target/sh4: Fix SUBV opcode

The documentation says:

  SUBV Rm, Rn        Rn - Rm -> Rn, underflow -> T

The overflow / underflow can be calculated as:

  T = ((Rn ^ Rm) & (Result ^ Rn)) >> 31

However we were using the incorrect:

  T = ((Rn ^ Rm) & (Result ^ Rm)) >> 31

Fix by using the Rn register instead of Rm.

Add tests provided by Paul Cercueil.

Cc: qemu-stable@nongnu.org
Fixes: ad8d25a11f ("target-sh4: implement addv and subv using TCG")
Reported-by: Paul Cercueil <paul@crapouillou.net>
Suggested-by: Paul Cercueil <paul@crapouillou.net>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2318
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Message-Id: <20240430163125.77430-3-philmd@linaro.org>

target/sh4: Fix ADDV opcode

The documentation says:

  ADDV Rm, Rn        Rn + Rm -> Rn, overflow -> T

But QEMU implementation was:

  ADDV Rm, Rn        Rn + Rm -> Rm, overflow -> T

Fix by filling the correct Rm register.

Add tests provided by Paul Cercueil.

Cc: qemu-stable@nongnu.org
Fixes: ad8d25a11f ("target-sh4: implement addv and subv using TCG")
Reported-by: Paul Cercueil <paul@crapouillou.net>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2317
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Message-Id: <20240430163125.77430-2-philmd@linaro.org>

MAINTAINERS: Update my email address

Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Acked-by: Paul Durrant <paul@xen.org>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>
Message-ID: <20240429154938.19340-1-anthony.perard@citrix.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

plugins: Update stale comment

"plugin_mask" was renamed as "event_mask" in commit c006147122
("plugins: create CPUPluginState and migrate plugin_mask").

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240427155714.53669-3-philmd@linaro.org>

plugins/api: Only include 'exec/ram_addr.h' with system emulation

"exec/ram_addr.h" shouldn't be used with user emulation.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240427155714.53669-4-philmd@linaro.org>

coverity: Update user emulation regexp

All user emulation headers are now under include/user/.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240428221450.26460-3-philmd@linaro.org>

user: Move 'thunk.h' from 'exec/user' to 'user'

Keep all user emulation headers under the same user/ directory.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240428221450.26460-2-philmd@linaro.org>

user: Move 'abitypes.h' from 'exec/user' to 'user'

Keep all user emulation headers under the same user/ directory.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240503125202.35667-1-philmd@linaro.org>

exec: Include missing license in 'exec/cpu-common.h'

Commit 1ad2134f91 ("Hardware convenience library") extracted
"cpu-common.h" from "cpu-all.h", which uses the LGPL-2.1+ license.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240427155714.53669-5-philmd@linaro.org>

accel/whpx: Fix NULL dereference in whpx_init_vcpu()

When mechanically moving the @dirty field to AccelCPUState
in commit 9ad49538c7, we neglected cpu->accel is still NULL
when we want to dereference it.

Fixes: 9ad49538c7 ("accel/whpx: Use accel-specific per-vcpu @dirty field")
Reported-by: Volker Rümelin <vr_qemu@t-online.de>
Suggested-by: Volker Rümelin <vr_qemu@t-online.de>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240429091918.27429-2-philmd@linaro.org>

accel/nvmm: Fix NULL dereference in nvmm_init_vcpu()

When mechanically moving the @dirty field to AccelCPUState
in commit 79f1926b2d, we neglected cpu->accel is still NULL
when we want to dereference it.

Reported-by: Volker Rümelin <vr_qemu@t-online.de>
Suggested-by: Volker Rümelin <vr_qemu@t-online.de>
Fixes: 79f1926b2d ("accel/nvmm: Use accel-specific per-vcpu @dirty field")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240429091918.27429-3-philmd@linaro.org>

Merge tag 'accel-20240426' of https://github.com/philmd/qemu into staging

Accelerators patches

A lot of trivial cleanups and simplifications (moving methods around,
adding/removing #include statements). Most notable changes:

- Rename NEED_CPU_H -> COMPILING_PER_TARGET
- Rename few template headers using the '.h.inc' suffix
- Extract some definitions / declarations into their own header:
  - accel/tcg/user-retaddr.h (helper_retaddr)
  - include/exec/abi_ptr.h (abi_ptr)
  - include/exec/breakpoint.h (CPUBreakpoint, CPUWatchpoint)
  - include/exec/mmu-access-type.h (MMUAccessType)
  - include/user/tswap-target.h (tswapl, bswaptls)

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE+qvnXhKRciHc/Wuy4+MsLN6twN4FAmYsAuEACgkQ4+MsLN6t
# wN78Rg//V9UoE0U9Lh6Sd2WpcSAYP9D1CBa+iGXhrmel0utER1sQLu022nvcLdHc
# XtCgtX3H0yECF8dPX02rVp8IbSlOv3c8N/a6BxD79cRGqgXBYR/dEUqfXqeLJn3l
# a58YU3i7sLNQ0l7VnwTiBnI0lw170/xJl2B2mcR1SvWuH3dr5vTeIXNureu36ORo
# rc0oqWHbw1Pyyn8ADE2kPyFCOiwPwvcOvAk8dXGfib+mNCwNVV+ZUtAPi711VD8d
# 9VW2gu2sXwnWdpROrSugSw+aPVF4UjltL9qJEl5bxoqWFmlET1Zn2NpKvsocUXmh
# CMQQS2Tr4LpaaVQJGxx0yUe0B65X5+gCkIhsMOubED7GRyTCjrkOPm6exz3ge6WV
# YmIboggFAk3OjAzLs7yZVkWsTK1Y3+3eX0u7AWPUsUu7rCT/Toc6QxDS7eT2hJfq
# UDXI355PGbImgiArQa+OsT7v1Le4/iQa+TfN4fdUDpxEdfaxhnijWh+E91CEp+w/
# Mq7db9Z1aMnhFKIKdkPYyfwB74yXQrmYchJ0QojZjbzqNGwkt9VeC7O9RcYjEaHM
# hMIexwccxexqGH22wn8vPd6ZVKtiLaG4AXO0v6Dn2YJ7/zb/ntcI6lRZqdBAHKNK
# MzkjTRRRR0wAfu8Lk8CaNNEqUP4Po43fbYymo6AZhIR8NqfApL0=
# =ifBx
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 26 Apr 2024 12:39:13 PM PDT
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.org>" [full]

