target/arm: Convert hint instruction space to decodetree

Convert the various instructions in the hint instruction space
to decodetree.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230602155223.2040685-3-peter.maydell@linaro.org

target/arm: Consistently use finalize_memop_asimd() for ASIMD loads/stores

In the recent refactoring we missed a few places which should be
calling finalize_memop_asimd() for ASIMD loads and stores but
instead are just calling finalize_memop(); fix these.

For the disas_ldst_single_struct() and disas_ldst_multiple_struct()
cases, this is not a behaviour change because there the size
is never MO_128 and the two finalize functions do the same thing.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Pass memop to gen_mte_check1_mmuidx() in reg_imm9 decode

In disas_ldst_reg_imm9() we missed one place where a call to
a gen_mte_check* function should now be passed the memop we
have created rather than just being passed the size. Fix this.

Fixes: 0a9091424d ("target/arm: Pass memop to gen_mte_check1*")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/arm: Return correct result for LDG when ATA=0

The LDG instruction loads the tag from a memory address (identified
by [Xn + offset]), and then merges that tag into the destination
register Xt. We implemented this correctly for the case when
allocation tags are enabled, but didn't get it right when ATA=0:
instead of merging the tag bits into Xt, we merged them into the
memory address [Xn + offset] and then set Xt to that.

Merge the tag bits into the old Xt value, as they should be.

Cc: qemu-stable@nongnu.org
Fixes: c15294c1e36a7dd9b25 ("target/arm: Implement LDG, STG, ST2G instructions")
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Fix return value from LDSMIN/LDSMAX 8/16 bit atomics

The atomic memory operations are supposed to return the old memory
data value in the destination register.  This value is not
sign-extended, even if the operation is the signed minimum or
maximum.  (In the pseudocode for the instructions the returned data
value is passed to ZeroExtend() to create the value in the register.)

We got this wrong because we were doing a 32-to-64 zero extend on the
result for 8 and 16 bit data values, rather than the correct amount
of zero extension.

Fix the bug by using ext8u and ext16u for the MO_8 and MO_16 data
sizes rather than ext32u.

Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230602155223.2040685-2-peter.maydell@linaro.org

Merge tag 'pull-loongarch-20230616' of https://gitlab.com/gaosong/qemu into staging

pull-loongarch-20230616

# -----BEGIN PGP SIGNATURE-----
#
# iLMEAAEIAB0WIQS4/x2g0v3LLaCcbCxAov/yOSY+3wUCZIwysgAKCRBAov/yOSY+
# 39FYA/465KtY2jDt4xG6AdwZDHckfxZQWlrfhyZvtapOkUG4AprOBV2nSS/ukyD4
# V8bg2/6cLS0GRKfDsqA3DcxSASWCAggIU4fTSj+DlYOZhNUIq14qzwqciZnO5CIH
# QDczSqu2LKRdP9j6MCtzIaZq/8pPDcOlgm7Dyct/kDo/64E2sg==
# =rD4j
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 16 Jun 2023 12:00:18 PM CEST
# 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-20230616' of https://gitlab.com/gaosong/qemu:
  target/loongarch: Fix CSR.DMW0-3.VSEG check
  hw/loongarch: Supplement cpu topology arguments
  hw/loongarch: Add numa support
  hw/intc: Set physical cpuid route for LoongArch ipi device
  hw/loongarch/virt: Add cpu arch_id support

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Merge tag 'xenpvh5-tag' of https://gitlab.com/sstabellini/qemu into staging

xenpvh5

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCgAdFiEE0E4zq6UfZ7oH0wrqiU+PSHDhrpAFAmSLo0QACgkQiU+PSHDh
# rpB1Gw/9H5Cx7wQZVyKfFwnyOoP2QTedCISxC0HL5qFmUGcJY21gXaJZ10JaU/HM
# zHEJj2M17EgVCTkZVqZeKuj+nzyjbRKatT3YmFqKqFNNt5M1yQxC9BfVgso4PND/
# SY0/8BvqumgJEqD3sf76KbQAILKwahPtA42LTM7S7r2ZsmQpvmOpdOhCVugpnqs/
# FheP8N6hdlZ7GnRGtXv9QnKxMVThuE3mRCUWCyYsV/Roz6uvPsvskrdSeC3LzzBd
# Ewq56vB+qQg+WbNTgK2BcVOzV/89k9tjWsUnamfhjD2lUxfHrne1FBclhKMcHhUv
# T53zjhxjlRfmzUxC4917Krt4Tw/AaDW7v1pn6RokUq5U059Wb8q0IjzL75FOeD3o
# e9DNp+RR8py44ejfi2WHR7jqayMPVIO86uJ3usshiZ9YgK5efFAtlwN/KNR5JX8k
# Y1BR9O8BebtRymljtiLWUFXlu3xywGSA23KotT7XtzXKEaTZkIHdI395YKksYPkG
# pil0C0bh5ZW3ZWd4M/CNcVOb69R53p15O77mjmKtjnkQYJAPD6Kbc9thZ1zdWwPR
# ivFPdiTJb0FElS0ywZwezKYRKXje6E9ejXgAzgFuZI/rFdeO0HfkifiNoro1NAxK
# g4V+LE5oPt09GpL2nuHrh/y9g9MnLlXyNBhPV0CRelU6fPKIk1w=
# =543t
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 16 Jun 2023 01:48:20 AM CEST
# gpg:                using RSA key D04E33ABA51F67BA07D30AEA894F8F4870E1AE90
# gpg: Good signature from "Stefano Stabellini <sstabellini@kernel.org>" [unknown]
# gpg:                 aka "Stefano Stabellini <stefano.stabellini@eu.citrix.com>" [undefined]
# 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: D04E 33AB A51F 67BA 07D3  0AEA 894F 8F48 70E1 AE90

* tag 'xenpvh5-tag' of https://gitlab.com/sstabellini/qemu:
  test/qtest: add xepvh to skip list for qtest
  meson.build: enable xenpv machine build for ARM
  hw/arm: introduce xenpvh machine
  meson.build: do not set have_xen_pci_passthrough for aarch64 targets
  hw/xen/xen-hvm-common: Use g_new and error_report
  hw/xen/xen-hvm-common: skip ioreq creation on ioreq registration failure
  include/hw/xen/xen_common: return error from xen_create_ioreq_server
  xen-hvm: reorganize xen-hvm and move common function to xen-hvm-common
  hw/i386/xen/xen-hvm: move x86-specific fields out of XenIOState
  hw/i386/xen: rearrange xen_hvm_init_pc
  hw/i386/xen/: move xen-mapcache.c to hw/xen/

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/loongarch: Fix CSR.DMW0-3.VSEG check

The previous code checks whether the highest 16 bits of virtual address
equal to that of CSR.DMW0-3. This is incorrect according to the spec,
and is corrected to compare only the highest four bits instead.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230614065556.2397513-1-c@jia.je>
Signed-off-by: Song Gao <gaosong@loongson.cn>

hw/loongarch: Supplement cpu topology arguments

Supplement LoongArch cpu topology arguments, including support socket
and threads per core.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Tianrui Zhao <zhaotianrui@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230613123251.2471878-1-zhaotianrui@loongson.cn>

hw/loongarch: Add numa support

1. Implement some functions for LoongArch numa support;
2. Implement fdt_add_memory_node() for fdt;
3. build_srat() fills node_id and adds build numa memory.

Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Tianrui Zhao <zhaotianrui@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230613122613.2471743-1-zhaotianrui@loongson.cn>

hw/intc: Set physical cpuid route for LoongArch ipi device

LoongArch ipi device uses physical cpuid to route to different
vcpus rather logical cpuid, and the physical cpuid is the same
with cpuid in acpi dsdt and srat table.

Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Tianrui Zhao <zhaotianrui@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230613120552.2471420-3-zhaotianrui@loongson.cn>

hw/loongarch/virt: Add cpu arch_id support

With acpi madt table, there is cpu physical coreid, which may
be different with logical id in qemu. This patch adds cpu arch_id
support, and fill madt table with arch_id. For the present cpu
arch_id is still equal to logical id.

Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Tianrui Zhao <zhaotianrui@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230613120552.2471420-2-zhaotianrui@loongson.cn>

Merge tag 'pull-aspeed-20230615' of https://github.com/legoater/qemu into staging

aspeed queue:

* extension of the rainier machine with VPD contents
* fixes for Coverity issues
* new "bmc-console" machine option
* new "vfp-d32" ARM CPU property

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEoPZlSPBIlev+awtgUaNDx8/77KEFAmSLQjcACgkQUaNDx8/7
# 7KE/XRAAkMZN7o+5vR7NocAbj9FasFq8G5Du8L5V52k7kjhmTIMtY2StKyXYBI4o
# MXe89P88gT5kmwAzyzFVkbLwZcS9wnA/71Pv+dc9Fe9fK9Q+1Tn9AGR+nJM7ZsOk
# qOP2SUfKqoeHBlFaWJAFSs09jPOzQr0elB55YErwXkjkJN9+PcI5l8E3aZGZz9qQ
# SZY1JpBYXidoF3bqEfQUs1SzszfFOrW9QYO5s93EYfyXTsV93JPfKVviy8DtNQjc
# EE+mSgvGSELu8ODMzk/b+O1OQ39S+AJ/qoqhYFZWrsxhROfLiFxV8ksAX52sdJeY
# z7hot8KUaAka2to+7OFQJn6N9i12MsSZ17XQhoOKC0IMwDtFIMFTIm7fkwxuJh9m
# ktiadL6MEWZCiiq8jLkM1sz/V9BcVEV/zI2WAJnOh8tniGG9U2xWGL7Ijf2Lnxs7
# P8cjT0XfBB6LzyEk1OSCScoBBdc3bHJD4xUpnr1ehscg58gyaPVMkkTziqWbeVH1
# CvxmGufS0jYXYS5R4/U4PEY5FQdBww77x/JTXx/NYc0ZWetgKCA/jqFtSgINgtdd
# jKFHPnvOv8NWDagy/+Vb0xZPHbXoYkliMtAV789FujI/6VzPrdW8YljPos/rX/oY
# b6/Yh1vCwuzVRut5wqMNefmX1ez36rdy3KDvg99Pu3Ln4QqBXhE=
# =qTHL
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 15 Jun 2023 06:54:15 PM CEST
# gpg:                using RSA key A0F66548F04895EBFE6B0B6051A343C7CFFBECA1
# gpg: Good signature from "Cédric Le Goater <clg@kaod.org>" [undefined]
# 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: A0F6 6548 F048 95EB FE6B  0B60 51A3 43C7 CFFB ECA1

* tag 'pull-aspeed-20230615' of https://github.com/legoater/qemu:
  target/arm: Allow users to set the number of VFP registers
  aspeed: Introduce a "bmc-console" machine option
  aspeed: Use the boot_rom region of the fby35 machine
  aspeed: Introduce a boot_rom region at the machine level
  aspeed/hace: Initialize g_autofree pointer
  hw/arm/aspeed: Add VPD data for Rainier machine

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

test/qtest: add xepvh to skip list for qtest

Like existing xen machines, xenpvh also cannot be used for qtest.

Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

meson.build: enable xenpv machine build for ARM

Add CONFIG_XEN for aarch64 device to support build for ARM targets.

Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

hw/arm: introduce xenpvh machine

Add a new machine xenpvh which creates a IOREQ server to register/connect with
Xen Hypervisor.

Optional: When CONFIG_TPM is enabled, it also creates a tpm-tis-device, adds a
TPM emulator and connects to swtpm running on host machine via chardev socket
and support TPM functionalities for a guest domain.

Extra command line for aarch64 xenpvh QEMU to connect to swtpm:
    -chardev socket,id=chrtpm,path=/tmp/myvtpm2/swtpm-sock \
    -tpmdev emulator,id=tpm0,chardev=chrtpm \
    -machine tpm-base-addr=0x0c000000 \

swtpm implements a TPM software emulator(TPM 1.2 & TPM 2) built on libtpms and
provides access to TPM functionality over socket, chardev and CUSE interface.
Github repo: https://github.com/stefanberger/swtpm
Example for starting swtpm on host machine:
    mkdir /tmp/vtpm2
    swtpm socket --tpmstate dir=/tmp/vtpm2 \
    --ctrl type=unixio,path=/tmp/vtpm2/swtpm-sock &

Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

meson.build: do not set have_xen_pci_passthrough for aarch64 targets

have_xen_pci_passthrough is only used for Xen x86 VMs.

Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

hw/xen/xen-hvm-common: Use g_new and error_report

Replace g_malloc with g_new and perror with error_report.

Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Reviewed-by: Paul Durrant <paul@xen.org>

hw/xen/xen-hvm-common: skip ioreq creation on ioreq registration failure

On ARM it is possible to have a functioning xenpv machine with only the
PV backends and no IOREQ server. If the IOREQ server creation fails continue
to the PV backends initialization.

Also, moved the IOREQ registration and mapping subroutine to new function
xen_do_ioreq_register().

Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Reviewed-by: Paul Durrant <paul@xen.org>

include/hw/xen/xen_common: return error from xen_create_ioreq_server

This is done to prepare for enabling xenpv support for ARM architecture.
On ARM it is possible to have a functioning xenpv machine with only the
PV backends and no IOREQ server. If the IOREQ server creation fails,
continue to the PV backends initialization.

Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Reviewed-by: Paul Durrant <paul@xen.org>

xen-hvm: reorganize xen-hvm and move common function to xen-hvm-common

This patch does following:
1. creates arch_handle_ioreq() and arch_xen_set_memory(). This is done in
    preparation for moving most of xen-hvm code to an arch-neutral location,
    move the x86-specific portion of xen_set_memory to arch_xen_set_memory.
    Also, move handle_vmport_ioreq to arch_handle_ioreq.

