migration: Document all migration_stats

Message-Id: <20230504103357.22130-2-quintela@redhat.com>
Reviewed-by: David Edmondson <david.edmondson@oracle.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration/rdma: Don't pass the QIOChannelRDMA as an opaque

We can calculate it from the QEMUFile like the caller.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230503131847.11603-6-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Fix block_bitmap_mapping migration

It is valid that params->has_block_bitmap_mapping is true and
params->block_bitmap_mapping is NULL.  So we can't use the trick of
having a single function.

Move to two functions one for each value and the tests are fixed.

Fixes: b804b35b1c8a0edfd127ac20819c234be55ac7fc
       migration: Create migrate_block_bitmap_mapping() function

Reported-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-Id: <20230503181036.14890-1-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Merge tag 'qga-pull-2023-05-04' of https://github.com/kostyanf14/qemu into staging

qga-pull-2023-05-04

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCgAdFiEEwsLBCepDxjwUI+uE711egWG6hOcFAmRTf4EACgkQ711egWG6
# hOdczhAAoMkw/bWZtpiEmeLIcliNVuF3gzouHRp2sBFeHCUtaE2jzmriuy4QqH81
# G+kgqdEogqv7G5Qq2LXNcbSl76eZnKbumtlj+N4XIDoukPWUmItIjuM9NYM0U84Y
# DAsj6o8Kw1W+GMBmk6AuLFNYMwv41GS6RyvH/daeYEGmFf0jYIARpeldTUxkxY86
# dfEylVPBfLjkctKfgl0h5GzRkYjmxyeisyigx4Wk8sVLvz5LLXx27sFVZ7//Rn16
# 7Ca88kzF9SUenF+ulV0HmVgR02b69w74izVII/Hh3pPkv4T2m98DeXlnHdyuc3Lb
# bWggM2pf1Yr/jUOPtscQ9IqKTdYTLKqaErVSQJYabG9lLXRerXLj8mhH/W070jzT
# hZwp+5VJzF/OykiM0PW/tI5a0Upg8/8w/LRPwuqP0nFsW5QYdAQXSFbtsUfVdPlE
# Htk66xoQTLJ2a38m5WCvaRL97psGHmVXuWq0VUByNCAlWv6EiAAbTZBimIo4FZj+
# Ps6e7Tnvhv5kai2qZ4ijxJq88n9/mv8t0s/oD/BoGQMXmNn+87bR18byY5NZWi/4
# CKk60zbmgo8fGZ/zzwk4T+48I3SLfM+wnHePYs8nrHgoqGVEUXsCdXUYifxIcpel
# Vnb1OHXN3n+U8LnleQ158oQcSVFicvwQMBkzM+mW16WMQlReRRM=
# =GvNm
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 04 May 2023 10:48:49 AM BST
# gpg:                using RSA key C2C2C109EA43C63C1423EB84EF5D5E8161BA84E7
# gpg: Good signature from "Kostiantyn Kostiuk (Upstream PR sign) <kkostiuk@redhat.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: C2C2 C109 EA43 C63C 1423  EB84 EF5D 5E81 61BA 84E7

* tag 'qga-pull-2023-05-04' of https://github.com/kostyanf14/qemu:
  qga: Fix suspend on Linux guests without systemd
  qga/commands-win32.c: Drop the check for _WIN32_WINNT >= 0x0601
  qga: test: Add tests for `merged` flag
  qga: Add `merged` variant to GuestExecCaptureOutputMode
  qga: Refactor guest-exec capture-output to take enum
  qga/linux: add usb support to guest-get-fsinfo

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

qga: Fix suspend on Linux guests without systemd

Allow the Linux guest agent to attempt each of the suspend methods
(systemctl, pm-* and writing to /sys) in turn.

Prior to this guests without systemd failed to suspend due to
`guest_suspend` returning early regardless of the return value of
`systemd_supports_mode`.

Signed-off-by: Mark Somerville <mark@qpok.net>
Reviewed-by: Konstantin Kostiuk <kkostiuk@redhat.com>
Signed-off-by: Konstantin Kostiuk <kkostiuk@redhat.com>

qga/commands-win32.c: Drop the check for _WIN32_WINNT >= 0x0601

All current versions of glib require _WIN32_WINNT set to 0x0601
or higher already, and we also use this value as a minimum in our
osdep.h header file, so there is no way to still compile this code
with an older version of the Windows ABI. Thus we can drop this
check now.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Konstantin Kostiuk <kkostiuk@redhat.com>
Signed-off-by: Konstantin Kostiuk <kkostiuk@redhat.com>

qga: test: Add tests for `merged` flag

This commit adds a test to ensure `merged` functions as expected.
We also add a negative test to ensure we haven't regressed previous
functionality.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
Signed-off-by: Konstantin Kostiuk <kkostiuk@redhat.com>

qga: Add `merged` variant to GuestExecCaptureOutputMode

Currently, any captured output (via `capture-output`) is segregated into
separate GuestExecStatus fields (`out-data` and `err-data`). This means
that downstream consumers have no way to reassemble the captured data
back into the original stream.

This is relevant for chatty and semi-interactive (ie. read only) CLI
tools.  Such tools may deliberately interleave stdout and stderr for
visual effect. If segregated, the output becomes harder to visually
understand.

This commit adds a new enum variant to the GuestExecCaptureOutputMode
qapi to merge the output streams such that consumers can have a pristine
view of the original command output.

Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Konstantin Kostiuk <kkostiuk@redhat.com>

qga: Refactor guest-exec capture-output to take enum

Previously capture-output was an optional boolean flag that either
captured all output or captured none. While this is OK in most cases, it
lacks flexibility for more advanced capture cases, such as wanting to
only capture stdout.

This commits refactors guest-exec qapi to take an enum for capture mode
instead while preserving backwards compatibility.

Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
Signed-off-by: Konstantin Kostiuk <kkostiuk@redhat.com>

qga/linux: add usb support to guest-get-fsinfo

Signed-off-by: Kfir Manor <kfir@daynix.com>
Reviewed-by: Konstantin Kostiuk <kkostiuk@redhat.com>
Signed-off-by: Konstantin Kostiuk <kkostiuk@redhat.com>

Merge tag 'pull-nbd-2023-05-03' of https://repo.or.cz/qemu/ericb into staging

nbd patches for 2023-05-03

- Eric Blake: clear LISTEN_FDNAMES when consuming systemd sockets
- Stefan Hajnoczi: clear export BlockDeviceOps in central location

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEccLMIrHEYCkn0vOqp6FrSiUnQ2oFAmRSr2gACgkQp6FrSiUn
# Q2qGIQf/Yg7Qwplxz9U+1VZ2sLwMU4NukzeYQFb5K3cdJUz0UzcZW7EZ4eL5Ll1F
# g6UUni+1y+xnTnN+isz062uSo05WCOoLdV0MKYPWjILqe2HpqYLK7Kt2Y8Syh/ET
# aiGoFl2T510A3M8QTNrQIGnul7g3JBRR1ar3mm7u5SD9gKxjlRFftEQLU8GrAEZP
# Bms122pN8Cvu/vRnBraeReNcTwdtWSQYtOvg2BHL+gROWcmS+FAtSpPOjXgjSSTa
# /K5QgBoph5I4H9sJVpN/nbi5aJkhm/A559tv/VR7SP862u6qz3n+2rJSoxee5S7x
# FgxJ/o03ZeTVFfIE3LdVycvOokPMJw==
# =DNEY
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 03 May 2023 08:00:56 PM BST
# 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-2023-05-03' of https://repo.or.cz/qemu/ericb:
  block/export: call blk_set_dev_ops(blk, NULL, NULL)
  systemd: Also clear LISTEN_FDNAMES during systemd socket activation

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

block/export: call blk_set_dev_ops(blk, NULL, NULL)

Most export types install BlockDeviceOps pointers. It is easy to forget
to remove them because that happens automatically via the "drive" qdev
property in hw/ but not block/export/.

Put blk_set_dev_ops(blk, NULL, NULL) calls in the core export.c code so
the export types don't need to remember.

