accel/tcg: Add cpu_atomic_*_mmu for 16-byte xchg, fetch_and, fetch_or

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250815122653.701782-4-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

qemu/atomic: Add atomic16 primitives for xchg, fetch_and, fetch_or

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250815122653.701782-3-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

qemu/atomic: Finish renaming atomic128-cas.h headers

The aarch64 header was not renamed with the others, meaning it
was skipped in favor of the generic version.

Cc: qemu-stable@nongnu.org
Fixes: 15606965400b ("qemu/atomic: Rename atomic128-cas.h headers using .h.inc suffix")
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250815122653.701782-2-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Correct condition of aa64_atomics feature function

The ARMv8.1-Atomics feature (renamed FEAT_LSE in more modern versions
of the Arm ARM) has always ben indicated by ID_AA64ISAR0.ATOMIC being
0b0010 or greater; 0b0001 is a reserved unused value.

We were incorrectly checking for != 0; this had no harmful effects
because all the CPUs set their value for this field to either 0
(for not having the feature) or 2 (if they do have it), but it's
better to match what the architecture specifies here.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250819145659.2165160-1-peter.maydell@linaro.org

MAINTAINERS: Put kernel-doc under the "docs build machinery" section

We never had a MAINTAINERS entry for the old kernel-doc script; add
the files for the new Python kernel-doc under "Sphinx documentation
configuration and build machinery", as the most appropriate
subsection.

Mauro has kindly volunteered to help with maintenance/review
of this area of the codebase, so add him as a maintainer.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250814171324.1614516-9-peter.maydell@linaro.org

scripts/kernel-doc: Delete the old Perl kernel-doc script

We can now delete the old Perl kernel-doc script. For posterity,
this is a complete diff of the local changes that we were carrying
between the kernel's Perl script as of kernel commit 72b97d0b911872ba
(the last time we synced it) and our local copy:

--- /tmp/kdoc   2025-08-14 10:42:47.620331939 +0100
+++ scripts/kernel-doc  2025-02-17 10:44:34.528421457 +0000
@@ -1,5 +1,5 @@
 #!/usr/bin/env perl
-# SPDX-License-Identifier: GPL-2.0
+# SPDX-License-Identifier: GPL-2.0-only

 use warnings;
 use strict;
@@ -224,12 +224,12 @@
 my $type_fp_param = '\@(\w+)\(\)';  # Special RST handling for func ptr params
 my $type_fp_param2 = '\@(\w+->\S+)\(\)';  # Special RST handling for structs with func ptr params
 my $type_env = '(\$\w+)';
-my $type_enum = '\&(enum\s*([_\w]+))';
-my $type_struct = '\&(struct\s*([_\w]+))';
-my $type_typedef = '\&(typedef\s*([_\w]+))';
-my $type_union = '\&(union\s*([_\w]+))';
-my $type_member = '\&([_\w]+)(\.|->)([_\w]+)';
-my $type_fallback = '\&([_\w]+)';
+my $type_enum = '#(enum\s*([_\w]+))';
+my $type_struct = '#(struct\s*([_\w]+))';
+my $type_typedef = '#(([A-Z][_\w]*))';
+my $type_union = '#(union\s*([_\w]+))';
+my $type_member = '#([_\w]+)(\.|->)([_\w]+)';
+my $type_fallback = '(?!)';    # this never matches
 my $type_member_func = $type_member . '\(\)';

 # Output conversion substitutions.
@@ -1745,6 +1745,9 @@
             )+
           \)\)\s+//x;

+    # Strip QEMU specific compiler annotations
+    $prototype =~ s/QEMU_[A-Z_]+ +//;
+
     # Yes, this truly is vile.  We are looking for:
     # 1. Return type (may be nothing if we're looking at a macro)
     # 2. Function name
@@ -2057,7 +2060,7 @@
     }
     elsif (/$doc_decl/o) {
        $identifier = $1;
-       if (/\s*([\w\s]+?)(\(\))?\s*-/) {
+       if (/\s*([\w\s]+?)(\s*-|:)/) {
            $identifier = $1;
        }