2. Pure code movement: move common functions to hw/xen/xen-hvm-common.c
    Extract common functionalities from hw/i386/xen/xen-hvm.c and move them to
    hw/xen/xen-hvm-common.c. These common functions are useful for creating
    an IOREQ server.

    xen_hvm_init_pc() contains the architecture independent code for creating
    and mapping a IOREQ server, connecting memory and IO listeners, initializing
    a xen bus and registering backends. Moved this common xen code to a new
    function xen_register_ioreq() which can be used by both x86 and ARM machines.

    Following functions are moved to hw/xen/xen-hvm-common.c:
        xen_vcpu_eport(), xen_vcpu_ioreq(), xen_ram_alloc(), xen_set_memory(),
        xen_region_add(), xen_region_del(), xen_io_add(), xen_io_del(),
        xen_device_realize(), xen_device_unrealize(),
        cpu_get_ioreq_from_shared_memory(), cpu_get_ioreq(), do_inp(),
        do_outp(), rw_phys_req_item(), read_phys_req_item(),
        write_phys_req_item(), cpu_ioreq_pio(), cpu_ioreq_move(),
        cpu_ioreq_config(), handle_ioreq(), handle_buffered_iopage(),
        handle_buffered_io(), cpu_handle_ioreq(), xen_main_loop_prepare(),
        xen_hvm_change_state_handler(), xen_exit_notifier(),
        xen_map_ioreq_server(), destroy_hvm_domain() and
        xen_shutdown_fatal_error()

3. Removed static type from below functions:
    1. xen_region_add()
    2. xen_region_del()
    3. xen_io_add()
    4. xen_io_del()
    5. xen_device_realize()
    6. xen_device_unrealize()
    7. xen_hvm_change_state_handler()
    8. cpu_ioreq_pio()
    9. xen_exit_notifier()

4. Replace TARGET_PAGE_SIZE with XC_PAGE_SIZE to match the page side with Xen.

Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>

hw/i386/xen/xen-hvm: move x86-specific fields out of XenIOState

In preparation to moving most of xen-hvm code to an arch-neutral location, move:
- shared_vmport_page
- log_for_dirtybit
- dirty_bitmap
- suspend
- wakeup

out of XenIOState struct as these are only used on x86, especially the ones
related to dirty logging.
Updated XenIOState can be used for both aarch64 and x86.

Also, remove free_phys_offset as it was unused.

Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Reviewed-by: Paul Durrant <paul@xen.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

hw/i386/xen: rearrange xen_hvm_init_pc

In preparation to moving most of xen-hvm code to an arch-neutral location,
move non IOREQ references to:
- xen_get_vmport_regs_pfn
- xen_suspend_notifier
- xen_wakeup_notifier
- xen_ram_init

towards the end of the xen_hvm_init_pc() function.

This is done to keep the common ioreq functions in one place which will be
moved to new function in next patch in order to make it common to both x86 and
aarch64 machines.

Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Reviewed-by: Paul Durrant <paul@xen.org>

hw/i386/xen/: move xen-mapcache.c to hw/xen/

xen-mapcache.c contains common functions which can be used for enabling Xen on
aarch64 with IOREQ handling. Moving it out from hw/i386/xen to hw/xen to make it
accessible for both aarch64 and x86.

Signed-off-by: Vikram Garhwal <vikram.garhwal@amd.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@amd.com>
Reviewed-by: Paul Durrant <paul@xen.org>

target/arm: Allow users to set the number of VFP registers

Cortex A7 CPUs with an FPU implementing VFPv4 without NEON support
have 16 64-bit FPU registers and not 32 registers. Let users set the
number of VFP registers with a CPU property.

The primary use case of this property is for the Cortex A7 of the
Aspeed AST2600 SoC.

Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>

aspeed: Introduce a "bmc-console" machine option

Most of the Aspeed machines use the UART5 device for the boot console,
and QEMU connects the first serial Chardev to this SoC device for this
purpose. See routine connect_serial_hds_to_uarts().

Nevertheless, some machines use another boot console, such as the fuji,
and commit 5d63d0c76c ("hw/arm/aspeed: Allow machine to set UART
default") introduced a SoC class attribute 'uart_default' and property
to be able to change the boot console device. It was later changed by
commit d2b3eaefb4 ("aspeed: Refactor UART init for multi-SoC machines").

The "bmc-console" machine option goes a step further and lets the user define
the UART device from the QEMU command line without introducing a new
machine definition. For instance, to use device UART3 (mapped on
/dev/ttyS2 under Linux) instead of the default UART5, one would use :

  -M ast2500-evb,bmc-console=uart3

Cc: Abhishek Singh Dagur <abhishek@drut.io>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Cédric Le Goater <clg@kaod.org>

aspeed: Use the boot_rom region of the fby35 machine

This change completes commits 5aa281d757 ("aspeed: Introduce a
spi_boot region under the SoC") and 8b744a6a47 ("aspeed: Add a
boot_rom overlap region in the SoC spi_boot container") which
introduced a spi_boot container at the SoC level to map the boot rom
region as an overlap.

It also fixes a Coverity report (CID 1508061) for a memory leak
warning when the QEMU process exits by using an bmc_boot_rom
MemoryRegion available at the machine level.

Cc: Peter Delevoryas <peter@pjd.dev>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Cédric Le Goater <clg@kaod.org>

aspeed: Introduce a boot_rom region at the machine level

This should also avoid Coverity to report a memory leak warning when
the QEMU process exits. See CID 1508061.

Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Cédric Le Goater <clg@kaod.org>

aspeed/hace: Initialize g_autofree pointer

As mentioned in docs/devel/style.rst "Automatic memory deallocation":

* Variables declared with g_auto* MUST always be initialized,
  otherwise the cleanup function will use uninitialized stack memory

This avoids QEMU to coredump when running the "hash test" command
under Zephyr.

Cc: Steven Lee <steven_lee@aspeedtech.com>
Cc: Joel Stanley <joel@jms.id.au>
Cc: qemu-stable@nongnu.org
Fixes: c5475b3f9a ("hw: Model ASPEED's Hash and Crypto Engine")
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com>
Message-Id: <20230421131547.2177449-1-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Cédric Le Goater <clg@kaod.org>

hw/arm/aspeed: Add VPD data for Rainier machine

The current modeling of Rainier machine creates zero filled VPDs(EEPROMs).
This makes some services and applications unhappy and causing them to fail.
Hence this drop adds some fabricated data for system and BMC FRU so that
vpd services are happy and active.

Tested:
   - The system-vpd.service is active.
   - VPD service related to bmc is active.

