Merge tag 'for-upstream' of https://repo.or.cz/qemu/kevin into staging

Block layer patches

- configure: Enable -Wthread-safety if present
- no_co_wrapper to fix bdrv_open*() calls from coroutine context
- curl fixes, including enablement of newer libcurl versions
- MAINTAINERS: drop Vladimir from parallels block driver
- hbitmap: fix hbitmap_status() return value for first dirty bit case
- file-posix: Fix assertion failure in write_zeroes after moving
  bdrv_getlength() to co_wrapper

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEE3D3rFZqa+V09dFb+fwmycsiPL9YFAmPvgm0RHGt3b2xmQHJl
# ZGhhdC5jb20ACgkQfwmycsiPL9ZxQg//ZWwwh/s/P1PnKAjInNZZNklAWKThNEbZ
# cF1S94w26IhEQqM0i6MflqcDsPU5t4xZtBUOizx++9M4G8amWnomJSdczUcKULla
# Az9yweFC1Gu6ENdw+ql5VOzCfpdH5Bn9Jkly5fxuI4vmnBz1PH1Dnd3P4wuLq2sL
# xna5dijVEhRc5mTKWjbp4nFfvQhucuEBPSNjgnZwEPbhciWxTMmB1GmyRvTxZy8v
# UY8PcoTlxdKeVQ6DTmkOirphpGj7HeNCEQnZppWs7vHys2oGi9kmR5qTKUNZxGrY
# 8yWiCiVDqbb50fhEC1srhph79bCij87QC1N33Bm+NuGjnjG4bKVx2B9DC8+6S/JS
# e3x6u+r0dd6/t0rjKnt1+inYqmM+i5lBJ7+R0yhWUQ+DYkvttNf5yiotD8qvccWJ
# Kcx14lfjPLK7siAMEY5K0bNMimhN4RR9oCLoPTOHei+vlxdfiMm2XPN61NNht5gD
# lYZ8JMBsEF/o2ebqTgsJrIHS+Q/8MqcwSunBc54fcXZoF+eiza3W2ArXLNfAEfGE
# U4JowNK2PrTIrpEjD+Vs0RsBBSmN5PcYIAz04ioODpDnYMq73/t3x9MKdVoxOT64
# AM7w58fSyWu8iwvkeA0d3XeVtSHFqZ49PqqIem4IegtnC/AXMUNrJ/VT99xHjeJY
# oLhOJz7LUg0=
# =FtaA
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 17 Feb 2023 13:34:37 GMT
# gpg:                using RSA key DC3DEB159A9AF95D3D7456FE7F09B272C88F2FD6
# gpg:                issuer "kwolf@redhat.com"
# gpg: Good signature from "Kevin Wolf <kwolf@redhat.com>" [full]
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74  56FE 7F09 B272 C88F 2FD6

* tag 'for-upstream' of https://repo.or.cz/qemu/kevin: (22 commits)
  hbitmap: fix hbitmap_status() return value for first dirty bit case
  block/file-posix: don't use functions calling AIO_WAIT_WHILE in worker threads
  MAINTAINERS: drop Vladimir from parallels block driver
  block: temporarily hold the new AioContext of bs_top in bdrv_append()
  block: Handle curl 7.55.0, 7.85.0 version changes
  block: Assert non-coroutine context for bdrv_open_inherit()
  block: Fix bdrv_co_create_opts_simple() to open images with no_co_wrapper
  vpc: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper
  vmdk: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper
  vhdx: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper
  vdi: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper
  qed: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper
  qcow2: Fix open/create to open images with no_co_wrapper
  qcow: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper
  parallels: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper
  luks: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper
  block: Create no_co_wrappers for open functions
  block-coroutine-wrapper: Introduce no_co_wrapper
  curl: Fix error path in curl_open()
  configure: Enable -Wthread-safety if present
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hbitmap: fix hbitmap_status() return value for first dirty bit case

The last return statement should return true, as we already evaluated that
start == next_dirty

Also, fix hbitmap_status() description in header

Cc: qemu-stable@nongnu.org
Fixes: a6426475a75 ("block/dirty-bitmap: introduce bdrv_dirty_bitmap_status()")
Signed-off-by: Andrey Zhadchenko <andrey.zhadchenko@virtuozzo.com>
Message-Id: <20230202181523.423131-1-andrey.zhadchenko@virtuozzo.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/file-posix: don't use functions calling AIO_WAIT_WHILE in worker threads

When calling bdrv_getlength() in handle_aiocb_write_zeroes(), the
function creates a new coroutine and then waits that it finishes using
AIO_WAIT_WHILE.
The problem is that this function could also run in a worker thread,
that has a different AioContext from main loop and iothreads, therefore
in AIO_WAIT_WHILE we will have in_aio_context_home_thread(ctx) == false
and therefore
assert(qemu_get_current_aio_context() == qemu_get_aio_context());
in the else branch will fail, crashing QEMU.

Aside from that, bdrv_getlength() is wrong also conceptually, because
it reads the BDS graph from another thread and is not protected by
any lock.

Replace it with raw_co_getlength, that doesn't create a coroutine and
doesn't read the BDS graph.

Reported-by: Ninad Palsule <ninad@linux.vnet.ibm.com>
Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20230209154522.1164401-1-eesposit@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

MAINTAINERS: drop Vladimir from parallels block driver

I have to admit this is out of my scope now. Still feel free to Cc me
directly if my help is needed :)

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-Id: <20230214182848.1564714-1-vsementsov@yandex-team.ru>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: temporarily hold the new AioContext of bs_top in bdrv_append()

bdrv_append() is called with bs_top AioContext held, but
bdrv_attach_child_noperm() could change the AioContext of bs_top.

bdrv_replace_node_noperm() calls bdrv_drained_begin() starting from
commit 2398747128 ("block: Don't poll in bdrv_replace_child_noperm()").
bdrv_drained_begin() can call BDRV_POLL_WHILE that assumes the new lock
is taken, so let's temporarily hold the new AioContext to prevent QEMU
from failing in BDRV_POLL_WHILE when it tries to release the wrong
AioContext.

Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2168209
Reported-by: Aihua Liang <aliang@redhat.com>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-Id: <20230214171621.11574-1-sgarzare@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Handle curl 7.55.0, 7.85.0 version changes

* 7.55.0 deprecates CURLINFO_CONTENT_LENGTH_DOWNLOAD in favour of a *_T
  version, which returns curl_off_t instead of a double.
* 7.85.0 deprecates CURLOPT_PROTOCOLS and CURLOPT_REDIR_PROTOCOLS in
  favour of *_STR variants, specifying the desired protocols via a
  string.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1440
Signed-off-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230123201431.23118-1-anjo@rev.ng>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Assert non-coroutine context for bdrv_open_inherit()

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-14-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Fix bdrv_co_create_opts_simple() to open images with no_co_wrapper

bdrv_co_create_opts_simple() runs in a coroutine. Therefore it is not
allowed to open images directly. Fix the call to use the corresponding
no_co_wrapper instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-13-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

vpc: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine. Therefore they are
not allowed to open images directly. Fix the calls to use the
corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-12-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

vmdk: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine. Therefore they are
not allowed to open images directly. Fix the calls to use the
corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-11-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

vhdx: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine. Therefore they are
not allowed to open images directly. Fix the calls to use the
corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-10-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

vdi: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine. Therefore they are
not allowed to open images directly. Fix the calls to use the
corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-9-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qed: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine. Therefore they are
not allowed to open images directly. Fix the calls to use the
corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-8-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qcow2: Fix open/create to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine, as does
qcow2_do_open(). Therefore they are not allowed to open images directly.
Fix the calls to use the corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-7-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qcow: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine. Therefore they are
not allowed to open images directly. Fix the calls to use the
corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-6-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

parallels: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine. Therefore they are
not allowed to open images directly. Fix the calls to use the
corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-5-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

luks: Fix .bdrv_co_create(_opts) to open images with no_co_wrapper

.bdrv_co_create implementations run in a coroutine. Therefore they are
not allowed to open images directly. Fix the calls to use the
corresponding no_co_wrappers instead.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-4-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Create no_co_wrappers for open functions

Images can't be opened in coroutine context because opening needs to
change the block graph. Add no_co_wrappers so that coroutines have a
simple way of opening images in a BH instead.

At the same time, mark the wrapped functions as no_coroutine_fn.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-3-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block-coroutine-wrapper: Introduce no_co_wrapper

Some functions must not be called from coroutine context. The common
pattern to use them anyway from a coroutine is running them in a BH and
letting the calling coroutine yield to be woken up when the BH is
completed.

Instead of manually writing such wrappers, add support for generating
them to block-coroutine-wrapper.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230126172432.436111-2-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

curl: Fix error path in curl_open()

g_hash_table_destroy() and g_hash_table_foreach_remove() (called by
curl_drop_all_sockets()) both require the table to be non-NULL, or will
print assertion failures (just print, no abort).

There are several paths in curl_open() that can lead to the out_noclean
label without s->sockets being allocated, so clean it only if it has
been allocated.

Example reproducer:
$ qemu-img info -f http ''
qemu-img: GLib: g_hash_table_foreach_remove: assertion 'hash_table != NULL' failed
qemu-img: GLib: g_hash_table_destroy: assertion 'hash_table != NULL' failed
qemu-img: Could not open '': http curl driver cannot handle the URL '' (does not start with 'http://')

Closes: https://gitlab.com/qemu-project/qemu/-/issues/1475
Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Hanna Czenczek <hreitz@redhat.com>
Message-Id: <20230206132949.92917-1-hreitz@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

configure: Enable -Wthread-safety if present

This enables clang's thread safety analysis (TSA), which we'll use to
statically check the block graph locking.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20221207131838.239125-9-kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230117135203.3049709-4-eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

bsd-user/mmap: use TSA_NO_TSA to suppress clang TSA warnings in FreeBSD

FreeBSD implements pthread headers using TSA (thread safety analysis)
annotations, therefore when an application is compiled with
-Wthread-safety there are some locking/annotation requirements that the
user of the pthread API has to follow.

This will also be the case in QEMU, since bsd-user/mmap.c uses the
pthread API. Therefore when building it with -Wthread-safety the
compiler will throw warnings because the functions are not properly
annotated. We need TSA to be enabled because it ensures that the
critical sections of an annotated variable are properly locked.

In order to make the compiler happy and avoid adding all the necessary
macros to all callers (lock functions should use TSA_ACQUIRE, while
unlock TSA_RELEASE, and this applies to all users of pthread_mutex_lock
and pthread_mutex_unlock), simply use TSA_NO_TSA to supppress such
warnings.

Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20230117135203.3049709-3-eesposit@redhat.com>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

util/qemu-thread-posix: use TSA_NO_TSA to suppress clang TSA warnings in FreeBSD

FreeBSD implements pthread headers using TSA (thread safety analysis)
annotations, therefore when an application is compiled with
-Wthread-safety there are some locking/annotation requirements that the
user of the pthread API has to follow.

This will also be the case in QEMU, since util/qemu-thread-posix.c uses
the pthread API. Therefore when building it with -Wthread-safety, the
compiler will throw warnings because the functions are not properly
annotated. We need TSA to be enabled because it ensures that the
critical sections of an annotated variable are properly locked.

In order to make the compiler happy and avoid adding all the necessary
macros to all callers (lock functions should use TSA_ACQUIRE, while
unlock TSA_RELEASE, and this applies to all users of pthread_mutex_lock
and pthread_mutex_unlock), simply use TSA_NO_TSA to supppress such
warnings.

Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20230117135203.3049709-2-eesposit@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

Merge tag 'pull-target-arm-20230216' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * Some mostly M-profile-related code cleanups
 * avocado: Retire the boot_linux.py AArch64 TCG tests
 * hw/arm/smmuv3: Add GBPA register
 * arm/virt: don't try to spell out the accelerator
 * hw/arm: Attach PSPI module to NPCM7XX SoC
 * Some cleanup/refactoring patches aiming towards
   allowing building Arm targets without CONFIG_TCG