@@ -2067,7 +2070,7 @@
        $contents = "";
        $section = $section_default;
        $new_start_line = $. + 1;
-       if (/-(.*)/) {
+       if (/[-:](.*)/) {
            # strip leading/trailing/multiple spaces
            $descr= $1;
            $descr =~ s/^\s*//;

These changes correspond to:
06e2329636f license: Update deprecated SPDX tag GPL-2.0 to GPL-2.0-only
(a bulk change which we won't bother to re-apply to this third-party script)
b30df2751e5 scripts/kernel-doc: strip QEMU_ from function definitions
4cf41794411 docs: tweak kernel-doc for QEMU coding standards

We have already applied the equivalent of these changes to the
Python code in libs/kdoc/ in the preceding commits.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

scripts/kerneldoc: Switch to the Python kernel-doc script

Change the Sphinx config to run the new Python kernel-doc script
instead of the Perl one.  The only difference between the two is that
the new script does not handle the -sphinx-version option, instead
assuming that Sphinx is always at least version 3: so we must
delete the code that passes that option to avoid the Python
script complaining about an unknown option.

QEMU's minimum Sphinx version is already 3.4.3, so this doesn't
change the set of versions we can handle.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250814171324.1614516-7-peter.maydell@linaro.org

scripts/kernel-doc: tweak for QEMU coding standards

This commit makes the equivalent changes to the Python script that we
had for the old Perl script in commit 4cf41794411f ("docs: tweak
kernel-doc for QEMU coding standards").  To repeat the rationale from
that commit:

    Surprisingly, QEMU does have a pretty consistent doc comment style and
    it is not very different from the Linux kernel's.  Of the documentation
    "sigils", only "#" separates the QEMU doc comment style from Linux's,
    and it has 200+ instances vs. 6 for the kernel's '&struct foo' (all in
    accel/tcg/translate-all.c), so it's clear that the two standards are
    different in this respect.  In addition, our structs are typedefed and
    recognized by CamelCase names.

Note that in 4cf41794411f we used '(?!)' as our type_fallback regex;
this is strictly not quite a replacement for the upstream
'\&([_\w]+)', because the latter includes a group that can later be
matched with \1, and the former does not.  The old perl script did
not care about this, but the python version does, so we must include
the extra set of brackets to ensure we have a group.

This commit does not include all the same changes that 4cf41794411f
did.  Of the missing pieces, some had already gone in an earlier
kernel-doc update; the parts we still had but do not include here are:

    @@ -2057,7 +2060,7 @@
         }
         elsif (/$doc_decl/o) {
            $identifier = $1;
    -       if (/\s*([\w\s]+?)(\(\))?\s*-/) {
    +       if (/\s*([\w\s]+?)(\s*-|:)/) {
                $identifier = $1;
            }

    @@ -2067,7 +2070,7 @@
            $contents = "";
            $section = $section_default;
            $new_start_line = $. + 1;
    -       if (/-(.*)/) {
    +       if (/[-:](.*)/) {
                # strip leading/trailing/multiple spaces
                $descr= $1;
                $descr =~ s/^\s*//;

The second of these is already in the upstream version: the line r =
KernRe("[-:](.*)") in process_name() matches the regex we have.  The
first change has been refactored into the doc_begin_data and
doc_begin_func changes.  Since the output HTML for QEMU's
documentation has no relevant changes with the new kerneldoc, we
assume that this too has been handled upstream.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250814171324.1614516-6-peter.maydell@linaro.org

scripts/kernel-doc: strip QEMU_ from function definitions

This commit is the Python version of our older commit
b30df2751e5 ("scripts/kernel-doc: strip QEMU_ from function definitions").