* tag 'accel-20240426' of https://github.com/philmd/qemu: (38 commits)
  plugins: Include missing 'qemu/bitmap.h' header
  hw/core: Avoid including the full 'hw/core/cpu.h' in 'tcg-cpu-ops.h'
  exec: Move CPUTLBEntry helpers to cputlb.c
  exec: Restrict inclusion of 'user/guest-base.h'
  exec: Rename 'exec/user/guest-base.h' as 'user/guest-base.h'
  exec: Restrict 'cpu_ldst.h' to TCG accelerator
  exec: Restrict TCG specific declarations of 'cputlb.h'
  exec: Declare CPUBreakpoint/CPUWatchpoint type in 'breakpoint.h' header
  exec: Declare MMUAccessType type in 'mmu-access-type.h' header
  exec: Declare abi_ptr type in its own 'abi_ptr.h' header
  exec/user: Do not include 'cpu.h' in 'abitypes.h'
  exec: Move [b]tswapl() declarations to 'exec/user/tswap-target.h'
  exec: Declare target_words_bigendian() in 'exec/tswap.h'
  exec/cpu-all: Remove unused tswapls() definitions
  exec/cpu-all: Remove unused 'qemu/thread.h' header
  exec/cpu-all: Reduce 'qemu/rcu.h' header inclusion
  accel/hvf: Use accel-specific per-vcpu @dirty field
  accel/nvmm: Use accel-specific per-vcpu @dirty field
  accel/whpx: Use accel-specific per-vcpu @dirty field
  accel/tcg: Rename helper-head.h -> helper-head.h.inc
  ...

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

plugins: Include missing 'qemu/bitmap.h' header

Since commit c006147122 ("plugins: create CPUPluginState and
migrate plugin_mask") "qemu/plugin.h" uses DECLARE_BITMAP(),
which is declared in "qemu/bitmap.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240418192525.97451-19-philmd@linaro.org>

Merge tag 'pull-nbd-2024-04-25' of https://repo.or.cz/qemu/ericb into staging

NBD patches for 2024-04-25

- Avoid calling poll() within coroutine

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEccLMIrHEYCkn0vOqp6FrSiUnQ2oFAmYqzkMACgkQp6FrSiUn
# Q2ol3wf9HbwiYkyHhqybb4ykEs75N8B2JPbOj6gYRSBn7rz90k1vElDCM2yQhlDN
# Ltuh8lTOaJb+Z4n2dKIF2m5hL2GTm/xtErIIpP7o6A+11mHW9ag/VLaAMdWJxmUr
# WEUIH6mVtuRcxTTCp01l/JAYpUxOoQs1fyQljONH5kg1MAZpTTD61/cuhrXlvPLU
# cVlrLfob90oYhydCq5o6ucW3GhaEYkaZzHIWFy7LphFySebMmnbnPhYf/JD6RZPL
# s5K7njMK1DOyguCLlOzSuRM4gIbYunnr0Ofr/orTlAUZvbhRGKUlH0RTMWVMzgek
# xArnEZYlsqF2wIvrz0GwMDL7BMmG7A==
# =vXJj
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 25 Apr 2024 02:42:27 PM PDT
# gpg:                using RSA key 71C2CC22B1C4602927D2F3AAA7A16B4A2527436A
# gpg: Good signature from "Eric Blake <eblake@redhat.com>" [full]
# gpg:                 aka "Eric Blake (Free Software Programmer) <ebb9@byu.net>" [full]
# gpg:                 aka "[jpeg image of size 6874]" [full]

* tag 'pull-nbd-2024-04-25' of https://repo.or.cz/qemu/ericb:
  nbd/server: Mark negotiation functions as coroutine_fn
  nbd/server: do not poll within a coroutine context

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

hw/core: Avoid including the full 'hw/core/cpu.h' in 'tcg-cpu-ops.h'

Only include what is required, avoiding the full
CPUState API from the huge "hw/core/cpu.h" header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240418192525.97451-4-philmd@linaro.org>

exec: Move CPUTLBEntry helpers to cputlb.c

The following CPUTLBEntry helpers are only used in accel/tcg/cputlb.c:
  - tlb_index()
  - tlb_entry()
  - tlb_read_idx()
  - tlb_addr_write()

Move them to this file, allowing to remove the huge "cpu.h" header
inclusion from "exec/cpu_ldst.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240418192525.97451-13-philmd@linaro.org>

exec: Restrict inclusion of 'user/guest-base.h'

Declare 'have_guest_base' in "user/guest-base.h".

Very few files require this header, so explicitly include
it there instead of "exec/cpu-all.h" which is used in many
source files.

Assert this user-specific header is only included from user
emulation.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231211212003.21686-23-philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>

exec: Rename 'exec/user/guest-base.h' as 'user/guest-base.h'

The include/user/ directory contains the user-emulation
specific headers. Move guest-base.h there too.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20240418192525.97451-15-philmd@linaro.org>

exec: Restrict 'cpu_ldst.h' to TCG accelerator

"exec/cpu_ldst.h" is specific to TCG, do not allow its
inclusion from other accelerators.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240418192525.97451-6-philmd@linaro.org>

exec: Restrict TCG specific declarations of 'cputlb.h'

Avoid TCG specific declarations being used from non-TCG accelerators.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240418192525.97451-5-philmd@linaro.org>

exec: Declare CPUBreakpoint/CPUWatchpoint type in 'breakpoint.h' header

The CPUBreakpoint and CPUWatchpoint structures are declared
in "hw/core/cpu.h", which contains declarations related to
CPUState and CPUClass. Some source files only require the
BP/WP definitions and don't need to pull in all CPU* API.
In order to simplify, create a new "exec/breakpoint.h" header.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20240418192525.97451-3-philmd@linaro.org>

exec: Declare MMUAccessType type in 'mmu-access-type.h' header

The MMUAccessType enum is declared in "hw/core/cpu.h".
"hw/core/cpu.h" contains declarations related to CPUState
and CPUClass. Some source files only require MMUAccessType
and don't need to pull in all CPU* declarations. In order
to simplify, create a new "exec/mmu-access-type.h" header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240418192525.97451-2-philmd@linaro.org>

exec: Declare abi_ptr type in its own 'abi_ptr.h' header

The abi_ptr type is declared in "exec/cpu_ldst.h" with all
the load/store helpers. Some source files requiring abi_ptr
type don't need the load/store helpers. In order to simplify,
create a new "exec/abi_ptr.h" header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231212123401.37493-21-philmd@linaro.org>

exec/user: Do not include 'cpu.h' in 'abitypes.h'

"exec/user/abitypes.h" requires:

 - "exec/cpu-defs.h"           (TARGET_LONG_BITS)
 - "exec/tswap.h"              (tswap32)

In order to avoid "cpu.h", pick the minimum required headers.

Assert this user-specific header is only included from user
emulation.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20231212123401.37493-20-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

exec: Move [b]tswapl() declarations to 'exec/user/tswap-target.h'

tswapl() and bswaptls() are target-dependent and only used
by user emulation. Move their definitions to a new header:
"exec/user/tswap-target.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20231212123401.37493-17-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

exec: Declare target_words_bigendian() in 'exec/tswap.h'

We usually check target endianess before swapping values,
so target_words_bigendian() declaration makes sense in
"exec/tswap.h" with the target swapping helpers.