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmPuY50ZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3oKND/9Cy7/XlMBOq1vlEM/TG7eK
# Jq582mUDsOqnebDZLEYheWe7wDzIHpoAZfCwOL1sLDbzjjglYXgQu7J3fQvuSofo
# QJBKkJiiATHCQb3I+A7VITl0pyW3lcRy9zZ7E5O9WdbbBcKIC8dwTNyiTAEmKyL9
# ldYxfCZiMl+j+za3uNUxnlEd3oC15UZZfGAi++KjdfdFH6TxyGRhkiAI3iP4QD7s
# aHo+9/TEyl1b9ApjXWymlgupwLeAOI3TwCq30lyBttxh9aU/hUkBttdc+oo+ymTH
# EuOB4IM2O7jCjGMKAeg7jlYfgLJRradaOsY4g0K0VxD/wGRTp6Qehmt//mzYwYYb
# izprDDb1UbJAH5OsFFPNevxQjJHfOOFYs+ObvadRgvLFTPuid8+1YNleknPxp7qm
# jRC9/DYekqcxU2mlEwuPApFNhZBoxwnJhFEHfmO8SCqre4+QfDNbGhUxPuHORt6V
# 0xLzaSC1L6RALLsDnhgYK/Iz8vA/2HmrcO+FC0TlIeKcGUV2xOMQB8+DRlsrzpEz
# KzrjXyuLqUOIWN6iGThxOG72qPd94BKO3SHue7r4BCYpDdeHIbQUC67cBq6b1LZ4
# QDOoeHwuTo5GRnAYoc+lQ7HyEfGiWy38wApQihZ3EOndot6DN/AD8nRC2632Hidu
# evQUClDbeXZDTv87AfIojg==
# =M2nc
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 16 Feb 2023 17:10:53 GMT
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [ultimate]
# gpg:                 aka "Peter Maydell <peter@archaic.org.uk>" [ultimate]
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* tag 'pull-target-arm-20230216' of https://git.linaro.org/people/pmaydell/qemu-arm: (30 commits)
  tests/qtest: Restrict tpm-tis-devices-{swtpm}-test to CONFIG_TCG
  tests/qtest: arm-cpu-features: Match tests to required accelerators
  target/arm: Use "max" as default cpu for the virt machine with KVM
  tests/avocado: Tag TCG tests with accel:tcg
  tests/avocado: Skip tests that require a missing accelerator
  target/arm: Move cpregs code out of cpu.h
  target/arm: Move PC alignment check
  target/arm: wrap call to aarch64_sve_change_el in tcg_enabled()
  target/arm: wrap psci call with tcg_enabled
  target/arm: rename handle_semihosting to tcg_handle_semihosting
  hw/arm/smmu-common: Fix TTB1 handling
  hw/arm/smmu-common: Support 64-bit addresses
  hw/arm: Attach PSPI module to NPCM7XX SoC
  hw/ssi: Add Nuvoton PSPI Module
  MAINTAINERS: Add myself to maintainers and remove Havard
  arm/virt: don't try to spell out the accelerator
  hw/arm: Add missing XLNX_ZYNQMP_ARM -> USB_DWC3 Kconfig dependency
  hw/arm/smmuv3: Add GBPA register
  tests/avocado: retire the Aarch64 TCG tests from boot_linux.py
  target/arm: Declare CPU <-> NVIC helpers in 'hw/intc/armv7m_nvic.h'
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/qtest: Restrict tpm-tis-devices-{swtpm}-test to CONFIG_TCG

These tests set -accel tcg, so restrict them to when TCG is present.

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

tests/qtest: arm-cpu-features: Match tests to required accelerators

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

target/arm: Use "max" as default cpu for the virt machine with KVM

Now that the cortex-a15 is under CONFIG_TCG, use as default CPU for a
KVM-only build the 'max' cpu.

Note that we cannot use 'host' here because the qtests can run without
any other accelerator (than qtest) and 'host' depends on KVM being
enabled.

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

tests/avocado: Tag TCG tests with accel:tcg

This allows the test to be skipped when TCG is not present in the QEMU
binary.

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

tests/avocado: Skip tests that require a missing accelerator

If a test was tagged with the "accel" tag and the specified
accelerator it not present in the qemu binary, cancel the test.

We can now write tests without explicit calls to require_accelerator,
just the tag is enough.

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

target/arm: Move cpregs code out of cpu.h

Since commit cf7c6d1004 ("target/arm: Split out cpregs.h") we now have
a cpregs.h header which is more suitable for this code.

Code moved verbatim.

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

target/arm: Move PC alignment check

Move this earlier to make the next patch diff cleaner. While here
update the comment slightly to not give the impression that the
misalignment affects only TCG.

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

target/arm: wrap call to aarch64_sve_change_el in tcg_enabled()

Signed-off-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: wrap psci call with tcg_enabled

for "all" builds (tcg + kvm), we want to avoid doing
the psci check if tcg is built-in, but not enabled.

Signed-off-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: rename handle_semihosting to tcg_handle_semihosting

make it clearer from the name that this is a tcg-only function.

Signed-off-by: Claudio Fontana <cfontana@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/smmu-common: Fix TTB1 handling

Addresses targeting the second translation table (TTB1) in the SMMU have
all upper bits set (except for the top byte when TBI is enabled). Fix
the TTB1 check.

Reported-by: Ola Hugosson <ola.hugosson@arm.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Message-id: 20230214171921.1917916-3-jean-philippe@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/smmu-common: Support 64-bit addresses

Addresses targeting the second translation table (TTB1) in the SMMU have
all upper bits set. Ensure the IOMMU region covers all 64 bits.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Message-id: 20230214171921.1917916-2-jean-philippe@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm: Attach PSPI module to NPCM7XX SoC

Signed-off-by: Hao Wu <wuhaotsh@google.com>
Reviewed-by: Titus Rwantare <titusr@google.com>
Reviewed-by: Philippe Mathieu-Daude <philmd@linaro.org>
Message-id: 20230208235433.3989937-4-wuhaotsh@google.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/ssi: Add Nuvoton PSPI Module

Nuvoton's PSPI is a general purpose SPI module which enables
connections to SPI-based peripheral devices.

Signed-off-by: Hao Wu <wuhaotsh@google.com>
Reviewed-by: Chris Rauer <crauer@google.com>
Reviewed-by: Philippe Mathieu-Daude <philmd@linaro.org>
Message-id: 20230208235433.3989937-3-wuhaotsh@google.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

MAINTAINERS: Add myself to maintainers and remove Havard

Havard is no longer working on the Nuvoton systems for a while
and won't be able to do any work on it in the future. So I'll
take over maintaining the Nuvoton system from him.

Signed-off-by: Hao Wu <wuhaotsh@google.com>
Acked-by: Havard Skinnemoen <hskinnemoen@google.com>
Reviewed-by: Philippe Mathieu-Daude <philmd@linaro.org>
Message-id: 20230208235433.3989937-2-wuhaotsh@google.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

arm/virt: don't try to spell out the accelerator

Just use current_accel_name() directly.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm: Add missing XLNX_ZYNQMP_ARM -> USB_DWC3 Kconfig dependency

Since commit acc0b8b05a when running the ZynqMP ZCU102 board with
a QEMU configured using --without-default-devices, we get:

  $ qemu-system-aarch64 -M xlnx-zcu102
  qemu-system-aarch64: missing object type 'usb_dwc3'
  Abort trap: 6

Fix by adding the missing Kconfig dependency.

Fixes: acc0b8b05a ("hw/arm/xlnx-zynqmp: Connect ZynqMP's USB controllers")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230216092327.2203-1-philmd@linaro.org
Reviewed-by: Francisco Iglesias <francisco.iglesias@amd.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/smmuv3: Add GBPA register

GBPA register can be used to globally abort all
transactions.

It is described in the SMMU manual in "6.3.14 SMMU_GBPA".
ABORT reset value is IMPLEMENTATION DEFINED, it is chosen to
be zero(Do not abort incoming transactions).