Signed-off-by: Ninad Palsule <ninad@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
[ clg: commit title cleanup ]
Signed-off-by: Cédric Le Goater <clg@kaod.org>

Merge tag 'pull-riscv-to-apply-20230614' of https://github.com/alistair23/qemu into staging

Second RISC-V PR for 8.1

* Skip Vector set tail when vta is zero
* Move zc* out of the experimental properties
* Mask the implicitly enabled extensions in isa_string based on priv version
* Rework CPU extension validation and validate MISA changes
* Fixup PMP TLB cacheing errors
* Writing to pmpaddr and MML/MMWP correctly triggers TLB flushes
* Fixup PMP bypass checks
* Deny access if access is partially inside a PMP entry
* Correct OpenTitanState parent type/size
* Fix QEMU crash when NUMA nodes exceed available CPUs
* Fix pointer mask transformation for vector address
* Updates and improvements for Smstateen
* Support disas for Zcm* extensions
* Support disas for Z*inx extensions
* Remove unused decomp_rv32/64 value for vector instructions
* Enable PC-relative translation
* Assume M-mode FW in pflash0 only when "-bios none"
* Support using pflash via -blockdev option
* Add vector registers to log
* Clean up reference of Vector MTYPE
* Remove the check for extra Vector tail elements
* Smepmp: Return error when access permission not allowed in PMP
* Fixes for smsiaddrcfg and smsiaddrcfgh in AIA

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEaukCtqfKh31tZZKWr3yVEwxTgBMFAmSJFRoACgkQr3yVEwxT
# gBMUkg/8Cuhqpx+zy7MeouVkyhEjUuhtCWyr0WVZBJzDkVEOrlY6TyR0hb5/o1Js
# LZf6ZMF6JQDN78bmUct8yFBZBGafey5tyonDCsnD7CNQuLPf2NSjTHhu9n5hKFqF
# F8Mpn9iFu6k1pr0iF7FbCccVWuDb3P4h2PaM0iFhmf4uz42BCMYdgJThhvv38xlt
# jr6A3dcjTpp8yB+iRCuhL2IU2XVee0XBiDUECqRXd0gmtOtqJNST8L+l8YkLy1VO
# WUMe8RCO6NMP7BLJ383WwCDeiFTo0mJebZQ0eR/G1xEhy7c8BBMh/CgQmq2F3wDZ
# Q0biaeozADgAaCC7aOAHI+1sAoMhOm1v2WhIVmh+XXUqT9856cKwc7DUPBmzb9Sj
# N5Zh+t9WCnZG7qpfxvkDF0Y/aRODMHZ1BW5L/ky9yBtyuRwXOJ6VycZTFyRkSwnN
# Gd/s9IClDOP1IP5s4TSMGGdelk4lH97x7fZE/2hxn59lp761JtMxbaEceBtqaBh8
# zNMTNN/KHs8LeiIBI2ZZ+nQav452Y6XYBivQ7OdsI8xkjnjG9gfgXXjvX1TIh0ow
# Hy5ZxtAtjXty49Gmjkx5VcBx4auJcnRDlLTzoZjTxq1te+gEWpw6O1EsEKasVLZe
# uN6PxTOxS3nHvRvPgQc1xNUdhDRqBaYsju6b9YmMxz1uefAjGM0=
# =fOTc
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 14 Jun 2023 03:17:14 AM CEST
# gpg:                using RSA key 6AE902B6A7CA877D6D659296AF7C95130C538013
# gpg: Good signature from "Alistair Francis <alistair@alistair23.me>" [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: 6AE9 02B6 A7CA 877D 6D65  9296 AF7C 9513 0C53 8013

* tag 'pull-riscv-to-apply-20230614' of https://github.com/alistair23/qemu: (60 commits)
  hw/intc: If mmsiaddrcfgh.L == 1, smsiaddrcfg and smsiaddrcfgh are read-only.
  target/riscv: Smepmp: Return error when access permission not allowed in PMP
  target/riscv/vector_helper.c: Remove the check for extra tail elements
  target/riscv/vector_helper.c: clean up reference of MTYPE
  target/riscv: Fix initialized value for cur_pmmask
  util/log: Add vector registers to log
  docs/system: riscv: Add pflash usage details
  riscv/virt: Support using pflash via -blockdev option
  hw/riscv: virt: Assume M-mode FW in pflash0 only when "-bios none"
  target/riscv: Remove pc_succ_insn from DisasContext
  target/riscv: Enable PC-relative translation
  target/riscv: Use true diff for gen_pc_plus_diff
  target/riscv: Change gen_set_pc_imm to gen_update_pc
  target/riscv: Change gen_goto_tb to work on displacements
  target/riscv: Introduce cur_insn_len into DisasContext
  target/riscv: Fix target address to update badaddr
  disas/riscv.c: Remove redundant parentheses
  disas/riscv.c: Fix lines with over 80 characters
  disas/riscv.c: Remove unused decomp_rv32/64 value for vector instructions
  disas/riscv.c: Support disas for Z*inx extensions
  ...

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Merge tag 'misc-20230613' of https://github.com/philmd/qemu into staging

Misc patches queue

- user emulation: Preserve environment variable order
- macos/darwin/hvf: Fix build warnings, slighly optimize DCache flush
- target/i386: Minor cleanups, rename  template headers with '.inc' suffix
- target/hppa: Avoid building int_helper.o on user emulation
- hw: Add 'name' property to pca954x, export ISAParallelState, silent warnings
- hw/vfio: Trace number of bitmap dirty pages
- exec/memory: Introduce RAM_NAMED_FILE to distinct block without named backing store

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE+qvnXhKRciHc/Wuy4+MsLN6twN4FAmSINukACgkQ4+MsLN6t
# wN5fEQ/+PEC2CoBtl5aIODSowEyaFVmNTHZWQFmRAUlBoPR2rU7NEpnh3j4BJWSS
# aKc6MV4GDExKvMS5GujXiruS3kYMyWUHSG+nGUqrkBmbWn/2tZdWfIa0uT20CTr/
# nxu8TWM/EO+uzvYdyTNrXdkBOkU5eKfAqFGcWP7uLtRkjXpvu2YcPxrKswQUFpTh
# XH5sCh6esaBkMZLfsl0I4ZQLxtBWpM3EmaDvlDxftMt4OZ3kwbzC55xhN1uk6KxY
# iarkROM54vqU7ZC+vWH6fT18teLvKWXo+9F1iD/Qio88xRCH/kuQRWBM8+XR9IDe
# CKuaRrJ1nOmPojG3Xngd99eM7FPjh1FR6bR+L62cm8/PzuzbQi5VBIdJvxXfvoTL
# OjiXA3stOVB/jZg37IAeZB2X644d1GJXcjpExpU+mk7qdkaZC16HVOTRYYQF/wFF
# QBUntZ9Oz0+XRtt4PC8ckynq0p0uEA/mZ14jMm04AJ3W8XtnOk2xp5rFTn/f4tsI
# fJhqHxNhfet2sU17UI/pRw/h3Ulg0TmAdmMZnz3jUQ13/6E0//+a6dqQED/5HeG5
# 6UvUEUdGfvHuyE8l1UT4p93pBcyshMuaVFfutxKP9qYGPLhYeiCvms8KfdOWhTkA
# NC1JlOScwdEqkn1hyMCeJHEUpnuZHZeiOtn4VV4J/SwyxkzVRS8=
# =sUgM
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 13 Jun 2023 11:29:13 AM CEST
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.org>" [full]

