Merge tag 'pull-trivial-patches' of https://gitlab.com/mjt0k/qemu into staging

trivial patches for 2023-09-21

# -----BEGIN PGP SIGNATURE-----
#
# iQFDBAABCAAtFiEEe3O61ovnosKJMUsicBtPaxppPlkFAmUL/84PHG1qdEB0bHMu
# bXNrLnJ1AAoJEHAbT2saaT5Zlz4H/iI7Rhmsw6E46WhQPz1oly8p5I3m6Tcxs5B3
# nagfaJC0EYjKyMZC1bsATJwRj8robCb5SDhZeUfudt1ytZYFfH3ulvlUrGYrMQRW
# YEfBFIDLexqrLpsykc6ovl2NB5BXQsK3n6NNbnYE1OxQt8Cy4kNQi1bStrZ8JzDE
# lIxvWZdwoQJ2K0VRDGRLrL6XG80qeONSXEoppXxJlfhk1Ar3Ruhijn3REzfQybvV
# 1zIa1/h80fSLuwOGSPuOLqVCt6JzTuOOrfYc9F+sjcmIQWHLECy6CwTHEbb921Tw
# 9HD6ah4rvkxoN2NWSPo/kM6tNW/pyOiYwYldx5rfWcQ5mhScuO8=
# =u6P0
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 21 Sep 2023 04:33:18 EDT
# gpg:                using RSA key 7B73BAD68BE7A2C289314B22701B4F6B1A693E59
# gpg:                issuer "mjt@tls.msk.ru"
# gpg: Good signature from "Michael Tokarev <mjt@tls.msk.ru>" [full]
# gpg:                 aka "Michael Tokarev <mjt@corpit.ru>" [full]
# gpg:                 aka "Michael Tokarev <mjt@debian.org>" [full]
# Primary key fingerprint: 6EE1 95D1 886E 8FFB 810D  4324 457C E0A0 8044 65C5
#      Subkey fingerprint: 7B73 BAD6 8BE7 A2C2 8931  4B22 701B 4F6B 1A69 3E59

* tag 'pull-trivial-patches' of https://gitlab.com/mjt0k/qemu:
  docs/devel/reset.rst: Correct function names
  docs/cxl: Cleanout some more aarch64 examples.
  hw/mem/cxl_type3: Add missing copyright and license notice
  hw/cxl: Fix out of bound array access
  docs/cxl: Change to lowercase as others
  hw/cxl/cxl_device: Replace magic number in CXLError definition
  hw/pci-bridge/cxl_upstream: Fix bandwidth entry base unit for SSLBIS
  hw/cxl: Fix CFMW config memory leak
  hw/i386/pc: fix code comment on cumulative flash size
  subprojects: Use the correct .git suffix in the repository URLs
  hw/other: spelling fixes
  hw/tpm: spelling fixes
  hw/pci: spelling fixes
  hw/net: spelling fixes
  i386: spelling fixes
  bsd-user: spelling fixes
  ppc: spelling fixes

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

Merge tag 'pull-parallels-2023-09-20-v2' of https://src.openvz.org/scm/~den/qemu into staging

Parallels format driver:
* regular calculation of cluster used bitmap of the image file
* cluster allocation on the base of that bitmap (effectively allocation of
  new clusters could be done inside the image if that offset space is unused)
* support of DISCARD and WRITE_ZEROES operations
* image check bugfixes
* unit tests fixes
* unit tests covering new functionality

# -----BEGIN PGP SIGNATURE-----
#
# iQHDBAABCgAtFiEE9vE2f3B8+RUZInytPzClrpN3nJ8FAmUL7u4PHGRlbkBvcGVu
# dnoub3JnAAoJED8wpa6Td5yfdaUL/RW+nOYlFNXlrjOVeasgGLkAKrKBja8O3/As
# aRo0DLZKITK8qbLEBAeTDyCpN9LLwy7WdUR1uT4V54FzE5zZP6HAdBEoj9AsaW/9
# wsTF+oyKeqmXw2y348t+lclp8eREHySecwiVhaxTpG9J2TQfDP/D2yhzRU88P7nH
# rbVZjOF2yOthzW6Y8h8e/LMd8rfODO053tYaMEBngjirBZnhESH3mAm1WB5mYs+q
# 2++4XQZcFFKWFp952MaEDphpwYdh80E65g4vth80JrDTyyMH0KZE9cQqbFb5UgZv
# aV1/DCaH0WTSDbjCaI/SrmqKXrO0Mkd/y/ShoQpTu7qJO/FbaClA58f+KfGE7VBd
# Fa5pM+JN12UVNxnNIF/Oe+wAiVUJYKtLaDMKibj+MUjM5sE/ZRLqzFLktDbQT0kS
# Qvs1u8HTvirJpvxOkJv4cEuNw07JERCzpl/qPF6XkS9rcKeIormhftaaRmjILxS/
# KEmDVNj63g1D0XDY3WTF7LHLNjtXpw==
# =FUWj
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 21 Sep 2023 03:21:18 EDT
# gpg:                using RSA key F6F1367F707CF91519227CAD3F30A5AE93779C9F
# gpg:                issuer "den@openvz.org"
# gpg: Good signature from "Denis V. Lunev <den@openvz.org>" [unknown]
# gpg: WARNING: The key's User ID is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: F6F1 367F 707C F915 1922  7CAD 3F30 A5AE 9377 9C9F

