tb-maint.c for system
accel/tcg: Use interval tree for user-only page tracking
accel/tcg: Move page_{get,set}_flags to user-exec.c
accel/tcg: Drop PAGE_RESERVED for CONFIG_BSD
accel/tcg: Use interval tree for TARGET_PAGE_DATA_SIZE
accel/tcg: Use interval tree for TBs in user-only mode
accel/tcg: Rename page_flush_tb
util: Add interval-tree.c
Peter Maydell [Tue, 20 Dec 2022 17:46:38 +0000 (17:46 +0000)]
block/io: Check for replay-enabled in bdrv_drain_all_begin()
In commit da0bd74434 we refactored bdrv_drain_all_begin() to pull out
the non-polling part into bdrv_drain_all_begin_nopoll(). This change
broke record-and-replay, because the "return early if replay enabled"
check is now in the sub-function bdrv_drain_all_begin_nopoll(), and
so it only causes us to return from that function, and not from the
calling bdrv_drain_all_begin().
Fix the regression by checking whether replay is enabled in both
functions.
The breakage and fix can be tested via 'make check-avocado': the
tests/avocado/reverse_debugging.py:ReverseDebugging_X86_64.test_x86_64_pc
tests/avocado/reverse_debugging.py:ReverseDebugging_AArch64.test_aarch64_virt
tests were both broken by this.
Khem Raj [Sun, 18 Dec 2022 22:07:40 +0000 (14:07 -0800)]
contrib/vhost-user-blk: Replace lseek64 with lseek
64bit off_t is already in use since build uses _FILE_OFFSET_BITS=64
already. Using lseek/off_t also makes it work with latest musl without
using _LARGEFILE64_SOURCE macro. This macro is implied with _GNU_SOURCE
when using glibc but not with musl.
Marcel Holtmann [Mon, 19 Dec 2022 17:53:37 +0000 (18:53 +0100)]
libvhost-user: Switch to unsigned int for inuse field in struct VuVirtq
It seems there is no need to keep the inuse field signed and end up with
compiler warnings for sign-compare.
CC libvhost-user.o
libvhost-user.c: In function ‘vu_queue_pop’:
libvhost-user.c:2763:19: error: comparison of integer expressions of different signedness: ‘int’ and ‘unsigned int’ [-Werror=sign-compare]
2763 | if (vq->inuse >= vq->vring.num) {
| ^~
libvhost-user.c: In function ‘vu_queue_rewind’:
libvhost-user.c:2808:13: error: comparison of integer expressions of different signedness: ‘unsigned int’ and ‘int’ [-Werror=sign-compare]
2808 | if (num > vq->inuse) {
| ^
Instead of casting the comparision to unsigned int, just make the inuse
field unsigned int in the fist place.
hw/virtio: Guard and restrict scope of qmp_virtio_feature_map_t[]
Commit f3034ad71f ("qmp: decode feature & status bits in
virtio-status") did not guard all qmp_virtio_feature_map_t
arrays with the corresponding #ifdef'ry used in
qmp_decode_features(). Fix that and reduce the arrays scope
by declaring them static.
virtio.c uses target_words_bigendian() which is declared in
"hw/core/cpu.h". Add the missing header to avoid when refactoring:
hw/virtio/virtio.c:2451:9: error: implicit declaration of function 'target_words_bigendian' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
if (target_words_bigendian()) {
^
Igor Mammedov [Tue, 29 Nov 2022 10:13:40 +0000 (11:13 +0100)]
remove DEC 21154 PCI bridge
Code has not been used practically since its inception (2004) f2aa58c6f4a20 UniNorth PCI bridge support
or maybe even earlier, but it was consuming contributors time
as QEMU was being rewritten.
Drop it for now. Whomever would like to actually
use the thing, can make sure it actually works/reintroduce
it back when there is a user.
PS:
I've stumbled upon this when replacing PCIDeviceClass::is_bridge
field with QOM cast to PCI_BRIDGE type. Unused DEC 21154
was the only one trying to use the field with plain PCIDevice.
It's not worth keeping the field around for the sake of the code
that was commented out 'forever'.
Jason Wang [Fri, 16 Dec 2022 03:35:52 +0000 (11:35 +0800)]
vhost: fix vq dirty bitmap syncing when vIOMMU is enabled
When vIOMMU is enabled, the vq->used_phys is actually the IOVA not
GPA. So we need to translate it to GPA before the syncing otherwise we
may hit the following crash since IOVA could be out of the scope of
the GPA log size. This could be noted when using virtio-IOMMU with
vhost using 1G memory.
Ani Sinha [Fri, 25 Nov 2022 04:41:38 +0000 (10:11 +0530)]
acpi/tests/avocado/bits: add SPDX license identifiers for bios bits tests
Added the SPDX license identifiers for biosbits tests.
Also added a comment on each of the test scripts to indicate that they run
from within the biosbits environment and hence are not subjected to the regular
maintenance activities for QEMU and is excluded from the dependency management
challenges in the host testing environment.