Some versions of Sphinx get confused if function attributes are
left on the C code from kernel-doc; strip out any QEMU_* prefixes
from function prototypes.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250814171324.1614516-5-peter.maydell@linaro.org

scripts: Import Python kerneldoc from Linux kernel

We last synced our copy of kerneldoc with Linux back in 2020.  In the
interim, upstream has entirely rewritten the script in Python, and
the new Python version is split into a main script plus some
libraries in the kernel's scripts/lib/kdoc.

Import all these files.  These are the versions as of kernel commit
0cc53520e68be, with no local changes.

We use the same lib/kdoc/ directory as the kernel does here, so we
can avoid having to edit the top-level script just to adjust a
pathname, even though it is probably not the naming we would have
picked if this was a purely QEMU script.

The Sphinx conf.py still points at the Perl version of the script,
so this Python code will not be invoked to build the docs yet.

NB: checkpatch complains about many things in this commit,
including the use of "GPL-2.0" rather than "GPL-2.0-only" in
the SPDX tags, but since this is a third party import we can
ignore this.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250814171324.1614516-4-peter.maydell@linaro.org

tests/qtest/libqtest.h: Remove stray space from doc comment

The doc comment for qtest_cb_for_every_machine has a stray
space at the start of its description, which makes kernel-doc
think that this line is part of the documentation of the
skip_old_versioned argument. The result is that the HTML
doesn't have a "Description" section and the text is instead
put in the wrong place.

Remove the stray space.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250814171324.1614516-3-peter.maydell@linaro.org

docs/sphinx/kerneldoc.py: Handle new LINENO syntax

The new upstream kernel-doc that we plan to update to uses a different
syntax for the LINENO directives that the Sphinx extension parses:
instead of
  #define LINENO 86
it has
  .. LINENO 86

Update the kerneldoc.py extension to handle both syntaxes, so
that it will work with both the old and the new kernel-doc.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250814171324.1614516-2-peter.maydell@linaro.org

scripts/kernel-doc: Avoid new Perl precedence warning

Newer versions of Perl (5.41.x and up) emit a warning for code in
kernel-doc:
 Possible precedence problem between ! and pattern match (m//) at /scripts/kernel-doc line 1597.

This is because the code does:
            if (!$param =~ /\w\.\.\.$/) {

In Perl, the !  operator has higher precedence than the =~
pattern-match binding, so the effect of this condition is to first
logically-negate the string $param into a true-or-false value and
then try to pattern match it against the regex, which in this case
will always fail.  This is almost certainly not what the author
intended.

In the new Python version of kernel-doc in the Linux kernel,
the equivalent code is written:

            if KernRe(r'\w\.\.\.$').search(param):
                # For named variable parameters of the form `x...`,
                # remove the dots
                param = param[:-3]
            else:
                # Handles unnamed variable parameters
                param = "..."

which is a more sensible way of writing the behaviour you would
get if you put in brackets to make the regex match first and
then negate the result.

Take this as the intended behaviour, and update the Perl to match.

For QEMU, this produces no change in output, presumably because we
never used the "unnamed variable parameters" syntax.

Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Message-id: 20250819115648.2125709-1-peter.maydell@linaro.org

hw/arm: add static NVDIMMs in device tree

NVDIMM is used for fast rootfs with EROFS, for example by kata
containers. To allow booting with static NVDIMM memory, add them to the
device tree in arm virt machine.

This allows users to boot directly with nvdimm memory devices without
having to rely on ACPI and hotplug.

Verified to work with command invocation:

./qemu-system-aarch64 \
  -M virt,nvdimm=on \
  -cpu cortex-a57 \
  -m 4G,slots=2,maxmem=8G \
  -object memory-backend-file,id=mem1,share=on,mem-path=/tmp/nvdimm,size=4G,readonly=off \
  -device nvdimm,id=nvdimm1,memdev=mem1,unarmed=off \
  -drive file=./debian-12-nocloud-arm64-commited.qcow2,format=qcow2 \
  -kernel ./vmlinuz-6.1.0-13-arm64 \
  -append "root=/dev/vda1 console=ttyAMA0,115200 acpi=off"
  -initrd ./initrd.img-6.1.0-13-arm64 \
  -nographic \
  -serial mon:stdio

