target/arm: Convert ERET, ERETAA, ERETAB to decodetree

Convert the exception-return insns ERET, ERETA and ERETB to
decodetree. These were the last insns left in the legacy
decoder function disas_uncond_reg_b(), which allows us to
remove it.

The old decoder explicitly decoded the DRPS instruction,
only in order to call unallocated_encoding() on it, exactly
as would have happened if it hadn't decoded it. This is
because this insn always UNDEFs unless the CPU is in
halting-debug state, which we don't emulate. So we list
the pattern in a comment in a64.decode, but don't actively
decode it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-21-peter.maydell@linaro.org

target/arm: Convert BRAA, BRAB, BLRAA, BLRAB to decodetree

Convert the last four BR-with-pointer-auth insns to decodetree.
The remaining cases in the outer switch in disas_uncond_b_reg()
all return early rather than leaving the case statement, so we
can delete the now-unused code at the end of that function.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-20-peter.maydell@linaro.org

target/arm: Convert BRA[AB]Z, BLR[AB]Z, RETA[AB] to decodetree

Convert the single-register pointer-authentication variants of BR,
BLR, RET to decodetree. (BRAA/BLRAA are in a different branch of
the legacy decoder and will be dealt with in the next commit.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-19-peter.maydell@linaro.org

target/arm: Convert BR, BLR, RET to decodetree

Convert the simple (non-pointer-auth) BR, BLR and RET insns
to decodetree.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-18-peter.maydell@linaro.org

target/arm: Convert conditional branch insns to decodetree

Convert the immediate conditional branch insn B.cond to
decodetree.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-17-peter.maydell@linaro.org

target/arm: Convert TBZ, TBNZ to decodetree

Convert the test-and-branch-immediate insns TBZ and TBNZ
to decodetree.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-16-peter.maydell@linaro.org

target/arm: Convert CBZ, CBNZ to decodetree

Convert the compare-and-branch-immediate insns CBZ and CBNZ
to decodetree.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-15-peter.maydell@linaro.org

target/arm: Convert unconditional branch immediate to decodetree

Convert the unconditional branch immediate insns B and BL to
decodetree.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-14-peter.maydell@linaro.org

target/arm: Convert Extract instructions to decodetree

Convert the EXTR instruction to decodetree (this is the
only one in the 'Extract" class). This is the last of
the dp-immediate insns in the legacy decoder, so we
can now remove disas_data_proc_imm().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-13-peter.maydell@linaro.org

target/arm: Convert Bitfield to decodetree

Convert the BFM, SBFM, UBFM instructions.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-12-peter.maydell@linaro.org
[PMM: Rebased]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Convert Move wide (immediate) to decodetree

Convert the MON, MOVZ, MOVK instructions.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-11-peter.maydell@linaro.org
[PMM: Rebased]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Convert Logical (immediate) to decodetree

Convert the ADD, ORR, EOR, ANDS (immediate) instructions.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-10-peter.maydell@linaro.org
[PMM: rebased]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Replace bitmask64 with MAKE_64BIT_MASK

Use the bitops.h macro rather than rolling our own here.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-9-peter.maydell@linaro.org

target/arm: Convert Add/subtract (immediate with tags) to decodetree

Convert the ADDG and SUBG (immediate) instructions.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-8-peter.maydell@linaro.org
[PMM: Rebased; use TRANS_FEAT()]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Convert Add/subtract (immediate) to decodetree

Convert the ADD and SUB (immediate) instructions.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-7-peter.maydell@linaro.org
[PMM: Rebased; adjusted to use translate.h's TRANS macro]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Split gen_add_CC and gen_sub_CC

Split out specific 32-bit and 64-bit functions.
These carry the same signature as tcg_gen_add_i64,
and so will be easier to pass as callbacks.

Retain gen_add_CC and gen_sub_CC during conversion.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-6-peter.maydell@linaro.org
[PMM: rebased]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Convert PC-rel addressing to decodetree

Convert the ADR and ADRP instructions.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-5-peter.maydell@linaro.org
[PMM: Rebased]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Pull calls to disas_sve() and disas_sme() out of legacy decoder

The SVE and SME decode is already done by decodetree.  Pull the calls
to these decoders out of the legacy decoder.  This doesn't change
behaviour because all the patterns in sve.decode and sme.decode
already require the bits that the legacy decoder is decoding to have
the correct values.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-4-peter.maydell@linaro.org

target/arm: Create decodetree skeleton for A64

The A64 translator uses a hand-written decoder for everything except
SVE or SME.  It's fairly well structured, but it's becoming obvious
that it's still more painful to add instructions to than the A32
translator, because putting a new instruction into the right place in
a hand-written decoder is much harder than adding new instruction
patterns to a decodetree file.

