Because we handle host IOMMU device creation in each container backend,
we know which type name to use, so hiod_typename property is useless
now, just remove it.
realize() is now moved after attachment, do the same for hiod creation.
Introduce a new function vfio_device_hiod_create_and_realize() to do
them all in one go.
Previously device attaching depends on realize() getting host IOMMU
capabilities to check dirty tracking support.
Now we have a separate call to ioctl(IOMMU_GET_HW_INFO) to get host
IOMMU capabilities and check that for dirty tracking support, there
is no dependency any more, move realize() call after attachment
succeed.
Suggested-by: Cédric Le Goater <clg@redhat.com> Suggested-by: Donald Dutile <ddutile@redhat.com> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Joao Martins <joao.m.martins@oracle.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Link: https://lore.kernel.org/qemu-devel/20250423072824.3647952-3-zhenzhong.duan@intel.com Signed-off-by: Cédric Le Goater <clg@redhat.com>
MAINTAINERS: Add a maintainer for util/vfio-helpers.c
The NVMe Block device driver makes use of a reduced VFIO library
managing the host interface. These routines are VFIO related and do
not have a maintainer. Move util/vfio-helpers.c under VFIO jurisdiction.
Cédric Le Goater [Wed, 26 Mar 2025 07:51:20 +0000 (08:51 +0100)]
vfio: Rename vfio-common.h to vfio-device.h
"hw/vfio/vfio-common.h" has been emptied of most of its declarations
by the previous changes and the only declarations left are related to
VFIODevice. Rename it to "hw/vfio/vfio-device.h" and make the
necessary adjustments.
Cédric Le Goater [Wed, 26 Mar 2025 07:51:17 +0000 (08:51 +0100)]
vfio: Introduce new files for VFIO MemoryListener
File "common.c" has been emptied of most of its definitions by the
previous changes and the only definitions left are related to the VFIO
MemoryListener handlers. Rename it to "listener.c" and introduce its
associated "vfio-listener.h" header file for the declarations.
Also rename vfio_devices_all_device_dirty_tracking_started() while at
it and use the prefix 'vfio_container_devices_' for routines simply
looping over the container's device list.
Cédric Le Goater [Wed, 26 Mar 2025 07:51:13 +0000 (08:51 +0100)]
vfio: Make vfio_container_query_dirty_bitmap() static
vfio_container_query_dirty_bitmap() is only used in "container-base.c".
Also, rename to vfio_container_iommu_query_dirty_bitmap() to reflect it
is using the VFIO IOMMU backend device ->query_dirty_bitmap() handler.
Cédric Le Goater [Wed, 26 Mar 2025 07:51:12 +0000 (08:51 +0100)]
vfio: Make vfio_devices_query_dirty_bitmap() static
vfio_devices_query_dirty_bitmap() is only used in "container-base.c".
Also, rename to vfio_container_devices_query_dirty_bitmap() to reflect
with the prefix 'vfio_container_devices_' that it simply loops over
the container's device list.
are all related to dirty page tracking directly at the container level
or at the container device level. Naming is a bit confusing. We will
propose new names in the following changes.
Cédric Le Goater [Wed, 26 Mar 2025 07:51:10 +0000 (08:51 +0100)]
vfio: Move vfio_reset_handler() into device.c
Pass-through devices of a VM are not necessarily in the same group and
all groups/address_spaces need to be scanned when the machine is
reset. Commit f16f39c3fc97 ("Implement PCI hot reset") introduced a VM
reset handler for this purpose. Move it under device.c
Also reintroduce the comment which explained the context and was lost
along the way.
Cédric Le Goater [Wed, 26 Mar 2025 07:51:07 +0000 (08:51 +0100)]
vfio: Move vfio_kvm_device_fd() into helpers.c
The vfio_kvm_device_add/del_fd() routines opening the VFIO pseudo
device are defined in "helpers.c". Move 'vfio_kvm_device_fd'
definition there and its declaration into "vfio-helpers.h" to reduce
exposure of VFIO internals in "hw/vfio/vfio-common.h".