This fixes the nbd and vhost-user-blk export types.

Fixes: fd6afc501a01 ("nbd/server: Use drained block ops to quiesce the server")
Fixes: ca858a5fe94c ("vhost-user-blk-server: notify client about disk resize")
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20230502211119.720647-1-stefanha@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

systemd: Also clear LISTEN_FDNAMES during systemd socket activation

Some time after systemd documented LISTEN_PID and LISTEN_FDS for
socket activation, they later added LISTEN_FDNAMES; now documented at:
https://www.freedesktop.org/software/systemd/man/sd_listen_fds.html

In particular, look at the implementation of sd_listen_fds_with_names():
https://github.com/systemd/systemd/blob/main/src/libsystemd/sd-daemon/sd-daemon.c

If we ever pass LISTEN_PID=xxx and LISTEN_FDS=n to a child process,
but leave LISTEN_FDNAMES=... unchanged as inherited from our parent
process, then our child process using sd_listen_fds_with_names() might
see a mismatch in the number of names (unexpected -EINVAL failure), or
even if the number of names matches the values of those names may be
unexpected (with even less predictable results).

Usually, this is not an issue - the point of LISTEN_PID is to tell
systemd socket activation to ignore all other LISTEN_* if they were
not directed to this particular pid.  But if we end up consuming a
socket directed to this qemu process, and later decide to spawn a
child process that also needs systemd socket activation, we must
ensure we are not leaking any stale systemd variables through to that
child.  The easiest way to do this is to wipe ALL LISTEN_* variables
at the time we consume a socket, even if we do not yet care about a
LISTEN_FDNAMES passed in from the parent process.

See also https://lists.freedesktop.org/archives/systemd-devel/2023-March/048920.html

Thanks: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20230324153349.1123774-1-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>

Merge tag 'migration-20230428-pull-request' of https://gitlab.com/juan.quintela/qemu into staging

Migraiton Pull request (20230428 take 2)

Hi

Dropped the compression cleanups to see if we find what is going on.

Please apply.

Later, Juan.

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEGJn/jt6/WMzuA0uC9IfvGFhy1yMFAmRSKKQACgkQ9IfvGFhy
# 1yN1Tw//ZBExxagVY5Uy7oUp+YBBexK3PkcWrU5xNVIu1SiFr654AWWO6DUoMIde
# pb0EsM1rU/7iSPw3PYOPJDAO2eHmzeeLxGTCbR98R4YSHgr5WBHcRmzKnY5irKIm
# JDn4xxyp8+grCcMTi5nJ6CMVDvtL8CAA1fZfn2zzJimJkuSRkTAYgGw1fIXXut05
# 911eOccmA3mUFsU81bOXLvWhmrPCyy5m2vrLhIhwEJGTUdVdQlk72at7aWq1pu9D
# WIZJG6hI/eGZW5DX6Xbj7tfLW2MeqBFpFijuCNsxU4jAGSoBHR7And9YO/i7kDCp
# 3RlQyL/sVf6NZ7STEXrLEIh7g7nH05AF3ZuuwBn8TN2Vf9NkVPNFUygdmwukzHTj
# i8/VoC/gSoVfnA4JXVhnVtzHwRpgzP6rlKSeHEW1IMMnKC0xvmss6Mq7+z9dZxx+
# /XbxdjjwmsjnYUxH+BNwUVX2PjPguF5xib8M1QhsQNGOZ2Z+4RuWj4h/ItWsGqVI
# jjI3sVPY6cw8aiSXiSL2AO9Hde9KRIkcITby/zembSkvfsRR3kIWcMNJsteAP4UI
# t6MQSGsFRjMUKeIyArU/ES3iYvnBXi9LSwOR94S9KR1vEPxmhefoGu5wnwJ0is3C
# nAJPaHkQCY4JCCsyNO18hpSStOlNH300fIlMoB+ftcbn+fnpeOo=
# =iGGF
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 03 May 2023 10:25:56 AM BST
# gpg:                using RSA key 1899FF8EDEBF58CCEE034B82F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>" [undefined]
# gpg:                 aka "Juan Quintela <quintela@trasno.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03  4B82 F487 EF18 5872 D723

* tag 'migration-20230428-pull-request' of https://gitlab.com/juan.quintela/qemu:
  migration: Drop unused parameter for migration_tls_client_create()
  migration: Drop unused parameter for migration_tls_get_creds()
  migration/rdma: Unfold last user of acct_update_position()
  migration/rdma: Split the zero page case from acct_update_position
  migration: Rename RAMStats to MigrationAtomicStats
  migration: Rename ram_counters to mig_stats
  migration: Move ram_stats to its own file migration-stats.[ch]
  multifd: We already account for this packet on the multifd thread

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

migration: Drop unused parameter for migration_tls_client_create()

It is not needed since we moved the accessor for tls properties to
options.c.

Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

migration: Drop unused parameter for migration_tls_get_creds()

It is not needed since we moved the accessor for tls properties to
options.c.

Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

migration/rdma: Unfold last user of acct_update_position()

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>

migration/rdma: Split the zero page case from acct_update_position

Now that we have atomic counters, we can do it on the place that we
need it, no need to do it inside ram.c.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>

migration: Rename RAMStats to MigrationAtomicStats

It is lousely based on MigrationStats, but that name is taken, so this
is the best one that I came with.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>
---

If you have any good suggestion for the name, I am all ears.

migration: Rename ram_counters to mig_stats

migration_stats is just too long, and it is going to have more than
ram counters in the near future.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>

migration: Move ram_stats to its own file migration-stats.[ch]

There is already include/qemu/stats.h, so stats.h was a bad idea.
We want this file to not depend on anything else, we will move all the
migration counters/stats to this struct.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>

multifd: We already account for this packet on the multifd thread

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>

Merge tag 'pull-tcg-20230502-2' of https://gitlab.com/rth7680/qemu into staging

Misc tcg-related patch queue.

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmRRb30dHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV+obAgAmL4F1gdkbUUPKnkv
# poYwutCX+c3kog22TF29BlKgC8vJa6UbRLMphz5q7v3wbCKQJMeNV/sKa+mhnHBK
# CB3wP8xXVAahWFARmWTIZEqlB3HQ/RIzhc5saKkiSzcGIrtXUj6fdfrz7mae+w/g
# kDGCbK8hGyuE580j9QAIPbpfqPoNhIPziECFA1AsNf5Krpxc1nDqIfZEuUzTLtLO
# 1WoSaUVbiGDQrTe2OVKF2mtrGbr2vWI1vnHJl67Lom6rG0LzOjb3W/8IN+n0+46E
# 7pMlUCDT1zeTxevRxBvDmwgCYA/QjFosd4enUuhVReTxTNhUc69+QyuOAhHO/IEq
# T0V3eA==
# =qZDQ
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 02 May 2023 09:15:57 PM BST
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [ultimate]

* tag 'pull-tcg-20230502-2' of https://gitlab.com/rth7680/qemu:
  tcg: Introduce tcg_out_movext2
  tcg/mips: Conditionalize tcg_out_exts_i32_i64
  tcg/loongarch64: Conditionalize tcg_out_exts_i32_i64
  accel/tcg: Add cpu_ld*_code_mmu
  migration/xbzrle: Use __attribute__((target)) for avx512
  qemu/int128: Re-shuffle Int128Alias members
  tcg: Add tcg_gen_gvec_rotrs
  tcg: Add tcg_gen_gvec_andcs
  qemu/host-utils.h: Add clz and ctz functions for lower-bit integers
  qemu/bitops.h: Limit rotate amounts
  accel/tcg: Uncache the host address for instruction fetch when tlb size < 1
  softmmu: Tidy dirtylimit_dirty_ring_full_time

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

tcg: Introduce tcg_out_movext2

This is common code in most qemu_{ld,st} slow paths, moving two
registers when there may be overlap between sources and destinations.
At present, this is only used by 32-bit hosts for 64-bit data,
but will shortly be used for more than that.