Remove "hw/core/cpu.h" when it was only included to get
the target_words_bigendian() declaration.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20231212123401.37493-16-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

exec/cpu-all: Remove unused tswapls() definitions

Last use of tswapls() was removed 2 years ago in commit
aee14c77f4 ("linux-user: Rewrite do_getdents, do_getdents64").

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20231212123401.37493-15-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

exec/cpu-all: Remove unused 'qemu/thread.h' header

Nothing is required from "qemu/thread.h" in "exec/cpu-all.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20231212123401.37493-13-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

exec/cpu-all: Reduce 'qemu/rcu.h' header inclusion

"exec/cpu-all.h" doesn't need definitions from "qemu/rcu.h",
however "exec/ram_addr.h" does.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231211212003.21686-17-philmd@linaro.org>

accel/hvf: Use accel-specific per-vcpu @dirty field

HVF has a specific use of the CPUState::vcpu_dirty field
(CPUState::vcpu_dirty is not used by common code).
To make this field accel-specific, add and use a new
@dirty variable in the AccelCPUState structure.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240424174506.326-4-philmd@linaro.org>

accel/nvmm: Use accel-specific per-vcpu @dirty field

NVMM has a specific use of the CPUState::vcpu_dirty field
(CPUState::vcpu_dirty is not used by common code).
To make this field accel-specific, add and use a new
@dirty variable in the AccelCPUState structure.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240424174506.326-3-philmd@linaro.org>

accel/whpx: Use accel-specific per-vcpu @dirty field

WHPX has a specific use of the CPUState::vcpu_dirty field
(CPUState::vcpu_dirty is not used by common code).
To make this field accel-specific, add and use a new
@dirty variable in the AccelCPUState structure.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240424174506.326-2-philmd@linaro.org>

accel/tcg: Rename helper-head.h -> helper-head.h.inc

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

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

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

Therefore rename "exec/helper-head.h" as "exec/helper-head.h.inc".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240424173333.96148-4-philmd@linaro.org>

accel/tcg: Rename load-extract/store-insert headers using .h.inc suffix

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

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

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

Therefore rename 'store-insert-al16.h' as 'store-insert-al16.h.inc'
and 'load-extract-al16-al8.h' as 'load-extract-al16-al8.h.inc'.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240424173333.96148-3-philmd@linaro.org>

accel/tcg: Include missing headers in 'tb-jmp-cache.h'

Due to missing headers, when including "tb-jmp-cache.h" we might get:

  accel/tcg/tb-jmp-cache.h:21:21: error: field ‘rcu’ has incomplete type
     21 |     struct rcu_head rcu;
        |                     ^~~
  accel/tcg/tb-jmp-cache.h:24:9: error: unknown type name ‘vaddr’
     24 |         vaddr pc;
        |         ^~~~~

Add the missing "qemu/rcu.h" and "exec/cpu-common.h" headers.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240111162442.43755-1-philmd@linaro.org>

accel/tcg: Include missing 'hw/core/cpu.h' header

tcg_cpu_init_cflags() accesses CPUState fields, so requires
"hw/core/cpu.h" to get its structure definition.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20231212123401.37493-12-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Un-inline retaddr helpers to 'user-retaddr.h'

set_helper_retaddr() is only used in accel/tcg/user-exec.c.

clear_helper_retaddr() is only used in accel/tcg/cpu-exec.c
and accel/tcg/user-exec.c.

No need to expose their definitions to all user-emulation
files including "exec/cpu_ldst.h", move them to a new
"user-retaddr.h" header (restricted to accel/tcg/).

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231211212003.21686-19-philmd@linaro.org>

target/i386: Include missing 'exec/exec-all.h' header

The XRSTOR instruction ends calling tlb_flush(), declared
in "exec/exec-all.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231211212003.21686-13-philmd@linaro.org>

target/sparc: Replace abi_ulong by uint32_t for TARGET_ABI32

We have abi_ulong == uint32_t for the 32-bit ABI.
Use the generic type to avoid to depend on the
"exec/user/abitypes.h" header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240418192525.97451-14-philmd@linaro.org>

target/ppc/excp_helper: Avoid 'abi_ptr' in system emulation

'abi_ptr' is a user specific type. The system emulation
equivalent is 'target_ulong'. Use it in ppc_ldl_code()
to emphasis this is not an user emulation function.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20231211212003.21686-18-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

target: Define TCG_GUEST_DEFAULT_MO in 'cpu-param.h'

accel/tcg/ files requires the following definitions:

  - TARGET_LONG_BITS
  - TARGET_PAGE_BITS
  - TARGET_PHYS_ADDR_SPACE_BITS
  - TCG_GUEST_DEFAULT_MO

The first 3 are defined in "cpu-param.h". The last one
in "cpu.h", with a bunch of definitions irrelevant for
TCG. By moving the TCG_GUEST_DEFAULT_MO definition to
"cpu-param.h", we can simplify various accel/tcg includes.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20231211212003.21686-4-philmd@linaro.org>

semihosting/guestfd: Remove unused 'semihosting/uaccess.h' header

Nothing in guestfd.c requires "semihosting/uaccess.h" nor "qemu.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20231212123401.37493-8-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

semihosting/uaccess: Avoid including 'cpu.h'

"semihosting/uaccess.h" only requires the following headers:

  - "exec/cpu-defs.h" for target_ulong,
  - "exec/cpu-common.h" for cpu_memory_rw_debug()
  - "exec/tswap.h" for tswap32() and tswap64().

Include them instead of the huge "cpu.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <42c6471e-8383-45e0-85ee-e20ca32ecbad@linaro.org>

gdbstub: Avoid including 'cpu.h' in 'gdbstub/helpers.h'

We only need the "exec/tswap.h" and "cpu-param.h" headers.
Only include "cpu.h" in the target gdbstub.c source files.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240418192525.97451-20-philmd@linaro.org>

gdbstub: Simplify #ifdef'ry in helpers.h

Slightly simplify by checking NEED_CPU_H definition in header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240322161439.6448-2-philmd@linaro.org>

gdbstub: Include missing 'hw/core/cpu.h' header

Functions such gdb_get_cpu_pid() dereference CPUState so
require the structure declaration from "hw/core/cpu.h":

  static uint32_t gdb_get_cpu_pid(CPUState *cpu)
  {
    ...
    return cpu->cluster_index + 1;
  }

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Message-Id: <20231211212003.21686-15-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

accel: Include missing 'exec/cpu_ldst.h' header

Theses files call cpu_ldl_code() which is declared
in "exec/cpu_ldst.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231211212003.21686-5-philmd@linaro.org>

exec: Expose 'target_page.h' API to user emulation

User-only objects might benefit from the "exec/target_page.h"
API, which allows to build some objects once for all targets.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231211212003.21686-3-philmd@linaro.org>

exec: Include 'cpu.h' before validating CPUArchState placement

CPUArchState 'env' field is defined within the ArchCPU structure,
so we need to include each target "cpu.h" header which defines it.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Message-Id: <20231211212003.21686-2-philmd@linaro.org>

exec: Reduce tlb_set_dirty() declaration scope

tlb_set_dirty() is only used in accel/tcg/cputlb.c,
where it is defined. Declare it statically, removing
the stub.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240418192525.97451-11-philmd@linaro.org>

exec: Rename NEED_CPU_H -> COMPILING_PER_TARGET

'NEED_CPU_H' guard target-specific code; it is defined by meson
altogether with the 'CONFIG_TARGET' definition. Rename NEED_CPU_H
as COMPILING_PER_TARGET to clarify its meaning.