Cédric Le Goater [Wed, 26 Mar 2025 07:51:06 +0000 (08:51 +0100)]
vfio: Introduce new files for CPR definitions and declarations
Gather all CPR related declarations into "vfio-cpr.h" to reduce exposure
of VFIO internals in "hw/vfio/vfio-common.h". These were introduced in
commit d9fa4223b30a ("vfio: register container for cpr").
Cédric Le Goater [Wed, 26 Mar 2025 07:51:02 +0000 (08:51 +0100)]
vfio: Move vfio_get_info_dma_avail() into helpers.c
vfio_get_info_dma_avail() is a low level routine similar to the other
routines extracting capabilities from 'struct vfio_iommu_type1_info'.
It belongs to file "helpers.c".
Cédric Le Goater [Wed, 26 Mar 2025 07:50:59 +0000 (08:50 +0100)]
vfio: Move VFIOAddressSpace helpers into container-base.c
VFIOAddressSpace is a common object used by VFIOContainerBase which is
declared in "hw/vfio/vfio-container-base.h". Move the VFIOAddressSpace
related services into "container-base.c".
Cédric Le Goater [Wed, 26 Mar 2025 07:50:57 +0000 (08:50 +0100)]
vfio: Introduce a new header file for VFIOcontainer declarations
Gather all VFIOcontainer related declarations into
"hw/vfio/vfio-container.h" to reduce exposure of VFIO internals in
"hw/vfio/vfio-common.h". These declarations were initially introduced
in commit 65501a745dba ("vfio: vfio-pci device assignment driver").
They are made available externally for PPC and s390x.
Cédric Le Goater [Wed, 26 Mar 2025 07:50:56 +0000 (08:50 +0100)]
vfio: Introduce new files for VFIORegion definitions and declarations
Gather all VFIORegion related declarations and definitions into their
own files to reduce exposure of VFIO internals in "hw/vfio/vfio-common.h".
They were introduced for 'vfio-platform' support in commits db0da029a185 ("vfio: Generalize region support") and a664477db8da
("hw/vfio/pci: Introduce VFIORegion").
To be noted that the 'vfio-platform' devices have been deprecated and
will be removed in QEMU 10.2. Until then, make the declarations
available externally for 'sysbus-fdt.c'.
Cédric Le Goater [Wed, 26 Mar 2025 07:50:55 +0000 (08:50 +0100)]
vfio: Introduce a new header file for VFIOIOMMUFD declarations
Gather all VFIOIOMMUFD related declarations introduced by commits 5ee3dc7af785 ("vfio/iommufd: Implement the iommufd backend") and 5b1e96e65403 ("vfio/iommufd: Introduce auto domain creation") into
"vfio-iommufd.h". This to reduce exposure of VFIO internals in
"hw/vfio/vfio-common.h".
Cédric Le Goater [Wed, 26 Mar 2025 07:50:48 +0000 (08:50 +0100)]
vfio: Introduce a new header file for external migration services
The migration core subsystem makes use of the VFIO migration API to
collect statistics on the number of bytes transferred. These services
are declared in "hw/vfio/vfio-common.h" which also contains VFIO
internal declarations. Move the migration declarations into a new
header file "hw/vfio/vfio-migration.h" to reduce the exposure of VFIO
internals.
While at it, use a 'vfio_migration_' prefix for these services.
To be noted, vfio_migration_add_bytes_transferred() is a VFIO
migration internal service which we will be moved in the subsequent
patches.
Amit Machhiwal [Tue, 8 Apr 2025 12:40:42 +0000 (18:10 +0530)]
vfio/spapr: Fix L2 crash with PCI device passthrough and memory > 128G
An L2 KVM guest fails to boot inside a pSeries LPAR when booted with a
memory more than 128 GB and PCI device passthrough. The L2 guest also
crashes when it is booted with a memory greater than 128 GB and a PCI
device is hotplugged later.