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

tcg/mips: Conditionalize tcg_out_exts_i32_i64

Since TCG_TYPE_I32 values are kept sign-extended in registers, we need not
extend if the register matches.  This is already relied upon by comparisons.

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

tcg/loongarch64: Conditionalize tcg_out_exts_i32_i64

Since TCG_TYPE_I32 values are kept sign-extended in registers,
via ".w" instructions, we need not extend if the register matches.
This is already relied upon by comparisons.

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

accel/tcg: Add cpu_ld*_code_mmu

At least RISC-V has the need to be able to perform a read
using execute permissions, outside of translation.
Add helpers to facilitate this.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Tested-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230325105429.1142530-9-richard.henderson@linaro.org>
Message-Id: <20230412114333.118895-9-richard.henderson@linaro.org>

migration/xbzrle: Use __attribute__((target)) for avx512

Use the attribute, which is supported by clang, instead of
the #pragma, which is not supported and, for some reason,
also not detected by the meson probe, so we fail by -Werror.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20230501210555.289806-1-richard.henderson@linaro.org>

qemu/int128: Re-shuffle Int128Alias members

Clang 14, with --enable-tcg-interpreter errors with

include/qemu/int128.h:487:16: error: alignment of field 'i' (128 bits)
  does not match the alignment of the first field in transparent union;
  transparent_union attribute ignored [-Werror,-Wignored-attributes]
    __int128_t i;
               ^
include/qemu/int128.h:486:12: note: alignment of first field is 64 bits
    Int128 s;
           ^
1 error generated.

By placing the __uint128_t member first, this is avoided.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230501204625.277361-1-richard.henderson@linaro.org>

tcg: Add tcg_gen_gvec_rotrs

Add tcg expander and helper functions for rotate right
vector with scalar operand.

Signed-off-by: Nazar Kazakov <nazar.kazakov@codethink.co.uk>
Message-Id: <20230428144757.57530-10-lawrence.hunter@codethink.co.uk>
[rth: Split out of larger patch; mask rotation count.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Add tcg_gen_gvec_andcs

Add tcg expander and helper functions for and-compliment
vector with scalar operand.

Signed-off-by: Nazar Kazakov <nazar.kazakov@codethink.co.uk>
Message-Id: <20230428144757.57530-10-lawrence.hunter@codethink.co.uk>
[rth: Split out of larger patch.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

qemu/host-utils.h: Add clz and ctz functions for lower-bit integers

This is for use in the RISC-V vclz and vctz instructions (implemented in
proceeding commit).

Signed-off-by: Kiran Ostrolenk <kiran.ostrolenk@codethink.co.uk>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230428144757.57530-11-lawrence.hunter@codethink.co.uk>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

qemu/bitops.h: Limit rotate amounts

Rotates have been fixed up to only allow for reasonable rotate amounts
(ie, no rotates >7 on an 8b value etc.)  This fixes a problem with riscv
vector rotate instructions.

Signed-off-by: Dickon Hood <dickon.hood@codethink.co.uk>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230428144757.57530-9-lawrence.hunter@codethink.co.uk>
[rth: Mask shifts in both directions.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Uncache the host address for instruction fetch when tlb size < 1

When PMP entry overlap part of the page, we'll set the tlb_size to 1, which
will make the address in tlb entry set with TLB_INVALID_MASK, and the next
access will again go through tlb_fill.However, this way will not work in
tb_gen_code() => get_page_addr_code_hostp(): the TLB host address will be
cached, and the following instructions can use this host address directly
which may lead to the bypass of PMP related check.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1542.

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230422130329.23555-6-liweiwei@iscas.ac.cn>

softmmu: Tidy dirtylimit_dirty_ring_full_time

Drop inline marker: let compiler decide.

Change return type to uint64_t: this matches the computation in the
return statement and the local variable assignment in the caller.

Rename local to dirty_ring_size_MB to fix typo.
Simplify conversion to MiB via qemu_target_page_bits and right shift.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Merge tag 'pull-target-arm-20230502-2' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * Support building Arm targets with CONFIG_TCG=no (ie KVM only)
 * hw/net: npcm7xx_emc: set MAC in register space
 * hw/arm/bcm2835_property: Implement "get command line" message
 * Deprecate the '-singlestep' command line option in favour of
   '-one-insn-per-tb' and '-accel one-insn-per-tb=on'
 * Deprecate 'singlestep' member of QMP StatusInfo struct
 * docs/about/deprecated.rst: Add "since 7.1" tag to dtb-kaslr-seed deprecation
 * hw/net/msf2-emac: Don't modify descriptor in-place in emac_store_desc()
 * raspi, aspeed: Write bootloader code correctly on big-endian hosts
 * hw/intc/allwinner-a10-pic: Fix bug on big-endian hosts
 * Fix bug in A32 ERET on big-endian hosts that caused guest crash
 * hw/sd/allwinner-sdhost: Correctly byteswap descriptor fields
 * hw/net/allwinner-sun8i-emac: Correctly byteswap descriptor fields

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmRRIqoZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3uHxEACkO5NYXhah9KnztwhEjAUH
# CnM6N9IHa4iEF7doWuiS39ZP4bXxCHnX4A1GqGRhsoub5PeiXxucTXlvkwWpOfct
# pX78wHW18pVNJ2jlaly0c9cEI71ojT4zbXC3kjD9f/qHx2qI3rs3Sgb6BYC8QNnx
# P/EHeefrnjwgGhBvzAJ8ATO/jpMYXukuxzjbIP0/7lqB5UxcNxv5rMTTansMO46r
# JN5NdOEM8M8DoJHrHR9K3+Y2Vr1XjOowDPSF8+4rRkJB2v/3415V9bInEsTAvhQu
# Ftua72vjVhTRfvgLXPc9C9S5sx6KHi+NdHfl/D7eL8B4aGHtBaUXv7pJrXUZYyLy
# XNztXUx5EuzxXVN2mg8PYyasnEGjNbTUckAD40iow+DSgemB+MVJRp8f7Rb2yRHh
# YuajDs77NUb9rVzozM+TTJkHfgLDkWCqX2Jm2kAea/gwowzdFCVosAs0cI+cDiBb
# xQUMpERGBE2QJk/+KKc5xmIHUnXZCFTC/ieY2mpr8G6upDspzP254EjNGUCgIZmW
# gYI/UTSX+f7M/+fYRgtSCdJ4LYkqdxUuGfyKccc4S2F2cCuQDGURRp4jHuI1cLyt
# lkrgD1Hj3d9d8ZiMwmXDtiEsJhxDmuVmikmviigfhjLZ0QBd7FdpNz7gQR3lfDwl
# YEGeFrhW2MHutjWSwxQWwQ==
# =ua3Q
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 02 May 2023 03:48:10 PM BST
# 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]

* tag 'pull-target-arm-20230502-2' of https://git.linaro.org/people/pmaydell/qemu-arm: (34 commits)
  hw/net/allwinner-sun8i-emac: Correctly byteswap descriptor fields
  hw/sd/allwinner-sdhost: Correctly byteswap descriptor fields
  target/arm: Add compile time asserts to load/store_cpu_field macros
  target/arm: Define and use new load_cpu_field_low32()
  hw/intc/allwinner-a10-pic: Don't use set_bit()/clear_bit()
  hw/arm/raspi: Use arm_write_bootloader() to write boot code
  hw/arm/aspeed: Use arm_write_bootloader() to write the bootloader
  hw/arm/boot: Make write_bootloader() public as arm_write_bootloader()
  hw/net/msf2-emac: Don't modify descriptor in-place in emac_store_desc()
  docs/about/deprecated.rst: Add "since 7.1" tag to dtb-kaslr-seed deprecation
  qmp: Deprecate 'singlestep' member of StatusInfo
  qapi/run-state.json: Fix missing newline at end of file
  hmp: Add 'one-insn-per-tb' command equivalent to 'singlestep'
  accel/tcg: Report one-insn-per-tb in 'info jit', not 'info status'
  Document that -singlestep command line option is deprecated
  bsd-user: Add '-one-insn-per-tb' option equivalent to '-singlestep'
  linux-user: Add '-one-insn-per-tb' option equivalent to '-singlestep'
  accel/tcg: Use one_insn_per_tb global instead of old singlestep global
  softmmu: Don't use 'singlestep' global in QMP and HMP commands
  make one-insn-per-tb an accel option
  ...