Signed-off-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Message-id: 20250807-nvdimm_arm64_virt-v2-1-b8054578bea8@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Enable FEAT_CSSC for -cpu max

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250803014019.416797-7-richard.henderson@linaro.org
[PMM: rebased to handle linux-user elfload.c refactor]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Implement CTZ, CNT, ABS

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250803014019.416797-6-richard.henderson@linaro.org
[PMM: fix tcg_rd/tcg_rn mixup]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Split out gen_wrap2_i32 helper

Wrapper to extract the low 32 bits, perform an operation,
and zero-extend back to 64 bits.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250803014019.416797-5-richard.henderson@linaro.org
[PMM: fixed wrong output-reg argument in callsites; add comment]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Implement MIN/MAX (register)

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250803014019.416797-4-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Implement MIN/MAX (immediate)

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250803014019.416797-3-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Add feature predicate for FEAT_CSSC

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20250803014019.416797-2-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Trap PMCR when MDCR_EL2.TPMCR is set

Trap PMCR_EL0 or PMCR accesses to EL2 when MDCR_EL2.TPMCR is set.
Similar to MDCR_EL2.TPM, MDCR_EL2.TPMCR allows trapping EL0 and EL1
accesses to the PMCR register to EL2.

Cc: qemu-stable@nongnu.org
Signed-off-by: Smail AIDER <smail.aider@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250811112143.1577055-2-smail.aider@huawei.com
Message-Id: <20250722131925.2119169-1-smail.aider@huawei.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3_kvm: preserve pending interrupts during cpr

Close a race condition that causes cpr-transfer to lose VFIO
interrupts on ARM.

CPR stops VCPUs but does not disable VFIO interrupts, which may continue
to arrive throughout the transition to new QEMU.

CPR calls kvm_irqchip_remove_irqfd_notifier_gsi in old QEMU to force
future interrupts to the producer eventfd, where they are preserved.
Old QEMU then destroys the old KVM instance.  However, interrupts may
already be pending in KVM state.  To preserve them, call ioctl
KVM_DEV_ARM_VGIC_SAVE_PENDING_TABLES to flush them to guest RAM, where
they will be picked up when the new KVM+VCPU instance is created.

Cc: qemu-stable@nongnu.org
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Message-id: 1754936384-278328-1-git-send-email-steven.sistare@oracle.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Implement FEAT_TCR2 and enable with -cpu max

Add FEAT_TCR2, which introduces the TCR2_EL1 and TCR2_EL2 registers.
These registers are extensions of the TCR_ELx registers and provide
top-level control of the EL10 and EL20 translation regimes.

Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250727074202.83141-5-richard.henderson@linaro.org
Message-ID: <20250711140828.1714666-5-gustavo.romero@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
[rth: Remove FEAT_MEC code; handle SCR and HCRX enable bits.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Implement FEAT_SCTLR2 and enable with -cpu max

Add FEAT_SCTLR2, which introduces the SCTLR2_EL1, SCTLR2_EL2, and
SCTLR2_EL3 registers. These registers are extensions of the SCTLR_ELx
ones.

Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250727074202.83141-4-richard.henderson@linaro.org
Message-ID: <20250711140828.1714666-4-gustavo.romero@linaro.org>
[rth: Remove FEAT_MEC code; handle SCR and HCRX enable bits.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/functional/test_aarch64_rme: update image

TF-A needs to be patched to enable support for FEAT_TCR2 and
FEAT_SCTLR2. This new image contains updated firmware.

Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20250727074202.83141-3-richard.henderson@linaro.org
Message-ID: <20250719035838.2284029-3-pierrick.bouvier@linaro.org>
[PMM: switch to os.makedirs(..., exist_ok=True) to improve
 robustness when re-run after test was cancelled midway]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/functional/test_aarch64_device_passthrough: update image

TF-A needs to be patched to enable support for FEAT_TCR2 and
FEAT_SCTLR2. This new image contains updated firmware.

Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250727074202.83141-2-richard.henderson@linaro.org
Message-ID: <20250719035838.2284029-2-pierrick.bouvier@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Clean up of register field definitions

Clean up the definitions of NSW and NSA fields in the VTCR register.
These two fields are already defined properly using FIELD() so they are
actually duplications. Also, define the NSW and NSA fields in the
VSTCR register using FIELD() and remove their definitions based on VTCR
fields.

Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Message-id: 20250725014755.2122579-1-gustavo.romero@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge tag 'pull-lu-20250830' of https://gitlab.com/rth7680/qemu into staging

linux-user: Finish elfload.c split
linux-user: Drop deprecated -p option
linux-user: Tidy print_socket_protocol
hw/core: Dump cpu_reset in the reset.exit phase
hw/core: Use qemu_log_trylock/unlock in cpu_common_reset_exit

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmiyKFIdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV/LBggAgMBSVMz1BwkPvckY
# paakdAwuOqRE5yF2YxQAHALJa3aH18Vqk06ENqM9R5iyqvHBGnvrw8fshIBVZnDP
# eQDjNFwnCtSrXuSMdfr0r8jZc+y9R8foQKs9j+KL0ESOi+4VNhORfzFe/yrIEu0y
# XM5XhBjBH0kK9+S20uy5x3WXhRkfqq2CZiUt1izqTOwtbdzYENxdvDj8iDk48FwL
# fkrXUSnlBBsdsltQCsjbrUbWi1Wqj7skswRIzI8KXsj+psy0JJL2kHthaWzm8tTS
# KIXVaOtxtU7LQWhTaknmpcNIkHpnjmEw2ZijxYT29V8WnZtstihVoqqCLbCv6u+7
# JwwQhQ==
# =kHx+
# -----END PGP SIGNATURE-----
# gpg: Signature made Sat 30 Aug 2025 08:23:14 AM AEST
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [ultimate]

* tag 'pull-lu-20250830' of https://gitlab.com/rth7680/qemu: (91 commits)
  linux-user: Remove target_pt_regs from target_syscall.h
  linux-user/sparc: Create target_ptrace.h
  linux-user: Remove a.out declarations from elfload.c
  linux-user: Move arch_parse_elf_property to aarch64/elfload.c
  linux-user: Remove MAP_DENYWRITE from elfload.c
  linux-user: Remove ELIBBAD from elfload.c
  linux-user: Rename elf_check_arch
  linux-user: Standardize on ELF_MACHINE not ELF_ARCH
  linux-user: Move elf parameters to hexagon/target_elf.h
  linux-user: Move elf parameters to xtensa/target_elf.h
  linux-user: Move elf parameters to hppa/target_elf.h
  linux-user: Move elf parameters to riscv/target_elf.h
  linux-user: Move elf parameters to s390x/target_elf.h
  linux-user: Move elf parameters to alpha/target_elf.h
  linux-user: Move elf parameters to m68k/target_elf.h
  linux-user: Move elf parameters to sh4/target_elf.h
  linux-user: Move elf parameters to openrisc/target_elf.h
  linux-user: Move elf parameters to microblaze/target_elf.h
  linux-user: Move elf parameters to {mips,mips64}/target_elf.h
  linux-user: Move elf parameters to loongarch64/target_elf.h
  ...

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

linux-user: Remove target_pt_regs from target_syscall.h

All target_pt_regs which have not been broken out to
target_ptrace.h by this point are unused.  Remove them.

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

linux-user/sparc: Create target_ptrace.h

Move target_pt_regs to target_ptrace.h.

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

linux-user: Remove a.out declarations from elfload.c

These should have been removed with the rest of the stub a.out
support in b9329d4b5321, though they were not in use even then.

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

linux-user: Move arch_parse_elf_property to aarch64/elfload.c

Rename the controlling macro to HAVE_ELF_GNU_PROPERTY to
match the other HAVE_* macros.