The issue arises from a conditional check for `levels > 1` in
`spapr_tce_create_table()` within L1 KVM. This check is meant to prevent
multi-level TCEs, which are not supported by the PowerVM hypervisor. As
a result, when QEMU makes a `VFIO_IOMMU_SPAPR_TCE_CREATE` ioctl call
with `levels > 1`, it triggers the conditional check and returns
`EINVAL`, causing the guest to crash with the following errors:
2025-03-04T06:36:36.133117Z qemu-system-ppc64: Failed to create a window, ret = -1 (Invalid argument)
2025-03-04T06:36:36.133176Z qemu-system-ppc64: Failed to create SPAPR window: Invalid argument
qemu: hardware error: vfio: DMA mapping failed, unable to continue
Fix this by checking the supported DDW "levels" returned by the
VFIO_IOMMU_SPAPR_TCE_GET_INFO ioctl before attempting the TCE create
ioctl in KVM.
The patch has been tested on KVM guests with memory configurations of up
to 390GB, and 450GB on PowerVM and bare-metal environments respectively.
Amit Machhiwal [Tue, 8 Apr 2025 12:40:41 +0000 (18:10 +0530)]
vfio/spapr: Enhance error handling in vfio_spapr_create_window()
Introduce an Error ** parameter to vfio_spapr_create_window() to enable
structured error reporting. This allows the function to propagate
detailed errors back to callers.
Hopefully, one day, we will be able to extend these callbacks with an
'Error **' parameter and avoid setting the global migration error.
Until then, it seems sensible to clearly identify the use cases, which
are limited, and open code vfio_migration_set_error(). One other
benefit is an improved error reporting when migration is running.
While at it, slightly modify error reporting to only report errors
when migration is not active and not always as is currently done.
Tomita Moeko [Thu, 13 Mar 2025 15:03:39 +0000 (23:03 +0800)]
vfio/igd: Update IGD passthrough documentation
A previous change made the OpRegion and LPC quirks independent of the
existing legacy mode, update the documentation accordingly. More related
topics, like creating EFI Option ROM of IGD for OVMF, how to solve the
VFIO_DMA_MAP Invalid Argument warning, as well as details on IGD memory
internals, are also added.
* tag 'pull-misc-2025-04-24' of https://repo.or.cz/qemu/armbru:
cleanup: Drop pointless label at end of function
cleanup: Drop pointless return at end of function
cleanup: Re-run return_directly.cocci
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
# -----BEGIN PGP SIGNATURE-----
#
# iLMEAAEKAB0WIQS4/x2g0v3LLaCcbCxAov/yOSY+3wUCaAmmRQAKCRBAov/yOSY+
# 3yZoA/4udi9ZmLsaiPqfKCS+0eF8XScIT493lVD359lFTBTT7mshh9PPhTLzdtiC
# 8fcfYi7jSjfC9gGTjPgnNCOzKIg3Gbdl61AFDgIwd8q/5HQAgonHAywTUtmqDaPK
# bXZ/JkkJQby2dla6015XKQS/d/EXWHgYjrcb1JZIRoaLworZPw==
# =zBCJ
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 23 Apr 2025 22:47:33 EDT
# gpg: using RSA key B8FF1DA0D2FDCB2DA09C6C2C40A2FFF239263EDF
# gpg: Good signature from "Song Gao <m17746591750@163.com>" [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: B8FF 1DA0 D2FD CB2D A09C 6C2C 40A2 FFF2 3926 3EDF
* tag 'pull-loongarch-20250424' of https://github.com/gaosong715/qemu:
target/loongarch: Guard BCEQZ/BCNEZ instructions with FP feature
target/loongarch: Add CRC feature flag and use it to gate CRC instructions
linux-user/loongarch64: Decode BRK break codes for FPE signals
target/loongarch: Move definition of TCG specified function to tcg directory
target/loongarch: Add static definition with function loongarch_tlb_search()
target/loongarch: Move function loongarch_tlb_search to directory tcg
target/loongarch: Define function loongarch_get_addr_from_tlb() non-static
