tcg/optimize: Split out arg_new_constant

Fixes a bug wherein raw uses of tcg_constant_internal
do not have their TempOptInfo initialized.

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

tcg: Eliminate duplicate env store operations

Notice when a constant is stored to the same location twice.

Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/optimize: Optimize env memory operations

Propagate stores to loads, loads to loads.

Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/optimize: Split out cmp_better_copy

Compare two temps for "better", split out from finding
the best from a whole list.  Use TCGKind, which already
gives the proper priority.

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

tcg/optimize: Pipe OptContext into reset_ts

Will be needed in the next patch.

Reviewed-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Don't free vector results

Avoid reusing vector temporaries so that we may re-use them
when propagating stores to loads.

Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Remove TCG_TARGET_HAS_neg_{i32,i64}

The movcond opcode is now mandatory for backends to implement.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-7-richard.henderson@linaro.org>

tcg/loongarch64: Implement neg opcodes

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-6-richard.henderson@linaro.org>

tcg/mips: Implement neg opcodes

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-5-richard.henderson@linaro.org>

tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64}

The movcond opcode is now mandatory for backends to implement.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-4-richard.henderson@linaro.org>

tcg/mips: Always implement movcond

Expand as branch over move if not supported in the ISA.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-3-richard.henderson@linaro.org>

tcg/mips: Split out tcg_out_setcond_int

Return the temp and a set of flags, to be used as a
primitive for setcond, brcond, movcond.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-2-richard.henderson@linaro.org>

tcg: Move tcg_temp_free_* out of line

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-12-richard.henderson@linaro.org>

tcg: Move tcg_temp_new_*, tcg_global_mem_new_* out of line

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-11-richard.henderson@linaro.org>

tcg: Move tcg_constant_* out of line

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-10-richard.henderson@linaro.org>

tcg: Unexport tcg_gen_op*_{i32,i64}

These functions are no longer used outside tcg-op.c.
There are several that are completely unused, so remove them.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-9-richard.henderson@linaro.org>

tcg: Move tcg_gen_opN declarations to tcg-internal.h

These are used within tcg-op.c and tcg-op-ldst.c.
There are no uses outside tcg/.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-8-richard.henderson@linaro.org>

tcg: Move vec_gen_* declarations to tcg-internal.h

These are used within tcg-op-vec.c and tcg/host/tcg-target.c.inc.
There are no uses outside tcg/.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-7-richard.henderson@linaro.org>

tcg: Move 64-bit expanders out of line

This one is more complicated, combining 32-bit and 64-bit
expansion with C if instead of preprocessor #if.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-6-richard.henderson@linaro.org>

tcg: Move 32-bit expanders out of line

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-5-richard.henderson@linaro.org>

tcg: Move generic expanders out of line

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-4-richard.henderson@linaro.org>

tcg: Move tcg_gen_op* out of line

In addition to moving out of line, with CONFIG_DEBUG_TCG
mark them all noinline.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-3-richard.henderson@linaro.org>

tcg: Mark tcg_gen_op* as noinline

Encourage the compiler to tail-call rather than inline
across the dozens of opcode expanders.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-2-richard.henderson@linaro.org>

accel/tcg: Fix condition for store_atom_insert_al16

Store bytes under a mask is fundamentally a cmpxchg, not a straight store.
Use HAVE_CMPXCHG128 instead of HAVE_ATOMIC128_RW.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230916220151.526140-8-richard.henderson@linaro.org>

accel/tcg: Remove redundant case in store_atom_16

We handled the HAVE_ATOMIC128_RW case with atomic16_set at the top of
the function; the only thing left for a host without that support is
to fall through to cpu_loop_exit_atomic.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230916220151.526140-7-richard.henderson@linaro.org>

host/include/loongarch64: Add atomic16 load and store

While loongarch64 does not have a 128-bit cmpxchg, it does
have 128-bit atomic load and store via the vector unit.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230916220151.526140-6-richard.henderson@linaro.org>

tcg/loongarch64: Use cpuinfo.h

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiajie Chen <c@jia.je>
Message-Id: <20230916220151.526140-5-richard.henderson@linaro.org>

util: Add cpuinfo for loongarch64

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiajie Chen <c@jia.je>
Message-Id: <20230916220151.526140-4-richard.henderson@linaro.org>

tcg/loongarch64: Use C_N2_I1 for INDEX_op_qemu_ld_a*_i128

Use new registers for the output, so that we never overlap
the input address, which could happen for user-only.
This avoids a "tmp = addr + 0" in that case.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiajie Chen <c@jia.je>
Message-Id: <20230916220151.526140-3-richard.henderson@linaro.org>

tcg: Add C_N2_I1

Constraint with two outputs, both in new registers.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiajie Chen <c@jia.je>
Message-Id: <20230916220151.526140-2-richard.henderson@linaro.org>

accel/tcg: Move HMP info jit and info opcount code

Move all of it into accel/tcg/monitor.c.  This puts everything
about tcg that is only used by the monitor in the same place.