Mechanical change running:

 $ sed -i s/NEED_CPU_H/COMPILING_PER_TARGET/g $(git grep -l NEED_CPU_H)

then manually add a /* COMPILING_PER_TARGET */ comment
after the '#endif' when the block is large.

Inspired-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240322161439.6448-4-philmd@linaro.org>

nbd/server: Mark negotiation functions as coroutine_fn

nbd_negotiate() is already marked coroutine_fn.  And given the fix in
the previous patch to have nbd_negotiate_handle_starttls not create
and wait on a g_main_loop (as that would violate coroutine
constraints), it is worth marking the rest of the related static
functions reachable only during option negotiation as also being
coroutine_fn.

Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240408160214.1200629-6-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
[eblake: drop one spurious coroutine_fn marking]
Signed-off-by: Eric Blake <eblake@redhat.com>

Merge tag 'hw-misc-20240425' of https://github.com/philmd/qemu into staging

Misc HW patch queue

- Script to compare machines compat_props[] (Maksim)
- Introduce 'module' CPU topology level (Zhao)
- Various cleanups (Thomas, Zhao, Inès, Bernhard)

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE+qvnXhKRciHc/Wuy4+MsLN6twN4FAmYqN3wACgkQ4+MsLN6t
# wN4hTw/9FHsItnEkme/864DRPSP7A9mCGa+JfzJmsL8oUb9fBjXXKm+lNchMLu3B
# uvzfXB2Ea24yf5vyrldo0XlU3i/4GDvqXTI6YFYqBvitGICauYBu+6n2NZh2Y/Pn
# zZCcVo167o0q7dHu2WSrZ6cSUchsF2C80HjuS07QaN2YZ7QMuN1+uqTjCQ/JHQWA
# MH4xHh7cXdfCbbv8iNhMWn6sa+Bw/UyfRcc2W6w9cF5Q5cuuTshgDyd0JBOzkM1i
# Mcul7TuKrSiLUeeeqfTjwtw3rtbNfkelV3ycgvgECFAlzPSjF5a6d/EGdO2zo3T/
# aFZnQBYrb4U0SzsmfXFHW7cSylIc1Jn2CCuZZBIvdVcu8TGDD5XsgZbGoCfKdWxp
# l67qbQJy1Mp3LrRzygJIaxDOfE8fhhRrcIxfK/GoTHaCkqeFRkGjTeiDTVBqAES2
# zs6kUYZyG/xGaa2tsMu+HbtSO5EEqPC2QCdHayY3deW42Kwjj/HFV50Ya8YgYSVp
# gEAjTDOle2dDjlkYud+ymTJz7LnGb3G7q0EZRI9DWolx/bu+uZGQqTSRRre4qFQY
# SgN576hsFGN4NdM7tyJWiiqD/OC9ZeqUx3gGBtmI52Q6obBCE9hcow0fPs55Tk95
# 1YzPrt/3IoPI5ZptCoA8DFiysQ46OLtpIsQO9YcrpJmxWyLDSr0=
# =tm+U
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 25 Apr 2024 03:59:08 AM PDT
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.org>" [full]

* tag 'hw-misc-20240425' of https://github.com/philmd/qemu: (22 commits)
  hw/core: Support module-id in numa configuration
  hw/core: Introduce module-id as the topology subindex
  hw/core/machine: Support modules in -smp
  hw/core/machine: Introduce the module as a CPU topology level
  hw/i386/pc_sysfw: Remove unused parameter from pc_isa_bios_init()
  hw/misc : Correct 5 spaces indents in stm32l4x5_exti
  hw/xtensa: Include missing 'exec/cpu-common.h' in 'bootparam.h'
  hw/elf_ops: Rename elf_ops.h -> elf_ops.h.inc
  hw/cxl/cxl-cdat: Make cxl_doe_cdat_init() return boolean
  hw/cxl/cxl-cdat: Make ct3_build_cdat() return boolean
  hw/cxl/cxl-cdat: Make ct3_load_cdat() return boolean
  hw: Add a Kconfig switch for the TYPE_CPU_CLUSTER device
  hw: Fix problem with the A*MPCORE switches in the Kconfig files
  hw/riscv/virt: Replace sprintf by g_strdup_printf
  hw/misc/imx: Replace sprintf() by snprintf()
  hw/misc/applesmc: Simplify DeviceReset handler
  target/i386: Move APIC related code to cpu-apic.c
  hw/core: Remove check on NEED_CPU_H in tcg-cpu-ops.h
  scripts: add script to compare compatibility properties
  python/qemu/machine: add method to retrieve QEMUMachine::binary field
  ...

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

Merge tag 'pull-target-arm-20240425' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * Implement FEAT_NMI and NMI support in the GICv3
 * hw/dma: avoid apparent overflow in soc_dma_set_request
 * linux-user/flatload.c: Remove unused bFLT shared-library and ZFLAT code
 * Add ResetType argument to Resettable hold and exit phase methods
 * Add RESET_TYPE_SNAPSHOT_LOAD ResetType
 * Implement STM32L4x5 USART