As the first step in conversion to decodetree, create the skeleton of
the decodetree decoder; where it does not handle instructions we will
fall back to the legacy decoder (which will be for everything at the
moment, since there are no patterns in a64.decode).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230512144106.3608981-3-peter.maydell@linaro.org

target/arm: Split out disas_a64_legacy

Split out all of the decode stuff from aarch64_tr_translate_insn.
Call it disas_a64_legacy to indicate it will be replaced.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230512144106.3608981-2-peter.maydell@linaro.org
[PMM: Rebased]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sbsa-ref: use Bochs graphics card instead of VGA

Bochs card is normal PCI Express card so it fits better in system with
PCI Express bus. VGA is simple legacy PCI card.

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Reviewed-by: Leif Lindholm <quic_llindhol@quicinc.com>
Message-id: 20230505120936.1097060-1-marcin.juszkiewicz@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: add RAZ/WI handling for DBGDTR[TX|RX]

The commit b3aa2f2128 (target/arm: provide stubs for more external
debug registers) was added to handle HyperV's unconditional usage of
Debug Communications Channel. It turns out that Linux will similarly
break if you enable CONFIG_HVC_DCC "ARM JTAG DCC console".

Extend the registers we RAZ/WI set to avoid this.

Cc: Anders Roxell <anders.roxell@linaro.org>
Cc: Evgeny Iakovlev <eiakovlev@linux.microsoft.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230516104420.407912-1-alex.bennee@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

arm/kvm: add support for MTE

Extend the 'mte' property for the virt machine to cover KVM as
well. For KVM, we don't allocate tag memory, but instead enable the
capability.

If MTE has been enabled, we need to disable migration, as we do not
yet have a way to migrate the tags as well. Therefore, MTE will stay
off with KVM unless requested explicitly.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230428095533.21747-2-cohuck@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Maintainers: add myself as reviewer for sbsa-ref

At Linaro I work on sbsa-ref, know direction it goes.

May not get code details each time.

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230515143753.365591-1-marcin.juszkiewicz@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Fix vd == vm overlap in sve_ldff1_z

If vd == vm, copy vm to scratch, so that we can pre-zero
the output and still access the gather indicies.

Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1612
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230504104232.1877774-1-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sbsa-ref: switch default cpu core to Neoverse-N1

The world outside moves to newer and newer cpu cores. Let move SBSA
Reference Platform to something newer as well.

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Reviewed-by: Leif Lindholm <quic_llindhol@quicinc.com>
Message-id: 20230506183417.1360427-1-marcin.juszkiewicz@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

qapi/parser: Drop two bad type hints for now

Two type hints fail centos-stream-8-x86_64 CI.  They are actually
broken.  Changing them to Optional[re.Match[str]] fixes them locally
for me, but then CI fails differently.  Drop them for now.

Fixes: 3e32dca3f0d1 (qapi: Rewrite parsing of doc comment section symbols and tags)
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20230517061600.1782455-1-armbru@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Merge tag 'linux-user-for-8.1-pull-request' of https://github.com/vivier/qemu into staging