* tag 'pull-parallels-2023-09-20-v2' of https://src.openvz.org/scm/~den/qemu: (22 commits)
  tests: extend test 131 to cover availability of the write-zeroes
  parallels: naive implementation of parallels_co_pwrite_zeroes
  tests: extend test 131 to cover availability of the discard operation
  parallels: naive implementation of parallels_co_pdiscard
  parallels: improve readability of allocate_clusters
  parallels: naive implementation of allocate_clusters with used bitmap
  parallels: update used bitmap in allocate_cluster
  parallels: accept multiple clusters in mark_used()
  tests: test self-cure of parallels image with duplicated clusters
  tests: fix broken deduplication check in parallels format test
  parallels: collect bitmap of used clusters at open
  parallels: add test which will validate data_off fixes through repair
  parallels: fix broken parallels_check_data_off()
  tests: ensure that image validation will not cure the corruption
  parallels: create mark_used() helper which sets bit in used bitmap
  parallels: refactor path when we need to re-check image in parallels_open
  parallels: return earlier from parallels_open() function on error
  parallels: return earler in fail_format branch in parallels_open()
  parallels: invent parallels_opts_prealloc() helper to parse prealloc opts
  parallels: fix memory leak in parallels_open()
  ...

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

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

Block layer patches

- Graph locking part 4 (node management)
- qemu-img map: report compressed data blocks
- block-backend: process I/O in the current AioContext

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEE3D3rFZqa+V09dFb+fwmycsiPL9YFAmULHnURHGt3b2xmQHJl
# ZGhhdC5jb20ACgkQfwmycsiPL9aB5hAAqH8To7WIUtg1rj1PY809ck78ghm18PKg
# TNdN7IbrXQghX5foh2VgPwVVl+JaW2CSrJYWQcAO6AbvFduNIi9iKzI6RT0xKXpb
# b8oQXS7zntFzwBv8ohOU5NSVJOgVmNP4h5qJIMmXgB9ZcLFG40zggVH2qQT7guUf
# 9MAc81kI/d5vvSHY0ZjdHjNOgwG4q1j8yytL7OFqWUfB8sXloUCA9lT7w4jIYD8L
# v2StUOLWB01Zts2o8SCNaFxuajs6wUee8b/DM1cyPyLy4KtOdXvLKhq2NlXpLo2i
# aZFr4PtizTVwrQZIJttA9jqM+QCsDOsiSat3BLNNsKUaCWHZB0rOGLCzMCtisyOo
# 4PzuL4UI21ik2zieO1qVM+Thqvw16kHtp6dD9pGk4X4ogGreGYEIxzBl79luR+AV
# NCRizoeFWTHKymS1tSoKrWT9ZNHcLmwemO6Tt1rMYk9jV3T4uY5e1NwxaUavEfsX
# f8dLfQjhNiySOoDknT1OSerBOVdTXURS2ri5H3GZxrxvJ4jOeFkn52C8r3YlZ3Wp
# Cr9LCUJZeXgwY+Q1JQ3D4VLY8aZ83txpw6XKEy0eTEv5wxkBj5LWhXx7hNb5F3lg
# bqaRYijVJn+P82wVxlftIzMfNeVBFHzFE90taPV5grJjr8lgrGBFmD7Puc97kfDX
# oTDBwRxJeew=
# =qTNA
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 20 Sep 2023 12:31:49 EDT
# 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: (28 commits)
  block: mark aio_poll as non-coroutine
  block-backend: process zoned requests in the current AioContext
  block-backend: process I/O in the current AioContext
  test-bdrv-drain: avoid race with BH in IOThread drain test
  block: remove AIOCBInfo->get_aio_context()
  qemu-img: map: report compressed data blocks
  block: add BDRV_BLOCK_COMPRESSED flag for bdrv_block_status()
  block: Mark bdrv_add/del_child() and caller GRAPH_WRLOCK
  block: Mark bdrv_unref_child() GRAPH_WRLOCK
  block: Mark bdrv_root_unref_child() GRAPH_WRLOCK
  block: Take graph rdlock in bdrv_change_aio_context()
  block: Take graph rdlock in bdrv_drop_intermediate()
  block: Mark bdrv_parent_cb_change_media() GRAPH_RDLOCK
  block: Mark bdrv_child_perm() GRAPH_RDLOCK
  block: Mark bdrv_get_cumulative_perm() and callers GRAPH_RDLOCK
  block: Mark bdrv_parent_perms_conflict() and callers GRAPH_RDLOCK
  block: Mark bdrv_attach_child() GRAPH_WRLOCK
  block: Call transaction callbacks with lock held
  block: Mark bdrv_attach_child_common() GRAPH_WRLOCK
  block: Mark bdrv_replace_child_tran() GRAPH_WRLOCK
  ...