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

linux-user: Remove MAP_DENYWRITE from elfload.c

The last use of this fallback was removed in bf858897b769.

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

linux-user: Remove ELIBBAD from elfload.c

The last use of this fallback was removed in 8e62a71738bc.

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

linux-user: Rename elf_check_arch

Rename to elf_check_machine to match ELF_MACHINE.
Remove the unnecessary definition for loongarch64.

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

linux-user: Standardize on ELF_MACHINE not ELF_ARCH

PowerPC was the one outlier that defined both ELF_ARCH and
ELF_MACHINE; ELF_ARCH was defined incorrectly, necessitating
the definition of elf_check_arch.

However, the elf file header field in question is called
e_machine, so ELF_MACHINE is in fact the better name.

Mechanically change most target/target_elf.h files,
then adjust ppc/target_elf.h manually.

Do not provide a default for ELF_MACHINE.

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

linux-user: Move elf parameters to hexagon/target_elf.h

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

linux-user: Move elf parameters to xtensa/target_elf.h

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

linux-user: Move elf parameters to hppa/target_elf.h

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

linux-user: Move elf parameters to riscv/target_elf.h

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

linux-user: Move elf parameters to s390x/target_elf.h

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

linux-user: Move elf parameters to alpha/target_elf.h

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

linux-user: Move elf parameters to m68k/target_elf.h

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

linux-user: Move elf parameters to sh4/target_elf.h

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

linux-user: Move elf parameters to openrisc/target_elf.h

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

linux-user: Move elf parameters to microblaze/target_elf.h

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

linux-user: Move elf parameters to {mips,mips64}/target_elf.h

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

linux-user: Move elf parameters to loongarch64/target_elf.h

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

linux-user: Move elf parameters to ppc/target_elf.h

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

linux-user: Move elf parameters to sparc/target_elf.h

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

linux-user: Move elf parameters to {arm,aarch64}/target_elf.h

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

linux-user: Move elf parameters to {i386,x86_64}/target_elf.h

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

linux-user: Remove redundant ELF_DATA definitons

We already provide ELF_DATA based on TARGET_BIG_ENDIAN.
Remove the extra definitions from openrisc and s390x.

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

linux-user: Remove ELF_EXEC_PAGESIZE

Use TARGET_PAGE_SIZE instead.  If the target page size may vary,
using a different fixed size is wrong.

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

linux-user: Move get_vdso_image_info to arm/elfload.c

Rename from vdso_image_info to avoid a symbol clash.
Define HAVE_VDSO_IMAGE_INFO to signal the external definition exists.
Provide fallback versions for other targets.

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

linux-user: Replace init_guest_commpage macro with function

Turn the fallback macro into a function.  This will produce
a link error if the other macros are set up incorrectly.

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

linux-user: Move init_guest_commpage to hppa/elfload.c

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

linux-user: Move init_guest_commpage to arm/elfload.c

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

linux-user: Move init_guest_commpage to x86_64/elfload.c

Rename INIT_GUEST_COMMPAGE to HAVE_GUEST_COMMPAGE
to match the other HAVE_* defines.

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

linux-user/xtensa: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/xtensa: Create target_ptrace.h

Remove the target_pt_regs; add target_user_pt_regs to
target_ptrace.h, which matches what is actually used.
Remove xtensa_reg_t and xtregs_opt_t.

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

linux-user/sh4: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/sh4: Create target_ptrace.h

Move target_pt_regs to target_ptrace.h.
Convert to abi_foo types.

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

linux-user/s390x: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
This lets us drop the ugly cast to uint32_t* in the middle.

Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/s390x: Create target_ptrace.h

Move target_psw_t to target_ptrace.h.  Note that abi_ulong already
has an attribute for 8-byte alignment, so there's no need to carry
another on target_psw_t.

Remove the target_pt_regs; add target_s390x_reg to target_ptrace.h,
which matches what is actually used.