linux-user pull request 20230512-v4

add open_tree(), move_mount()
add /proc/cpuinfo for riscv
fixes and cleanup

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEEzS913cjjpNwuT1Fz8ww4vT8vvjwFAmRkiZISHGxhdXJlbnRA
# dml2aWVyLmV1AAoJEPMMOL0/L748FdIP/RC1JaCftkP7ajAstNbZLMLegMxjUYHV
# TrdhsMOsm804ZmLgTqqfS3bJ080mIHup0xUnHBckcEtUcwaz54cJ1BAR2WlM3/8A
# t3fHMt3PDkh3OPd/3AnmpLE8XRh7yBztirPYfZc6SKqnFzT0TZrwBoQnwprEnZ5r
# c0gbrgLZLunZhrWU1BbQmuIufW1qDoQo4PzwnyZeux1fHA1/v/dx3wgSLpv3V4k6
# x0Kj8TvtMUU4/io2RqYF4jKopfhwsh0jnr9rlOmydOExalKq1VbRptJI2UC4KVOY
# MZuApF1EaZfrW+v/WSlvmzaZ/zRzP1L0X3Xh0wB4J9Rj3057/elXr6bi+R+rM46p
# xGTcti9ahWKP2J4/xrazRw2lfPsLcw/YbqVGG79AX1xLJPCiWq6lamzc/g3ptFnx
# F/RRETe65z7apzF/nzU7SDOsMdN5p4/fMb1SysLuAov5OepNVjNVWyiTgqOHB5uC
# ye+lOYkkvk+qRdMbls/fIcjDQ3C4AjoBWj4QlgRc0/Qf6ac4TkVjzPa70Y6eyzzS
# LEV9D4fXD8EZgYWENNGmbbKPNbtfqc9uR6gXdgkEsKDx/rf5IMf1d6r1C99dhB3A
# nbu0JpFKKY2lhD2oGVPDE3UQMW9DXXhZpDApUBsLNiEwfuoXZee+apH+6jc8tbn6
# r+8LFB1mM9os
# =NfIV
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 17 May 2023 01:00:18 AM PDT
# gpg:                using RSA key CD2F75DDC8E3A4DC2E4F5173F30C38BD3F2FBE3C
# gpg:                issuer "laurent@vivier.eu"
# gpg: Good signature from "Laurent Vivier <lvivier@redhat.com>" [undefined]
# gpg:                 aka "Laurent Vivier <laurent@vivier.eu>" [undefined]
# gpg:                 aka "Laurent Vivier (Red Hat) <lvivier@redhat.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: CD2F 75DD C8E3 A4DC 2E4F  5173 F30C 38BD 3F2F BE3C

* tag 'linux-user-for-8.1-pull-request' of https://github.com/vivier/qemu:
  linux-user: fix getgroups/setgroups allocations
  linux-user: Fix mips fp64 executables loading
  linux-user: Don't require PROT_READ for mincore
  linux-user: Add new flag VERIFY_NONE
  linux-user/main: Use list_cpus() instead of cpu_list()
  linux-user: Add open_tree() syscall
  linux-user: Add move_mount() syscall
  linux-user: report ENOTTY for unknown ioctls
  linux-user: Emulate /proc/cpuinfo output for riscv

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

linux-user: fix getgroups/setgroups allocations

linux-user getgroups(), setgroups(), getgroups32() and setgroups32()
used alloca() to allocate grouplist arrays, with unchecked gidsetsize
coming from the "guest".  With NGROUPS_MAX being 65536 (linux, and it
is common for an application to allocate NGROUPS_MAX for getgroups()),
this means a typical allocation is half the megabyte on the stack.
Which just overflows stack, which leads to immediate SIGSEGV in actual
system getgroups() implementation.

An example of such issue is aptitude, eg
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=811087#72

Cap gidsetsize to NGROUPS_MAX (return EINVAL if it is larger than that),
and use heap allocation for grouplist instead of alloca().  While at it,
fix coding style and make all 4 implementations identical.

Try to not impose random limits - for example, allow gidsetsize to be
negative for getgroups() - just do not allocate negative-sized grouplist
in this case but still do actual getgroups() call.  But do not allow
negative gidsetsize for setgroups() since its argument is unsigned.

Capping by NGROUPS_MAX seems a bit arbitrary, - we can do more, it is
not an error if set size will be NGROUPS_MAX+1. But we should not allow
integer overflow for the array being allocated. Maybe it is enough to
just call g_try_new() and return ENOMEM if it fails.

Maybe there's also no need to convert setgroups() since this one is
usually smaller and known beforehand (KERN_NGROUPS_MAX is actually 63, -
this is apparently a kernel-imposed limit for runtime group set).

The patch fixes aptitude segfault mentioned above.

Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Message-Id: <20230409105327.1273372-1-mjt@msgid.tls.msk.ru>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

linux-user: Fix mips fp64 executables loading

If a program requires fr1, we should set the FR bit of CP0 control status
register and add F64 hardware flag. The corresponding `else if` branch
statement is copied from the linux kernel sources (see `arch_check_elf` function
in linux/arch/mips/kernel/elf.c).