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

Merge tag 'pull-sp-20231105' of https://gitlab.com/rth7680/qemu into staging

target/sparc: Explicitly compute condition codes

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmVH9oodHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV/M8QgAgPTp/wFLVnSRFLaN
# fBoelVhM4WTWMQ+SUwZMtCvqcMHaBxIMu+hyk5MI11hFOUi9N+vWvRb+NZ6JbK+1
# sqWcx0NdYfNdOeoi1dgzGgcCkFA8u9zW/K7Ih0W8WuU20uiJ4Zw/qmnEELIl/mZR
# 5Ft1mhLMhQSYsH0KSypugLWBxR9SFNH1cV3C1SG2q+6snm/mhKk9NN18zJGFdmmY
# 4CQThx159P/DaPUONZbSAMN94opu6K8FSymELPDUZBYwJRq7fyGKYuDUGRvN1kxx
# I8p/MF1V5Vcth9lvGyBYulFWjo9BDMpkIdmWzXZLOWfzZVAed8PcglxoQqgMbU5u
# eyY/Cw==
# =Tv1h
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 06 Nov 2023 04:09:46 HKT
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* tag 'pull-sp-20231105' of https://gitlab.com/rth7680/qemu: (21 commits)
  target/sparc: Check for invalid cond in gen_compare_reg
  target/sparc: Implement UDIV inline
  target/sparc: Implement UDIVX and SDIVX inline
  target/sparc: Discard cpu_cond at the end of each insn
  target/sparc: Record entire jump condition in DisasContext
  target/sparc: Merge gen_op_next_insn into only caller
  target/sparc: Pass displacement to advance_jump_cond
  target/sparc: Merge advance_jump_uncond_{never,always} into advance_jump_cond
  target/sparc: Merge gen_branch2 into advance_pc
  target/sparc: Do flush_cond in advance_jump_cond
  target/sparc: Always copy conditions into a new temporary
  target/sparc: Change DisasCompare.c2 to int
  target/sparc: Remove DisasCompare.is_bool
  target/sparc: Remove CC_OP leftovers
  target/sparc: Remove CC_OP_TADDTV, CC_OP_TSUBTV
  target/sparc: Remove CC_OP_SUB, CC_OP_SUBX, CC_OP_TSUB
  target/sparc: Remove CC_OP_ADD, CC_OP_ADDX, CC_OP_TADD
  target/sparc: Remove CC_OP_DIV
  target/sparc: Remove CC_OP_LOGIC
  target/sparc: Split psr and xcc into components
  ...

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'migration-20231103-pull-request' of https://gitlab.com/juan.quintela/qemu into staging

Migration Pull request (20231103)

Hi

In this PULL:
- dirty limit fixes (hyman)
- coverity issues (juan)

Please apply.

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEGJn/jt6/WMzuA0uC9IfvGFhy1yMFAmVE4dgACgkQ9IfvGFhy
# 1yPBgxAAvrsCHwU6/m9y+XGokyHTKKKIVysLip/14jAjDL+viLYgxdVyOGQKQGBa
# +yV+XHTaEyKdihKG4Z5nWuC0yM+sdZQpWXQAcHJLPaPs5wDGICICpdAFY2LbAWSK
# jtX9uq7crywIL4mVKiX+HOjRUPCAYUx/2TcqJf2+0+MKDEVC33ikxNbcx8ZELY+Q
# +hGyOws3mkHSQjyaNUVgnnQtGzikYqcNO2efa+zVPdXYd+TUWW2e9I++Qf48r0Hv
# OqeZAB7bSAb39PNRuj0I1gt4d3WTHzHt7BSpX1OuFqQnzLw8vS5iDQH943WAyGkY
# NblZVb8pyzSg1Jy18H/SmrJDXeufRwqFwD+1NHyxGjsF89KOuVUqGrGpRXhMBtmA
# DSzdgn5jqW5lI1po9FqGdlPTFlhstpMH3DSfPQWurvJh42oM38gmSEHLBNpc4tXo
# 8udMYI09H/kHUoNMTZNGjnZO9LfarGsag6eOJP1bMMublhRlKCaL9RIyV9oOHycE
# IeOeQFeBP/BmYFLWbVPeZej7uiqsEc7VPDJK2QXns210UYanaWmggkmpdAr0I0EV
# pEKHSfVv1qlIlFH4d7MhcJzP2/rY62EC5tYQjT0UaBnCRcDInKrNWa3kbDL0akwr
# 0aJgpbT5ipknVChtwnMWJlbqpeW/VUF5g0jVpYQ3jbe/Zf+OtmU=
# =Pv8z
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 03 Nov 2023 20:04:40 HKT
# gpg:                using RSA key 1899FF8EDEBF58CCEE034B82F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>" [full]
# gpg:                 aka "Juan Quintela <quintela@trasno.org>" [full]
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03  4B82 F487 EF18 5872 D723

* tag 'migration-20231103-pull-request' of https://gitlab.com/juan.quintela/qemu:
  migration: Unlock mutex in error case
  docs/migration: Add the dirty limit section
  tests/migration: Introduce dirty-limit into guestperf
  tests/migration: Introduce dirty-ring-size option into guestperf
  tests: Add migration dirty-limit capability test
  system/dirtylimit: Drop the reduplicative check
  system/dirtylimit: Fix a race situation

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'dump-pull-request' of https://gitlab.com/marcandre.lureau/qemu into staging

dump queue

Hi

The "dump" queue, with:
- [PATCH v3 qemu 0/3] Allow dump-guest-memory to output standard kdump format
- [PATCH v2 0/5] dump: Minor fixes & improvements