Other fields have default values of Use Incoming.

If UPDATE is not set, the write is ignored. This is the only permitted
behavior in SMMUv3.2 and later.(6.3.14.1 Update procedure)

As this patch adds a new state to the SMMU (GBPA), it is added
in a new subsection for forward migration compatibility.
GBPA is only migrated if its value is different from the reset value.
It does this to be backward migration compatible if SW didn't write
the register.

Signed-off-by: Mostafa Saleh <smostafa@google.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Message-id: 20230214094009.2445653-1-smostafa@google.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/avocado: retire the Aarch64 TCG tests from boot_linux.py

The two TCG tests for GICv2 and GICv3 are very heavy weight distros
that take a long time to boot up, especially for an --enable-debug
build. The total code coverage they give is:

  Overall coverage rate:
    lines......: 11.2% (59584 of 530123 lines)
    functions..: 15.0% (7436 of 49443 functions)
    branches...: 6.3% (19273 of 303933 branches)

We already get pretty close to that with the machine_aarch64_virt
tests which only does one full boot (~120s vs ~600s) of alpine. We
expand the kernel+initrd boot (~8s) to test both GICs and also add an
RNG device and a block device to generate a few IRQs and exercise the
storage layer. With that we get to a coverage of:

  Overall coverage rate:
    lines......: 11.0% (58121 of 530123 lines)
    functions..: 14.9% (7343 of 49443 functions)
    branches...: 6.0% (18269 of 303933 branches)

which I feel is close enough given the massive time saving. If we want
to target any more sub-systems we can use lighter weight more directed
tests.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230203181632.2919715-1-alex.bennee@linaro.org
Cc: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Declare CPU <-> NVIC helpers in 'hw/intc/armv7m_nvic.h'

While dozens of files include "cpu.h", only 3 files require
these NVIC helper declarations.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230206223502.25122-12-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Store CPUARMState::nvic as NVICState*

There is no point in using a void pointer to access the NVIC.
Use the real type to avoid casting it while debugging.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230206223502.25122-11-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Restrict CPUARMState::nvic to sysemu

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230206223502.25122-10-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Restrict CPUARMState::arm_boot_info to sysemu

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230206223502.25122-9-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Restrict CPUARMState::gicv3state to sysemu

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230206223502.25122-8-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Avoid resetting CPUARMState::eabi field

Although the 'eabi' field is only used in user emulation where
CPU reset doesn't occur, it doesn't belong to the area to reset.
Move it after the 'end_reset_fields' for consistency.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230206223502.25122-7-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Convert CPUARMState::eabi to boolean

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230206223502.25122-6-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Constify ID_PFR1 on user emulation

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230206223502.25122-5-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Reduce arm_v7m_mmu_idx_[all/for_secstate_and_priv]() scope

arm_v7m_mmu_idx_all() and arm_v7m_mmu_idx_for_secstate_and_priv()
are only used for system emulation in m_helper.c.
Move the definitions to avoid prototype forward declarations.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230206223502.25122-4-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Simplify arm_v7m_mmu_idx_for_secstate() for user emulation

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230206223502.25122-3-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/armv7m_nvic: Use OBJECT_DECLARE_SIMPLE_TYPE() macro

Manually convert to OBJECT_DECLARE_SIMPLE_TYPE() macro,
similarly to automatic conversion from commit 8063396bf3
("Use OBJECT_DECLARE_SIMPLE_TYPE when possible").

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230206223502.25122-2-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge tag 'migration-20230215-pull-request' of https://gitlab.com/juan.quintela/qemu into staging

Migration Pull request

This pull request contains:

* Add qemu_file_get_to_fd() a.k.a. make vfio happy(Avihai)
* migration/block is now DPRINTF() free zone (Philippe)
* remove res_compat and improve docs (me)

Please apply.

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEGJn/jt6/WMzuA0uC9IfvGFhy1yMFAmPtOxgACgkQ9IfvGFhy
# 1yM+8w/+Nrm40o8B+0M1X/2O8SPaXE0Oxll5qk5AgfC+2Ef/n5661ZzOM1xQcqzv
# CN3bj3g7dKN05nGhqW5ZFrEZliJCGksx2FBKMq9H6R5YBSMo/+GHT4ugPqsujudl
# RBafIHzp4h4nTh/4aJ4Kr2Vn/bVQzGVyv5c6yUdUhKhE1QfCLu9AMC5vN1hfWxBz
# IQBvMkLNXXhWOTdkVAElVlFOM6GZzWb1UYDkBaQHjjw/dN0MdCFFj7tIaBIyU2Eb
# KyQCoYeJjjWrljsrfzfCucrW+mbVGmCe1TiCEnaAmmgzOI0NwARmUcdYdCyZ4Nhh
# flNjpqQB6H4/ekI0mus2YPMPEkHyVgBMuF8fHV3YDwxcQ25U39RLymVfAoKPNg88
# S1EcQ9h2QVqdML7bR9vTmPOZKk5vD25H7B8LQIJpT8PHVg3ZzzRaOIR8UHmmFXx/
# PSL1rGfPcCLJ3P1XlVjwBApebMHyid9c+Sohda9L9HPer0jlB+mGQUEV8J67DJiT
# xlffFgJmYu3+AJu9BoqrG0AUW6zA78jSs4Tbp78qbIVA4uzGkWI9JvROMfG9Qcc7
# vruuL0sUXHVUhzW2F+0M+b1rVhaHHQYvFo3mIqdHrd5mGicdgh1HUbaJuEJxeJnh
# oFt6SZbIx3N/pllcvMsxQa+rFI/po+FEkRwIF02SzTRWYHB0qos=
# =KpKO
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 15 Feb 2023 20:05:44 GMT
# gpg:                using RSA key 1899FF8EDEBF58CCEE034B82F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>" [full]
# gpg:                 aka "Juan Quintela <quintela@trasno.org>" [full]
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03  4B82 F487 EF18 5872 D723

* tag 'migration-20230215-pull-request' of https://gitlab.com/juan.quintela/qemu:
  migration: Rename res_{postcopy,precopy}_only
  migration: Remove unused res_compatible
  migration: In case of postcopy, the memory ends in res_postcopy_only
  migration/block: Convert remaining DPRINTF() debug macro to trace events
  migration/qemu-file: Add qemu_file_get_to_fd()

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

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

* block/iscsi: fix double-free on BUSY or similar statuses
* catch [accel] entry without accelerator
* target/i386: various fixes for BMI and ADX instructions
* make the contents of meson-buildoptions.sh stable