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

Merge tag 'pull-testing-200923-1' of https://gitlab.com/stsquad/qemu into staging

testing updates:

  - update most Debian to bookworm
  - fix some typos
  - update loongarch toolchain
  - fix microbit test
  - handle GitLab/Cirrus timeout discrepancy
  - improve avocado console handling
  - disable mips avocado images pending bugfix

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEEZoWumedRZ7yvyN81+9DbCVqeKkQFAmUK/RcACgkQ+9DbCVqe
# KkQtYwf/Qu0eQZ8ZM4PsKcW07O76qn3cOphTFeZM01hICeiiMGRnwBVtBGIsdx1r
# MaXd7o35tnJRMbUnlGCFUDMWDZaafQIKIlsFwAGTMqgQ+kv+GB22MHpNySRZ9pXl
# ed1tOyz8maId4b3ECvGJqJSNOBB1P3tw7rdbFEhuSyXFZKJc79w1nCYjJyEtNpST
# CT7AYXJiVLiB4jSB7XH9XrkVTvw4k+PjzmLUFRJoGlik0O7xj2i7zfGO5+VxCm9t
# VluEcrlQ5w3JNL69yRhqTtrHAC7bBKqUOaF1bEA//ELNNQn2heuxDeHlDAgOtpV/
# VkShHgWJAwGLishFlv/+tmp5fbU5CQ==
# =3Lsz
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 20 Sep 2023 10:09:27 EDT
# gpg:                using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <alex.bennee@linaro.org>" [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: 6685 AE99 E751 67BC AFC8  DF35 FBD0 DB09 5A9E 2A44

* tag 'pull-testing-200923-1' of https://gitlab.com/stsquad/qemu:
  tests/avocado: Disable MIPS Malta tests due to GitLab issue #1884
  tests/avocado: Fix console data loss
  gitlab: make Cirrus CI jobs gating
  gitlab: make Cirrus CI timeout explicit
  qtest: kill orphaned qtest QEMU processes on FreeBSD
  microbit: add missing qtest_quit() call
  tests/docker: Update docker-loongarch-cross toolchain
  gitlab: fix typo/spelling in comments
  tests: update most Debian images to Bookworm

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

Merge tag 'pull-block-2023-09-01' of https://gitlab.com/hreitz/qemu into staging

Block patches

- Fix for file-posix's zoning code crashing on I/O errors
- Throttling refactoring

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEEy2LXoO44KeRfAE00ofpA0JgBnN8FAmTxnMISHGhyZWl0ekBy
# ZWRoYXQuY29tAAoJEKH6QNCYAZzfYkUP+gMG9hhzvgjj/tw9rEBQjciihzcQmqQJ
# 2Mm37RH2jj5bnnTdaTbMkcRRwVhncYSCwK9q5EYVbZmU9C/v4YJmsSEQlcl7wVou
# hbPUv6NHaBrJZX9nxNSa2RHui6pZMLKa/D0rJVB7NjYBrrRtiPo7kiLVQYjYXa2g
# kcCCfY4t3Z2RxOP31mMXRjYlhJE9bIuZdTEndrKme8KS2JGPZEJ9xjkoW1tj96EX
# oc/Cg2vk7AEtsFYA0bcD8fTFkBDJEwyYl3usu7Tk24pvH16jk7wFSqRVSsDMfnER
# tG8X3mHLIY0hbSkpzdHJdXINvZ6FWpQb0CGzIKr+pMiuWVdWr1HglBr0m4pVF+Y4
# A6AI6VX2JJgtacypoDyCZC9mzs1jIdeiwq9v5dyuikJ6ivTwEEoeoSLnLTN3AjXn
# 0mtQYzgCg5Gd6+rTo7XjSO9SSlbaVrDl/B2eXle6tmIFT5k+86fh0hc+zTmP8Rkw
# Knbc+5Le95wlMrOUNx2GhXrTGwX510hLxKboho/LITxtAzqvXnEJKrYbnkm3WPnw
# wfHnR5VQH1NKEpiH/p33og6OV/vu9e7vgp0ZNZV136SnzC90C1zMUwg2simJW701
# 34EtN0XBX8XBKrxfe7KscV9kRE8wrWWJVbhp+WOcQEomGI8uraxzWqDIk/v7NZXv
# m4XBscaB+Iri
# =oKgk
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 01 Sep 2023 04:11:46 EDT
# gpg:                using RSA key CB62D7A0EE3829E45F004D34A1FA40D098019CDF
# gpg:                issuer "hreitz@redhat.com"
# gpg: Good signature from "Hanna Reitz <hreitz@redhat.com>" [unknown]
# gpg: WARNING: The key's User ID is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: CB62 D7A0 EE38 29E4 5F00  4D34 A1FA 40D0 9801 9CDF