* tag 'misc-20230613' of https://github.com/philmd/qemu:
  exec/memory: Introduce RAM_NAMED_FILE flag
  hw/vfio: Add number of dirty pages to vfio_get_dirty_bitmap tracepoint
  exec/ram_addr: Return number of dirty pages in cpu_physical_memory_set_dirty_lebitmap()
  hw/char/parallel-isa: Export struct ISAParallelState
  hw/char/parallel: Export struct ParallelState
  hw/scsi/megasas: Silent GCC duplicated-cond warning
  hw/ide/ahci: Remove stray backslash
  hw/i2c: Enable an id for the pca954x devices
  target/i386: Rename helper template headers as '.h.inc'
  target/i386/helper: Shuffle do_cpu_init()
  target/i386/helper: Remove do_cpu_sipi() stub for user-mode emulation
  target/hppa/meson: Only build int_helper.o with system emulation
  accel/hvf: Report HV_DENIED error
  util/cacheflush: Avoid possible redundant dcache flush on Darwin
  util/cacheflush: Use declarations from <OSCacheControl.h> on Darwin
  cocoa: Fix warnings about invalid prototype declarations
  linux-user, bsd-user: Preserve incoming order of environment variables in the target

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

hw/intc: If mmsiaddrcfgh.L == 1, smsiaddrcfg and smsiaddrcfgh are read-only.

According to the `The RISC-V Advanced Interrupt Architecture`
document, if register `mmsiaddrcfgh` of the domain has bit L set
to one, then `smsiaddrcfg` and `smsiaddrcfgh` are locked as
read-only alongside `mmsiaddrcfg` and `mmsiaddrcfgh`.

Signed-off-by: Tommy Wu <tommy.wu@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Anup Patel <anup@brainfault.org>
Message-Id: <20230609055936.3925438-1-tommy.wu@sifive.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

exec/memory: Introduce RAM_NAMED_FILE flag

migrate_ignore_shared() is an optimization that avoids copying memory
that is visible and can be mapped on the target.  However, a
memory-backend-ram or a memory-backend-memfd block with the RAM_SHARED
flag set is not migrated when migrate_ignore_shared() is true.  This is
wrong, because the block has no named backing store, and its contents will
be lost.  To fix, ignore shared memory iff it is a named file.  Define a
new flag RAM_NAMED_FILE to distinguish this case.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <1686151116-253260-1-git-send-email-steven.sistare@oracle.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/vfio: Add number of dirty pages to vfio_get_dirty_bitmap tracepoint

Include the number of dirty pages on the vfio_get_dirty_bitmap tracepoint.
These are fetched from the newly added return value in
cpu_physical_memory_set_dirty_lebitmap().

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
Message-Id: <20230530180556.24441-3-joao.m.martins@oracle.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

exec/ram_addr: Return number of dirty pages in cpu_physical_memory_set_dirty_lebitmap()

In preparation for including the number of dirty pages in the
vfio_get_dirty_bitmap() tracepoint, return the number of dirty pages in
cpu_physical_memory_set_dirty_lebitmap() similar to
cpu_physical_memory_sync_dirty_bitmap().

To avoid counting twice when GLOBAL_DIRTY_RATE is enabled, stash the
number of bits set per bitmap quad in a variable (@nbits) and reuse it
there.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230530180556.24441-2-joao.m.martins@oracle.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/char/parallel-isa: Export struct ISAParallelState

Allows the struct to be embedded directly into device models without additional
allocation.

Suggested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230612081238.1742-3-shentey@gmail.com>
[PMD: Update MAINTAINERS entry and use SPDX license identifier]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/char/parallel: Export struct ParallelState

Exporting ParallelState is a precondition for exporing TYPE_ISA_PARALLEL to be
performed in the next patch.

Suggested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230612081238.1742-2-shentey@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/scsi/megasas: Silent GCC duplicated-cond warning

GCC9 is confused when building with CFLAG -O3:

  hw/scsi/megasas.c: In function ‘megasas_scsi_realize’:
  hw/scsi/megasas.c:2387:26: error: duplicated ‘if’ condition [-Werror=duplicated-cond]
   2387 |     } else if (s->fw_sge >= 128 - MFI_PASS_FRAME_SIZE) {
  hw/scsi/megasas.c:2385:19: note: previously used here
   2385 |     if (s->fw_sge >= MEGASAS_MAX_SGE - MFI_PASS_FRAME_SIZE) {
  cc1: all warnings being treated as errors

When this device was introduced in commit e8f943c3bcc, the author
cared about modularity, using a definition for the firmware limit.

However if the firmware limit isn't changed (MEGASAS_MAX_SGE = 128),
the code ends doing the same check twice.

Per the maintainer [*]:

> The original code assumed that one could change MFI_PASS_FRAME_SIZE,
> but it turned out not to be possible as it's being hardcoded in the
> drivers themselves (even though the interface provides mechanisms to
> query it). So we can remove the duplicate lines.

Add the 'MEGASAS_MIN_SGE' definition for the '64' magic value,
slightly rewrite the condition check to simplify a bit the logic
and remove the unnecessary / duplicated check.

[*] https://lore.kernel.org/qemu-devel/e0029fc5-882f-1d63-15e3-1c3dbe9b6a2c@suse.de/

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Message-Id: <20230328210126.16282-1-philmd@linaro.org>

hw/ide/ahci: Remove stray backslash

This backslash obviously does not belong here, so remove it.

Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-Id: <20230601134434.519805-2-nks@flawful.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i2c: Enable an id for the pca954x devices

This allows the devices to be more readily found and specified.
Without setting the name field, they can only be found by device type
name, which doesn't let you specify the second of the same device type
behind a bus.

Tested: Verified that by default the device was findable with the name
'pca954x[77]', for an instance attached at that address.

Signed-off-by: Patrick Venture <venture@google.com>
Reviewed-by: Hao Wu <wuhaotsh@google.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Corey Minyard <cminyard@mvista.com>
Message-Id: <20230322172136.48010-1-venture@google.com>
[PMD: Fix typo in property name]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/i386: Rename helper template headers as '.h.inc'

Since commit 139c1837db ("meson: rename included C source files
to .c.inc"), QEMU standard procedure for included C files is to
use *.c.inc.

Besides, since commit 6a0057aa22 ("docs/devel: make a statement
about includes") this is documented as the Coding Style:

  If you do use template header files they should be named with
  the ``.c.inc`` or ``.h.inc`` suffix to make it clear they are
  being included for expansion.