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmPpDTcUHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroODWAgAhWi0XSBe91+34ahQqx6JoLStNX9e
# xy1Cm/5LPDrDquD/4T6gnzb1Min/X2AsVu9c+VIbuHbL/rFc7aNIEL6c7KzwSFLs
# vEDF7tSnlIMK0ClDsDYBz7HUIgBgqE2crAJmTJus2Cqd+Ef5bMxhQi5Imrk6qtmO
# HRdVYEGasJ7CO50oUB91AMrNInWQw0qBnBOB8TnwTLTcvE1txa+xnZuZgQA2HrGX
# OweLdKRcYPvRYvBB3wLMbwqEKbD1eYbdbNt7T2KkuVQpYcgfoCpayHIIMroD+hvu
# BImmG9wWieDKH4Brs765gH6/3VF5UZKbgDQo9Wz+W/5QqnqRSsOYBiMkmA==
# =G3PN
# -----END PGP SIGNATURE-----
# gpg: Signature made Sun 12 Feb 2023 16:00:55 GMT
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [full]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>" [full]
# 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:
  libqtest: ensure waitpid() is only called once
  libqtest: split qtest_spawn_qemu function
  target/i386: fix ADOX followed by ADCX
  target/i386: Fix C flag for BLSI, BLSMSK, BLSR
  target/i386: Fix BEXTR instruction
  tests/tcg/i386: Introduce and use reg_t consistently
  vl: catch [accel] entry without accelerator
  block/iscsi: fix double-free on BUSY or similar statuses
  remove unnecessary extern "C" blocks
  build: make meson-buildoptions.sh stable

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

migration: Rename res_{postcopy,precopy}_only

Once that res_compatible is removed, they don't make sense anymore.
We remove the _only preffix.  And to make things clearer we rename
them to must_precopy and can_postcopy.

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

migration: Remove unused res_compatible

Nothing assigns to it after previous commit.

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

migration: In case of postcopy, the memory ends in res_postcopy_only

So remove last assignation of res_compatible.

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

Merge tag 'misc-next-pull-request' of https://gitlab.com/berrange/qemu into staging

 * Document 'password-secret' option for -iscsi
 * Deprecate iSCSI 'password' in favour of 'password-secret'
 * Remove deprecated 'password' option for SPICE
 * Fix handling of cached read buffers with TLS

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE2vOm/bJrYpEtDo4/vobrtBUQT98FAmPtGoUACgkQvobrtBUQ
# T99RsA/7BQqQCyZcpTxYXYnGace4qnVx/VQLX7gESjnbsXlA0qUAtxepg2YdsU3W
# GxyZHflP48TesSAsxsA9sBwoFaJZvsq+fZnq62clNNiZTIRynSi65KVUdbqQ/a3F
# 3A9XZlqLzXJMPpwyowzcjMCx+/4+cIr9DA0hKUzQJtNder1I17ySXFGGsO98iM24
# KR/Pulk2NogWLez1Tr7qF01+OK8/cBPpD/QEHzKXsyVcbgnFLFnBIpbpLtlK4BEb
# g6csk665kbWZXud25YXdyG9pUPOrNM4l+7fQcDhjy3IoS2P6/bCTp1jsn3txTbEj
# yTKFJEEM18UWNQCvn4wyHiRCK6+KrFigocdj0jdsyG5DtNUTxqNkdscJa8V/d0hR
# 4y56Z33yK5JQIPaHE28f5vy/w2014hUoqpoYHt4JOeUuUJq9tvMqY6TCThc3ZvvP
# 08DF/0MwIj0NzP5hiEtNxUwFNjwmVRZCSzmNX+22SXOlICwZeQpejwtpEz8YMgDo
# S4vEUjhcNYXoTkL6qZFfBuwn1Gkqqg1DRx8amh3WTB7JDiFA9YwefmBiblAqEK0D
# AG1vPo53uHFytoYqMtsEI/JVE7wfZCBHGBXz2euOOKzUqWS0OGL0h71s5D0yDY+U
# /8YBy86qj7Vm7QuUz+Z2OaWH+miXYy8/kMjdfEcGgqbORReZWrI=
# =sAEk
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 15 Feb 2023 17:46:45 GMT
# gpg:                using RSA key DAF3A6FDB26B62912D0E8E3FBE86EBB415104FDF
# gpg: Good signature from "Daniel P. Berrange <dan@berrange.com>" [full]
# gpg:                 aka "Daniel P. Berrange <berrange@redhat.com>" [full]
# Primary key fingerprint: DAF3 A6FD B26B 6291 2D0E  8E3F BE86 EBB4 1510 4FDF

* tag 'misc-next-pull-request' of https://gitlab.com/berrange/qemu:
  ui: remove deprecated 'password' option for SPICE
  block: deprecate iSCSI 'password' in favour of 'password-secret'
  block: mention 'password-secret' option for -iscsi
  io/channel-tls: fix handling of bigger read buffers
  crypto: TLS: introduce `check_pending`

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

migration/block: Convert remaining DPRINTF() debug macro to trace events

Finish the conversion from commit fe80c0241d
("migration: using trace_ to replace DPRINTF").

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration/qemu-file: Add qemu_file_get_to_fd()

Add new function qemu_file_get_to_fd() that allows reading data from
QEMUFile and writing it straight into a given fd.

This will be used later in VFIO migration code.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

ui: remove deprecated 'password' option for SPICE

This has been replaced by the 'password-secret' option,
which references a 'secret' object instance.

Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

block: deprecate iSCSI 'password' in favour of 'password-secret'

Support for referencing secret objects was added in

  commit b189346eb1784df95ed6fed610411dbf23d19e1f
  Author: Daniel P. Berrangé <berrange@redhat.com>
  Date:   Thu Jan 21 14:19:21 2016 +0000

    iscsi: add support for getting CHAP password via QCryptoSecret API

The existing 'password' option is overdue for deprecation and
subsequent removal.

Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

block: mention 'password-secret' option for -iscsi

The 'password-secret' option was added

  commit b189346eb1784df95ed6fed610411dbf23d19e1f
  Author: Daniel P. Berrangé <berrange@redhat.com>
  Date:   Thu Jan 21 14:19:21 2016 +0000

    iscsi: add support for getting CHAP password via QCryptoSecret API

but was not mentioned in the command line docs

Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

io/channel-tls: fix handling of bigger read buffers

Since the TLS backend can read more data from the underlying QIOChannel
we introduce a minimal child GSource to notify if we still have more
data available to be read.

Signed-off-by: Antoine Damhet <antoine.damhet@shadow.tech>
Signed-off-by: Charles Frey <charles.frey@shadow.tech>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

crypto: TLS: introduce `check_pending`

The new `qcrypto_tls_session_check_pending` function allows the caller
to know if data have already been consumed from the backend and is
already available.