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

hw/net/allwinner-sun8i-emac: Correctly byteswap descriptor fields

In allwinner-sun8i-emac we just read directly from guest memory into
a host FrameDescriptor struct and back.  This only works on
little-endian hosts.  Reading and writing of descriptors is already
abstracted into functions; make those functions also handle the
byte-swapping so that TransferDescriptor structs as seen by the rest
of the code are always in host-order, and fix two places that were
doing ad-hoc descriptor reading without using the functions.

Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230424165053.1428857-3-peter.maydell@linaro.org

hw/sd/allwinner-sdhost: Correctly byteswap descriptor fields

In allwinner_sdhost_process_desc() we just read directly from
guest memory into a host TransferDescriptor struct and back.
This only works on little-endian hosts. Abstract the reading
and writing of descriptors into functions that handle the
byte-swapping so that TransferDescriptor structs as seen by
the rest of the code are always in host-order.

This fixes a failure of one of the avocado tests on s390.

Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230424165053.1428857-2-peter.maydell@linaro.org

target/arm: Add compile time asserts to load/store_cpu_field macros

Add some compile-time asserts to the load_cpu_field() and store_cpu_field()
macros that the struct field being accessed is the expected size. This
lets us catch cases where we incorrectly tried to do a 32-bit load
from a 64-bit struct field.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230424153909.1419369-3-peter.maydell@linaro.org

target/arm: Define and use new load_cpu_field_low32()

In several places in the 32-bit Arm translate.c, we try to use
load_cpu_field() to load from a CPUARMState field into a TCGv_i32
where the field is actually 64-bit. This works on little-endian
hosts, but gives the wrong half of the register on big-endian.

Add a new load_cpu_field_low32() which loads the low 32 bits
of a 64-bit field into a TCGv_i32. The new macro includes a
compile-time check against accidentally using it on a field
of the wrong size. Use it to fix the two places in the code
where we were using load_cpu_field() on a 64-bit field.

This fixes a bug where on big-endian hosts the guest would
crash after executing an ERET instruction, and a more corner
case one where some UNDEFs for attempted accesses to MSR
banked registers from Secure EL1 might go to the wrong EL.

Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230424153909.1419369-2-peter.maydell@linaro.org

hw/intc/allwinner-a10-pic: Don't use set_bit()/clear_bit()

The Allwinner PIC model uses set_bit() and clear_bit() to update the
values in its irq_pending[] array when an interrupt arrives.  However
it is using these functions wrongly: they work on an array of type
'long', and it is passing an array of type 'uint32_t'.  Because the
code manually figures out the right array element, this works on
little-endian hosts and on 32-bit big-endian hosts, where bits 0..31
in a 'long' are in the same place as they are in a 'uint32_t'.
However it breaks on 64-bit big-endian hosts.

Remove the use of set_bit() and clear_bit() in favour of using
deposit32() on the array element.  This fixes a bug where on
big-endian 64-bit hosts the guest kernel would hang early on in
bootup.

Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230424152833.1334136-1-peter.maydell@linaro.org

hw/arm/raspi: Use arm_write_bootloader() to write boot code

When writing the secondary-CPU stub boot loader code to the guest,
use arm_write_bootloader() instead of directly calling
rom_add_blob_fixed().  This fixes a bug on big-endian hosts, because
arm_write_bootloader() will correctly byte-swap the host-byte-order
array values into the guest-byte-order to write into the guest
memory.

Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Tested-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230424152717.1333930-4-peter.maydell@linaro.org

hw/arm/aspeed: Use arm_write_bootloader() to write the bootloader

When writing the secondary-CPU stub boot loader code to the guest,
use arm_write_bootloader() instead of directly calling
rom_add_blob_fixed().  This fixes a bug on big-endian hosts, because
arm_write_bootloader() will correctly byte-swap the host-byte-order
array values into the guest-byte-order to write into the guest
memory.

Cc: qemu-stable@nongnu.org
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Tested-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230424152717.1333930-3-peter.maydell@linaro.org
[PMM: Moved the "make arm_write_bootloader() function public" part
 to its own patch; updated commit message to note that this fixes
 an actual bug; adjust to the API changes noted in previous commit]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/boot: Make write_bootloader() public as arm_write_bootloader()

The arm boot.c code includes a utility function write_bootloader()
which assists in writing a boot-code fragment into guest memory,
including handling endianness and fixing it up with entry point
addresses and similar things.  This is useful not just for the boot.c
code but also in board model code, so rename it to
arm_write_bootloader() and make it globally visible.

Since we are making it public, make its API a little neater: move the
AddressSpace* argument to be next to the hwaddr argument, and allow
the fixupcontext array to be const, since we never modify it in this
function.

Cc: qemu-stable@nongnu.org
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Tested-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230424152717.1333930-2-peter.maydell@linaro.org
[PMM: Split out from another patch by Cédric, added doc comment]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/net/msf2-emac: Don't modify descriptor in-place in emac_store_desc()

The msf2-emac ethernet controller has functions emac_load_desc() and
emac_store_desc() which read and write the in-memory descriptor
blocks and handle conversion between guest and host endianness.

As currently written, emac_store_desc() does the endianness
conversion in-place; this means that it effectively consumes the
input EmacDesc struct, because on a big-endian host the fields will
be overwritten with the little-endian versions of their values.
Unfortunately, in all the callsites the code continues to access
fields in the EmacDesc struct after it has called emac_store_desc()
-- specifically, it looks at the d.next field.

The effect of this is that on a big-endian host networking doesn't
work because the address of the next descriptor is corrupted.

We could fix this by making the callsite avoid using the struct; but
it's more robust to have emac_store_desc() leave its input alone.

(emac_load_desc() also does an in-place conversion, but here this is
fine, because the function is supposed to be initializing the
struct.)

Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-id: 20230424151919.1333299-1-peter.maydell@linaro.org

docs/about/deprecated.rst: Add "since 7.1" tag to dtb-kaslr-seed deprecation

In commit 5242876f37ca we deprecated the dtb-kaslr-seed property of
the virt board, but forgot the "since n.n" tag in the documentation
of this in deprecated.rst.

This deprecation note first appeared in the 7.1 release, so
retrospectively add the correct "since 7.1" annotation to it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-id: 20230420122256.1023709-1-peter.maydell@linaro.org

qmp: Deprecate 'singlestep' member of StatusInfo

The 'singlestep' member of StatusInfo has never done what the QMP
documentation claims it does.  What it actually reports is whether
TCG is working in "one guest instruction per translation block" mode.

We no longer need this field for the HMP 'info status' command, as
we've moved that information to 'info jit'.  It seems unlikely that
anybody is monitoring the state of this obscure TCG setting via QMP,
especially since QMP provides no means for changing the setting.  So
simply deprecate the field, without providing any replacement.

Until we do eventually delete the member, correct the misstatements
in the QAPI documentation about it.

If we do find that there are users for this, then the most likely way
we would provide replacement access to the information would be to
put the accelerator QOM object at a well-known path such as
/machine/accel, which could then be used with the existing qom-set
and qom-get commands.

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: Markus Armbruster <armbru@redhat.com>
Message-id: 20230417164041.684562-11-peter.maydell@linaro.org

qapi/run-state.json: Fix missing newline at end of file

The run-state.json file is missing a trailing newline; add it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230417164041.684562-10-peter.maydell@linaro.org

hmp: Add 'one-insn-per-tb' command equivalent to 'singlestep'

The 'singlestep' HMP command is confusing, because it doesn't
actually have anything to do with single-stepping the CPU.  What it
does do is force TCG emulation to put one guest instruction in each
TB, which can be useful in some situations.

Create a new HMP command  'one-insn-per-tb', so we can document that
'singlestep' is just a deprecated synonym for it, and eventually
perhaps drop it.