* tag 'pull-block-2023-09-01' of https://gitlab.com/hreitz/qemu:
  tests/file-io-error: New test
  file-posix: Simplify raw_co_prw's 'out' zone code
  file-posix: Fix zone update in I/O error path
  file-posix: Check bs->bl.zoned for zone info
  file-posix: Clear bs->bl.zoned on error
  block/throttle-groups: Use ThrottleDirection instread of bool is_write
  fsdev: Use ThrottleDirection instread of bool is_write
  throttle: use THROTTLE_MAX/ARRAY_SIZE for hard code
  throttle: use enum ThrottleDirection instead of bool is_write
  cryptodev: use NULL throttle timer cb for read direction
  test-throttle: test read only and write only
  throttle: support read-only and write-only
  test-throttle: use enum ThrottleDirection
  throttle: introduce enum ThrottleDirection

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

docs/devel/reset.rst: Correct function names

resettable_class_set_parent_phases() was mistakenly called
resettable_class_set_parent_reset_phases() in some places.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

docs/cxl: Cleanout some more aarch64 examples.

These crossed with the previous fix to get rid of examples
using aarch64 for which support is not yet upstream.

Reviewed-by: Fan Ni <fan.ni@samsung.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1892
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/mem/cxl_type3: Add missing copyright and license notice

This has been missing from the start. Assume it should match
with cxl/cxl-component-utils.c as both were part of early
postings from Ben.

Reported-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Dave Jiang <dave.jiang@intel.com>
Acked-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Fan Ni <fan.ni@samsung.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/cxl: Fix out of bound array access

According to cxl_interleave_ways_enc(), fw->num_targets is allowed to be up
to 16. This also corresponds to CXL r3.0 spec. So, the fw->target_hbs[]
array is iterated from 0 to 15. But it is statically declared of length 8.
Thus, out of bound array access may occur.

Fixes: c28db9e000 ("hw/pci-bridge: Make PCIe and CXL PXB Devices inherit from TYPE_PXB_DEV")
Signed-off-by: Dmitry Frolov <frolov@swemel.ru>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Link: https://lore.kernel.org/r/20230913101055.754709-1-frolov@swemel.ru
Cc: qemu-stable@nongnu.org
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

docs/cxl: Change to lowercase as others

Using the same style as elsewhere for topology / topo

Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Link: https://lore.kernel.org/r/20230519085802.2106900-2-lizhijian@cn.fujitsu.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/cxl/cxl_device: Replace magic number in CXLError definition

Replace the magic number 32 with CXL_RAS_ERR_HEADER_NUM for better code
readability and maintainability.

Signed-off-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/pci-bridge/cxl_upstream: Fix bandwidth entry base unit for SSLBIS

According to ACPI spec 6.5 5.2.28.4 System Locality Latency and Bandwidth
Information Structure, if the "Entry Base Unit" is 1024 for BW and the
matrix entry has the value of 100, the BW is 100 GB/s. So the
entry_base_unit should be changed from 1000 to 1024 given the comment notes
it's 16GB/s for .latency_bandwidth.

Fixes: 882877fc359d ("hw/pci-bridge/cxl-upstream: Add a CDAT table access DOE")
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/cxl: Fix CFMW config memory leak

Allocate targets and targets[n] resources when all sanity checks are
passed to avoid memory leaks.

Cc: qemu-stable@nongnu.org
Suggested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Fan Ni <fan.ni@samsung.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/i386/pc: fix code comment on cumulative flash size

- The comment is incorrectly indented / formatted.

- The comment states a 8MB limit, even though the code enforces a 16MB
  limit.

Both of these warts come from commit 0657c657eb37 ("hw/i386/pc: add max
combined fw size as machine configuration option", 2020-12-09); clean them
up.