Signed-off-by: Antoine Damhet <antoine.damhet@shadow.tech>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

Merge tag 'pull-request-2023-02-14' of https://gitlab.com/thuth/qemu into staging

* Bump minimum Clang version to 10.0
* Improve the handling of the libdw library
* Deprecate --enable-gprof builds and remove them from CI
* Remove the deprecated "sga" device
* Some header #include clean-ups
* Make qtests more flexible with regards to missing devices
* Some small s390x-related fixes/improvements

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmPra8ARHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbW5zQ/+LRyPlCB9/Zz+40QUNIl9M9pcpjwCno9w
# pWIeOXMftBu8vCa/+o58Y22NzmWq7cFypfsIvjrlCqCIZQx1shez5+bvE/ZSyECd
# vy3I+ybKRSaXgE/ighAifwOR7eOhqLBHyXJkAN5grMx0kuu3oKqXBpyJNqTHjq6o
# VgKTu8+o7Ddb+CuzwGahKzyzTqD7gY+9+Y5Gnlq0DvbNEq9k6tcjQZ8P3MqOmjWJ
# WnBZH9Rrq9itnlcZXMl8xY6w572GLTq+aJbtc/iS3cvNV2WMwTFMj4lJZjuSkzX0
# af2Aw/4e23uKpjnD44cI17XkRSQbUcT756b3KtdHlKHRvKSL+zwjsraimEN6sleO
# Y5ibaFamqvoR7IuOqJ7/cbpBN+P46JU1ryQ17BvAN/HnNSEmkn6FLGtDhyvabIHq
# C69TBb1IX0GApo0txeg0d0nFrSvVACpX+/D1bRedmI5SD3EukTcBr6UELiBqXqLH
# O75tWQydMwxuP8ay7/zU4g4PApif/RBbR41pw+oVsQNIx1p5QilNy069m/V2nT3k
# gMciT2+U8bIu3GUULuW0zPuD/o88XJk+ocjFHIE8BbwCx41iL6jKhM+xHwUeIkv4
# dZT+8BlbBzNZlSAgpMu9wt0T2hyHrKYtmLxvrVi/qq6ZyO7ZUVBfOGfqAd3QtvQX
# hIERpFunzFY=
# =GNLD
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 14 Feb 2023 11:08:48 GMT
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [full]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [full]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [full]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* tag 'pull-request-2023-02-14' of https://gitlab.com/thuth/qemu: (22 commits)
  hw/s390x/event-facility: Replace DO_UPCAST(SCLPEvent) by SCLP_EVENT()
  tests/tcg/s390x: Use -nostdlib for softmmu tests
  tests/qtest: Don't build virtio-serial-test.c if device not present
  tests/qtest: bios-tables-test: Skip if missing configs
  tests/qemu-iotests: Require virtio-scsi-pci
  tests/qtest: Do not include hexloader-test if loader device is not present
  tests/qtest: Check for devices in bios-tables-test
  tests/qtest: drive_del-test: Skip tests that require missing devices
  tests/qtest: Skip unplug tests that use missing devices
  test/qtest: Fix coding style in device-plug-test.c
  tests/qtest: hd-geo-test: Check for missing devices
  tests/qtest: Add dependence on PCIE_PORT for virtio-net-failover.c
  tests/qtest: Do not run lsi53c895a test if device is not present
  tests/qtest: Skip PXE tests for missing devices
  Do not include "qemu/error-report.h" in headers that do not need it
  include/hw: Do not include "hw/registerfields.h" in headers that don't need it
  hw/misc/sga: Remove the deprecated "sga" device
  tests/qtest/npcm7xx_pwm-test: Be less verbose unless V=2
  build: deprecate --enable-gprof builds and remove from CI
  meson: Disable libdw for static builds by default
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/s390x/event-facility: Replace DO_UPCAST(SCLPEvent) by SCLP_EVENT()

Use the SCLP_EVENT() QOM type-checking macro to avoid DO_UPCAST().

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230212225144.58660-16-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Use -nostdlib for softmmu tests

The code currently uses -nostartfiles, but this does not prevent
linking with libc. On Fedora there is no cross-libc, so the linking
step fails.

Fix by using the more comprehensive -nostdlib (that's also what
probe_target_compiler() checks for as well).

Fixes: 503e549e441e ("tests/tcg/s390x: Test unaligned accesses to lowcore")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230131182057.2261614-1-iii@linux.ibm.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Don't build virtio-serial-test.c if device not present

The virtconsole device might not be present in the QEMU build that is
being tested.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230213210738.9719-5-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: bios-tables-test: Skip if missing configs

If we build with --without-default-devices, CONFIG_HPET and
CONFIG_PARALLEL are set to N, which makes the respective devices go
missing from acpi tables.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230208194700.11035-13-farosas@suse.de>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qemu-iotests: Require virtio-scsi-pci

Check that virtio-scsi-pci is present in the QEMU build before running
the tests.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230208194700.11035-12-farosas@suse.de>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Do not include hexloader-test if loader device is not present

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230208194700.11035-11-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Check for devices in bios-tables-test

Do not include tests that require devices that are not available in
the QEMU build.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <20230208194700.11035-10-farosas@suse.de>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: drive_del-test: Skip tests that require missing devices

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230208194700.11035-9-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Skip unplug tests that use missing devices

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230208194700.11035-8-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

test/qtest: Fix coding style in device-plug-test.c

We should not mix declarations and statements in QEMU code.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230208194700.11035-7-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: hd-geo-test: Check for missing devices

Don't include tests that require devices not available in the QEMU
binary.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230208194700.11035-6-farosas@suse.de>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Add dependence on PCIE_PORT for virtio-net-failover.c

This test depends on the presence of the pcie-root-port device. Add a
build time dependency.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230208194700.11035-4-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Do not run lsi53c895a test if device is not present

The tests are built once for all the targets, so as long as one QEMU
binary is built with CONFIG_LSI_SCSI_PCI=y, this test will
run. However some binaries might not include the device. So check this
again in runtime.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230208194700.11035-3-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Skip PXE tests for missing devices

Check if the devices we're trying to add are present in the QEMU
binary. They could have been removed from the build via Kconfig or the
--without-default-devices option.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230208194700.11035-2-farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Do not include "qemu/error-report.h" in headers that do not need it

Include it in the .c files instead that use the error reporting
functions.

Message-Id: <20230210111931.1115489-1-thuth@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

include/hw: Do not include "hw/registerfields.h" in headers that don't need it

Include "hw/registerfields.h" in the .c files instead (if needed).

Message-Id: <20230210112315.1116966-1-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

hw/misc/sga: Remove the deprecated "sga" device

It's been deprecated since QEMU v6.2, so it should be OK to
finally remove this now.