We aren't obliged to do deprecate-and-drop for HMP commands,
but it's easy enough to do so, so we do.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230417164041.684562-9-peter.maydell@linaro.org

accel/tcg: Report one-insn-per-tb in 'info jit', not 'info status'

Currently we report whether the TCG accelerator is in
'one-insn-per-tb' mode in the 'info status' output.  This is a pretty
minor piece of TCG specific information, and we want to deprecate the
'singlestep' field of the associated QMP command.  Move the
'one-insn-per-tb' reporting to 'info jit'.

We don't need a deprecate-and-drop period for this because the
HMP interface has no stability guarantees.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230417164041.684562-8-peter.maydell@linaro.org

Document that -singlestep command line option is deprecated

Document that the -singlestep command line option is now
deprecated, as it is replaced by either the TCG accelerator
property 'one-insn-per-tb' for system emulation or the new
'-one-insn-per-tb' option for usermode emulation, and remove
the only use of the deprecated syntax from a README.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230417164041.684562-7-peter.maydell@linaro.org

bsd-user: Add '-one-insn-per-tb' option equivalent to '-singlestep'

The '-singlestep' option is confusing, because it doesn't actually
have anything to do with single-stepping the CPU. What it does do
is force TCG emulation to put one guest instruction in each TB,
which can be useful in some situations.

Create a new command line argument -one-insn-per-tb, so we can
document that -singlestep is just a deprecated synonym for it,
and eventually perhaps drop it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230417164041.684562-6-peter.maydell@linaro.org

linux-user: Add '-one-insn-per-tb' option equivalent to '-singlestep'

The '-singlestep' option is confusing, because it doesn't actually
have anything to do with single-stepping the CPU. What it does do
is force TCG emulation to put one guest instruction in each TB,
which can be useful in some situations.

Create a new command line argument -one-insn-per-tb, so we can
document that -singlestep is just a deprecated synonym for it,
and eventually perhaps drop it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230417164041.684562-5-peter.maydell@linaro.org

accel/tcg: Use one_insn_per_tb global instead of old singlestep global

The only place left that looks at the old 'singlestep' global
variable is the TCG curr_cflags() function.  Replace the old global
with a new 'one_insn_per_tb' which is defined in tcg-all.c and
declared in accel/tcg/internal.h.  This keeps it restricted to the
TCG code, unlike 'singlestep' which was available to every file in
the system and defined in multiple different places for softmmu vs
linux-user vs bsd-user.

While we're making this change, use qatomic_read() and qatomic_set()
on the accesses to the new global, because TCG will read it without
holding a lock.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230417164041.684562-4-peter.maydell@linaro.org

softmmu: Don't use 'singlestep' global in QMP and HMP commands

The HMP 'singlestep' command, the QMP 'query-status' command and the
HMP 'info status' command (which is just wrapping the QMP command
implementation) look at the 'singlestep' global variable. Make them
access the new TCG accelerator 'one-insn-per-tb' property instead.

This leaves the HMP and QMP command/field names and output strings
unchanged; we will clean that up later.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230417164041.684562-3-peter.maydell@linaro.org

make one-insn-per-tb an accel option

This commit adds 'one-insn-per-tb' as a property on the TCG
accelerator object, so you can enable it with
   -accel tcg,one-insn-per-tb=on

It has the same behaviour as the existing '-singlestep' command line
option.  We use a different name because 'singlestep' has always been
a confusing choice, because it doesn't have anything to do with
single-stepping the CPU.  What it does do is force TCG emulation to
put one guest instruction in each TB, which can be useful in some
situations (such as analysing debug logs).

The existing '-singlestep' commandline options are decoupled from the
global 'singlestep' variable and instead now are syntactic sugar for
setting the accel property.  (These can then go away after a
deprecation period.)

The global variable remains for the moment as:
 * what the TCG code looks at to change its behaviour
 * what HMP and QMP use to query and set the behaviour

In the following commits we'll clean those up to not directly
look at the global variable.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230417164041.684562-2-peter.maydell@linaro.org

hw/arm/bcm2835_property: Implement "get command line" message

This query copies the kernel command line into the message buffer. It
was previously stubbed out to return empty, this commit makes it reflect
the arguments specified with `-append`.

I observed the following peculiarities on my Pi 3B+:
- If the buffer is shorter than the string, the response header gives
  the full length, but no data is actually copied.
- No NUL terminator is added: even if the buffer is long enough to fit
  one, the buffer's original contents are preserved past the string's
  end.
- The VC firmware adds the following extra parameters beside the
  user-supplied ones (via /boot/cmdline.txt): `video`, `vc_mem.mem_base`
  and `vc_mem.mem_size`. This is currently not implemented in qemu.

Signed-off-by: Daniel Bertalan <dani@danielbertalan.dev>
Message-id: 20230425103250.56653-1-dani@danielbertalan.dev
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: added comment about NUL and short-buffer behaviour]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/net: npcm7xx_emc: set MAC in register space

The MAC address set from Qemu wasn't being saved into the register space.

Reviewed-by: Hao Wu <wuhaotsh@google.com>
Signed-off-by: Patrick Venture <venture@google.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: moved variable declaration to top of function]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge tag 'pull-request-2023-05-02' of https://gitlab.com/thuth/qemu into staging

* Fix the failing FreeBSD job in our CI
* Run the tpm-tis-i2c-test only if TCG is enabled
* Fix a use-after-free problem in the new reentracy checking code

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmRQ4vERHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbXo8g//dVRM+LCeCqNfjqtQLwGUUAt3ve1vruLD
# v9BT4ooTNIGFScQlc9NRnweCPPBvmdSyXfxdfa3ITW62BEKEcCaf9tXiNeXE9jUE
# VAQMA3u5/w0HSsWK9uqCZ/5azI4mwo5c2ykocX7XsdyP0LoFvlgydYBZ9S5PxRit
# JsX78x740qTGjOSxD2O5wYocDwRUVsCcY7nAIC1Cj52gxi9Vedxqzb8MW30AU+oh
# W9h49nVJAOvcx1UBNtYC0t6LCdJvMkhLPmewp4g4o0iiQqnGBjjD0pT2SUYhl1A5
# pZYjsq7IBFGPDEyQOD3R4VeaMPpo22NLifLbU8Kt+BepyHJHJgCkjX/WUnCEzwu9
# jmxaIUubT5/UanDzOX5qa/JDgNnUxkLfoVgZ8GOeNKoJ+Ik52xATnMm5Dqod369E
# MydBN8gqnT+I7Qb7KxSS5Q18YugUS3uUd17LqpZ846yxyqMjoM2Zee2J2K7x4clO
# cJ5h1AP7D6ZzyPzBMD0U9fanA+2M/qFiV3NJ9IDqiQYglafnU+gZOohE+p1Id4Dq
# IeSITK+OaaYPdlADfKcVJNLy3qBvyL/ZsJ/2X1zRO5vhPVal4tMfo4Fv8Su0vx34
# mXiUJO8SWk/GiPtJy7fNhxfiBBhwDTiyVX97In/+jbFC3ZKwPnwo/nq1DeBg4YI7
# Rj2rHq4HS7o=
# =zdof
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 02 May 2023 11:16:17 AM BST
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [undefined]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [undefined]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* tag 'pull-request-2023-05-02' of https://gitlab.com/thuth/qemu:
  async: avoid use-after-free on re-entrancy guard
  tests/qtest: Restrict tpm-tis-i2c-test to CONFIG_TCG
  tests/qtest: Disable the spice test of readconfig-test on FreeBSD

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

tests/qtest: Restrict tpm-tis-i2c-test to CONFIG_TCG

The test set -accel tcg, so restrict it to when TCG is present.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230426180013.14814-13-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

arm/Kconfig: Do not build TCG-only boards on a KVM-only build

Move all the CONFIG_FOO=y from default.mak into "default y if TCG"
statements in Kconfig. That way they won't be selected when
CONFIG_TCG=n.

I'm leaving CONFIG_ARM_VIRT in default.mak because it allows us to
keep the two default.mak files not empty and keep aarch64-default.mak
including arm-default.mak. That way we don't surprise anyone that's
used to altering these files.