Arguably, it's also better to be consistent with the binary units (such as
"MiB") that QEMU uses nowadays.

Cc: "Michael S. Tsirkin" <mst@redhat.com> (supporter:PC)
Cc: Marcel Apfelbaum <marcel.apfelbaum@gmail.com> (supporter:PC)
Cc: Paolo Bonzini <pbonzini@redhat.com> (maintainer:X86 TCG CPUs)
Cc: Richard Henderson <richard.henderson@linaro.org> (maintainer:X86 TCG CPUs)
Cc: Eduardo Habkost <eduardo@habkost.net> (maintainer:X86 TCG CPUs)
Cc: qemu-trivial@nongnu.org
Fixes: 0657c657eb37
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

subprojects: Use the correct .git suffix in the repository URLs

This avoids the warnings à la:
"warning: redirecting to https://gitlab.com/qemu-project/xyz.git/"

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/other: spelling fixes

Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

tests: extend test 131 to cover availability of the write-zeroes

This patch contains test which minimally tests write-zeroes on top of
working discard.

The following checks are added:
* write 2 clusters, write-zero to the first allocated cluster
* write 2 cluster, write-zero to the half the first allocated cluster

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: naive implementation of parallels_co_pwrite_zeroes

The zero flag is missed in the Parallels format specification. We can
resort to discard if we have no backing file.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

tests: extend test 131 to cover availability of the discard operation

This patch contains test which minimally tests discard and new cluster
allocation logic.

The following checks are added:
* write 2 clusters, discard the first allocated
* write another cluster, check that the hole is filled
* write 2 clusters, discard the first allocated, write 1 cluster at
  non-aligned to cluster offset (2 new clusters should be allocated)

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: naive implementation of parallels_co_pdiscard

* Discarding with backing stores is not supported by the format.
* There is no buffering/queueing of the discard operation.
* Only operations aligned to the cluster are supported.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: improve readability of allocate_clusters

Replace 'space' representing the amount of data to preallocate with
'bytes'.

Rationale:
* 'space' at each place is converted to bytes
* the unit is more close to the variable name

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: naive implementation of allocate_clusters with used bitmap

The access to the bitmap is not optimized completely.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: update used bitmap in allocate_cluster

We should extend the bitmap if the file is extended and set the bit in
the image used bitmap once the cluster is allocated. Sanity check at
that moment also looks like a good idea.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: accept multiple clusters in mark_used()

This would be useful in the next patch in allocate_clusters(). This
change would not imply serious performance drawbacks as usually image
is full of data or are at the end of the bitmap.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

tests: test self-cure of parallels image with duplicated clusters

The test is quite similar with the original one for duplicated clusters.
There is the only difference in the operation which should fix the
image.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

tests: fix broken deduplication check in parallels format test

Original check is broken as supposed reading from 2 different clusters
results in read from the same file offset twice. This is definitely
wrong.

We should be sure that
* the content of both clusters is correct after repair
* clusters are at the different offsets after repair
In order to check the latter we write some content into the first one
and validate that fact.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: collect bitmap of used clusters at open

If the operation is failed, we need to check image consistency if the
problem is not about memory allocation.

Bitmap adjustments in allocate_cluster are not performed yet.
They worth to be separate. This was proven useful during debug of this
series. Kept as is for future bissecting.

It should be specifically noted that used bitmap must be recalculated
if data_off has been fixed during image consistency check.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: add test which will validate data_off fixes through repair

We have only check through self-repair and that proven to be not enough.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: fix broken parallels_check_data_off()

Once we have repaired data_off field in the header we should update
s->data_start which is calculated on the base of it.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

tests: ensure that image validation will not cure the corruption

Since
    commit cfce1091d55322789582480798a891cbaf66924e
    Author: Alexander Ivanov <alexander.ivanov@virtuozzo.com>
    Date:   Tue Jul 18 12:44:29 2023 +0200
    parallels: Image repairing in parallels_open()
there is a potential pit fall with calling
    qemu-io -c "read"
The image is opened in read-write mode and thus could be potentially
repaired. This could ruin testing process.

The patch forces read-only opening for reads. In that case repairing
is impossible.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: create mark_used() helper which sets bit in used bitmap

This functionality is used twice already and next patch will add more
code with it.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: refactor path when we need to re-check image in parallels_open

More conditions follows thus the check should be more scalable.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: return earlier from parallels_open() function on error

At the beginning of the function we can return immediately until we
really allocate s->header.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: return earler in fail_format branch in parallels_open()

We do not need to perform any deallocation/cleanup if wrong format is
detected.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: invent parallels_opts_prealloc() helper to parse prealloc opts

This patch creates above mentioned helper and moves its usage to the
beginning of parallels_open(). This simplifies parallels_open() a bit.