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmYqMhMZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3uVlD/47U3zYP33y4+wJcRScC0QI
# jYd82jS7GhD5YP5QPrIEMaSbDwtYGi4Rez1taaHvZ2fWLg2gE973iixmTaM2mXCd
# xPEqMsRXkFrQnC89K5/v9uR04AvHxoM8J2mD2OKnUT0RVBs38WxCUMLETBsD18/q
# obs1RzDRhEs5BnwwPMm5HI1iQeVvDRe/39O3w3rZfA8DuqerrNOQWuJd43asHYjO
# Gc1QzCGhALlXDoqk11IzjhJ7es8WbJ5XGvrSNe9QLGNJwNsu9oi1Ez+5WK2Eht9r
# eRvGNFjH4kQY1YCShZjhWpdzU9KT0+80KLirMJFcI3vUztrYZ027/rMyKLHVOybw
# YAqgEUELwoGVzacpaJg73f77uknKoXrfTH25DfoLX0yFCB35JHOPcjU4Uq1z1pfV
# I80ZcJBDJ95mXPfyKLrO+0IyVBztLybufedK2aiH16waEGDpgsJv66FB2QRuQBYW
# O0i6/4DEUZmfSpOmr8ct+julz7wCWSjbvo6JFWxzzxvD0M5T3AFKXZI244g1SMdh
# LS8V7WVCVzVJ5mK8Ujp2fVaIIxiBzlXVZrQftWv5rhyDOiIIeP8pdekmPlI6p5HK
# 3/2efzSYNL2UCDZToIq24El/3md/7vHR6DBfBT1/pagxWUstqqLgkJO42jQtTG0E
# JY1cZ/EQY7cqXGrww8lhWA==
# =WEsU
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 25 Apr 2024 03:36:03 AM PDT
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [full]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [full]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [full]
# gpg:                 aka "Peter Maydell <peter@archaic.org.uk>" [unknown]

* tag 'pull-target-arm-20240425' of https://git.linaro.org/people/pmaydell/qemu-arm: (37 commits)
  tests/qtest: Add tests for the STM32L4x5 USART
  hw/arm: Add the USART to the stm32l4x5 SoC
  hw/char/stm32l4x5_usart: Add options for serial parameters setting
  hw/char/stm32l4x5_usart: Enable serial read and write
  hw/char: Implement STM32L4x5 USART skeleton
  reset: Add RESET_TYPE_SNAPSHOT_LOAD
  docs/devel/reset: Update to new API for hold and exit phase methods
  hw, target: Add ResetType argument to hold and exit phase methods
  scripts/coccinelle: New script to add ResetType to hold and exit phases
  allwinner-i2c, adm1272: Use device_cold_reset() for software-triggered reset
  hw/misc: Don't special case RESET_TYPE_COLD in npcm7xx_clk, gcr
  linux-user/flatload.c: Remove unused bFLT shared-library and ZFLAT code
  hw/dma: avoid apparent overflow in soc_dma_set_request
  hw/arm/virt: Enable NMI support in the GIC if the CPU has FEAT_NMI
  target/arm: Add FEAT_NMI to max
  hw/intc/arm_gicv3: Report the VINMI interrupt
  hw/intc/arm_gicv3: Report the NMI interrupt in gicv3_cpuif_update()
  hw/intc/arm_gicv3: Implement NMI interrupt priority
  hw/intc/arm_gicv3: Handle icv_nmiar1_read() for icc_nmiar1_read()
  hw/intc/arm_gicv3: Add NMI handling CPU interface registers
  ...

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

Merge tag 'pull-request-2024-04-25' of https://gitlab.com/thuth/qemu into staging

* Update OpenBSD CI image to 7.5
* Update/remove Ubuntu 20.04 CI jobs
* Update (most) CentOS 8 CI jobs to CentOS 9
* Some clean-ups and improvements to travis.yml
* Minor test fixes
* s390x header clean-ups
* Doc updates

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmYqbu4RHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbWqeA//cFAzjjmayCRZzuwwCFH0ILPrMNViRLFc
# ZuslNEFygDPl1H1wnw3MKzhHy1hbaC3gf30MjtejU61OMMyxS4exZd1rvw94a0cm
# OEisE8UG52kRsqPwKPktB2bybgX3BZbrFEwp1P0DsvpLTX7wI5nOZyNR4zWOf5Ym
# mODN/MjMFOhWjONOnNDRn4TbySQqolIQBTCq+f1J5Ej74V+p17aC5Fe3xhlFp8Ip
# aRockPW6dLpNt26zx6kKvQSYtkyLgQJSeUUyUCgcla03yzNSuV/kJPUoW0ewa+q8
# DZg1Ru5WJ6O8lELQdYq630cmdwg3e9EeI6q/U/1A11auuLaafOBi0eZW9LdPlrqD
# 6a+zwVn+ipyRdz8eRGZVRGdhJ6XT27YfFuKxdiu4BxnS0LRks4vDcXreIcQmiIUN
# bg/zSp6snCpYf7+GlUReZXWnVx401nu59+BNNKUV0qIxdORNm8kwd9ZpSQwXP/nF
# BMPhj2hoqvWb4C4r3WlTaSPlkJGhkb2lMLucjCbeGrdnmna0RFOFB301fllbpnVm
# 11SRipMEfrj/G5qp4giPLcruzesvRaZm85nmwDyOQWxr5Q0KWWfBVXZMt+qqOckR
# 2SUtLPd9nWruCy7KN15BrOWkmXc+OU8UFUqXIOvflkI6aF1bmFYRyrXgqX2q7QDT
# kEfWnBvBqxw=
# =1uVo
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 25 Apr 2024 07:55:42 AM PDT
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [full]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [full]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [full]

* tag 'pull-request-2024-04-25' of https://gitlab.com/thuth/qemu:
  target/s390x: Remove KVM stubs in cpu_models.h
  tests/unit: Remove debug statements in test-nested-aio-poll.c
  docs/devel: fix minor typo in submitting-a-patch.rst
  hw/s390x: Include missing 'cpu.h' header
  tests: Update our CI to use CentOS Stream 9 instead of 8
  tests/docker/dockerfiles: Run lcitool-refresh after the lcitool update
  tests/lcitool/libvirt-ci: Update to the latest master branch
  tests: Remove Ubuntu 20.04 container
  .travis.yml: Do some more testing with Clang
  .travis.yml: Update the jobs to Ubuntu 22.04
  .travis.yml: Remove the unused UNRELIABLE environment variable
  Revert ".travis.yml: Cache Avocado cache"
  tests/vm: update openbsd image to 7.5
  docs: i386: pc: Update maximum CPU numbers for PC Q35
  tests/qtest : Use `g_assert_cmphex` instead of `g_assert_cmpuint`
  MAINTAINERS: update email of Peter Lieven

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

target/s390x: Remove KVM stubs in cpu_models.h

Since the calls are elided when KVM is not available,
we can remove the stubs (which are never compiled).