With this change we can start building with --disable-tcg.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230426180013.14814-12-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

arm/Kconfig: Always select SEMIHOSTING when TCG is present

We are about to enable the build without TCG, so CONFIG_SEMIHOSTING
and CONFIG_ARM_COMPATIBLE_SEMIHOSTING cannot be unconditionally set in
default.mak anymore. So reflect the change in a Kconfig.

Instead of using semihosting/Kconfig, use a target-specific file, so
that the change doesn't affect other architectures which might
implement semihosting in a way compatible with KVM.

The selection from ARM_v7M needs to be removed to avoid a cycle during
parsing.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230426180013.14814-11-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/avocado: Pass parameters to migration test

The migration tests are currently broken for an aarch64 host because
the tests pass no 'machine' and 'cpu' options on the QEMU command
line.

Add a separate class to each architecture so that we can specify
'machine' and 'cpu' options instead of relying on defaults.

Add a skip decorator to keep the current behavior of only running
migration tests when the qemu target matches the host architecture.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-id: 20230426180013.14814-10-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/qtest: Fix tests when no KVM or TCG are present

It is possible to have a build with both TCG and KVM disabled due to
Xen requiring the i386 and x86_64 binaries to be present in an aarch64
host.

If we build with --disable-tcg on the aarch64 host, we will end-up
with a QEMU binary (x86) that does not support TCG nor KVM.

Skip tests that crash or hang in the above scenario. Do not include
any test cases if TCG and KVM are missing.

Make sure that calls to qtest_has_accel are placed after g_test_init
in similar fashion to commit ae4b01b349 ("tests: Ensure TAP version is
printed before other messages") to avoid TAP parsing errors.

Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230426180013.14814-9-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: move cpu_tcg to tcg/cpu32.c

move the module containing cpu models definitions
for 32bit TCG-only CPUs to tcg/ and rename it for clarity.

Signed-off-by: Claudio Fontana <cfontana@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230426180013.14814-8-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/qtest: Adjust and document query-cpu-model-expansion test for arm

We're about to move the 32-bit CPUs under CONFIG_TCG, so adjust the
query-cpu-model-expansion test to check against the cortex-a7, which
is already under CONFIG_TCG. That allows the next patch to contain
only code movement. (All the test cares about is that the CPU type
it's checking is one which definitely doesn't work under KVM.)

While here add comments clarifying what we're testing.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Thomas Huth <thuth@redhat.com>
Message-id: 20230426180013.14814-7-farosas@suse.de
Suggested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Move 64-bit TCG CPUs into tcg/

Move the 64-bit CPUs that are TCG-only:
- cortex-a35
- cortex-a55
- cortex-a72
- cortex-a76
- a64fx
- neoverse-n1

Keep the CPUs that can be used with KVM:
- cortex-a57
- cortex-a53
- max
- host

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230426180013.14814-6-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Do not expose all -cpu max features to qtests

We're about to move the TCG-only -cpu max configuration code under
CONFIG_TCG. To be able to do that we need to make sure the qtests
still have some cpu configured even when no other accelerator is
available.

Delineate now what is used with TCG-only and what is also used with
qtests to make the subsequent patches cleaner.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230426180013.14814-5-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Extract TCG -cpu max code into a function

Introduce aarch64_max_tcg_initfn that contains the TCG-only part of
-cpu max configuration. We'll need that to be able to restrict this
code to a TCG-only config in the next patches.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-id: 20230426180013.14814-4-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Remove dead code from cpu_max_set_sve_max_vq

The sve-max-vq property has been removed from the -cpu max used with
KVM, so code under kvm_enabled in cpu_max_set_sve_max_vq is not
reachable.

Fixes: 0baa21be49 ("target/arm: Make KVM -cpu max exactly like -cpu host")
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-id: 20230426180013.14814-3-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Move cortex sysregs into a separate file

The file cpu_tcg.c is about to be moved into the tcg/ directory, so
move the register definitions into a new file.

Also move the function declaration to the more appropriate cpregs.h.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230426180013.14814-2-farosas@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

async: avoid use-after-free on re-entrancy guard

A BH callback can free the BH, causing a use-after-free in aio_bh_call.
Fix that by keeping a local copy of the re-entrancy guard pointer.

Buglink: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=58513
Fixes: 9c86c97f12 ("async: Add an optional reentrancy guard to the BH API")
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20230501141956.3444868-1-alxndr@bu.edu>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Restrict tpm-tis-i2c-test to CONFIG_TCG

The test set -accel tcg, so restrict it to when TCG is present.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230426180013.14814-13-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Disable the spice test of readconfig-test on FreeBSD

The spice test is currently hanging on FreeBSD. It likely was
never working before, since in the past, our configure script
was failing to detect this feature due to a bug in the spice
package there (it just got enabled recently by the commit
https://cgit.freebsd.org/ports/commit/?id=cf16b1c9063351325f0 ).
To get the CI working again, let's disable the failing test for
now until someone has enough spare time to debug and fix the real
underlying problem.

Message-Id: <20230428151351.1365822-1-thuth@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging

* Fix compilation issues under Debian 10
* Update kernel headers to 6.3rc5
* Suppress GCC13 false positive in aio_bh_poll()
* Add new x86 feature bits
* Coverity fixes
* More steps towards removing qatomic_mb_set/read
* Fix reduced-phys-bits value for AMD SEV

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmRNC0IUHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroNo0wgArWNGKZpbmQ0e5L6ajMvaaPmg4mVL
# a2SJGU0TwTp0fUgZr14z2iwzIpSqQrsqhzTIAzOTs0OICDBPBuNvnRucMa+SVQGO
# Tc89YAwBVDo66dAKhWi+WR9tx7sTFCso0nbsBfczzdnwAw3g1MJ87Ueqc5tlPGBK
# E7YSAD6l4UuogoN5BLU7bSsG/X7bwcyzeUXRB4ik+Z9abWd4DH9qiROnBKLMmBLK
# nAi47h8b8MltWORpO+wf6HtkMKi37SAzl9VLHVuHcRhIdY/JhWCRhYSo0HXhgX66
# JLVkyxFpIndT0dUW/xnqATGez92FRZyTxHbxbAcWM0SoC1jOVfUXB+7Gdw==
# =vxou
# -----END PGP SIGNATURE-----
# gpg: Signature made Sat 29 Apr 2023 01:19:14 PM BST
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [undefined]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* tag 'for-upstream' of https://gitlab.com/bonzini/qemu:
  cpus-common: stop using mb_set/mb_read
  async: Suppress GCC13 false positive in aio_bh_poll()
  tests: vhost-user-test: release mutex on protocol violation
  Update linux headers to v6.3rc5
  update-linux-headers.sh: Add missing kernel headers.
  Fix libvhost-user.c compilation.
  target/i386: Add support for PREFETCHIT0/1 in CPUID enumeration
  target/i386: Add support for AVX-NE-CONVERT in CPUID enumeration
  target/i386: Add support for AVX-VNNI-INT8 in CPUID enumeration
  target/i386: Add support for AVX-IFMA in CPUID enumeration
  target/i386: Add support for AMX-FP16 in CPUID enumeration
  target/i386: Add support for CMPCCXADD in CPUID enumeration
  i386/cpu: Update how the EBX register of CPUID 0x8000001F is set
  i386/sev: Update checks and information related to reduced-phys-bits
  qemu-options.hx: Update the reduced-phys-bits documentation
  qapi, i386/sev: Change the reduced-phys-bits value from 5 to 1

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

cpus-common: stop using mb_set/mb_read

Use a store-release at the end of the work item, and a load-acquire when
waiting for the item to be completed.  This is the standard message passing
pattern and is both enough and clearer than mb_read/mb_set.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

async: Suppress GCC13 false positive in aio_bh_poll()

GCC13 reports an error :

../util/async.c: In function ‘aio_bh_poll’:
include/qemu/queue.h:303:22: error: storing the address of local variable ‘slice’ in ‘*ctx.bh_slice_list.sqh_last’ [-Werror=dangling-pointer=]
  303 |     (head)->sqh_last = &(elm)->field.sqe_next;                          \
      |     ~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~
../util/async.c:169:5: note: in expansion of macro ‘QSIMPLEQ_INSERT_TAIL’
  169 |     QSIMPLEQ_INSERT_TAIL(&ctx->bh_slice_list, &slice, next);
      |     ^~~~~~~~~~~~~~~~~~~~
../util/async.c:161:17: note: ‘slice’ declared here
  161 |     BHListSlice slice;
      |                 ^~~~~
../util/async.c:161:17: note: ‘ctx’ declared here

But the local variable 'slice' is removed from the global context list
in following loop of the same routine. Add a pragma to silent GCC.

Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Tested-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20230420202939.1982044-1-clg@kaod.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge tag 'pull-qapi-2023-04-28' of https://repo.or.cz/qemu/armbru into staging

QAPI patches patches for 2023-04-28

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEENUvIs9frKmtoZ05fOHC0AOuRhlMFAmRLlpsSHGFybWJydUBy
# ZWRoYXQuY29tAAoJEDhwtADrkYZTTjwP/iezbnBFv55HxaRFisEpBom7skXakV1r
# 0KTyYNo0BatS01yVlX7IOh37ZVmoBHrpbs0SLb2V5/uVALQBvmlnhUCCQ+BJwhJF
# JON6yjZRKUikVKn8CGj0ql9KDF6bC/T1ByFEotGc7SBMCx+FWuGXnmVzwOKOK2O3
# XAHG1P04wlxuyHMt3V7qqbc5IyzZIEHLbEitDhQM/6quQmwx4NzFR+JwvzUXCok1
# F7cEt0cKbE3JZHSXWIVTOuseeZR8vnk3giKBgFOetqhVorGusiTPMQwpDPvvgKGn
# fkg2l5SZQz9hSDfuzCKpJSQDqO8ji+62fHci9+PPpG10raoMiwCiuhP9yYECYekM
# 7BcAY7fItrUB+x+vjm8JRY15JxOT8ouHLfyknNLOwow8QwtsiNn35VlRJU7O+k+D
# gk9qvwl1RoLUSXzOo6Ye1NvzF8oMZ8o8f0vmYC+W3sdg89U9QnRrpCaX8YCytLnd
# rEXJiL26mg54yXrWShIzLvC33//sz7AvH6IjwrRt0ZQu7ANRJsNmYY9S1THAJBKW
# b9hY3Aosw8Advt0RUFkYr/PndZp/gxDpbtOCIkKZkti6XRkY89ExPk6zVmeo8miw
# DeEo1kRtzKYDC7ZU0yp73MZDSmUt17L1AfdgS0QuCQCfPHsa8LwtZoQmRIihT/Bo
# GZTLIZPiVC79
# =mkwX
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 28 Apr 2023 10:49:15 AM BST
# gpg:                using RSA key 354BC8B3D7EB2A6B68674E5F3870B400EB918653
# gpg:                issuer "armbru@redhat.com"
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>" [undefined]
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* tag 'pull-qapi-2023-04-28' of https://repo.or.cz/qemu/armbru:
  docs/devel/qapi-code-gen: Describe some doc markup pitfalls
  qapi storage-daemon/qapi: Fix documentation section structure
  qapi: Format since information the conventional way: (since X.Y)
  qapi: Fix misspelled section tags in doc comments
  qapi: Replace ad hoc "since" documentation by member documentation
  qapi: Fix argument documentation markup
  qga/qapi-schema: Fix member documentation markup
  qapi: Fix unintended definition lists in documentation
  qapi: Fix bullet list markup in documentation
  qapi: Delete largely misleading "Stability Considerations"
  qapi: Tidy up examples
  qapi: @foo should be used to reference, not ``foo``
  qapi/block-core: Clean up after removal of dirty bitmap @status
  qapi: Fix up references to long gone error classes
  qapi: Fix misspelled references
  qga/qapi-schema: Fix a misspelled reference
  qga/qapi-schema: Tidy up documentation of guest-fsfreeze-status

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

tests: vhost-user-test: release mutex on protocol violation

chr_read() is printing an error message and returning with s->data_mutex taken.
This can potentially cause a hang.  Reported by Coverity.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Update linux headers to v6.3rc5

commit 7e364e56293bb98cae1b55fd835f5991c4e96e7d

Signed-off-by: David 'Digit' Turner <digit@google.com>
Message-Id: <20230405172109.3081788-4-digit@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

update-linux-headers.sh: Add missing kernel headers.

Add <linux/memfd.h>, used by hw/display/virtio-gpu-udmabuf.c
Add <linux/nvme_ioctl.h>, used by qga/commands-posix.c
Add <linux/const.h> used by kvm-all.c, which requires
the _BITUL() macro definition to be available.

Without these, QEMU will not compile on Debian 10 systems.

Signed-off-by: David 'Digit' Turner <digit@google.com>
Message-Id: <20230405172109.3081788-3-digit@google.com>
[Add <linux/stddef.h> for __DECLARE_FLEX_ARRAY. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Fix libvhost-user.c compilation.

The source file uses VIRTIO_F_VERSION_1 which is
not defined by <linux/virtio_config.h> on Debian 10.

The system-provided <linux/virtio_config.h> which
does not include the macro definition is included
through <linux/vhost.h>, so fix the issue by including
the standard-headers version before that.

Signed-off-by: David 'Digit' Turner <digit@google.com>
Message-Id: <20230405125920.2951721-2-digit@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Add support for PREFETCHIT0/1 in CPUID enumeration

Latest Intel platform Granite Rapids has introduced a new instruction -
PREFETCHIT0/1, which moves code to memory (cache) closer to the
processor depending on specific hints.

The bit definition:
CPUID.(EAX=7,ECX=1):EDX[bit 14]

Add CPUID definition for PREFETCHIT0/1.

Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-7-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Add support for AVX-NE-CONVERT in CPUID enumeration

AVX-NE-CONVERT is a new set of instructions which can convert low
precision floating point like BF16/FP16 to high precision floating point
FP32, as well as convert FP32 elements to BF16. This instruction allows
the platform to have improved AI capabilities and better compatibility.

The bit definition:
CPUID.(EAX=7,ECX=1):EDX[bit 5]

Add CPUID definition for AVX-NE-CONVERT.

Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-6-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Add support for AVX-VNNI-INT8 in CPUID enumeration

AVX-VNNI-INT8 is a new set of instructions in the latest Intel platform
Sierra Forest, aims for the platform to have superior AI capabilities.
This instruction multiplies the individual bytes of two unsigned or
unsigned source operands, then adds and accumulates the results into the
destination dword element size operand.

The bit definition:
CPUID.(EAX=7,ECX=1):EDX[bit 4]

AVX-VNNI-INT8 is on a new feature bits leaf. Add a CPUID feature word
FEAT_7_1_EDX for this leaf.

Add CPUID definition for AVX-VNNI-INT8.

Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-5-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Add support for AVX-IFMA in CPUID enumeration

AVX-IFMA is a new instruction in the latest Intel platform Sierra
Forest. This instruction packed multiplies unsigned 52-bit integers and
adds the low/high 52-bit products to Qword Accumulators.

The bit definition:
CPUID.(EAX=7,ECX=1):EAX[bit 23]

Add CPUID definition for AVX-IFMA.

Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-4-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Add support for AMX-FP16 in CPUID enumeration

Latest Intel platform Granite Rapids has introduced a new instruction -
AMX-FP16, which performs dot-products of two FP16 tiles and accumulates
the results into a packed single precision tile. AMX-FP16 adds FP16
capability and allows a FP16 GPU trained model to run faster without
loss of accuracy or added SW overhead.

The bit definition:
CPUID.(EAX=7,ECX=1):EAX[bit 21]

Add CPUID definition for AMX-FP16.

Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-3-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Add support for CMPCCXADD in CPUID enumeration

CMPccXADD is a new set of instructions in the latest Intel platform
Sierra Forest. This new instruction set includes a semaphore operation
that can compare and add the operands if condition is met, which can
improve database performance.