The patch also ensures that we store prealloc_size on block driver state
always in sectors. This makes code cleaner and avoids wrong opinion at
the assignment that the value is in bytes.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: fix memory leak in parallels_open()

We should free opts allocated through qemu_opts_create() at the end.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: mark driver as supporting CBT

Parallels driver indeed support Parallels Dirty Bitmap Feature in
read-only mode. The patch adds bdrv_supports_persistent_dirty_bitmap()
callback which always return 1 to indicate that.

This will allow to copy CBT from Parallels image with qemu-img.

Note: read-write support is signalled through
bdrv_co_can_store_new_dirty_bitmap() and is different.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

parallels: fix formatting in bdrv_parallels initialization

Old code is ugly and contains tabulations. There are no functional
changes in this patch.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alexander Ivanov <alexander.ivanov@virtuozzo.com>

Merge tag 'pull-loongarch-20230920' of https://gitlab.com/gaosong/qemu into staging

Add LASX instructions support.

# -----BEGIN PGP SIGNATURE-----
#
# iLMEAAEIAB0WIQS4/x2g0v3LLaCcbCxAov/yOSY+3wUCZQqV7wAKCRBAov/yOSY+
# 35GTA/9rXGbr9pIUnlGstUnWzIJb0vs6f4kt9DaKRPF1zyxaF/59sgl3gqCNAjBA
# eAKfm5W4B8ABJ+PYR3ZVAg9AcAP9AOEi+qV6DgRwvYPPK3WbGqIpJL7i+7gNMMUs
# gppv+IfJEkri8YLXXa7GWffuGOebqdqyD6Pl1B2eiKS4KYSRGw==
# =fNr2
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 20 Sep 2023 02:49:19 EDT
# gpg:                using RSA key B8FF1DA0D2FDCB2DA09C6C2C40A2FFF239263EDF
# gpg: Good signature from "Song Gao <m17746591750@163.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B8FF 1DA0 D2FD CB2D A09C  6C2C 40A2 FFF2 3926 3EDF

* tag 'pull-loongarch-20230920' of https://gitlab.com/gaosong/qemu: (57 commits)
  target/loongarch: CPUCFG support LASX
  target/loongarch: Move simply DO_XX marcos togther
  target/loongarch: Implement xvld xvst
  target/loongarch: Implement xvshuf xvperm{i} xvshuf4i
  target/loongarch: Implement xvpack xvpick xvilv{l/h}
  target/loongarch: Implement xvreplve xvinsve0 xvpickve
  target/loongarch: Implement xvinsgr2vr xvpickve2gr
  target/loongarch: Implement xvbitsel xvset
  target/loongarch: Implement xvfcmp
  target/loongarch: Implement xvseq xvsle xvslt
  target/loongarch: Implement LASX fpu fcvt instructions
  target/loongarch: Implement LASX fpu arith instructions
  target/loongarch: Implement xvfrstp
  target/loongarch: Implement xvbitclr xvbitset xvbitrev
  target/loongarch: Implement xvpcnt
  target/loongarch: Implement xvclo xvclz
  target/loongarch: Implement xvssrlrn xvssrarn
  target/loongarch: Implement xvssrln xvssran
  target/loongarch: Implement xvsrlrn xvsrarn
  target/loongarch: Implement xvsrln xvsran
  ...

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

Merge tag 'hppa-btlb-pull-request' of https://github.com/hdeller/qemu-hppa into staging

Block-TLB support and linux-user fixes for hppa target

All 32-bit hppa CPUs allow a fixed number of TLB entries to have a
different page size than the default 4k.
Those are called "Block-TLBs" and are created at startup by the
operating system and managed by the firmware of hppa machines
through the firmware PDC_BLOCK_TLB call.

This patchset adds the necessary glue to SeaBIOS-hppa and
qemu to allow up to 16 BTLB entries in the emulation.

Two patches from Mikulas Patocka fix signal delivery issues
in linux-user on hppa.