# -----BEGIN PGP SIGNATURE-----
#
# iQJQBAABCAA6FiEEh6m9kz+HxgbSdvYt2ujhCXWWnOUFAmVEmsEcHG1hcmNhbmRy
# ZS5sdXJlYXVAcmVkaGF0LmNvbQAKCRDa6OEJdZac5YNkD/sFnz+I75mn6+DIdC3x
# aSVUU87JxAvWkt+G3KYGS+de2+g2YkRkPwwrIsIceiX7mlL4Es350AVcTl7+fXpu
# Jl9k9I32QI+U3pNXo9BStIqjKUMBxmmKs4aLCh9OHJ6oliTCG+aJTUmSl/dABIuw
# fAcW9vjhyR4ogAp8x7WhR6PKEAAb6OE/9k0w/z0GV2K09N/R0pPAvObQ36VQJ/Cl
# 6DN8tRRytl0IQmC/mZZ+MQPQ5cvamK78X3DmnYCGtyN9HTQERfUFMSSgD/sHLvNi
# rMKuwhXiGQfDs/xQ9Z6Vh2AL7JfAwbIQwUstepb78M/5GBLaZfwFYG4+eCohJE82
# s0GOQ45Yks+AOTGj6lNyOfJ8PIf0SocCTbnLWZicpdHIfoEkSmmL0VZ5w+w0EpDO
# WOZJRpANJGTLhKNb//X3A3OJ05LoavN3/criokhC19DW/yE/VEGd3dXlP6yvFOku
# vGUINGivg1bw7yO0S/rzXNw4+cHCPgBCXbKCNuMI6B+dxL5pUR5Zr4OqcYgwejqE
# RWMdqsHA4ohpzc3AfbuHLFilXJNAgLR3jAEiVUXyrz9U1FiYEiq/8RNuupe9Uveq
# pO1PDZ9fher0Zda4y28bHl/e5M9hVeCFqElcVk0FQGt97T5olVvSaL/hFUPf65ls
# 8A3lN6WaAT9dvM33pkeswZvGxg==
# =eSbp
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 03 Nov 2023 15:01:21 HKT
# gpg:                using RSA key 87A9BD933F87C606D276F62DDAE8E10975969CE5
# gpg:                issuer "marcandre.lureau@redhat.com"
# gpg: Good signature from "Marc-André Lureau <marcandre.lureau@redhat.com>" [full]
# gpg:                 aka "Marc-André Lureau <marcandre.lureau@gmail.com>" [full]
# Primary key fingerprint: 87A9 BD93 3F87 C606 D276  F62D DAE8 E109 7596 9CE5

* tag 'dump-pull-request' of https://gitlab.com/marcandre.lureau/qemu:
  dump: Drop redundant check for empty dump
  dump: Improve some dump-guest-memory error messages
  dump: Recognize "fd:" protocols on Windows hosts
  dump: Fix g_array_unref(NULL) in dump-guest-memory
  dump: Rename qmp_dump_guest_memory() parameter to match QAPI schema
  dump: Add command interface for kdump-raw formats
  dump: Allow directly outputting raw kdump format
  dump: Pass DumpState to write_ functions

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'pull-loongarch-20231103' of https://gitlab.com/gaosong/qemu into staging

pull-loongarch-20231103

# -----BEGIN PGP SIGNATURE-----
#
# iLMEAAEKAB0WIQS4/x2g0v3LLaCcbCxAov/yOSY+3wUCZUSQIgAKCRBAov/yOSY+
# 31aIBADj5FzdUxyFB813SouAiEiyMdI4bN98AunomAk3Kt8PF1XPoP8kPzcjxcMI
# kCW4eoHb12MVs9OclkqFY3VyaxtSD3YSG/h8W9YxaDyU+L/q89RS+J4r6CAZ8ylg
# J4uxs3Lv8nwPEvRb4zITAt8JQllLey1100j/uu4fU0Rx7vUcMA==
# =9RMx
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 03 Nov 2023 14:16:02 HKT
# 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-20231103' of https://gitlab.com/gaosong/qemu:
  linux-user/loongarch64: Add LASX sigcontext save/restore
  linux-user/loongarch64: Add LSX sigcontext save/restore
  linux-user/loongarch64: Use abi_{ulong,uint} types
  linux-user/loongarch64: setup_sigframe() set 'end' context size 0
  linux-user/loongarch64: Fix setup_extcontext alloc wrong fpu_context size
  linux-user/loongarch64: Use traps to track LSX/LASX usage
  target/loongarch: Support 4K page size
  target/loongarch: Implement query-cpu-model-expansion
  target/loongarch: Allow user enable/disable LSX/LASX features
  target/loongarch: Add cpu model 'max'

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

target/sparc: Check for invalid cond in gen_compare_reg

Consolidate the test here; drop the "inverted logic".
Fix MOVr and FMOVR, which were missing the invalid test.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Implement UDIV inline

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Implement UDIVX and SDIVX inline

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Discard cpu_cond at the end of each insn

If the insn raises no exceptions, there will be no path in which
cpu_cond is used, and so the computation may be optimized away.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Record entire jump condition in DisasContext

Use the original condition instead of consuming cpu_cond,
which will now only be live along exception paths.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Merge gen_op_next_insn into only caller

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Pass displacement to advance_jump_cond

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Merge advance_jump_uncond_{never,always} into advance_jump_cond

Handle these via TCG_COND_{ALWAYS,NEVER}.
Allow dc->npc to be variable, using gen_mov_pc_npc.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Merge gen_branch2 into advance_pc

The function had only one caller.  Canonicalize the cpu_cond
test to TCG_COND_NE, the "natural" sense of its value.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Do flush_cond in advance_jump_cond