Signed-off-by: Daniil Kovalev <dkovalev@compiler-toolchain-for.me>
Reviewed-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Message-Id: <20230404052153.16617-1-dkovalev@compiler-toolchain-for.me>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

linux-user: Don't require PROT_READ for mincore

The kernel does not require PROT_READ for addresses passed to mincore.
For example the fincore(1) tool from util-linux uses PROT_NONE and
currently does not work under qemu-user.

Example (with fincore(1) from util-linux 2.38):

$ fincore /proc/self/exe
RES PAGES  SIZE FILE
24K     6 22.1K /proc/self/exe

$ qemu-x86_64 /usr/bin/fincore /proc/self/exe
fincore: failed to do mincore: /proc/self/exe: Cannot allocate memory

With this patch:

$ ./build/qemu-x86_64 /usr/bin/fincore /proc/self/exe
RES PAGES  SIZE FILE
24K     6 22.1K /proc/self/exe

Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20230422100314.1650-3-thomas@t-8ch.de>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

linux-user: Add new flag VERIFY_NONE

This can be used to validate that an address range is mapped but without
being readable or writable.

It will be used by an updated implementation of mincore().

Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20230422100314.1650-2-thomas@t-8ch.de>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

linux-user/main: Use list_cpus() instead of cpu_list()

This way we can get rid of the if'deffery and the XXX comment
here (it's repeated in the list_cpus() function anyway).

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230424122126.236586-1-thuth@redhat.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

linux-user: Add open_tree() syscall

Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20230424153429.276788-2-thomas@t-8ch.de>
[lv: move declaration at the beginning of the block,
     define syscall]
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

linux-user: Add move_mount() syscall

Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
[lv: define syscall]
Message-Id: <20230424153429.276788-1-thomas@t-8ch.de>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

linux-user: report ENOTTY for unknown ioctls

The correct error number for unknown ioctls is ENOTTY.

ENOSYS would mean that the ioctl() syscall itself is not implemented,
which is very improbable and unexpected for userspace.

ENOTTY means "Inappropriate ioctl for device". This is what the kernel
returns on unknown ioctls, what qemu is trying to express and what
userspace is prepared to handle.

Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230426070659.80649-1-thomas@t-8ch.de>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

linux-user: Emulate /proc/cpuinfo output for riscv

RISC-V does not expose all extensions via hwcaps, thus some userspace
applications may want to query these via /proc/cpuinfo.

Currently when querying this file the host's file is shown instead
which is slightly confusing. Emulate a basic /proc/cpuinfo file
with mmu info and an ISA string.

Signed-off-by: Afonso Bordado <afonsobordado@gmail.com>
Reviewed-by: Palmer Dabbelt <palmer@rivosinc.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Message-Id: <167873059442.9885.15152085316575248452-0@git.sr.ht>
[lv: removed the test that fails in CI for unknown reason]
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

Merge tag 'pull-tcg-20230516-3' of https://gitlab.com/rth7680/qemu into staging

tcg/i386: Fix tcg_out_addi_ptr for win64
tcg: Implement atomicity for TCGv_i128
tcg: First quarter of cleanups for building tcg once

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmRkWC8dHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV/I+wf8CUF+J/E9u0EuurrB
# 1asDicANUJIACnqlcEpSPKuSMtbzo1RDTQUR+d3GWJjyLASvSJZFZTQqWBdACRpc
# sNuDz3/1a6FbiM14CwIVmPpcjQXa+18Ck670Chmw51KyEt2xyDJTySFIGEqjiuTf
# YVDBbOs8neFZdcDvAs1qNUTjhRj4nNtkpQoBpv0tGH7E0CzPp6OcvxwfieVyLOIa
# Cy1ELM3aMyVN5MTjnORYLK70Pa9emdjB88SlypZx363ARKC7B50lzYPQ4E5zrOZq
# FKrOq5nFWLCtn4BID0R+jUmuUP6znR/hTlToDmf/9B4j9TUivERWlc54lz3YU6Gn
# su3FKg==
# =LVOb
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 16 May 2023 09:29:35 PM PDT
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [ultimate]