Alex Bennée [Wed, 23 Nov 2022 15:21:33 +0000 (15:21 +0000)]
include/hw: attempt to document VirtIO feature variables
We have a bunch of variables associated with the device and the vhost
backend which are used inconsistently throughout the code base. Lets
start trying to bring some order by agreeing what each variable is
for.
Eugenio Pérez [Thu, 15 Dec 2022 11:31:43 +0000 (12:31 +0100)]
vdpa: add shadow_data to vhost_vdpa
The memory listener that thells the device how to convert GPA to qemu's
va is registered against CVQ vhost_vdpa. memory listener translations
are always ASID 0, CVQ ones are ASID 1 if supported.
Let's tell the listener if it needs to register them on iova tree or
not.
Eugenio Pérez [Thu, 15 Dec 2022 11:31:42 +0000 (12:31 +0100)]
vdpa: store x-svq parameter in VhostVDPAState
CVQ can be shadowed two ways:
- Device has x-svq=on parameter (current way)
- The device can isolate CVQ in its own vq group
QEMU needs to check for the second condition dynamically, because CVQ
index is not known before the driver ack the features. Since this is
dynamic, the CVQ isolation could vary with different conditions, making
it possible to go from "not isolated group" to "isolated".
Saving the cmdline parameter in an extra field so we never disable CVQ
SVQ in case the device was started with x-svq cmdline.
Eugenio Pérez [Thu, 15 Dec 2022 11:31:41 +0000 (12:31 +0100)]
vdpa: add asid parameter to vhost_vdpa_dma_map/unmap
So the caller can choose which ASID is destined.
No need to update the batch functions as they will always be called from
memory listener updates at the moment. Memory listener updates will
always update ASID 0, as it's the passthrough ASID.
Eugenio Pérez [Thu, 15 Dec 2022 11:31:38 +0000 (12:31 +0100)]
vdpa: request iova_range only once
Currently iova range is requested once per queue pair in the case of
net. Reduce the number of ioctls asking it once at initialization and
reusing that value for each vhost_vdpa.
Eugenio Pérez [Thu, 15 Dec 2022 11:31:36 +0000 (12:31 +0100)]
vhost: move iova_tree set to vhost_svq_start
Since we don't know if we will use SVQ at qemu initialization, let's
allocate iova_tree only if needed. To do so, accept it at SVQ start, not
at initialization.
This will avoid to create it if the device does not support SVQ.
Eugenio Pérez [Thu, 15 Dec 2022 11:31:34 +0000 (12:31 +0100)]
vhost: set SVQ device call handler at SVQ start
By the end of this series CVQ is shadowed as long as the features
support it.
Since we don't know at the beginning of qemu running if this is
supported, move the event notifier handler setting to the start of the
SVQ, instead of the start of qemu run. This will avoid to create them if
the device does not support SVQ.
An 'ICH9-LPC.enable_tco' property has been exposed for a
very long time, but attempts to set it have never been
honoured.
Originally, any user provided 'enable_tco' value was force
replaced by a value passed from the machine type setup
code that was determine by machine type compat properties.
ich9: Remove enable_tco arguments from init functions
The enable_tco arguments are always true, so they are not needed
anymore.
Leaving the ich9_pm_init to just force set 'enable_tco' to true.
This still overrides any user specified property. The initialization
of property defaults should be done when properties are first
registered, rather than during object construction.
hw/isa: enable TCO watchdog reboot pin strap by default
The TCO watchdog implementation default behaviour from POV of the
guest OS relies on the initial values for two I/O ports:
* TCO1_CNT == 0x0
Since bit 11 (TCO Timer Halt) is clear, the watchdog state
is considered to be initially running
* GCS == 0x20
Since bit 5 (No Reboot) is set, the watchdog will not trigger
when the timer expires
This is a safe default, because the No Reboot bit will prevent the
watchdog from triggering if the guest OS is unaware of its existance,
or is slow in configuring it. When a Linux guest initializes the TCO
watchdog, it will attempt to clear the "No Reboot" flag, and read the
value back. If the clear was honoured, the driver will treat this as
an indicator that the watchdog is functional and create the guest
watchdog device.
QEMU implements a second "no reboot" flag, however, via pin straps
which overrides the behaviour of the guest controlled "no reboot"
flag:
This second 'noreboot' pin was defaulted to high, which also inhibits
triggering of the requested watchdog actions, unless QEMU is launched
with the magic flag "-global ICH9-LPC.noreboot=false".
This is a bad default as we are exposing a watchdog to every guest OS
using the q35 machine type, but preventing it from actually doing what
it is designed to do. What is worse is that the guest OS and its apps
have no way to know that the watchdog is never going to fire, due to
this second 'noreboot' pin.
If a guest OS had no watchdog device at all, then apps whose operation
and/or data integrity relies on a watchdog can refuse to launch, and
alert the administrator of the problematic deployment. With Q35 machines
unconditionally exposing a watchdog though, apps will think their
deployment is correct but in fact have no protection at all.