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240419090631.48055-1-philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/unit: Remove debug statements in test-nested-aio-poll.c

We have been running this test for almost a year; it
is safe to remove its debug statements, which clutter
CI jobs output:

  ▶  88/100 /nested-aio-poll                      OK
  io_read 0x16bb26158
  io_poll_true 0x16bb26158
  > io_poll_ready
  io_read 0x16bb26164
  < io_poll_ready
  io_poll_true 0x16bb26158
  io_poll_false 0x16bb26164
  > io_poll_ready
  io_poll_false 0x16bb26164
  io_poll_false 0x16bb26164
  io_poll_false 0x16bb26164
  io_poll_false 0x16bb26164
  io_poll_false 0x16bb26164
  io_poll_false 0x16bb26164
  io_poll_false 0x16bb26164
  io_poll_false 0x16bb26164
  io_poll_false 0x16bb26164
  io_read 0x16bb26164
  < io_poll_ready
  88/100 qemu:unit / test-nested-aio-poll        OK

Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20240422112246.83812-1-philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

docs/devel: fix minor typo in submitting-a-patch.rst

s/Resolved:/Resolves:/

Signed-off-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Message-ID: <20240422124128.4034482-1-manos.pitsidianakis@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

hw/s390x: Include missing 'cpu.h' header

"cpu.h" is implicitly included. Include it explicitly to
avoid the following error when refactoring headers:

  hw/s390x/s390-stattrib.c:86:40: error: use of undeclared identifier 'TARGET_PAGE_SIZE'
      len = sac->peek_stattr(sas, addr / TARGET_PAGE_SIZE, buflen, vals);
                                         ^
  hw/s390x/s390-stattrib.c:94:58: error: use of undeclared identifier 'TARGET_PAGE_MASK'
                     addr / TARGET_PAGE_SIZE, len, addr & ~TARGET_PAGE_MASK);
                                                         ^
  hw/s390x/s390-stattrib.c:224:40: error: use of undeclared identifier 'TARGET_PAGE_BITS'
          qemu_put_be64(f, (start_gfn << TARGET_PAGE_BITS) | STATTR_FLAG_MORE);
                                         ^
  In file included from hw/s390x/s390-virtio-ccw.c:17:
  hw/s390x/s390-virtio-hcall.h:22:27: error: unknown type name 'CPUS390XState'
  int s390_virtio_hypercall(CPUS390XState *env);
                            ^

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Acked-by: Eric Farman <farman@linux.ibm.com>
Message-ID: <20240322162822.7391-1-philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

nbd/server: do not poll within a coroutine context

Coroutines are not supposed to block. Instead, they should yield.

The client performs TLS upgrade outside of an AIOContext, during
synchronous handshake; this still requires g_main_loop.  But the
server responds to TLS upgrade inside a coroutine, so a nested
g_main_loop is wrong.  Since the two callbacks no longer share more
than the setting of data.complete and data.error, it's just as easy to
use static helpers instead of trying to share a common code path.  It
is also possible to add assertions that no other code is interfering
with the eventual path to qio reaching the callback, whether or not it
required a yield or main loop.

Fixes: f95910f ("nbd: implement TLS support in the protocol negotiation")
Signed-off-by: Zhu Yangyang <zhuyangyang14@huawei.com>
[eblake: move callbacks to their use point, add assertions]
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240408160214.1200629-5-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

hw/core: Support module-id in numa configuration

Module is a level above the core, thereby supporting numa
configuration on the module level can bring user more numa flexibility.

This is the natural further support for module level.

Add module level support in numa configuration.

Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Tested-by: Babu Moger <babu.moger@amd.com>
Message-ID: <20240424154929.1487382-5-zhao1.liu@intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/core: Introduce module-id as the topology subindex

Add module-id in CpuInstanceProperties, to locate the CPU with module
level.

Suggested-by: Xiaoyao Li <xiaoyao.li@intel.com>
Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Tested-by: Babu Moger <babu.moger@amd.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240424154929.1487382-4-zhao1.liu@intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/core/machine: Support modules in -smp

Add "modules" parameter parsing support in -smp.

Suggested-by: Xiaoyao Li <xiaoyao.li@intel.com>
Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Tested-by: Babu Moger <babu.moger@amd.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240424154929.1487382-3-zhao1.liu@intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/core/machine: Introduce the module as a CPU topology level

In x86, module is the topology level above core, which contains a set
of cores that share certain resources (in current products, the resource
usually includes L2 cache, as well as module scoped features and MSRs).

Though smp.clusters could also share the L2 cache resource [1], there
are following reasons that drive us to introduce the new smp.modules:

  * As the CPU topology abstraction in device tree [2], cluster supports
    nesting (though currently QEMU hasn't support that). In contrast,
    (x86) module does not support nesting.

  * Due to nesting, there is great flexibility in sharing resources
    on cluster, rather than narrowing cluster down to sharing L2 (and
    L3 tags) as the lowest topology level that contains cores.

  * Flexible nesting of cluster allows it to correspond to any level
    between the x86 package and core.

  * In Linux kernel, x86's cluster only represents the L2 cache domain
    but QEMU's smp.clusters is the CPU topology level. Linux kernel will
    also expose module level topology information in sysfs for x86. To
    avoid cluster ambiguity and keep a consistent CPU topology naming
    style with the Linux kernel, we introduce module level for x86.

The module is, in existing hardware practice, the lowest layer that
contains the core, while the cluster is able to have a higher
topological scope than the module due to its nesting.

Therefore, place the module between the cluster and the core:

    drawer/book/socket/die/cluster/module/core/thread

With the above topological hierarchy order, introduce module level
support in MachineState and MachineClass.

[1]: https://lore.kernel.org/qemu-devel/c3d68005-54e0-b8fe-8dc1-5989fe3c7e69@huawei.com/
[2]: https://www.kernel.org/doc/Documentation/devicetree/bindings/cpu/cpu-topology.txt

Suggested-by: Xiaoyao Li <xiaoyao.li@intel.com>
Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Tested-by: Babu Moger <babu.moger@amd.com>
Message-ID: <20240424154929.1487382-2-zhao1.liu@intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i386/pc_sysfw: Remove unused parameter from pc_isa_bios_init()

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240422200625.2768-2-shentey@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/misc : Correct 5 spaces indents in stm32l4x5_exti

Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240421141455.116548-1-ines.varhol@telecom-paris.fr>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/xtensa: Include missing 'exec/cpu-common.h' in 'bootparam.h'

cpu_physical_memory_write() is declared in "exec/cpu-common.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20240418192525.97451-21-philmd@linaro.org>

hw/elf_ops: Rename elf_ops.h -> elf_ops.h.inc

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

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

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

Therefore rename "hw/elf_ops.h" as "hw/elf_ops.h.inc".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240424173333.96148-2-philmd@linaro.org>

hw/cxl/cxl-cdat: Make cxl_doe_cdat_init() return boolean

As error.h suggested, the best practice for callee is to return
something to indicate success / failure.

With returned boolean, there's no need to dereference @errp to check
failure case.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Message-ID: <20240418100433.1085447-4-zhao1.liu@linux.intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/cxl/cxl-cdat: Make ct3_build_cdat() return boolean

As error.h suggested, the best practice for callee is to return
something to indicate success / failure.

So make ct3_build_cdat() return boolean, and this is the preparation for
cxl_doe_cdat_init() returning boolean.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Message-ID: <20240418100433.1085447-3-zhao1.liu@linux.intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/cxl/cxl-cdat: Make ct3_load_cdat() return boolean

As error.h suggested, the best practice for callee is to return
something to indicate success / failure.