Do this here instead of in each caller.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Always copy conditions into a new temporary

This will allow the condition to live across changes to
the global cc variables.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Change DisasCompare.c2 to int

We don't require c2 to be variable, so emphasize that.
We don't currently require c2 to be non-zero, but that will change.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Remove DisasCompare.is_bool

Since we're going to feed cpu_cond to another comparison, we don't
reqire a boolean value -- anything non-zero is sufficient.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Remove CC_OP leftovers

All instructions have been converted to generate
full condition codes explicitly.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Remove CC_OP_TADDTV, CC_OP_TSUBTV

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Remove CC_OP_SUB, CC_OP_SUBX, CC_OP_TSUB

These are all related and implementable with common code.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Remove CC_OP_ADD, CC_OP_ADDX, CC_OP_TADD

These are all related and implementable with common code.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Remove CC_OP_DIV

Return both result and overflow from helper_[us]div.
Compute all flags explicitly in gen_op_[us]divcc.

Marginally improve the INT64_MIN special case in helper_sdiv.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Remove CC_OP_LOGIC

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Split psr and xcc into components

Step in removing CC_OP: change the representation of CC_OP_FLAGS.
The 8 bits are distributed between 6 variables, which should make
it easy to keep up to date.

The code within cc_helper.c is quite ugly but is only temporary.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/sparc: Introduce cpu_put_psr_icc

Isolate linux-user from changes to icc representation.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

migration: Unlock mutex in error case

We were not unlocking bitmap mutex on the error case.  To fix it
forever change to enclose the code with WITH_QEMU_LOCK_GUARD().
Coverity CID 1523750.

Fixes: a2326705e5 ("migration: Stop migration immediately in RDMA error paths")
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231103074245.55166-1-quintela@redhat.com>

docs/migration: Add the dirty limit section

The dirty limit feature has been introduced since the 8.1
QEMU release but has not reflected in the document, add a
section for that.

Signed-off-by: Hyman Huang <yong.huang@smartx.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <0f2b2c63fec22ea23e4926cdeb567b7a0ebd8152.1698847223.git.yong.huang@smartx.com>

tests/migration: Introduce dirty-limit into guestperf

Currently, guestperf does not cover the dirty-limit
migration, support this feature.

Note that dirty-limit requires 'dirty-ring-size' set.

To enable dirty-limit, setting x-vcpu-dirty-limit-period
as 500ms and x-vcpu-dirty-limit as 10MB/s:
$ ./tests/migration/guestperf.py \
    --dirty-ring-size 4096 \
    --dirty-limit --x-vcpu-dirty-limit-period 500 \
    --vcpu-dirty-limit 10 --output output.json \

To run the entire standardized set of dirty-limit-enabled
comparisons, with unix migration:
$ ./tests/migration/guestperf-batch.py \
    --dirty-ring-size 4096 \
    --dst-host localhost --transport unix \
    --filter compr-dirty-limit* --output outputdir

Signed-off-by: Hyman Huang <yong.huang@smartx.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <516e7a55dfc6e33d33510be37eb24223de5dc072.1697815117.git.yong.huang@smartx.com>
Message-ID: <e1283565b00b34b0377bbd27bee4bb8fc7c255a8.1698847223.git.yong.huang@smartx.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

tests/migration: Introduce dirty-ring-size option into guestperf

Dirty ring size configuration is not supported by guestperf tool.

Introduce dirty-ring-size (ranges in [1024, 65536]) option so
developers can play with dirty-ring and dirty-limit feature easier.

To set dirty ring size with 4096 during migration test:
$ ./tests/migration/guestperf.py --dirty-ring-size 4096 xxx

Signed-off-by: Hyman Huang <yong.huang@smartx.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <8a388cec5c1f73a34d42515bbc43837e97ee3839.1698847223.git.yong.huang@smartx.com>

tests: Add migration dirty-limit capability test

Add migration dirty-limit capability test if kernel support
dirty ring.

Migration dirty-limit capability introduce dirty limit
capability, two parameters: x-vcpu-dirty-limit-period and
vcpu-dirty-limit are introduced to implement the live
migration with dirty limit.

The test case does the following things:
1. start src, dst vm and enable dirty-limit capability
2. start migrate and set cancel it to check if dirty limit
   stop working.
3. restart dst vm
4. start migrate and enable dirty-limit capability
5. check if migration satisfy the convergence condition
   during pre-switchover phase.

Note that this test case involves many passes, so it runs
in slow mode only.

Signed-off-by: Hyman Huang <yong.huang@smartx.com>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <e55a302df9da7dbc00ad825f47f57c1a756d303e.1698847223.git.yong.huang@smartx.com>

system/dirtylimit: Drop the reduplicative check

Checking if dirty limit is in service is done by the
dirtylimit_query_all function, drop the reduplicative
check in the qmp_query_vcpu_dirty_limit function.

Signed-off-by: Hyman Huang <yong.huang@smartx.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <df9c3514933ff6750ef88068af18d3054bedf746.1698847223.git.yong.huang@smartx.com>

system/dirtylimit: Fix a race situation

Fix a race situation for global variable dirtylimit_state.

Also, replace usleep by g_usleep to increase platform
accessibility to the sleep function.