This patch flips the default of the second 'no reboot' flag, so that
configured watchdog actions will be honoured out of the box for the
7.2 Q35 machine type onwards, if the guest enables use of the watchdog.
accel/tcg: Restrict page_collection structure to system TB maintainance
Only the system emulation part of TB maintainance uses the
page_collection structure. Restrict its declaration (and the
functions requiring it) to tb-maint.c.
Convert the 'len' argument of tb_invalidate_phys_page_fast__locked()
from signed to unsigned.
Commit d9bb58e510 ("tcg: move tcg related files into accel/tcg/
subdirectory") introduced accel/tcg/trace-events, so we don't
need to use the root trace-events anymore.
accel/tcg: Move remainder of page locking to tb-maint.c
The only thing that still touches PageDesc in translate-all.c
are some locking routines related to tb-maint.c which have not
yet been moved. Do so now.
Move some code up in tb-maint.c as well, to untangle the maze
of ifdefs, and allow a sensible final ordering.
Move some declarations from exec/translate-all.h to internal.h,
as they are only used within accel/tcg/.
accel/tcg: Move PageDesc tree into tb-maint.c for system
Now that PageDesc is not used for user-only, and for system
it is only used for tb maintenance, move the implementation
into tb-main.c appropriately ifdefed.
We have not yet eliminated all references to PageDesc for
user-only, so retain a typedef to the structure without definition.
accel/tcg: Move page_{get,set}_flags to user-exec.c
This page tracking implementation is specific to user-only,
since the system softmmu version is in cputlb.c. Move it
out of translate-all.c to user-exec.c.
Make bsd-user match linux-user in not marking host pages
as reserved. This isn't especially effective anyway, as
it doesn't take into account any heap memory that qemu
may allocate after startup.
accel/tcg: Use interval tree for TBs in user-only mode
Begin weaning user-only away from PageDesc.
Since, for user-only, all TB (and page) manipulation is done with
a single mutex, and there is no virtual/physical discontinuity to
split a TB across discontinuous pages, place all of the TBs into
a single IntervalTree. This makes it trivial to find all of the
TBs intersecting a range.
Retain the existing PageDesc + linked list implementation for
system mode. Move the portion of the implementation that overlaps
the new user-only code behind the common ifdef.
Peter Maydell [Tue, 20 Dec 2022 21:35:09 +0000 (21:35 +0000)]
Merge tag 'qga-pull-2022-12-20' of github.com:kostyanf14/qemu into staging
qga-pull-2022-12-20
# gpg: Signature made Tue 20 Dec 2022 13:57:59 GMT
# gpg: using RSA key C2C2C109EA43C63C1423EB84EF5D5E8161BA84E7
# gpg: Good signature from "Kostiantyn Kostiuk (Upstream PR sign) <[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: C2C2 C109 EA43 C63C 1423 EB84 EF5D 5E81 61BA 84E7
* tag 'qga-pull-2022-12-20' of github.com:kostyanf14/qemu:
qga-win: choose the right libpcre version to include in MSI package
qga: map GLib log levels to system levels
qga-win: add logging to Windows event log
qga: Add initial OpenBSD and NetBSD support
qga:/qga-win: skip getting pci info for USB disks
qga:/qga-win: adding a empty PCI address creation function
Andrey Drobyshev [Tue, 13 Dec 2022 15:13:43 +0000 (17:13 +0200)]
qga-win: choose the right libpcre version to include in MSI package
According to GLib changelog [1], since version 2.73.2 GLib is using
libpcre2 instead of libpcre. As a result, qemu-ga MSI installation
fails due to missing DLL when linked with the newer GLib.
This commit makes wixl to put the right libpcre version into the MSI
bundle: either libpcre-1.dll or libpcre2-8-0.dll, depending on the
present version of GLib.
Andrey Drobyshev [Tue, 29 Nov 2022 17:38:09 +0000 (19:38 +0200)]
qga: map GLib log levels to system levels
This patch translates GLib-specific log levels to system ones, so that
they may be used by both *nix syslog() (as a "priority" argument) and
Windows ReportEvent() (as a "wType" argument).
Currently the only codepath to write to "syslog" domain is slog()
function. However, this patch allows the interface to be extended.
Note that since slog() is using G_LOG_LEVEL_INFO level, its behaviour
doesn't change.
Andrey Drobyshev [Tue, 29 Nov 2022 17:38:08 +0000 (19:38 +0200)]
qga-win: add logging to Windows event log
This commit allows QGA to write to Windows event log using Win32 API's
ReportEvent() [1], much like syslog() under *nix guests.
In order to generate log message definitions we use a very basic message
text file [2], so that every QGA's message gets ID 1. The tools
"windmc" and "windres" respectively are used to generate ".rc" file and
COFF object file, and then the COFF file is linked into qemu-ga.exe.
projects / qemu.git / log