* tag 'pull-tcg-20230516-3' of https://gitlab.com/rth7680/qemu: (74 commits)
  tcg: Split out exec/user/guest-base.h
  tcg: Add tlb_dyn_max_bits to TCGContext
  tcg: Add page_bits and page_mask to TCGContext
  tcg: Remove TARGET_LONG_BITS, TCG_TYPE_TL
  tcg/mips: Remove TARGET_LONG_BITS, TCG_TYPE_TL
  tcg/loongarch64: Remove TARGET_LONG_BITS, TCG_TYPE_TL
  tcg/aarch64: Remove TARGET_LONG_BITS, TCG_TYPE_TL
  tcg/aarch64: Remove USE_GUEST_BASE
  tcg/arm: Remove TARGET_LONG_BITS
  tcg/i386: Remove TARGET_LONG_BITS, TCG_TYPE_TL
  tcg/i386: Adjust type of tlb_mask
  tcg/i386: Conditionalize tcg_out_extu_i32_i64
  tcg/i386: Always enable TCG_TARGET_HAS_extr[lh]_i64_i32
  tcg/tci: Elimnate TARGET_LONG_BITS, target_ulong
  tcg: Split INDEX_op_qemu_{ld,st}* for guest address size
  tcg: Remove TCGv from tcg_gen_atomic_*
  tcg: Remove TCGv from tcg_gen_qemu_{ld,st}_*
  tcg: Add addr_type to TCGContext
  accel/tcg: Widen plugin_gen_empty_mem_callback to i64
  tcg: Reduce copies for plugin_gen_mem_callbacks
  ...

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

tcg: Split out exec/user/guest-base.h

TCG will need this declaration, without all of the other
bits that come with cpu-all.h.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Add tlb_dyn_max_bits to TCGContext

Disconnect guest tlb parameters from TCG compilation.

Reviewed-by: Anton Johansson <anjo@rev.ng>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Add page_bits and page_mask to TCGContext

Disconnect guest page size from TCG compilation.
While this could be done via exec/target_page.h, we want to cache
the value across multiple memory access operations, so we might
as well initialize this early.

The changes within tcg/ are entirely mechanical:

    sed -i s/TARGET_PAGE_BITS/s->page_bits/g
    sed -i s/TARGET_PAGE_MASK/s->page_mask/g

Reviewed-by: Anton Johansson <anjo@rev.ng>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Remove TARGET_LONG_BITS, TCG_TYPE_TL

All uses replaced with TCGContext.addr_type.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/mips: Remove TARGET_LONG_BITS, TCG_TYPE_TL

All uses replaced with TCGContext.addr_type.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/loongarch64: Remove TARGET_LONG_BITS, TCG_TYPE_TL

All uses replaced with TCGContext.addr_type.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/aarch64: Remove TARGET_LONG_BITS, TCG_TYPE_TL

All uses replaced with TCGContext.addr_type.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/aarch64: Remove USE_GUEST_BASE

Eliminate the test vs TARGET_LONG_BITS by considering this
predicate to be always true, and simplify accordingly.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/arm: Remove TARGET_LONG_BITS

All uses can be infered from the INDEX_op_qemu_*_a{32,64}_*
opcode being used.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/i386: Remove TARGET_LONG_BITS, TCG_TYPE_TL

All uses can be infered from the INDEX_op_qemu_*_a{32,64}_* opcode
being used.  Add a field into TCGLabelQemuLdst to record the usage.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/i386: Adjust type of tlb_mask

Because of its use on tgen_arithi, this value must be a signed
32-bit quantity, as that is what may be encoded in the insn.
The truncation of the value to unsigned for 32-bit guests is
done via the REX bit via 'trexw'.

Removes the only uses of target_ulong from this tcg backend.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/i386: Conditionalize tcg_out_extu_i32_i64

Since TCG_TYPE_I32 values are kept zero-extended in registers, via
omission of the REXW bit, we need not extend if the register matches.
This is already relied upon by qemu_{ld,st}.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/i386: Always enable TCG_TARGET_HAS_extr[lh]_i64_i32

Keep all 32-bit values zero extended in the register, not solely when
addresses are 32 bits.  This eliminates a dependency on TARGET_LONG_BITS.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/tci: Elimnate TARGET_LONG_BITS, target_ulong

We now have the address size as part of the opcode, so
we no longer need to test TARGET_LONG_BITS.  We can use
uint64_t for target_ulong, as passed into load/store helpers.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Split INDEX_op_qemu_{ld,st}* for guest address size

For 32-bit hosts, we cannot simply rely on TCGContext.addr_bits,
as we need one or two host registers to represent the guest address.