Signed-off-by: Hyman Huang <yong.huang@smartx.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <27c86239e21eda03d11ce5a3d07da3c229f562e3.1698847223.git.yong.huang@smartx.com>

linux-user/loongarch64: Add LASX sigcontext save/restore

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231101030816.2353416-7-gaosong@loongson.cn>

linux-user/loongarch64: Add LSX sigcontext save/restore

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231101030816.2353416-6-gaosong@loongson.cn>

linux-user/loongarch64: Use abi_{ulong,uint} types

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231101030816.2353416-5-gaosong@loongson.cn>

linux-user/loongarch64: setup_sigframe() set 'end' context size 0

See:
  https://github.com/torvalds/linux/blob/master/arch/loongarch/kernel/signal.c

The kernel setup_sigcontext() set end context size 0.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231101030816.2353416-4-gaosong@loongson.cn>

linux-user/loongarch64: Fix setup_extcontext alloc wrong fpu_context size

See:
 https://github.com/torvalds/linux/blob/master/arch/loongarch/kernel/signal.c

The alloc size is sizeof(struct target_fpu_context).

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231101030816.2353416-3-gaosong@loongson.cn>

linux-user/loongarch64: Use traps to track LSX/LASX usage

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231101030816.2353416-2-gaosong@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>

target/loongarch: Support 4K page size

The LoongArch kernel supports 4K page size.
Change TARGET_PAGE_BITS to 12.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20231023024059.3858349-1-gaosong@loongson.cn>

target/loongarch: Implement query-cpu-model-expansion

Add support for the query-cpu-model-expansion QMP command to LoongArch.
We support query the cpu features.

  e.g
    la464 and max cpu support LSX/LASX, default enable,
    la132 not support LSX/LASX.

    1. start with '-cpu max,lasx=off'

    (QEMU) query-cpu-model-expansion type=static  model={"name":"max"}
    {"return": {"model": {"name": "max", "props": {"lasx": false, "lsx": true}}}}

    2. start with '-cpu la464,lasx=off'
    (QEMU) query-cpu-model-expansion type=static  model={"name":"la464"}
    {"return": {"model": {"name": "max", "props": {"lasx": false, "lsx": true}}}

    3. start with '-cpu la132,lasx=off'
    qemu-system-loongarch64: can't apply global la132-loongarch-cpu.lasx=off: Property 'la132-loongarch-cpu.lasx' not found

    4. start with '-cpu max,lasx=off' or start with '-cpu la464,lasx=off' query cpu model la132
    (QEMU) query-cpu-model-expansion type=static  model={"name":"la132"}
    {"return": {"model": {"name": "la132"}}}

Acked-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231020084925.3457084-4-gaosong@loongson.cn>

target/loongarch: Allow user enable/disable LSX/LASX features

Some users may not need LSX/LASX, this patch allows the user
enable/disable LSX/LASX features.

 e.g
 '-cpu max,lsx=on,lasx=on'   (default);
 '-cpu max,lsx=on,lasx=off'  (enabled LSX);
 '-cpu max,lsx=off,lasx=on'  (enabled LASX, LSX);
 '-cpu max,lsx=off'          (disable LSX and LASX).

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231020084925.3457084-3-gaosong@loongson.cn>

target/loongarch: Add cpu model 'max'

We use cpu la464 for the 'max' cpu.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231020084925.3457084-2-gaosong@loongson.cn>

Merge tag 'pull-target-arm-20231102' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * linux-user/elfload: Add missing arm64 hwcap values
 * stellaris-gamepad: Convert to qdev
 * docs/specs: Convert various txt docs to rST
 * MAINTAINERS: Make sure that gicv3_internal.h is covered, too
 * hw/arm/pxa2xx_gpio: Pass CPU using QOM link property
 * hw/watchdog/wdt_imx2: Trace MMIO access and timer activity
 * hw/misc/imx7_snvs: Trace MMIO access
 * hw/misc/imx6_ccm: Convert DPRINTF to trace events
 * hw/i2c/pm_smbus: Convert DPRINTF to trace events
 * target/arm: Enable FEAT_MOPS insns in user-mode emulation
 * linux-user: Report AArch64 hwcap2 fields above bit 31
 * target/arm: Make FEAT_MOPS SET* insns handle Xs == XZR correctly
 * target/arm: Fix SVE STR increment
 * hw/char/stm32f2xx_usart: implement TX interrupts
 * target/arm: Correctly propagate stage 1 BTI guarded bit in a two-stage walk
 * xlnx-versal-virt: Add AMD/Xilinx TRNG device

* tag 'pull-target-arm-20231102' of https://git.linaro.org/people/pmaydell/qemu-arm: (33 commits)
  tests/qtest: Introduce tests for AMD/Xilinx Versal TRNG device
  hw/arm: xlnx-versal-virt: Add AMD/Xilinx TRNG device
  hw/misc: Introduce AMD/Xilix Versal TRNG device
  target/arm: Correctly propagate stage 1 BTI guarded bit in a two-stage walk
  hw/char/stm32f2xx_usart: Add more definitions for CR1 register
  hw/char/stm32f2xx_usart: Update IRQ when DR is written
  hw/char/stm32f2xx_usart: Extract common IRQ update code to update_irq()
  target/arm: Fix SVE STR increment
  target/arm: Make FEAT_MOPS SET* insns handle Xs == XZR correctly
  linux-user: Report AArch64 hwcap2 fields above bit 31
  target/arm: Enable FEAT_MOPS insns in user-mode emulation
  hw/i2c/pm_smbus: Convert DPRINTF to trace events
  hw/misc/imx6_ccm: Convert DPRINTF to trace events
  hw/misc/imx7_snvs: Trace MMIO access
  hw/watchdog/wdt_imx2: Trace timer activity
  hw/watchdog/wdt_imx2: Trace MMIO access
  hw/arm/pxa2xx_gpio: Pass CPU using QOM link property
  MAINTAINERS: Make sure that gicv3_internal.h is covered, too
  docs/specs/vmgenid: Convert to rST
  docs/specs/vmcoreinfo: Convert to rST
  ...