target/loongarch: Set function loongarch_map_address() with common code
target/loongarch: Add stub function loongarch_get_addr_from_tlb
target/loongarch: Move function get_dir_base_width to common directory
target/loongarch: Add function loongarch_get_addr_from_tlb
target/loongarch: Move header file helper.h to directory tcg
hw/intc/loongarch_pch_msi: Remove gpio input handler
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Stefan Hajnoczi [Thu, 24 Apr 2025 17:44:28 +0000 (13:44 -0400)]
Merge tag 'pull-tcg-20250423' of https://gitlab.com/rth7680/qemu into staging
meson: Introduce top-level libuser_ss and libsystem_ss
meson: Add hw_common_arch dictionary
accel/tcg: Lots of cleanups to enable build once for:
user-exec-stub.c,
plugin-gen.c,
translator.c
page-vary: Restrict scope of TARGET_PAGE_BITS_MIN
tcg: Always define TARGET_INSN_START_EXTRA_WORDS
tcg: Convert TARGET_GUEST_DEFAULT_MO to TCGCPUOps::guest_default_memory_order
tcg: Convert TARGET_SUPPORTS_MTTCG to TCGCPUOps::mttcg_supported field
target/riscv: Do not expose rv128 CPU on user mode emulation
* tag 'pull-tcg-20250423' of https://gitlab.com/rth7680/qemu: (148 commits)
tcg: Convert TARGET_SUPPORTS_MTTCG to TCGCPUOps::mttcg_supported field
target/i386: Remove AccelCPUClass::cpu_class_init need
target/riscv: Remove AccelCPUClass::cpu_class_init need
accel/tcg: Move mttcg warning to tcg_init_machine
tcg: Convert TCGState::mttcg_enabled to TriState
accel/tcg: Remove mttcg_enabled
tcg: Move qemu_tcg_mttcg_enabled() to 'system/tcg.h'
tcg: Pass max_threads not max_cpus to tcg_init
tcg: Move cpu_req_mo() macro to target-agnostic 'backend-ldst.h'
tcg: Remove the TCG_GUEST_DEFAULT_MO definition globally
tcg: Have tcg_req_mo() use TCGCPUOps::guest_default_memory_order
tcg: Propagate CPUState argument to cpu_req_mo()
tcg: Remove use of TCG_GUEST_DEFAULT_MO in tb_gen_code()
tcg: Define guest_default_memory_order in TCGCPUOps
tcg: Simplify tcg_req_mo() macro
tcg: Always define TCG_GUEST_DEFAULT_MO
exec: Do not include 'accel/tcg/cpu-ldst.h' in 'exec-all.h'
exec: Restrict 'cpu_ldst.h' to accel/tcg/
exec: Restrict 'cpu-ldst-common.h' to accel/tcg/
tcg: Always define TARGET_INSN_START_EXTRA_WORDS
...
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
A few functions now end with a label. The next commit will clean them
up.
Signed-off-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250407082643.2310002-3-armbru@redhat.com>
[Straightforward conflict with commit 988ad4ccebb6 (hw/loongarch/virt:
Fix cpuslot::cpu set at last in virt_cpu_plug()) resolved]
Coccinelle's indentation of virt_create_plic() results in a long line.
Avoid that by mimicking the old indentation manually.
Don't touch tests/tcg/mips/user/. I'm not sure these files are ours
to make style cleanups on. They might be imported third-party code,
which we should leave as is to not complicate future updates.
Signed-off-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250407082643.2310002-2-armbru@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
WANG Rui [Fri, 18 Apr 2025 08:21:02 +0000 (16:21 +0800)]
target/loongarch: Guard BCEQZ/BCNEZ instructions with FP feature
The BCEQZ and BCNEZ instructions depend on access to condition codes
from floating-point comparisons. Previously, these instructions were
unconditionally enabled for 64-bit targets.
This patch updates their translation to be gated under the `FP` feature
flag instead, ensuring they are only available when the floating-point
unit is present.
This improves correctness for CPUs lacking floating-point support.