Message-Id: <20230209161540.1054669-1-thuth@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest/npcm7xx_pwm-test: Be less verbose unless V=2

The npcm7xx_pwm-test produces a lot of output at V=1, which
means that on our CI tests the log files exceed the gitlab
500KB limit. Suppress the messages about exactly what is
being tested unless at V=2 and above.

This follows the pattern we use with qom-test.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20230209135047.1753081-1-peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

build: deprecate --enable-gprof builds and remove from CI

As gprof relies on instrumentation you rarely get useful data compared
to a real optimised build. Lets deprecate the build option and
simplify the CI configuration as a result.

Buglink: https://gitlab.com/qemu-project/qemu/-/issues/1338
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230131094224.861621-1-alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

meson: Disable libdw for static builds by default

Static QEMU build fails on Debian Bullseye:

    /usr/bin/ld: /usr/lib/x86_64-linux-gnu/libdw.a(debuginfod-client.o): in function `__libdwfl_debuginfod_init':
    (.text.startup+0x17): undefined reference to `dlopen'

The reason is that pkg-config does not suggest -ldl for libdw, and
adding --extra-ldflags="-ldl" resolves the issue. However, static
linking with libdw is an unclear topic:

* Linux perf does it.
* Debian's libdw-dev description says:

      Only link to the static version for special cases and when you
      don't need anything from the ebl backends.

* As the error message above indicates, -ldl is also needed for
  debuginfod support.

The functionality provided by libdw is needed for analyzing performance
of JITed code, which is mostly useful to developers and researchers.
Therefore, in order to avoid unpleasant surprises for people who don't
need this, simply disable libdw for static builds by default. It can
still be enabled explicitly if needed.

Reported-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230210005208.438142-2-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

meson: Add missing libdw knobs

Add the missing meson infrastructure bits for the new libdw
dependency. Model them after the existing capstone knobs.

Fixes: 7c10cb38ccb8 ("accel/tcg: Add debuginfo support")
Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230210005208.438142-1-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

configure: Bump minimum Clang version to 10.0

Anthony Perard recently reported some problems with Clang v6.0 from
Ubuntu Bionic (with regards to the -Wmissing-braces configure test).
Since we're not officially supporting that version of Ubuntu anymore,
we should better bump our minimum version check in the configure script
instead of using our time to fix problems of unsupported compilers.
According to repology.org, our supported distros ship these versions
of Clang (looking at the highest version only):

              Fedora 36: 14.0.5
      CentOS 8 (RHEL-8): 12.0.1
              Debian 11: 13.0.1
     OpenSUSE Leap 15.4: 13.0.1
       Ubuntu LTS 20.04: 12.0.0
          FreeBSD Ports: 15.0.7
          NetBSD pkgsrc: 15.0.7
               Homebrew: 15.0.7
            MSYS2 mingw: 15.0.7
            Haiku ports: 12.0.1

While it seems like we could update to v12.0.0 from that point of view,
the default version on Ubuntu 20.04 is still v10.0, and we use that for
our CI tests based via the tests/docker/dockerfiles/ubuntu2004.docker
file.

Thus let's make v10.0 our minimum version now (which corresponds to
Apple Clang version v12.0). The -Wmissing-braces check can then be
removed, too, since both our minimum GCC and our minimum Clang version
now handle this correctly.

Message-Id: <20230131180239.1582302-1-thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Merge tag 'migration-20230213-pull-request' of https://gitlab.com/juan.quintela/qemu into staging

Migration Pull request (take3)

Hi

In this PULL request:
- Added to leonardo fixes:
Fixes: b5eea99ec2 ("migration: Add yank feature")
Reported-by: Li Xiaohui <xiaohli@redhat.com>
Please apply.

[take 2]
- rebase to latest upstream
- fix compilation of linux-user (if have_system was missing) (me)
- cleanup multifd_load_cleanup(leonardo)
- Document RAM flags (me)

Please apply.

[take 1]
This are all the reviewed patches for migration:
- AVX512 support for xbzrle (Ling Xu)
- /dev/userfaultd support (Peter Xu)
- Improve ordering of channels (Peter Xu)
- multifd cleanups (Li Zhang)
- Remove spurious files from last merge (me)
  Rebase makes that to you
- Fix mixup between state_pending_{exact,estimate} (me)
- Cache RAM size during migration (me)
- cleanup several functions (me)

Please apply.

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEGJn/jt6/WMzuA0uC9IfvGFhy1yMFAmPppZYACgkQ9IfvGFhy
# 1yPLvQ//f8D6txzFawaxrfzpSAHnq70Gx+B5GkIwgwB8nlPIC3QELEf5uooM/RGA
# nSaUctUNOJUWqVGK3vp3jDIep02DzdIUrlOfy96h+pnTMpyMWFC2BexDfveVMUId
# dw8WCWZkGCFDfIWuKF+GA8eTu6HM1ouzgCJrRmPtCqsikzAPkogPm60hQSTAQxm9
# Kzdp1SXV1HmyA440vy8rtYf71BKpvb9OJFmwgZ+ICy0rc1aUmgJbKxkwyOgiI2lq
# ONekpbOg7lzlFVAQu1QHTmYN13bsn4uzwUgdifn1PixFQyRE3AVs4jdTmqeLnoPe
# Ac6j8v3pDOw/Xf4kpRWUmhkqTMEJt8/lyneJzu1mQkw0wwiUtDvknFgPG8wJsa+J
# ZQr1cBXQj4IjtkN6+ixF7XYvx3T6pWz0L+/w2+TbFBdLWIrPgFH0yPYjhx7FdDid
# cjUHyS1a0w9ngnXOxRG8+UNHWCpPOUhXeeiyNioogYZNKu77PFxJVDMe3eB6dXAB
# pDfl4P129PloKAPafcz9E6Sxr+lIgrETZmsRJlRz4czg18TxlIukMlDtyrepNWti
# GtIf9xTpP3JKjpHnKbWLaxP5VeFC7kQd0qas4VxD+tDjbJdUZdZMfHcOSS0SMRGe
# q5LVEzMMIPCQJQIqiLEJ0HTUUOtB8i+bGoirNEbDqhLa/oZwPP8=
# =TDnO
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 13 Feb 2023 02:51:02 GMT
# gpg:                using RSA key 1899FF8EDEBF58CCEE034B82F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>" [full]
# gpg:                 aka "Juan Quintela <quintela@trasno.org>" [full]
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03  4B82 F487 EF18 5872 D723