Conflicts:
  hw/input/stellaris_input.c
  The qdev conversion in this pull request ("stellaris-gamepad: Convert
  to qdev") eliminates the vmstate_register() call that was converted to
  vmstate_register_any() in the conflicting migration pull request.
  vmstate_register_any() is no longer necessary now that this device has
  been converted to qdev, so take this pull request's version of
  stellaris_gamepad.c over the previous pull request's
  stellaris_input.c (the file was renamed).

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'migration-20231102-pull-request' of https://gitlab.com/juan.quintela/qemu into staging

Migration Pull request (20231102)

Hi

In this pull request:

- migration reboot mode (steve)
  * I disabled the test because our CI don't like programs using so
    much shared memory.  Searching for a fix.
- test for postcopy recover (fabiano)
- MigrateAddress QAPI (het)
- better return path error handling (peter)
- traces for downtime (peter)
- vmstate_register() check for duplicates (juan)
  thomas find better solutions for s390x and ipmi.
  now also works on s390x

Please, apply.

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEGJn/jt6/WMzuA0uC9IfvGFhy1yMFAmVDipMACgkQ9IfvGFhy
# 1yNYnQ/9E5Cywsoqljqa/9FiKBSII2qMrmkfu6JLKqePnsh5pFZiukbudYRuJCCe
# ZTDEmD0NmKRJbDx2xRU1qx/e6gKJy+gz37KP89Buuh/WwZHPboPYtxQpGvCSiH26
# J3i+1+TgaqmkLzcO35wa8tp6gneQclWeAwKgMvdb4cm2pJEhgWRKI62ccyLzxeve
# UCzFQn60t55ETyVZGnRD4YwdTQvGKH+DPlyTuJOLR3DePuvZd8EdH+ypvB4RLAy7
# 3+CuQOxmF5LRXPbpJuAeOsudbmhhHzrO/yL7ZmsiKQTthsJv+SzC1bO94jhQrawZ
# Q7GCii5KpGq0KnRTRKZRGk6XKwxcYRduXMX3R5tXuVmDmCZsjhXzziU8yEdftph8
# 5TJdk1o0Gb043EFu81mrsQYS+9yJqe6sy6m3PTJaec54cAty5ln+c17WOvpAOaSV
# +1phe05ftuVPmQ3KWhbIR/tCmavNLwEZxpVIfyaKJx04bFbtQ9gRpRyURORX4KXc
# s4WXvNirQEohxYBnP4TPvA09xBTW3V08pk/wRDwt0YDXnLiqCltOuxD8r05K8K4B
# MkCLcWj0g7he2tBkF60oz1KSIE0oTB81um9AzLIv5F2YSYLaJM5BIcoC437MR2f4
# MOR7drR1fP5GsRu/SeU5BWvhVq3IvdOxR7G2MLNRJJvl7ZtGXDc=
# =uaqL
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 02 Nov 2023 19:40:03 HKT
# gpg:                using RSA key 1899FF8EDEBF58CCEE034B82F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>" [full]
# gpg:                 aka "Juan Quintela <quintela@trasno.org>" [full]
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03  4B82 F487 EF18 5872 D723

* tag 'migration-20231102-pull-request' of https://gitlab.com/juan.quintela/qemu: (40 commits)
  migration: modify test_multifd_tcp_none() to use new QAPI syntax.
  migration: Implement MigrateChannelList to hmp migration flow.
  migration: Implement MigrateChannelList to qmp migration flow.
  migration: modify migration_channels_and_uri_compatible() for new QAPI syntax
  migration: New migrate and migrate-incoming argument 'channels'
  migration: Convert the file backend to the new QAPI syntax
  migration: convert exec backend to accept MigrateAddress.
  migration: convert rdma backend to accept MigrateAddress
  migration: convert socket backend to accept MigrateAddress
  migration: convert migration 'uri' into 'MigrateAddress'
  migration: New QAPI type 'MigrateAddress'
  migration: Change ram_dirty_bitmap_reload() retval to bool
  tests/migration-test: Add a test for postcopy hangs during RECOVER
  migration: Allow network to fail even during recovery
  migration: Refactor error handling in source return path
  tests/qtest: migration: add reboot mode test
  cpr: reboot mode
  cpr: relax vhost migration blockers
  cpr: relax blockdev migration blockers
  migration: per-mode blockers
  ...

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

tests/qtest: Introduce tests for AMD/Xilinx Versal TRNG device

Signed-off-by: Tong Ho <tong.ho@amd.com>
Message-id: 20231031184611.3029156-4-tong.ho@amd.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm: xlnx-versal-virt: Add AMD/Xilinx TRNG device

Connect the support for Versal True Random Number Generator
(TRNG) device.

Warning: unlike the TRNG component in a real device from the
Versal device familiy, the connected TRNG model is not of
cryptographic grade and is not intended for use cases when
cryptograpically strong TRNG is needed.