Signed-off-by: WANG Rui <wangrui@loongson.cn> Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Message-Id: <20250418082103.447780-3-wangrui@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
WANG Rui [Fri, 18 Apr 2025 08:21:01 +0000 (16:21 +0800)]
target/loongarch: Add CRC feature flag and use it to gate CRC instructions
This patch replaces the obsolete IOCSR_BRD bit with CRC in cpucfg1[25],
in both LA464 and LA132 CPU initialization functions. The corresponding
field macro in `cpu.h` is updated to reflect this change.
Additionally, the availability macro `avail_CRC()` is introduced in
`translate.h` to check the CRC feature flag.
All CRC-related instruction translations are updated to be gated by
the new CRC feature flag instead of hardcoded CPU features.
This ensures correctness and configurability when enabling CRC
instructions based on hardware capabilities.
Signed-off-by: WANG Rui <wangrui@loongson.cn> Reviewed-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250418082103.447780-2-wangrui@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
WANG Rui [Mon, 14 Apr 2025 07:49:52 +0000 (15:49 +0800)]
linux-user/loongarch64: Decode BRK break codes for FPE signals
Handle specific LoongArch BRK break codes in user-mode emulation
to deliver accurate floating-point exception signals. Specifically,
BRK_OVERFLOW (6) triggers TARGET_FPE_INTOVF, and BRK_DIVZERO (7)
triggers TARGET_FPE_INTDIV. Other BRK codes fall back to a generic
SIGTRAP.
This improves correctness for programs that rely on BRK to signal
overflow or divide-by-zero conditions.
Signed-off-by: WANG Rui <wangrui@loongson.cn> Acked-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20250414074952.6253-1-wangrui@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:17 +0000 (16:04 +0800)]
target/loongarch: Move definition of TCG specified function to tcg directory
Function loongarch_cpu_tlb_fill() only works in TCG mode, move its
definition from header file internals.h to file tcg/tcg_loongarch.h
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-10-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:16 +0000 (16:04 +0800)]
target/loongarch: Add static definition with function loongarch_tlb_search()
Function loongarch_tlb_search() is only referenced in file tcg/tlb_helper.c,
define this function with static attribution.
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-9-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:15 +0000 (16:04 +0800)]
target/loongarch: Move function loongarch_tlb_search to directory tcg
Function loongarch_tlb_search() and loongarch_map_tlb_entry() works
only in TCG mode, move these functions to directory tcg.
There is no any function change, only code moving.
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-8-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:14 +0000 (16:04 +0800)]
target/loongarch: Define function loongarch_get_addr_from_tlb() non-static
Define function loongarch_get_addr_from_tlb() non-static, and add its
definition in header file tcg/tcg_loongarch.h
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-7-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:13 +0000 (16:04 +0800)]
target/loongarch: Set function loongarch_map_address() with common code
Function loongarch_map_address is to get physical address from virtual
address, it is used by qmp commands to dump memory from virtual
address.
It is used by kvm mode also, here move function loongarch_map_address()
out of macro CONFIG_TCG. And it is common code, the similar with
function loongarch_page_table_walker().
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-6-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:12 +0000 (16:04 +0800)]
target/loongarch: Add stub function loongarch_get_addr_from_tlb
Stub function loongarch_get_addr_from_tlb() is added if option
CONFIG_TCG is not enabled, so this function can be called in KVM
only mode.
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-5-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:11 +0000 (16:04 +0800)]
target/loongarch: Move function get_dir_base_width to common directory
Function get_dir_base_width() is used by loongarch_page_table_walker(),
so it is used by KVM mode also, here move this function from directory
tcg to common directory.
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-4-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:10 +0000 (16:04 +0800)]
target/loongarch: Add function loongarch_get_addr_from_tlb
Function loongarch_get_addr_from_tlb() is added to get physical address
from TLB tables. TLB table only works in TCG mode, in future this
function will be moved to TCG directory.
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-3-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
Bibo Mao [Wed, 23 Apr 2025 08:04:09 +0000 (16:04 +0800)]
target/loongarch: Move header file helper.h to directory tcg
Header file helper.h is specified for tcg mode, move this file to
directory tcg. And create new file helper.h to include header
file in tcg mode.