Create the new opcodes and update all users.  Since we have not
yet eliminated TARGET_LONG_BITS, only one of the two opcodes will
ever be used, so we can get away with treating them the same in
the backends.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Remove TCGv from tcg_gen_atomic_*

Expand from TCGv to TCGTemp inline in the translators,
and validate that the size matches tcg_ctx->addr_type.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Remove TCGv from tcg_gen_qemu_{ld,st}_*

Expand from TCGv to TCGTemp inline in the translators,
and validate that the size matches tcg_ctx->addr_type.
These inlines will eventually be seen only by target-specific code.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Add addr_type to TCGContext

This will enable replacement of TARGET_LONG_BITS within tcg/.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Widen plugin_gen_empty_mem_callback to i64

Since we do this inside gen_empty_mem_cb anyway, let's
do this earlier inside tcg expansion.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Reduce copies for plugin_gen_mem_callbacks

We only need to make copies for loads, when the destination
overlaps the address.  For now, only eliminate the copy for
stores and 128-bit loads.

Rename plugin_prep_mem_callbacks to plugin_maybe_preserve_addr,
returning NULL if no copy is made.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Merge do_gen_mem_cb into caller

As do_gen_mem_cb is called once, merge it into gen_empty_mem_cb.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Merge gen_mem_wrapped with plugin_gen_empty_mem_callback

As gen_mem_wrapped is only used in plugin_gen_empty_mem_callback,
we can avoid the curiosity of union mem_gen_fn by inlining it.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Widen tcg_gen_code pc_start argument to uint64_t

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Widen helper_atomic_* addresses to uint64_t

Always pass the target address as uint64_t.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Widen helper_{ld,st}_i128 addresses to uint64_t

Always pass the target address as uint64_t.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Widen tcg-ldst.h addresses to uint64_t

Always pass the target address as uint64_t.
Adjust tcg_out_{ld,st}_helper_args to match.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Widen gen_insn_data to uint64_t

We already pass uint64_t to restore_state_to_opc; this changes all
of the other uses from insn_start through the encoding to decoding.

Reviewed-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Split out memory ops to tcg-op-ldst.c

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/sparc64: Use atom_and_align_for_opc

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

tcg/s390x: Use atom_and_align_for_opc

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

tcg/riscv: Use atom_and_align_for_opc

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

tcg/ppc: Use atom_and_align_for_opc

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

tcg/mips: Use atom_and_align_for_opc

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

tcg/loongarch64: Use atom_and_align_for_opc

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

tcg/arm: Use atom_and_align_for_opc

No change to the ultimate load/store routines yet, so some atomicity
conditions not yet honored, but plumbs the change to alignment through
the relevant functions.

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

tcg/aarch64: Use atom_and_align_for_opc

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

tcg/i386: Use atom_and_align_for_opc

No change to the ultimate load/store routines yet, so some atomicity
conditions not yet honored, but plumbs the change to alignment through
the relevant functions.

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

tcg: Introduce atom_and_align_for_opc

Examine MemOp for atomicity and alignment, adjusting alignment
as required to implement atomicity on the host.

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

tcg: Support TCG_TYPE_I128 in tcg_out_{ld,st}_helper_{args,ret}

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

tcg: Merge tcg_out_helper_load_regs into caller

Now that tcg_out_helper_load_regs is not recursive, we can
merge it into its only caller, tcg_out_helper_load_slots.

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

tcg: Introduce tcg_out_movext3

With x86_64 as host, we do not have any temporaries with which to
resolve cycles, but we do have xchg.   As a side bonus, the set of
graphs that can be made with 3 nodes and all nodes conflicting is
small: two.  We can solve the cycle with a single temp.

This is required for x86_64 to handle stores of i128: 1 address
register and 2 data registers.

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

tcg: Add INDEX_op_qemu_{ld,st}_i128

Add opcodes for backend support for 128-bit memory operations.

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

tcg: Introduce tcg_target_has_memory_bswap

Replace the unparameterized TCG_TARGET_HAS_MEMORY_BSWAP macro
with a function with a memop argument.