Signed-off-by: Tong Ho <tong.ho@amd.com>
Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20231031184611.3029156-3-tong.ho@amd.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc: Introduce AMD/Xilix Versal TRNG device

This adds a non-cryptographic grade implementation of the
model for the True Random Number Generator (TRNG) component
in AMD/Xilinx Versal device family.

This implements all 3 modes defined by the actual hardware
specs, all of which selectable by guest software at will
at anytime:
1) PRNG mode, in which the generated sequence is required to
   be reproducible after reseeded by the same 384-bit value
   as supplied by guest software.
2) Test mode, in which the generated sequence is required to
   be reproducible ater reseeded by the same 128-bit test
   seed supplied by guest software.
3) TRNG mode, in which non-reproducible sequence is generated
   based on periodic reseed by a suitable entropy source.

This model is only intended for non-real world testing of
guest software, where cryptographically strong PRNG or TRNG
is not needed.

This model supports versions 1 & 2 of the device, with
default to be version 2; the 'hw-version' uint32 property
can be set to 0x0100 to override the default.

Other implemented properties:
- 'forced-prng', uint64
  When set to non-zero, mode 3's entropy source is implemented
  as a deterministic sequence based on the given value and other
  deterministic parameters.
  This option allows the emulation to test guest software using
  mode 3 and to reproduce data-dependent defects.

- 'fips-fault-events', uint32, bit-mask
  bit 3: Triggers the SP800-90B entropy health test fault irq
  bit 1: Triggers the FIPS 140-2 continuous test fault irq

Signed-off-by: Tong Ho <tong.ho@amd.com>
Message-id: 20231031184611.3029156-2-tong.ho@amd.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

dump: Drop redundant check for empty dump

dump_init() first computes the size of the dump, taking the filter
area into account, and fails if its zero.  It then looks for memory in
the filter area, and fails if there is none.

This is redundant: if the size of the dump is zero, there is no
memory, and vice versa.  Delete this check.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20231031104531.3169721-6-armbru@redhat.com>

dump: Improve some dump-guest-memory error messages

Zero @length is rejected with "Invalid parameter 'length'".  Improve
to "parameter 'length' expects a non-zero length".

qemu_open_old() is a wrapper around qemu_open_internal() that throws
away error information.  Switch to the wrapper that doesn't:
qemu_create().  Example improvement:

    (qemu) dump-guest-memory /dev/fdset/x 0 1
    Error: Could not open '/dev/fdset/x': Invalid argument

becomes

    Error: Could not parse fdset /dev/fdset/x

@protocol values not starting with "fd:" or "file:" are rejected with
"Invalid parameter 'protocol'".  Improve to "parameter 'protocol' must
start with 'file:' or 'fd:'".

While there, make the conditional checking @protocol a little more
obvious.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20231031104531.3169721-5-armbru@redhat.com>

dump: Recognize "fd:" protocols on Windows hosts

A few QMP command can work with named file descriptors.

The only way to create a named file descriptor used to be QMP command
getfd, which only works on POSIX hosts.  Thus, named file descriptors
were actually usable only there.

They became usable on Windows hosts when we added QMP command
get-win32-socket (commit 4cda177c601 "qmp: add 'get-win32-socket'").

Except in dump-guest-memory, because qmp_dump_guest_memory() compiles
its named file descriptor code only #if !defined(WIN32).

Compile it unconditionally, like we do for the other commands
supporting them.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20231031104531.3169721-4-armbru@redhat.com>

dump: Fix g_array_unref(NULL) in dump-guest-memory

When dump_init()'s check for non-zero @length fails, dump_cleanup()
passes null s->string_table_buf to g_array_unref(), which spews "GLib:
g_array_unref: assertion 'array' failed" to stderr.

Guard the g_array_unref().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20231031104531.3169721-3-armbru@redhat.com>

dump: Rename qmp_dump_guest_memory() parameter to match QAPI schema

The name of the second parameter differs between QAPI schema and C
implementation: it's @protocol in the former and @file in the latter.
Potentially confusing.  Change the C implementation to match the QAPI
schema.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20231031104531.3169721-2-armbru@redhat.com>

dump: Add command interface for kdump-raw formats

The QMP dump API represents the dump format as an enumeration. Add three
new enumerators, one for each supported kdump compression, each named
"kdump-raw-*".

For the HMP command line, rather than adding a new flag corresponding to
each format, it seems more human-friendly to add a single flag "-R" to
switch the kdump formats to "raw" mode. The choice of "-R" also
correlates nicely to the "makedumpfile -R" option, which would serve to
reassemble a flattened vmcore.

Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[ Marc-André: replace loff_t with off_t, indent fixes ]
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230918233233.1431858-4-stephen.s.brennan@oracle.com>

dump: Allow directly outputting raw kdump format

The flattened format (currently output by QEMU) is used by makedumpfile
only when it is outputting a vmcore to a file which is not seekable. The
flattened format functions essentially as a set of instructions of the
form "seek to the given offset, then write the given bytes out".

The flattened format can be reconstructed using makedumpfile -R, or
makedumpfile-R.pl, but it is a slow process because it requires copying
the entire vmcore. The flattened format can also be directly read by
crash, but still, it requires a lengthy reassembly phase.