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

linux-user/ppc: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/ppc: Create target_ptrace.h

Move the target_pt_regs structure from target_syscall.h.

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

linux-user/openrisc: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/openrisc: Create target_ptrace.h

Move the target_pt_regs structure from target_syscall.h and
rename to target_user_regs_struct, obviating the comment.

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

linux-user/mips: Use target_ulong for target_elf_greg_t

Make use of the fact that target_elf_gregset_t is a proper structure.
The target_ulong type matches the abi_ulong/abi_ullong
selection within mips64/target_elf.h.

Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/mips: Create target_ptrace.h

Move the target_pt_regs structure from target_syscall.h.
Fix the incorrect ordering of the fields.

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

linux-user/microblaze: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/microblaze: Fold target_pt_regs.r* to an array

Separately enumerating all 32 registers is not helpful.

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

linux-user/microblaze: Create target_ptrace.h

Move the target_pt_regs structure from target_syscall.h.

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

linux-user/m68k: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/loongarch64: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
Note that the kernel's uses an array, and then it has a bunch of defines
to create symbolic offsets.  Modulo some reserved fields, which we do
not implement here, this is the same layout as struct user_pt_regs.

Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/loongarch64: Create target_ptrace.h

Remove the target_pt_regs structure from target_syscall.h.
Add target_user_pt_regs to target_ptrace.h, which matches
what is actually used on loongarch64.

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

linux-user/arm: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a proper structure.
Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/arm: Create target_ptrace.h

Move the target_pt_regs structure from target_syscall.h.
Replace the array with proper structure fields.

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

linux-user/aarch64: Expand target_elf_gregset_t

Make use of the fact that target_elf_gregset_t is a
proper structure by using target_user_pt_regs.

Drop ELF_NREG, target_elf_greg_t, and tswapreg.

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

linux-user/aarch64: Create target_ptrace.h

Move the target_pt_regs structure from target_syscall.h, and
rename to target_user_pt_regs, to match what's in ptrace.h.

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

linux-user/x86_64: Fix dump of fs_base, gs_base

We were storing the selector, not the base.

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

linux-user/x86_64: Expand target_elf_gregset_t

The comment re ELF_NREG is incorrect or out-of-date.
Make use of the fact that target_elf_gregset_t is a
proper structure by using target_user_regs_struct.

Drop target_elf_greg_t and tswapreg.

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

linux-user/x86_64: Create target_ptrace.h

Remove the target_pt_regs structure from target_syscall.h.
Add target_user_regs_struct to target_ptrace.h, which matches
what is actually used on x86_64.

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

linux-user/i386: Expand target_elf_gregset_t

The comment re ELF_NREG is incorrect or out-of-date.
Make use of the fact that target_elf_gregset_t is a
proper structure by using target_user_regs_struct.

Drop target_elf_greg_t and tswapreg.

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

linux-user/i386: Create target_ptrace.h

Remove the target_pt_regs structure from target_syscall.h.
Add target_user_regs_struct to target_ptrace.h, which is
what is actually used by ELF_CORE_COPY_REGS; the layout
of the two structure definitions is identical.

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

linux-user: Remove target_elf_greg_t, tswapreg from elfload.c

These are no longer used within the generic file.

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

linux-user: Move elf_core_copy_regs to xtensa/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to s390x/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to m68k/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to sh4/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to openrisc/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to microblaze/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to mips/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to loongarch64/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to ppc/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to aarch64/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to arm/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Move elf_core_copy_regs to {i386,x86_64}/elfload.c

Move elf_core_copy_regs to elfload.c.
Move HAVE_ELF_CORE_DUMP, ELF_NREGS, target_elf_gregset_t to target_elf.h.
For now, duplicate the definitions of target_elf_greg_t and tswapreg.

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

linux-user: Rename USE_ELF_CORE_DUMP to HAVE_ELF_CORE_DUMP

The other knobs in target_elf.h are all HAVE_*.
Rename this USE_ELF_CORE_DUMP to match.

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