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

tcg/riscv: Support softmmu unaligned accesses

The system is required to emulate unaligned accesses, even if the
hardware does not support it.  The resulting trap may or may not
be more efficient than the qemu slow path.  There are linux kernel
patches in flight to allow userspace to query hardware support;
we can re-evaluate whether to enable this by default after that.

In the meantime, softmmu now matches useronly, where we already
assumed that unaligned accesses are supported.

Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/loongarch64: Support softmmu unaligned accesses

Test the final byte of an unaligned access.
Use BSTRINS.D to clear the range of bits, rather than AND.

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

tcg/loongarch64: Check the host supports unaligned accesses

This should be true of all loongarch64 running Linux.

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

accel/tcg: Remove helper_unaligned_{ld,st}

These functions are now unused.

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

tcg/sparc64: Use standard slow path for softmmu

Drop the target-specific trampolines for the standard slow path.
This lets us use tcg_out_helper_{ld,st}_args, and handles the new
atomicity bits within MemOp.

At the same time, use the full load/store helpers for user-only mode.
Drop inline unaligned access support for user-only mode, as it does
not handle atomicity.

Use TCG_REG_T[1-3] in the tlb lookup, instead of TCG_REG_O[0-2].
This allows the constraints to be simplified.

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

tcg/sparc64: Split out tcg_out_movi_s32

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

tcg/sparc64: Rename tcg_out_movi_imm32 to tcg_out_movi_u32

Emphasize that the constant is unsigned.

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

target/sparc64: Remove tcg_out_movi_s13 case from tcg_out_movi_imm32

Shuffle the order in tcg_out_movi_int to check s13 first, and
drop this check from tcg_out_movi_imm32.  This might make the
sequence for in_prologue larger, but not worth worrying about.

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

tcg/sparc64: Rename tcg_out_movi_imm13 to tcg_out_movi_s13

Emphasize that the constant is signed.

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

tcg/sparc64: Allocate %g2 as a third temporary

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

tcg/s390x: Use full load/store helpers in user-only mode

Instead of using helper_unaligned_{ld,st}, use the full load/store helpers.
This will allow the fast path to increase alignment to implement atomicity
while not immediately raising an alignment exception.

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

tcg/mips: Use full load/store helpers in user-only mode

Instead of using helper_unaligned_{ld,st}, use the full load/store helpers.
This will allow the fast path to increase alignment to implement atomicity
while not immediately raising an alignment exception.

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

tcg/arm: Use full load/store helpers in user-only mode

Instead of using helper_unaligned_{ld,st}, use the full load/store helpers.
This will allow the fast path to increase alignment to implement atomicity
while not immediately raising an alignment exception.

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

tcg/arm: Adjust constraints on qemu_ld/st

Always reserve r3 for tlb softmmu lookup.  Fix a bug in user-only
ALL_QLDST_REGS, in that r14 is clobbered by the BLNE that leads
to the misaligned trap.  Remove r0+r1 from user-only ALL_QLDST_REGS;
I believe these had been reserved for bswap, which we no longer
perform during qemu_st.

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

tcg/riscv: Use full load/store helpers in user-only mode

Instead of using helper_unaligned_{ld,st}, use the full load/store helpers.
This will allow the fast path to increase alignment to implement atomicity
while not immediately raising an alignment exception.

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

tcg/loongarch64: Use full load/store helpers in user-only mode

Instead of using helper_unaligned_{ld,st}, use the full load/store helpers.
This will allow the fast path to increase alignment to implement atomicity
while not immediately raising an alignment exception.

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

tcg/ppc: Use full load/store helpers in user-only mode

Instead of using helper_unaligned_{ld,st}, use the full load/store helpers.
This will allow the fast path to increase alignment to implement atomicity
while not immediately raising an alignment exception.

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

tcg/aarch64: Use full load/store helpers in user-only mode

Instead of using helper_unaligned_{ld,st}, use the full load/store helpers.
This will allow the fast path to increase alignment to implement atomicity
while not immediately raising an alignment exception.

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

tcg/i386: Use full load/store helpers in user-only mode

Instead of using helper_unaligned_{ld,st}, use the full load/store helpers.
This will allow the fast path to increase alignment to implement atomicity
while not immediately raising an alignment exception.

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