To sum up, the flattened format is not an ideal one: it should only be
used on files which are actually not seekable. This is the exact
strategy which makedumpfile uses, as seen in the implementation of
"write_buffer()" in makedumpfile [1]. However, QEMU has always used the
flattened format. For compatibility it is best not to change the default
output format without warning. So, add a flag to DumpState which changes
the output to use the normal (i.e. raw) format. This flag will be added
to the QMP and HMP commands in the next change.

[1]: https://github.com/makedumpfile/makedumpfile/blob/f23bb943568188a2746dbf9b6692668f5a2ac3b6/makedumpfile.c#L5008-L5040

Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[ Marc-André: replace loff_t with off_t ]
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230918233233.1431858-3-stephen.s.brennan@oracle.com>

dump: Pass DumpState to write_ functions

For the next patch, we need a reference to DumpState when writing data.

Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230918233233.1431858-2-stephen.s.brennan@oracle.com>

target/arm: Correctly propagate stage 1 BTI guarded bit in a two-stage walk

In a two-stage translation, the result of the BTI guarded bit should
be the guarded bit from the first stage of translation, as there is
no BTI guard information in stage two.  Our code tried to do this,
but got it wrong, because we currently have two fields where the GP
bit information might live (ARMCacheAttrs::guarded and
CPUTLBEntryFull::extra::arm::guarded), and we were storing the GP bit
in the latter during the stage 1 walk but trying to copy the former
in combine_cacheattrs().

Remove the duplicated storage, and always use the field in
CPUTLBEntryFull; correctly propagate the stage 1 value to the output
in get_phys_addr_twostage().

Note for stable backports: in v8.0 and earlier the field is named
result->f.guarded, not result->f.extra.arm.guarded.

Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1950
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20231031173723.26582-1-peter.maydell@linaro.org

hw/char/stm32f2xx_usart: Add more definitions for CR1 register

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Hans-Erik Floryd <hans-erik.floryd@rt-labs.com>
Message-id: 20231030151528.1138131-4-hans-erik.floryd@rt-labs.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/char/stm32f2xx_usart: Update IRQ when DR is written

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Hans-Erik Floryd <hans-erik.floryd@rt-labs.com>
Message-id: 20231030151528.1138131-3-hans-erik.floryd@rt-labs.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/char/stm32f2xx_usart: Extract common IRQ update code to update_irq()

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Hans-Erik Floryd <hans-erik.floryd@rt-labs.com>
Message-id: 20231030151528.1138131-2-hans-erik.floryd@rt-labs.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Fix SVE STR increment

The previous change missed updating one of the increments and
one of the MemOps.  Add a test case for all vector lengths.

Cc: qemu-stable@nongnu.org
Fixes: e6dd5e782be ("target/arm: Use tcg_gen_qemu_{ld, st}_i128 in gen_sve_{ld, st}r")
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20231031143215.29764-1-richard.henderson@linaro.org
[PMM: fixed checkpatch nit]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Make FEAT_MOPS SET* insns handle Xs == XZR correctly

Most of the registers used by the FEAT_MOPS instructions cannot use
31 as a register field value; this is CONSTRAINED UNPREDICTABLE to
NOP or UNDEF (we UNDEF).  However, it is permitted for the "source
value" register for the memset insns SET* to be 31, which (as usual
for most data-processing insns) means it should be the zero register
XZR. We forgot to handle this case, with the effect that trying to
set memory to zero with a "SET* Xd, Xn, XZR" sets the memory to
the value that happens to be in the low byte of SP.

Handle XZR when getting the SET* data value from the register file.

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>
Message-id: 20231030174000.3792225-4-peter.maydell@linaro.org

linux-user: Report AArch64 hwcap2 fields above bit 31

The AArch64 ELF hwcap2 field is 64 bits, but our get_elf_hwcap2()
works with uint32_t, so it accidentally fails to report any hwcaps
over bit 31.  Use uint64_t here.

The Arm hwcap2 is only 32 bits (because the ELF format makes these
fields be the size of "long" in the ABI), but since it shares the
prototype declaration for get_elf_hwcap2() it is easier to also
expand it to 64 bits.

The only hwcap fields we implement already that are affected by this
are the HBC and MOPS ones, neither of which were implemented in a
previous release, so this doesn't need backporting to older stable
branches.

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>
Message-id: 20231030174000.3792225-3-peter.maydell@linaro.org

target/arm: Enable FEAT_MOPS insns in user-mode emulation

In user-mode emulation, we need to set the SCTLR_EL1.MSCEn
bit to avoid all the FEAT_MOPS insns UNDEFing.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20231030174000.3792225-2-peter.maydell@linaro.org

hw/i2c/pm_smbus: Convert DPRINTF to trace events

Let the trace messages slightly deviate from the function names
("smb" -> "smbus") being traced in order to avoid conflights with the SMB
protocol.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Corey Minyard <cminyard@mvista.com>
Message-id: 20231028122415.14869-6-shentey@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc/imx6_ccm: Convert DPRINTF to trace events

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20231028122415.14869-5-shentey@gmail.com
[PMM: Add "Hz" unit indicator to frequency traces]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc/imx7_snvs: Trace MMIO access

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20231028122415.14869-4-shentey@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/watchdog/wdt_imx2: Trace timer activity

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-id: 20231028122415.14869-3-shentey@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/watchdog/wdt_imx2: Trace MMIO access

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20231028122415.14869-2-shentey@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/pxa2xx_gpio: Pass CPU using QOM link property

Instead of passing the CPU index and resolving it,
use a QOM link to directly pass the CPU.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20231030083706.63685-1-philmd@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>