So make ct3_load_cdat() return boolean, and this is the preparation for
cxl_doe_cdat_init() returning boolean.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Message-ID: <20240418100433.1085447-2-zhao1.liu@linux.intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw: Add a Kconfig switch for the TYPE_CPU_CLUSTER device

The cpu-cluster device is only needed for some few arm and riscv
machines. Let's avoid compiling and linking it if it is not really
necessary.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240415065655.130099-3-thuth@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw: Fix problem with the A*MPCORE switches in the Kconfig files

A9MPCORE, ARM11MPCORE and A15MPCORE are defined twice, once in
hw/cpu/Kconfig and once in hw/arm/Kconfig. This is only possible
by accident, since hw/cpu/Kconfig is never included from hw/Kconfig.
Fix it by declaring the switches only in hw/cpu/Kconfig (since the
related files reside in the hw/cpu/ folder) and by making sure that
the file hw/cpu/Kconfig is now properly included from hw/Kconfig.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240415065655.130099-2-thuth@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/riscv/virt: Replace sprintf by g_strdup_printf

sprintf() is deprecated on Darwin since macOS 13.0 / XCode 14.1.
Use g_strdup_printf instead.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
[rth: Use g_strdup_printf]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240412073346.458116-26-richard.henderson@linaro.org>

hw/misc/imx: Replace sprintf() by snprintf()

sprintf() is deprecated on Darwin since macOS 13.0 / XCode 14.1,
resulting in painful developer experience. Use snprintf() instead.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20240411104340.6617-6-philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

hw/misc/applesmc: Simplify DeviceReset handler

Have applesmc_find_key() return a const pointer.
Since the returned buffers are not modified in
applesmc_io_data_write(), it is pointless to
delete and re-add the keys in the DeviceReset
handler. Add them once in DeviceRealize, and
discard them in the DeviceUnrealize handler.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20240410180819.92332-1-philmd@linaro.org>

tests/qtest: Add tests for the STM32L4x5 USART

Test:
- read/write from/to the usart registers
- send/receive a character/string over the serial port

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240329174402.60382-6-arnaud.minier@telecom-paris.fr
[PMM: fix checkpatch nits, remove commented out code]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm: Add the USART to the stm32l4x5 SoC

Add the USART to the SoC and connect it to the other implemented devices.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240329174402.60382-5-arnaud.minier@telecom-paris.fr
[PMM: fixed a few checkpatch nits]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/char/stm32l4x5_usart: Add options for serial parameters setting

Add a function to change the settings of the
serial connection.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240329174402.60382-4-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/char/stm32l4x5_usart: Enable serial read and write

Implement the ability to read and write characters to the
usart using the serial port.

The character transmission is based on the
cmsdk-apb-uart implementation.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240329174402.60382-3-arnaud.minier@telecom-paris.fr
[PMM: fixed a few checkpatch nits]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/char: Implement STM32L4x5 USART skeleton

Add the basic infrastructure (register read/write, type...)
to implement the STM32L4x5 USART.

Also create different types for the USART, UART and LPUART
of the STM32L4x5 to deduplicate code and enable the
implementation of different behaviors depending on the type.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240329174402.60382-2-arnaud.minier@telecom-paris.fr
[PMM: update to new reset hold method signature;
 fixed a few checkpatch nits]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

reset: Add RESET_TYPE_SNAPSHOT_LOAD

Some devices and machines need to handle the reset before a vmsave
snapshot is loaded differently -- the main user is the handling of
RNG seed information, which does not want to put a new RNG seed into
a ROM blob when we are doing a snapshot load.

Currently this kind of reset handling is supported only for:
 * TYPE_MACHINE reset methods, which take a ShutdownCause argument
 * reset functions registered with qemu_register_reset_nosnapshotload

To allow a three-phase-reset device to also distinguish "snapshot
load" reset from the normal kind, add a new ResetType
RESET_TYPE_SNAPSHOT_LOAD. All our existing reset methods ignore
the reset type, so we don't need to update any device code.

Add the enum type, and make qemu_devices_reset() use the
right reset type for the ShutdownCause it is passed. This
allows us to get rid of the device_reset_reason global we
were using to implement qemu_register_reset_nosnapshotload().

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>
Reviewed-by: Luc Michel <luc.michel@amd.com>
Message-id: 20240412160809.1260625-7-peter.maydell@linaro.org

docs/devel/reset: Update to new API for hold and exit phase methods

Update the reset documentation's example code to match the new API
for the hold and exit phase method APIs where they take a ResetType
argument.

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>
Reviewed-by: Luc Michel <luc.michel@amd.com>
Message-id: 20240412160809.1260625-6-peter.maydell@linaro.org

hw, target: Add ResetType argument to hold and exit phase methods

We pass a ResetType argument to the Resettable class enter
phase method, but we don't pass it to hold and exit, even though
the callsites have it readily available. This means that if
a device cared about the ResetType it would need to record it
in the enter phase method to use later on. Pass the type to
all three of the phase methods to avoid having to do that.

Commit created with

  for dir in hw target include; do \
      spatch --macro-file scripts/cocci-macro-file.h \
             --sp-file scripts/coccinelle/reset-type.cocci \
             --keep-comments --smpl-spacing --in-place \
             --include-headers --dir $dir; done

and no manual edits.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Luc Michel <luc.michel@amd.com>
Message-id: 20240412160809.1260625-5-peter.maydell@linaro.org

scripts/coccinelle: New script to add ResetType to hold and exit phases

We pass a ResetType argument to the Resettable class enter phase
method, but we don't pass it to hold and exit, even though the
callsites have it readily available.  This means that if a device
cared about the ResetType it would need to record it in the enter
phase method to use later on.  We should pass the type to all three
of the phase methods to avoid having to do that.

This coccinelle script adds the ResetType argument to the hold and
exit phases of the Resettable interface.

The first part of the script (rules holdfn_assigned, holdfn_defined,
exitfn_assigned, exitfn_defined) update implementations of the
interface within device models, both to change the signature of their
method implementations and to pass on the reset type when they invoke
reset on some other device.

The second part of the script is various special cases:
 * method callsites in resettable_phase_hold(), resettable_phase_exit()
   and device_phases_reset()
 * updating the typedefs for the methods
 * isl_pmbus_vr.c has some code where one device's reset method directly
   calls the implementation of a different device's method

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Luc Michel <luc.michel@amd.com>
Message-id: 20240412160809.1260625-4-peter.maydell@linaro.org

allwinner-i2c, adm1272: Use device_cold_reset() for software-triggered reset

Rather than directly calling the device's implementation of its 'hold'
reset phase, call device_cold_reset(). This means we don't have to
adjust this callsite when we add another argument to the function
signature for the hold and exit reset methods.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Luc Michel <luc.michel@amd.com>
Message-id: 20240412160809.1260625-3-peter.maydell@linaro.org

hw/misc: Don't special case RESET_TYPE_COLD in npcm7xx_clk, gcr

The npcm7xx_clk and npcm7xx_gcr device reset methods look at
the ResetType argument and only handle RESET_TYPE_COLD,
producing a warning if another reset type is passed. This
is different from how every other three-phase-reset method
we have works, and makes it difficult to add new reset types.