# -----BEGIN PGP SIGNATURE-----
#
# iHUEABYKAB0WIQS86RI+GtKfB8BJu973ErUQojoPXwUCZQnz0wAKCRD3ErUQojoP
# X6NDAP9F1Huhceot8peohGodRDOhnXWfDcjQZSDvadieKv/rJQEA60Z5QV5VlQgw
# SyUT4AcoiB7N4nvS+iDa+6dKfRH/YQM=
# =kqqt
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 19 Sep 2023 15:17:39 EDT
# gpg:                using EDDSA key BCE9123E1AD29F07C049BBDEF712B510A23A0F5F
# gpg: Good signature from "Helge Deller <deller@gmx.de>" [unknown]
# gpg:                 aka "Helge Deller <deller@kernel.org>" [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: 4544 8228 2CD9 10DB EF3D  25F8 3E5F 3D04 A7A2 4603
#      Subkey fingerprint: BCE9 123E 1AD2 9F07 C049  BBDE F712 B510 A23A 0F5F

* tag 'hppa-btlb-pull-request' of https://github.com/hdeller/qemu-hppa:
  linux-user/hppa: lock both words of function descriptor
  linux-user/hppa: clear the PSW 'N' bit when delivering signals
  target/hppa: Wire up diag instruction to support BTLB
  target/hppa: Extract diagnose immediate value
  target/hppa: Add BTLB support to hppa TLB functions
  target/hppa: Report and clear BTLBs via fw_cfg at startup
  target/hppa: Allow up to 16 BTLB entries
  target/hppa: Update to SeaBIOS-hppa version 9

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

block: mark aio_poll as non-coroutine

It is forbidden to block on the event loop during a coroutine, as that
can cause deadlocks due to recursive locking.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <20230908075458.527013-1-pbonzini@redhat.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block-backend: process zoned requests in the current AioContext

Process zoned requests in the current thread's AioContext instead of in
the BlockBackend's AioContext.

There is no need to use the BlockBackend's AioContext thanks to CoMutex
bs->wps->colock, which protects zone metadata.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230912231037.826804-5-stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block-backend: process I/O in the current AioContext

Switch blk_aio_*() APIs over to multi-queue by using
qemu_get_current_aio_context() instead of blk_get_aio_context(). This
change will allow devices to process I/O in multiple IOThreads in the
future.

I audited existing blk_aio_*() callers:
- migration/block.c: blk_mig_lock() protects the data accessed by the
  completion callback.
- The remaining emulated devices and exports run with
  qemu_get_aio_context() == blk_get_aio_context().

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230912231037.826804-4-stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

test-bdrv-drain: avoid race with BH in IOThread drain test

This patch fixes a race condition in test-bdrv-drain that is difficult
to reproduce. test-bdrv-drain sometimes fails without an error message
on the block pull request sent by Kevin Wolf on Sep 4, 2023. I was able
to reproduce it locally and found that "block-backend: process I/O in
the current AioContext" (in this patch series) is the first commit where
it reproduces.

I do not know why "block-backend: process I/O in the current AioContext"
exposes this bug. It might be related to the fact that the test's preadv
request runs in the main thread instead of IOThread a after my commit.
That might simply change the timing of the test.

Now on to the race condition in test-bdrv-drain. The main thread
schedules a BH in IOThread a and then drains the BDS:

  aio_bh_schedule_oneshot(ctx_a, test_iothread_main_thread_bh, &data);

  /* The request is running on the IOThread a. Draining its block device
   * will make sure that it has completed as far as the BDS is concerned,
   * but the drain in this thread can continue immediately after
   * bdrv_dec_in_flight() and aio_ret might be assigned only slightly
   * later. */
  do_drain_begin(drain_type, bs);

If the BH completes before do_drain_begin() then there is nothing to
worry about.

If the BH invokes bdrv_flush() before do_drain_begin(), then
do_drain_begin() waits for it to complete.

The problematic case is when do_drain_begin() runs before the BH enters
bdrv_flush(). Then do_drain_begin() misses the BH and the drain
mechanism has failed in quiescing I/O.

Fix this by incrementing the in_flight counter so that do_drain_begin()
waits for test_iothread_main_thread_bh().

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230912231037.826804-3-stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: remove AIOCBInfo->get_aio_context()

The synchronous bdrv_aio_cancel() function needs the acb's AioContext so
it can call aio_poll() to wait for cancellation.

It turns out that all users run under the BQL in the main AioContext, so
this callback is not needed.

Remove the callback, mark bdrv_aio_cancel() GLOBAL_STATE_CODE just like
its blk_aio_cancel() caller, and poll the main loop AioContext.

The purpose of this cleanup is to identify bdrv_aio_cancel() as an API
that does not work with the multi-queue block layer.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230912231037.826804-2-stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-img: map: report compressed data blocks

Right now "qemu-img map" reports compressed blocks as containing data
but having no host offset.  This is not very informative.  Instead,
let's add another boolean field named "compressed" in case JSON output
mode is specified.  This is achieved by utilizing new allocation status
flag BDRV_BLOCK_COMPRESSED for bdrv_block_status().

Also update the expected qemu-iotests outputs to contain the new field.