The bit definition:
CPUID.(EAX=7,ECX=1):EAX[bit 7]

Add CPUID definition for CMPCCXADD.

Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-2-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

i386/cpu: Update how the EBX register of CPUID 0x8000001F is set

Update the setting of CPUID 0x8000001F EBX to clearly document the ranges
associated with fields being set.

Fixes: 6cb8f2a663 ("cpu/i386: populate CPUID 0x8000_001F when SEV is active")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <5822fd7d02b575121380e1f493a8f6d9eba2b11a.1664550870.git.thomas.lendacky@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

i386/sev: Update checks and information related to reduced-phys-bits

The value of the reduced-phys-bits parameter is propogated to the CPUID
information exposed to the guest. Update the current validation check to
account for the size of the CPUID field (6-bits), ensuring the value is
in the range of 1 to 63.

Maintain backward compatibility, to an extent, by allowing a value greater
than 1 (so that the previously documented value of 5 still works), but not
allowing anything over 63.

Fixes: d8575c6c02 ("sev/i386: add command to initialize the memory encryption context")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <cca5341a95ac73f904e6300f10b04f9c62e4e8ff.1664550870.git.thomas.lendacky@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

qemu-options.hx: Update the reduced-phys-bits documentation

A guest only ever experiences, at most, 1 bit of reduced physical
addressing. Update the documentation to reflect this as well as change
the example value on the reduced-phys-bits option.

Fixes: a9b4942f48 ("target/i386: add Secure Encrypted Virtualization (SEV) object")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <13a62ced1808546c1d398e2025cf85f4c94ae123.1664550870.git.thomas.lendacky@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

qapi, i386/sev: Change the reduced-phys-bits value from 5 to 1

A guest only ever experiences, at most, 1 bit of reduced physical
addressing. Change the query-sev-capabilities json comment to use 1.

Fixes: 31dd67f684 ("sev/i386: qmp: add query-sev-capabilities command")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <cb96d8e09154533af4b4e6988469bc0b32390b65.1664550870.git.thomas.lendacky@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge tag 'pull-request-2023-04-28' of https://gitlab.com/thuth/qemu into staging

* Prevent reentrant DMA accesses by default
* Only compile hw/rdma code when necessary
* Fix a potential locking issue in the vhost-user-test
* Offer more registers in GDB for s390x TCG

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmRLlDkRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbU98RAApVbehXIXMF4lm0B4/pTRlP7CEjbKvNdQ
# Ca/xuqiCctjisJWQMLf8WDdymzaA02ycSYo0X9diO4uisKog4kN4uYje3VTbSJnx
# 7a0MlOpUqP6AR3hZYy5y/wv1M7Yjfhnj7yTUsAo/mQYoZmvqRviNwJ6ekdIU9J4W
# V2EfY+R/ebNami9VvuHTqudcZxmmR+4S5m5JipeHvjFE0DJucES0S8sN3W+iSCxG
# ydqG02CUoZtoorkJZccy3FPyeTVHgvcNqJbhAA5pFBgo30L+mx0llqztzLnWnLCe
# GWvpW2lZ2jC5MOg8oEpbiR807eB3ITpzCu9SX0IUOxMUHpGtl49sx1WNFrrlgD4Z
# b+mZSwJx4grwkU8iD/nF0apQoqEIAH6VSyRAIq1yLAsq1H4/K4c8M+bZ/jAD+gw+
# +DujZIPJ6Gcu3BeqWgOcOY1rRol3ksmRYlU8GrHTR/zIU+q3h7Gk3aH4+Ore1qnE
# hI/YzyyZlQgFMvENqakPJIS9IH+h41gCltwE8YxqJyUWIBfRs7iBMzdxQuUW+c2b
# n6OPYqeDlQt4hEpq4mUtZ+aJV1ey6kyk5ESLTj8QaJiZtcT1+lmvZX2ug2oeyrJm
# ys2458ulKZu0rGjXWWXNFB8u3MM9xN/PLqeD/wKShyLzkit5Nfx4azw5LPbmjrbj
# Twmf1HGeUiY=
# =v84A
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 28 Apr 2023 10:39:05 AM BST
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [undefined]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [undefined]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* tag 'pull-request-2023-04-28' of https://gitlab.com/thuth/qemu:
  apic: disable reentrancy detection for apic-msi
  raven: disable reentrancy detection for iomem
  bcm2835_property: disable reentrancy detection for iomem
  lsi53c895a: disable reentrancy detection for script RAM
  hw: replace most qemu_bh_new calls with qemu_bh_new_guarded
  checkpatch: add qemu_bh_new/aio_bh_new checks
  async: Add an optional reentrancy guard to the BH API
  memory: prevent dma-reentracy issues
  tests: vhost-user-test: release mutex on protocol violation
  hw/rdma: VMW_PVRDMA should depend on VMXNET3_PCI
  hw/rdma: Compile target-independent parts of the rdma code only once
  hw/rdma: Remove unused macros PG_DIR_SZ and PG_TBL_SZ
  s390x/gdb: Split s390-virt.xml

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

docs/devel/qapi-code-gen: Describe some doc markup pitfalls

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20230427095346.1238913-1-armbru@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

qapi storage-daemon/qapi: Fix documentation section structure

In the QEMU QMP Reference Manual, subsection "Block core (VM
unrelated)" is empty.  Its contents is at the end of subsection
"Background jobs" instead.  That's because qapi/job.json is included
first from qapi/block-core.json, which makes qapi/job.json's
documentation go between qapi/block-core.json's subsection heading and
contents.

In the QEMU Storage Daemon QMP Reference Manual, section "Block
Devices" contains nothing but an empty subsection "Block core (VM
unrelated)".  The latter's contents is at the end section "Socket data
types", along with subsection "Block device exports".  Subsection
"Background jobs" is at the end of section "Cryptography".  All this
is because storage-daemon/qapi/qapi-schema.json includes modules in a
confused order.

Fix both as follows.

Turn subsection "Background jobs" into a section.

Move it before section "Block devices" in the QEMU QMP Reference
Manual, by including qapi/jobs.json right before qapi/block.json.

Reorder include directives in storage-daemon/qapi/qapi-schema.json to
match the order in qapi/qapi-schema.json, so that the QEMU Storage
Daemon QMP Reference Manual's section structure the QEMU QMP Reference
Manual's.

In the QEMU QMP Reference Manual, qapi/cryptodev.json's documentation
is at the end of section "Virtio devices".  That's because it lacks a
section heading, and therefore gets squashed into whatever section
happens to precede it.

Add section heading so it's in section "Cryptography devices".

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Acked-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230425064223.820979-17-armbru@redhat.com>

qapi: Format since information the conventional way: (since X.Y)

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230425064223.820979-16-armbru@redhat.com>

qapi: Fix misspelled section tags in doc comments

Section tags are case sensitive and end with a colon.  Screwing up
either gets them interpreted as ordinary paragraph.  Fix a few.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230425064223.820979-15-armbru@redhat.com>

qapi: Replace ad hoc "since" documentation by member documentation

MemoryDeviceInfoKind, NetClientDriver, and GuestPanicAction mention
some members only in ad hoc since documentation.  The generated
documentation shows these members as "Not documented".

Replace by formal member documentation.

Add actual documentation text for the GuestPanicAction members, to
match existing member documentation there.  For the others, merely
move existing "since" information.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230425064223.820979-14-armbru@redhat.com>

qapi: Fix argument documentation markup

Member / argument documentation of BlockdevAmendOptionsQcow2,
job-resume, and RDMA_GID_STATUS_CHANGED is parsed as ordinary text due
to missing colon or space before the colon.  The generated
documentation shows these members / arguments as "Not documented".

The fix is obvious: add missing colons, delete extra spaces.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230425064223.820979-13-armbru@redhat.com>

qga/qapi-schema: Fix member documentation markup

GuestDiskStatsInfo's member documentation is parsed as ordinary text
due to missing colons.  The generated documentation shows these
members as "Not documented".

The fix is obvious: add the missing colons.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230425064223.820979-12-armbru@redhat.com>