A better pattern is "assume that any reset type you don't know
about should be handled like RESET_TYPE_COLD"; switch these
devices to do that. Then adding a new reset type will only
need to touch those devices where its behaviour really needs
to be different from the standard cold reset.

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>
Reviewed-by: Luc Michel <luc.michel@amd.com>
Message-id: 20240412160809.1260625-2-peter.maydell@linaro.org

linux-user/flatload.c: Remove unused bFLT shared-library and ZFLAT code

Ever since the bFLT format support was added in 2006, there has been
a chunk of code in the file guarded by CONFIG_BINFMT_SHARED_FLAT
which is supposedly for shared library support.  This is not enabled
and it's not possible to enable it, because if you do you'll run into
the "#error needs checking" in the calc_reloc() function.

Similarly, CONFIG_BINFMT_ZFLAT exists but can't be enabled because of
an "#error code needs checking" in load_flat_file().

This code is obviously unfinished and has never been used; nobody in
the intervening 18 years has complained about this or fixed it, so
just delete the dead code.  If anybody ever wants the feature they
can always pull it out of git, or (perhaps better) write it from
scratch based on the current Linux bFLT loader rather than the one of
18 years ago.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240411115313.680433-1-peter.maydell@linaro.org

hw/dma: avoid apparent overflow in soc_dma_set_request

In soc_dma_set_request() we try to set a bit in a uint64_t, but we
do it with "1 << ch->num", which can't set any bits past 31;
any use for a channel number of 32 or more would fail due to
integer overflow.

This doesn't happen in practice for our current use of this code,
because the worst case is when we call soc_dma_init() with an
argument of 32 for the number of channels, and QEMU builds with
-fwrapv so the shift into the sign bit is well-defined. However,
it's obviously not the intended behaviour of the code.

Add casts to force the shift to be done as 64-bit arithmetic,
allowing up to 64 channels.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: afbb5194d4 ("Handle on-chip DMA controllers in one place, convert OMAP DMA to use it.")
Signed-off-by: Anastasia Belova <abelova@astralinux.ru>
Message-id: 20240409115301.21829-1-abelova@astralinux.ru
[PMM: Edit commit message to clarify that this doesn't actually
 bite us in our current usage of this code.]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/virt: Enable NMI support in the GIC if the CPU has FEAT_NMI

If the CPU implements FEAT_NMI, then turn on the NMI support in the
GICv3 too.  It's permitted to have a configuration with FEAT_NMI in
the CPU (and thus NMI support in the CPU interfaces too) but no NMI
support in the distributor and redistributor, but this isn't a very
useful setup as it's close to having no NMI support at all.

We don't need to gate the enabling of NMI in the GIC behind a
machine version property, because none of our current CPUs
implement FEAT_NMI, and '-cpu max' is not something we maintain
migration compatibility across versions for. So we can always
enable the GIC NMI support when the CPU has it.

Neither hvf nor KVM support NMI in the GIC yet, so we don't enable
it unless we're using TCG.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240407081733.3231820-25-ruanjinjie@huawei.com
[PMM: Update comment and commit message]
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Add FEAT_NMI to max

Enable FEAT_NMI on the 'max' CPU.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-24-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3: Report the VINMI interrupt

In vCPU Interface, if the vIRQ has the non-maskable property, report
vINMI to the corresponding vPE.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-23-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3: Report the NMI interrupt in gicv3_cpuif_update()

In CPU Interface, if the IRQ has the non-maskable property, report NMI to
the corresponding PE.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-22-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3: Implement NMI interrupt priority

If GICD_CTLR_DS bit is zero and the NMI is non-secure, the NMI priority is
higher than 0x80, otherwise it is higher than 0x0. And save the interrupt
non-maskable property in hppi.nmi to deliver NMI exception. Since both GICR
and GICD can deliver NMI, it is both necessary to check whether the pending
irq is NMI in gicv3_redist_update_noirqset and gicv3_update_noirqset.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-21-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3: Handle icv_nmiar1_read() for icc_nmiar1_read()

Implement icv_nmiar1_read() for icc_nmiar1_read(), so add definition for
ICH_LR_EL2.NMI and ICH_AP1R_EL2.NMI bit.

If FEAT_GICv3_NMI is supported, ich_ap_write() should consider ICV_AP1R_EL1.NMI
bit. In icv_activate_irq() and icv_eoir_write(), the ICV_AP1R_EL1.NMI bit
should be set or clear according to the Non-maskable property. And the RPR
priority should also update the NMI bit according to the APR priority NMI bit.

By the way, add gicv3_icv_nmiar1_read trace event.

If the hpp irq is a NMI, the icv iar read should return 1022 and trap for
NMI again

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
[PMM: use cs->nmi_support instead of cs->gic->nmi_support]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-20-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3: Add NMI handling CPU interface registers

Add the NMIAR CPU interface registers which deal with acknowledging NMI.

When introduce NMI interrupt, there are some updates to the semantics for the
register ICC_IAR1_EL1 and ICC_HPPIR1_EL1. For ICC_IAR1_EL1 register, it
should return 1022 if the intid has non-maskable property. And for
ICC_NMIAR1_EL1 register, it should return 1023 if the intid do not have
non-maskable property. Howerever, these are not necessary for ICC_HPPIR1_EL1
register.

And the APR and RPR has NMI bits which should be handled correctly.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
[PMM: Separate out whether cpuif supports NMI from whether the
 GIC proper (IRI) supports NMI]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-19-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3: Implement GICD_INMIR

Add GICD_INMIR, GICD_INMIRnE register and support access GICD_INMIR0.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-18-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3_redist: Implement GICR_INMIR0

Add GICR_INMIR0 register and support access GICR_INMIR0.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-17-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3: Add irq non-maskable property

A SPI, PPI or SGI interrupt can have non-maskable property. So maintain
non-maskable property in PendingIrq and GICR/GICD. Since add new device
state, it also needs to be migrated, so also save NMI info in
vmstate_gicv3_cpu and vmstate_gicv3.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240407081733.3231820-16-ruanjinjie@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3_kvm: Not set has-nmi=true for the KVM GICv3

So far, there is no FEAT_GICv3_NMI support in the in-kernel GIC, so make it
an error to try to set has-nmi=true for the KVM GICv3.

Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Message-id: 20240407081733.3231820-15-ruanjinjie@huawei.com
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>