Therefore move the included templates in the tcg/ directory and
rename as '.h.inc'.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230608133108.72655-5-philmd@linaro.org>

target/i386/helper: Shuffle do_cpu_init()

Move the #ifdef'ry inside do_cpu_init() instead of
declaring an empty stub for user emulation.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230602224628.59546-3-philmd@linaro.org>

target/i386/helper: Remove do_cpu_sipi() stub for user-mode emulation

Since commit  604664726f ("target/i386: Restrict cpu_exec_interrupt()
handler to sysemu"), do_cpu_sipi() isn't called anymore on user
emulation. Remove the now pointless stub.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230602224628.59546-2-philmd@linaro.org>

target/hppa/meson: Only build int_helper.o with system emulation

int_helper.c only contains system emulation code:
remove the #ifdef'ry and move the file to the meson
softmmu source set.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230602223016.58647-1-philmd@linaro.org>

accel/hvf: Report HV_DENIED error

On MacOS 11 and subsequent versions, in case the resulting binary is not
signed with the proper entitlement, handle and report the HV_DENIED
error.

Signed-off-by: Antonio Caggiano <quic_acaggian@quicinc.com>
Message-Id: <20230608123014.28715-1-quic_acaggian@quicinc.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

util/cacheflush: Avoid possible redundant dcache flush on Darwin

<libkern/OSCacheControl.h> describes sys_icache_invalidate() as
"equivalent to sys_cache_control(kCacheFunctionPrepareForExecution)",
having kCacheFunctionPrepareForExecution defined as:

  /* Prepare memory for execution.  This should be called
   * after writing machine instructions to memory, before
   * executing them.  It syncs the dcache and icache. [...]
   */

Since the dcache is also sync'd, we can avoid the sys_dcache_flush()
call when both rx/rw pointers are equal.

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20230605195911.96033-1-philmd@linaro.org>

util/cacheflush: Use declarations from <OSCacheControl.h> on Darwin

Per the cache(3) man page, sys_icache_invalidate() and
sys_dcache_flush() are declared in <libkern/OSCacheControl.h>.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230605175647.88395-2-philmd@linaro.org>

cocoa: Fix warnings about invalid prototype declarations

Fix the following Cocoa trivial warnings:

  C compiler for the host machine: cc (clang 14.0.0 "Apple clang version 14.0.0 (clang-1400.0.29.202)")
  Objective-C compiler for the host machine: clang (clang 14.0.0)

  [100/334] Compiling Objective-C object libcommon.fa.p/net_vmnet-bridged.m.o
  net/vmnet-bridged.m:40:31: warning: a function declaration without a prototype is deprecated in all versions of C [-Wstrict-prototypes]
  static char* get_valid_ifnames()
                                ^
                                 void
  [742/1436] Compiling Objective-C object libcommon.fa.p/ui_cocoa.m.o
  ui/cocoa.m:1937:22: warning: a function declaration without a prototype is deprecated in all versions of C [-Wstrict-prototypes]
  static int cocoa_main()
                       ^
                        void

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20230425192820.34063-1-philmd@linaro.org>

linux-user, bsd-user: Preserve incoming order of environment variables in the target

Do not reverse the order of environment variables in the target environ
array relative to the incoming environ order.  Some testsuites depend on a
specific order, even though it is not defined by any standard.

Signed-off-by: Andreas Schwab <schwab@suse.de>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <mvmlejfsivd.fsf@suse.de>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/riscv: Smepmp: Return error when access permission not allowed in PMP

On an address match, skip checking for default permissions and return error
based on access defined in PMP configuration.

v3 Changes:
o Removed explicit return of boolean value from comparision
  of priv/allowed_priv

v2 Changes:
o Removed goto to return in place when address matches
o Call pmp_hart_has_privs_default at the end of the loop

Fixes: 90b1fafce06 ("target/riscv: Smepmp: Skip applying default rules when address matches")
Signed-off-by: Himanshu Chauhan <hchauhan@ventanamicro.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Message-Id: <20230605164548.715336-1-hchauhan@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv/vector_helper.c: Remove the check for extra tail elements

Commit 752614cab8e6 ("target/riscv: rvv: Add tail agnostic for vector
load / store instructions") added an extra check for LMUL fragmentation,
intended for setting the "rest tail elements" in the last register for a
segment load insn.

Actually, the max_elements derived in vext_ld*() won't be a fraction of
vector register size, since the lmul encoded in desc is emul, which has
already been adjusted to 1 for LMUL fragmentation case by vext_get_emul()
in trans_rvv.c.inc, for ld_stride(), ld_us(), ld_index() and ldff().

Besides, vext_get_emul() has also taken EEW/SEW into consideration, so no
need to call vext_get_total_elems() which would base on the emul to derive
another emul, the second emul would be incorrect when esz differs from sew.

Thus this patch removes the check for extra tail elements.

Fixes: 752614cab8e6 ("target/riscv: rvv: Add tail agnostic for vector load / store instructions")

Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Message-Id: <20230607091646.4049428-1-xiao.w.wang@intel.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv/vector_helper.c: clean up reference of MTYPE

There's no code using MTYPE, which was a concept used in older vector
implementation.

Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Message-Id: <20230608053517.4102648-1-xiao.w.wang@intel.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Fix initialized value for cur_pmmask

We initialize cur_pmmask as -1(UINT32_MAX/UINT64_MAX) and regard it
as if pointer mask is disabled in current implementation. However,
the addresses for vector load/store will be adjusted to zero in this
case and -1(UINT32_MAX/UINT64_MAX) is valid value for pmmask when
pointer mask is enabled.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Message-Id: <20230610094651.43786-1-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

util/log: Add vector registers to log

Added QEMU option 'vpu' to log vector extension registers such as gpr\fpu.

Signed-off-by: Ivan Klokov <ivan.klokov@syntacore.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230410124451.15929-2-ivan.klokov@syntacore.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

docs/system: riscv: Add pflash usage details

pflash devices can be used in virt machine for different
purposes like for ROM code or S-mode FW payload. Add a
section in the documentation on how to use pflash devices
for different purposes.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230601045910.18646-4-sunilvl@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

riscv/virt: Support using pflash via -blockdev option

Currently, pflash devices can be configured only via -pflash
or -drive options. This is the legacy way and the
better way is to use -blockdev as in other architectures.
libvirt also has moved to use -blockdev method.

To support -blockdev option, pflash devices need to be
created in instance_init itself. So, update the code to
move the virt_flash_create() to instance_init. Also, use
standard interfaces to detect whether pflash0 is
configured or not.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reported-by: Andrea Bolognani <abologna@redhat.com>
Tested-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230601045910.18646-3-sunilvl@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: virt: Assume M-mode FW in pflash0 only when "-bios none"

Currently, virt machine supports two pflash instances each with
32MB size. However, the first pflash is always assumed to
contain M-mode firmware and reset vector is set to this if
enabled. Hence, for S-mode payloads like EDK2, only one pflash
instance is available for use. This means both code and NV variables
of EDK2 will need to use the same pflash.