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250423080417.3739809-2-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
MSI interrupt is triggered by writing message on specified memory address.
In generic it is used by PCI devices, and no device is connected pch MSI
irqchip with GPIO pin line method, here remove gpio input setting for MSI
controller.
Signed-off-by: Bibo Mao <maobibo@loongson.cn> Reviewed-by: Song Gao <gaosong@loongson.cn> Tested-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20250410085004.3577627-1-maobibo@loongson.cn> Signed-off-by: Song Gao <gaosong@loongson.cn>
tcg: Convert TARGET_SUPPORTS_MTTCG to TCGCPUOps::mttcg_supported field
Instead of having a compile-time TARGET_SUPPORTS_MTTCG definition,
have each target set the 'mttcg_supported' field in the TCGCPUOps
structure.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Anton Johansson <anjo@rev.ng> Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250405161320.76854-17-philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
target/i386: Remove AccelCPUClass::cpu_class_init need
Expose x86_tcg_ops symbol, then directly set it as
CPUClass::tcg_ops in TYPE_X86_CPU's class_init(),
using CONFIG_TCG #ifdef'ry. No need for the
AccelCPUClass::cpu_class_init() handler anymore.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250405161320.76854-3-philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
target/riscv: Remove AccelCPUClass::cpu_class_init need
Expose riscv_tcg_ops symbol, then directly set it as
CPUClass::tcg_ops in TYPE_RISCV_CPU's class_init(),
using CONFIG_TCG #ifdef'ry. No need for the
AccelCPUClass::cpu_class_init() handler anymore.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250405161320.76854-2-philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Since the TCGState instance is zero-initialized, the
mttcg_enabled is initialzed as AUTO (ON_OFF_AUTO_AUTO).
In tcg_init_machine(), if mttcg_enabled is still AUTO,
set a default value (effectively inlining the
default_mttcg_enabled() method content).
In the tcg_get_thread() getter, consider AUTO / OFF states
as "single", otherwise ON is "multi".
Reviewed-by: Anton Johansson <anjo@rev.ng> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg: Move qemu_tcg_mttcg_enabled() to 'system/tcg.h'
qemu_tcg_mttcg_enabled() is specific to 1/ TCG and
2/ system emulation. Move the prototype declaration
to "system/tcg.h", reducing 'mttcg_enabled' variable
scope.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250403220420.78937-17-philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg: Move cpu_req_mo() macro to target-agnostic 'backend-ldst.h'
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg: Remove the TCG_GUEST_DEFAULT_MO definition globally
By directly using TCGCPUOps::guest_default_memory_order,
we don't need the TCG_GUEST_DEFAULT_MO definition anymore.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Anton Johansson <anjo@rev.ng> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg: Have tcg_req_mo() use TCGCPUOps::guest_default_memory_order
In order to use TCG with multiple targets, replace the
compile time use of TCG_GUEST_DEFAULT_MO by a runtime access
to TCGCPUOps::guest_default_memory_order via CPUState.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
In preparation of having tcg_req_mo() access CPUState in
the next commit, pass it to cpu_req_mo(), its single caller.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg: Remove use of TCG_GUEST_DEFAULT_MO in tb_gen_code()
Use TCGCPUOps::guest_default_memory_order to set TCGContext::guest_mo.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg: Define guest_default_memory_order in TCGCPUOps
Add the TCGCPUOps::guest_default_memory_order field and have
each target initialize it.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Anton Johansson <anjo@rev.ng> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Now that TCG_GUEST_DEFAULT_MO is always defined,
simplify the tcg_req_mo() macro.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Anton Johansson <anjo@rev.ng> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
We only require the TCG_GUEST_DEFAULT_MO for MTTCG-enabled
frontends, otherwise we use a default value of TCG_MO_ALL.