* tag 'migration-20230213-pull-request' of https://gitlab.com/juan.quintela/qemu: (22 commits)
  ram: Document migration ram flags
  migration/multifd: Move load_cleanup inside incoming_state_destroy
  migration/multifd: Join all multifd threads in order to avoid leaks
  migration/multifd: Remove unnecessary assignment on multifd_load_cleanup()
  migration/multifd: Change multifd_load_cleanup() signature and usage
  migration: Postpone postcopy preempt channel to be after main
  migration: Add a semaphore to count PONGs
  migration: Cleanup postcopy_preempt_setup()
  migration: Rework multi-channel checks on URI
  Update bench-code for addressing CI problem
  AVX512 support for xbzrle_encode_buffer
  migration: I messed state_pending_exact/estimate
  migration: Make ram_save_target_page() a pointer
  migration: Calculate ram size once
  migration: Split ram_bytes_total_common() in two functions
  migration: Make find_dirty_block() return a single parameter
  migration: Simplify ram_find_and_save_block()
  util/userfaultfd: Support /dev/userfaultfd
  linux-headers: Update to v6.1
  multifd: Remove some redundant code
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

ram: Document migration ram flags

0x80 is RAM_SAVE_FLAG_HOOK, it is in qemu-file now.
Bigger usable flag is 0x200, noticing that.
We can reuse RAM_SAVe_FLAG_FULL.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration/multifd: Move load_cleanup inside incoming_state_destroy

Currently running migration_incoming_state_destroy() without first running
multifd_load_cleanup() will cause a yank error:

qemu-system-x86_64: ../util/yank.c:107: yank_unregister_instance:
Assertion `QLIST_EMPTY(&entry->yankfns)' failed.
(core dumped)

The above error happens in the target host, when multifd is being used
for precopy, and then postcopy is triggered and the migration finishes.
This will crash the VM in the target host.

To avoid that, move multifd_load_cleanup() inside
migration_incoming_state_destroy(), so that the load cleanup becomes part
of the incoming state destroying process.

Running multifd_load_cleanup() twice can become an issue, though, but the
only scenario it could be ran twice is on process_incoming_migration_bh().
So removing this extra call is necessary.

On the other hand, this multifd_load_cleanup() call happens way before the
migration_incoming_state_destroy() and having this happening before
dirty_bitmap_mig_before_vm_start() and vm_start() may be a need.

So introduce a new function multifd_load_shutdown() that will mainly stop
all multifd threads and close their QIOChannels. Then use this function
instead of multifd_load_cleanup() to make sure nothing else is received
before dirty_bitmap_mig_before_vm_start().

Fixes: b5eea99ec2 ("migration: Add yank feature")
Reported-by: Li Xiaohui <xiaohli@redhat.com>
Signed-off-by: Leonardo Bras <leobras@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration/multifd: Join all multifd threads in order to avoid leaks

Current approach will only join threads that are still running.

For the threads not joined, resources or private memory are always kept in
the process space and never reclaimed before process end, and this risks
serious memory leaks.

This should usually not represent a big problem, since multifd migration
is usually just ran at most a few times, and after it succeeds there is
not much to be done before exiting the process.

Yet still, it should not hurt performance to join all of them.

Fixes: b5eea99ec2 ("migration: Add yank feature")
Reported-by: Li Xiaohui <xiaohli@redhat.com>
Signed-off-by: Leonardo Bras <leobras@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration/multifd: Remove unnecessary assignment on multifd_load_cleanup()

Before assigning "p->quit = true" for every multifd channel,
multifd_load_cleanup() will call multifd_recv_terminate_threads() which
already does the same assignment, while protected by a mutex.

So there is no point doing the same assignment again.

Fixes: b5eea99ec2 ("migration: Add yank feature")
Reported-by: Li Xiaohui <xiaohli@redhat.com>
Signed-off-by: Leonardo Bras <leobras@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration/multifd: Change multifd_load_cleanup() signature and usage

Since it's introduction in commit f986c3d256 ("migration: Create multifd
migration threads"), multifd_load_cleanup() never returned any value
different than 0, neither set up any error on errp.

Even though, on process_incoming_migration_bh() an if clause uses it's
return value to decide on setting autostart = false, which will never
happen.

In order to simplify the codebase, change multifd_load_cleanup() signature
to 'void multifd_load_cleanup(void)', and for every usage remove error
handling or decision made based on return value != 0.

Fixes: b5eea99ec2 ("migration: Add yank feature")
Reported-by: Li Xiaohui <xiaohli@redhat.com>
Signed-off-by: Leonardo Bras <leobras@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Postpone postcopy preempt channel to be after main

Postcopy with preempt-mode enabled needs two channels to communicate.  The
order of channel establishment is not guaranteed.  It can happen that the
dest QEMU got the preempt channel connection request before the main
channel is established, then the migration may make no progress even during
precopy due to the wrong order.

To fix it, create the preempt channel only if we know the main channel is
established.

For a general postcopy migration, we delay it until postcopy_start(),
that's where we already went through some part of precopy on the main
channel.  To make sure dest QEMU has already established the channel, we
wait until we got the first PONG received.  That's something we do at the
start of precopy when postcopy enabled so it's guaranteed to happen sooner
or later.

For a postcopy recovery, we delay it to qemu_savevm_state_resume_prepare()
where we'll have round trips of data on bitmap synchronizations, which
means the main channel must have been established.

Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Add a semaphore to count PONGs

This is mostly useless, but useful for us to know whether the main channel
is correctly established without changing the migration protocol.

Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Cleanup postcopy_preempt_setup()

Since we just dropped the only case where postcopy_preempt_setup() can
return an error, it doesn't need a retval anymore because it never fails.
Move the preempt check to the caller, preparing it to be used elsewhere to
do nothing but as simple as kicking the async connection.

Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Rework multi-channel checks on URI

The whole idea of multi-channel checks was not properly done, IMHO.

Currently we check multi-channel in a lot of places, but actually that's
not needed because we only need to check it right after we get the URI and
that should be it.

If the URI check succeeded, we should never need to check it again because
we must have it.  If it check fails, we should fail immediately on either
the qmp_migrate or qmp_migrate_incoming, instead of failingg it later after
the connection established.

Neither should we fail any set capabiliities like what we used to do here:

5ad15e8614 ("migration: allow enabling mutilfd for specific protocol only", 2021-10-19)

Because logically the URI will only be set later after the capability is
set, so it doesn't make a lot of sense to check the URI type when setting
the capability, because we're checking the cap with an old URI passed in,
and that may not even be the URI we're going to use later.

This patch mostly reverted all such checks for before, dropping the
variable migrate_allow_multi_channels and helpers.  Instead, add a common
helper to check URI for multi-channels for either qmp_migrate and
qmp_migrate_incoming and that should do all the proper checks.  The failure
will only trigger with the "migrate" or "migrate_incoming" command, or when
user specified "-incoming xxx" where "xxx" is not "defer".

Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>