The OS distros keep the EDK2 FW code as readonly. When non-volatile
variables also need to share the same pflash, it is not possible
to keep it as readonly since variables need write access.

To resolve this issue, the code and NV variables need to be separated.
But in that case we need an extra flash. Hence, modify the convention
for non-KVM guests such that, pflash0 will contain the M-mode FW
only when "-bios none" option is used. Otherwise, pflash0 will contain
the S-mode payload FW. This enables both pflash instances available
for EDK2 use.

When KVM is enabled, pflash0 is always assumed to contain the
S-mode payload firmware only.

Example usage:
1) pflash0 containing M-mode FW
qemu-system-riscv64 -bios none -pflash <mmode_fw> -machine virt
or
qemu-system-riscv64 -bios none \
-drive file=<mmode_fw>,if=pflash,format=raw,unit=0 -machine virt

2) pflash0 containing S-mode payload like EDK2
qemu-system-riscv64 -pflash <smode_fw_code> -pflash <smode_vars> -machine  virt
or
qemu-system-riscv64 -bios <opensbi_fw> \
-pflash <smode_fw_code> \
-pflash <smode_vars> \
-machine  virt
or
qemu-system-riscv64 -bios <opensbi_fw> \
-drive file=<smode_fw_code>,if=pflash,format=raw,unit=0,readonly=on \
-drive file=<smode_fw_vars>,if=pflash,format=raw,unit=1 \
-machine virt

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reported-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Tested-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230601045910.18646-2-sunilvl@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Remove pc_succ_insn from DisasContext

pc_succ_insn is no longer useful after the introduce of cur_insn_len
and all pc related value use diff value instead of absolute value.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230526072124.298466-8-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Enable PC-relative translation

Add a base pc_save for PC-relative translation(CF_PCREL).
Diable the directly sync pc from tb by riscv_cpu_synchronize_from_tb.
Use gen_pc_plus_diff to get the pc-relative address.
Enable CF_PCREL in System mode.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230526072124.298466-7-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Use true diff for gen_pc_plus_diff

Reduce reliance on absolute values by using true pc difference for
gen_pc_plus_diff() to prepare for PC-relative translation.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230526072124.298466-6-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Change gen_set_pc_imm to gen_update_pc

Reduce reliance on absolute values(by passing pc difference) to
prepare for PC-relative translation.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230526072124.298466-5-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Change gen_goto_tb to work on displacements

Reduce reliance on absolute value to prepare for PC-relative translation.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230526072124.298466-4-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Introduce cur_insn_len into DisasContext

Use cur_insn_len to store the length of the current instruction to
prepare for PC-relative translation.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230526072124.298466-3-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Fix target address to update badaddr

Compute the target address before storing it into badaddr
when mis-aligned exception is triggered.
Use a target_pc temp to store the target address to avoid
the confusing operation that udpate target address into
cpu_pc before misalign check, then update it into badaddr
and restore cpu_pc to current pc if exception is triggered.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230526072124.298466-2-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

disas/riscv.c: Remove redundant parentheses

Remove redundant parenthese and fix multi-line comments.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230523093539.203909-9-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

disas/riscv.c: Fix lines with over 80 characters

Fix lines with over 80 characters.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230523093539.203909-8-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

disas/riscv.c: Remove unused decomp_rv32/64 value for vector instructions

Currently decomp_rv32 and decomp_rv64 value in opcode_data for vector
instructions are the same op index as their own. And they have no
functional decomp_data. So they have no functional difference from just
leaving them as zero.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230523093539.203909-7-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

disas/riscv.c: Support disas for Z*inx extensions

Support disas for Z*inx instructions only when Zfinx extension is supported.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230523093539.203909-6-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

disas/riscv.c: Support disas for Zcm* extensions

Support disas for Zcmt* instructions only when related extensions
are supported.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230523093539.203909-5-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Pass RISCVCPUConfig as target_info to disassemble_info

Pass RISCVCPUConfig as disassemble_info.target_info to support disas
of conflict instructions related to specific extensions.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230523093539.203909-4-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Split RISCVCPUConfig declarations from cpu.h into cpu_cfg.h

Split RISCVCPUConfig declarations to prepare for passing it to disas.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230523093539.203909-3-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

disas: Change type of disassemble_info.target_info to pointer

Use pointer to pass more information of target to disasembler,
such as pass cpu.cfg related information in following commits.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230523093539.203909-2-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: smstateen knobs

Add knobs to allow users to enable smstateen and also export it via the
ISA extension string.

Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Reviewed-by: Weiwei Li<liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230518175058.2772506-4-mchitale@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Reuse tb->flags.FS

When misa.F is 0 tb->flags.FS field is unused and can be used to save
the current state of smstateen0.FCSR check which is needed by the
floating point translation routines.

Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Message-Id: <20230518175058.2772506-3-mchitale@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: smstateen check for fcsr

Implement the s/h/mstateen.fcsr bit as defined in the smstateen spec
and check for it when accessing the fcsr register and its fields.

Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230518175058.2772506-2-mchitale@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Update cur_pmmask/base when xl changes

write_mstatus() can only change current xl when in debug mode.
And we need update cur_pmmask/base in this case.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Message-Id: <20230524015933.17349-3-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Fix pointer mask transformation for vector address

actual_address = (requested_address & ~mpmmask) | mpmbase.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Message-Id: <20230524015933.17349-2-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: qemu crash when NUMA nodes exceed available CPUs

Command "qemu-system-riscv64 -machine virt
-m 2G -smp 1 -numa node,mem=1G -numa node,mem=1G"
would trigger this problem.Backtrace with:
 #0  0x0000555555b5b1a4 in riscv_numa_get_default_cpu_node_id  at ../hw/riscv/numa.c:211
 #1  0x00005555558ce510 in machine_numa_finish_cpu_init  at ../hw/core/machine.c:1230
 #2  0x00005555558ce9d3 in machine_run_board_init  at ../hw/core/machine.c:1346
 #3  0x0000555555aaedc3 in qemu_init_board  at ../softmmu/vl.c:2513
 #4  0x0000555555aaf064 in qmp_x_exit_preconfig  at ../softmmu/vl.c:2609
 #5  0x0000555555ab1916 in qemu_init  at ../softmmu/vl.c:3617
 #6  0x000055555585463b in main  at ../softmmu/main.c:47
This commit fixes the issue by adding parameter checks.

Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Yin Wang <yin.wang@intel.com>
Message-Id: <20230519023758.1759434-1-yin.wang@intel.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv/opentitan: Correct OpenTitanState parent type/size

OpenTitanState is the 'machine' (or 'board') state: it isn't
a SysBus device, but inherits from the MachineState type.
Correct the instance size.
Doing so we  avoid leaking an OpenTitanState pointer in
opentitan_machine_init().