In order to simplify, require the definition for all targets,
defining it for hexagon, m68k, rx, sh4 and tricore.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Anton Johansson <anjo@rev.ng> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
exec: Do not include 'accel/tcg/cpu-ldst.h' in 'exec-all.h'
Only 2 files requiring "accel/tcg/cpu-ldst.h" API do not
include it:
- accel/tcg/cpu-exec.c
- target/arm/tcg/sve_helper.c
Include it there and remove it from "exec/exec-all.h".
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Do not define TARGET_INSN_START_EXTRA_WORDS under the
hood, have each target explicitly define it.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg: Declare TARGET_INSN_START_EXTRA_WORDS in 'cpu-param.h'
To avoid including the huge "cpu.h" for a simple definition,
move TARGET_INSN_START_EXTRA_WORDS to "cpu-param.h".
Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tb_check_watchpoint() calls cpu_get_tb_cpu_state(),
which is declared in each "cpu.h" header. It is indirectly
included via "tcg/insn-start-words.h". Since we want to
rework "tcg/insn-start-words.h", removing "cpu.h" in the
next commit, add the missing header now, otherwise we'd
get:
accel/tcg/translate-all.c:598:9: error: call to undeclared function 'cpu_get_tb_cpu_state' [-Wimplicit-function-declaration]
598 | cpu_get_tb_cpu_state(env, &pc, &cs_base, &flags);
| ^
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
target/riscv: Do not expose rv128 CPU on user mode emulation
As Richard mentioned:
We should allow RV128 in user-mode at all until there's a
kernel abi for it.
Remove the experimental 'x-rv128' CPU on user emulation
(since it is experimental, no deprecation period is required).
Reported-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Pierrick Bouvier [Tue, 25 Mar 2025 04:59:08 +0000 (21:59 -0700)]
meson: add common hw files
Those files will be compiled once per base architecture ("arm" in this
case), instead of being compiled for every variant/bitness of
architecture.
We make sure to not include target cpu definitions (exec/cpu-defs.h) by
defining header guard directly. This way, a given compilation unit can
access a specific cpu definition, but not access to compile time defines
associated.
Previous commits took care to clean up some headers to not rely on
cpu-defs.h content.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250325045915.994760-24-pierrick.bouvier@linaro.org>
At worst, for 32-bit arm binary, using these methods will
now produce a link time error, instead of a compile time one.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20250403235821.9909-37-philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Pierrick Bouvier [Tue, 25 Mar 2025 04:59:06 +0000 (21:59 -0700)]
target/arm/cpu: define same set of registers for aarch32 and aarch64
To eliminate TARGET_AARCH64, we need to make various definitions common
between 32 and 64 bit Arm targets.
Added registers are used only by aarch64 code, and the only impact is on
the size of CPUARMState, and added zarray
(ARMVectorReg zarray[ARM_MAX_VQ * 16]) member (+64KB)
It could be eventually possible to allocate this array only for aarch64
emulation, but I'm not sure it's worth the hassle to save a few KB per
vcpu. Running qemu-system takes already several hundreds of MB of
(resident) memory, and qemu-user takes dozens of MB of (resident) memory
anyway.
As part of this, we define ARM_MAX_VQ once for aarch32 and aarch64,
which will affect zregs field for aarch32.
This field is used for MVE and SVE implementations. MVE implementation
is clipping index value to 0 or 1 for zregs[*].d[],
so we should not touch the rest of data in this case anyway.
This change is safe regarding migration, because aarch64 registers still
have the same size, and for aarch32, only zregs is modified.
Migration code explicitly specify a size of 2 for env.vfp.zregs[0].d,
VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[0].d, ARMCPU, 0, 2). So extending
the storage size has no impact.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250325045915.994760-22-pierrick.bouvier@linaro.org>
Pierrick Bouvier [Tue, 25 Mar 2025 04:59:05 +0000 (21:59 -0700)]
target/arm/cpu: flags2 is always uint64_t
Do not rely on target dependent type, but use a fixed type instead.
Since the original type is unsigned, it is safe to extend its size
without any side effect.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20250325045915.994760-21-pierrick.bouvier@linaro.org>
projects / thirdparty / qemu.git / log