Fixes: fe0fe4735e ("riscv: Initial commit of OpenTitan machine")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230520054510.68822-6-philmd@linaro.org>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv/opentitan: Explicit machine type definition

Expand the DEFINE_MACHINE() macro, converting the class_init()
handler.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230520054510.68822-5-philmd@linaro.org>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv/opentitan: Add TYPE_OPENTITAN_MACHINE definition

QOM type names are usually defined as TYPE_FOO.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230520054510.68822-4-philmd@linaro.org>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv/opentitan: Declare QOM types using DEFINE_TYPES() macro

When multiple QOM types are registered in the same file,
it is simpler to use the the DEFINE_TYPES() macro. Replace
the type_init() / type_register_static() combination. This
is in preparation of adding the OpenTitan machine type to
this array in a pair of commits.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230520054510.68822-3-philmd@linaro.org>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv/opentitan: Rename machine_[class]_init() functions

Follow QOM style which declares FOO_init() as instance
initializer and FOO_class_init() as class initializer:
rename the OpenTitan machine class/instance init()
accordingly.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230520054510.68822-2-philmd@linaro.org>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Deny access if access is partially inside the PMP entry

Access will fail if access is partially inside the PMP entry.
However,only setting ret = false doesn't really mean pmp violation
since pmp_hart_has_privs_default() may return true at the end of
pmp_hart_has_privs().

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-13-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Separate pmp_update_rule() in pmpcfg_csr_write

Use pmp_update_rule_addr() and pmp_update_rule_nums() separately to
update rule nums only once for each pmpcfg_csr_write. Then remove
pmp_update_rule() since it become unused.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-12-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Flush TLB only when pmpcfg/pmpaddr really changes

TLB needn't be flushed when pmpcfg/pmpaddr don't changes.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Message-Id: <20230517091519.34439-11-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Flush TLB when pmpaddr is updated

TLB should be flushed not only for pmpcfg csr changes, but also for
pmpaddr csr changes.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Message-Id: <20230517091519.34439-10-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Update the next rule addr in pmpaddr_csr_write()

Currently only the rule addr of the same index of pmpaddr is updated
when pmpaddr CSR is modified. However, the rule addr of next PMP entry
may also be affected if its A field is PMP_AMATCH_TOR. So we should
also update it in this case.

Write to pmpaddr CSR will not affect the rule nums, So we needn't update
call pmp_update_rule_nums()  in pmpaddr_csr_write().

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-9-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Flush TLB when MMWP or MML bits are changed

MMWP and MML bits may affect the allowed privs of PMP entries and the
default privs, both of which may change the allowed privs of exsited
TLB entries. So we need flush TLB when they are changed.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-8-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Remove unused paramters in pmp_hart_has_privs_default()

The addr and size parameters in pmp_hart_has_privs_default() are unused.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-7-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Make RLB/MML/MMWP bits writable only when Smepmp is enabled

RLB/MML/MMWP bits in mseccfg CSR are introduced by Smepmp extension.
So they can only be writable and set to 1s when cfg.epmp is true.
Then we also need't check on epmp in pmp_hart_has_privs_default().

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-6-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Change the return type of pmp_hart_has_privs() to bool

We no longer need the pmp_index for matched PMP entry now.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-5-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Make the short cut really work in pmp_hart_has_privs

Return the result directly for short cut, since We needn't do the
following check on the PMP entries if there is no PMP rules.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-4-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Move pmp_get_tlb_size apart from get_physical_address_pmp

pmp_get_tlb_size can be separated from get_physical_address_pmp and is only
needed when ret == TRANSLATE_SUCCESS.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-3-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Update pmp_get_tlb_size()

PMP entries before (including) the matched PMP entry may only cover partial
of the TLB page, and this may split the page into regions with different
permissions. Such as for PMP0 (0x80000008~0x8000000F, R) and PMP1 (0x80000000~
0x80000FFF, RWX), write access to 0x80000000 will match PMP1. However we cannot
cache the translation result in the TLB since this will make the write access
to 0x80000008 bypass the check of PMP0. So we should check all of them instead
of the matched PMP entry in pmp_get_tlb_size() and set the tlb_size to 1 in
this case.
Set tlb_size to TARGET_PAGE_SIZE if PMP is not support or there is no PMP rules.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-2-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rework write_misa()

write_misa() must use as much common logic as possible. We want to open
code just the bits that are exclusive to the CSR write operation and TCG
internals.

Our validation is done with riscv_cpu_validate_set_extensions(), but we
need a small tweak first. When enabling RVG we're doing:

        env->misa_ext |= RVI | RVM | RVA | RVF | RVD;
        env->misa_ext_mask = env->misa_ext;

This works fine for realize() time but this can potentially overwrite
env->misa_ext_mask if we reutilize the function for write_misa().

Instead of doing misa_ext_mask = misa_ext, sum up the RVG extensions in
misa_ext_mask as well. This won't change realize() time behavior
(misa_ext_mask will be == misa_ext) and will ensure that write_misa()
won't change misa_ext_mask by accident.

After that, rewrite write_misa() to work as follows:

- mask the write using misa_ext_mask to avoid enabling unsupported
  extensions;

- suppress RVC if the next insn isn't aligned;

- disable RVG if any of RVG dependencies are being disabled by the user;

- assign env->misa_ext and run riscv_cpu_validate_set_extensions(). On
  error, rollback env->misa_ext to its original value, logging a
  GUEST_ERROR to inform the user about the failed write;

- handle RVF and MSTATUS_FS and continue as usual.

Let's keep write_misa() as experimental for now until this logic gains
enough mileage.

Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230517135714.211809-12-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv/cpu.c: remove cfg setup from riscv_cpu_init()

We have 4 config settings being done in riscv_cpu_init(): ext_ifencei,
ext_icsr, mmu and pmp. This is also the constructor of the "riscv-cpu"
device, which happens to be the parent device of every RISC-V cpu.

The result is that these 4 configs are being set every time, and every
other CPU should always account for them. CPUs such as sifive_e need to
disable settings that aren't enabled simply because the parent class
happens to be enabling it.

Moving all configurations from the parent class to each CPU will
centralize the config of each CPU into its own init(), which is clearer
than having to account to whatever happens to be set in the parent
device. These settings are also being set in register_cpu_props() when
no 'misa_ext' is set, so for these CPUs we don't need changes. Named
CPUs will receive all cfgs that the parent were setting into their
init().

Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517135714.211809-11-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv/cpu.c: validate extensions before riscv_timer_init()

There is no need to init timers if we're not even sure that our
extensions are valid. Execute riscv_cpu_validate_set_extensions() before
riscv_timer_init().

Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517135714.211809-10-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv/cpu.c: add riscv_cpu_validate_misa_mxl()

Let's remove more code that is open coded in riscv_cpu_realize() and put
it into a helper. Let's also add an error message instead of just
asserting out if env->misa_mxl_max != env->misa_mlx.

Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517135714.211809-9-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>