Signed-off-by: Andrey Drobyshev <andrey.drobyshev@virtuozzo.com>
Message-ID: <20230907210226.953821-3-andrey.drobyshev@virtuozzo.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: add BDRV_BLOCK_COMPRESSED flag for bdrv_block_status()

Functions qcow2_get_host_offset(), get_cluster_offset(),
vmdk_co_block_status() explicitly report compressed cluster types when data
is compressed.  However, this information is never passed further.  Let's
make use of it by adding new BDRV_BLOCK_COMPRESSED flag for
bdrv_block_status(), so that caller may know that the data range is
compressed.  In particular, we're going to use this flag to tweak
"qemu-img map" output.

This new flag is only being utilized by qcow, qcow2 and vmdk formats, as only
those support compression.

Reviewed-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Andrey Drobyshev <andrey.drobyshev@virtuozzo.com>
Message-ID: <20230907210226.953821-2-andrey.drobyshev@virtuozzo.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_add/del_child() and caller GRAPH_WRLOCK

The functions read the parents list in the generic block layer, so we
need to hold the graph lock already there. The BlockDriver
implementations actually modify the graph, so it has to be a writer
lock.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-22-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_unref_child() GRAPH_WRLOCK

Instead of taking the writer lock internally, require callers to already
hold it when calling bdrv_unref_child(). These callers will typically
already hold the graph lock once the locking work is completed, which
means that they can't call functions that take it internally.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-21-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_root_unref_child() GRAPH_WRLOCK

Instead of taking the writer lock internally, require callers to already
hold it when calling bdrv_root_unref_child(). These callers will
typically already hold the graph lock once the locking work is
completed, which means that they can't call functions that take it
internally.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-20-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Take graph rdlock in bdrv_change_aio_context()

The function reads the parents list, so it needs to hold the graph lock.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-ID: <20230911094620.45040-19-kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Take graph rdlock in bdrv_drop_intermediate()

The function reads the parents list, so it needs to hold the graph lock.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-ID: <20230911094620.45040-18-kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_parent_cb_change_media() GRAPH_RDLOCK

The function reads the parents list, so it needs to hold the graph lock.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-17-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_child_perm() GRAPH_RDLOCK

This adds GRAPH_RDLOCK annotations to declare that callers of
bdrv_child_perm() need to hold a reader lock for the graph because
some implementations access the children list of a node.

The callers of bdrv_child_perm() conveniently already hold the lock.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-16-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_get_cumulative_perm() and callers GRAPH_RDLOCK

The function reads the parents list, so it needs to hold the graph lock.

This happens to result in BlockDriver.bdrv_set_perm() to be called with
the graph lock held. For consistency, make it the same for all of the
BlockDriver callbacks for updating permissions and annotate the function
pointers with GRAPH_RDLOCK_PTR.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-15-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_parent_perms_conflict() and callers GRAPH_RDLOCK

The function reads the parents list, so it needs to hold the graph lock.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-14-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_attach_child() GRAPH_WRLOCK

Instead of taking the writer lock internally, require callers to already
hold it when calling bdrv_attach_child_common(). These callers will
typically already hold the graph lock once the locking work is
completed, which means that they can't call functions that take it
internally.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-13-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Call transaction callbacks with lock held

In previous patches, we changed some transactionable functions to be
marked as GRAPH_WRLOCK, but required that tran_finalize() is still
called without the lock. This was because all callbacks that can be in
the same transaction need to follow the same convention.

Now that we don't have conflicting requirements any more, we can switch
all of the transaction callbacks to be declared GRAPH_WRLOCK, too, and
call tran_finalize() with the lock held.

Document for each of these transactionable functions that the lock needs
to be held when completing the transaction, and make sure that all
callers down to the place where the transaction is finalised actually
have the writer lock.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-12-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_attach_child_common() GRAPH_WRLOCK

Instead of taking the writer lock internally, require callers to already
hold it when calling bdrv_attach_child_common(). These callers will
typically already hold the graph lock once the locking work is
completed, which means that they can't call functions that take it
internally.

Note that the transaction callbacks still take the lock internally, so
tran_finalize() must be called without the lock held. This is because
bdrv_append() also calls bdrv_replace_node_noperm(), which currently
requires the transaction callbacks to be called unlocked. In the next
step, both of them can be switched to locked tran_finalize() calls
together.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-11-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_replace_child_tran() GRAPH_WRLOCK

Instead of taking the writer lock internally, require callers to already
hold it when calling bdrv_replace_child_tran(). These callers will
typically already hold the graph lock once the locking work is
completed, which means that they can't call functions that take it
internally.

While a graph lock is held, polling is not allowed. Therefore draining
the necessary nodes can no longer be done in bdrv_remove_child() and
bdrv_replace_node_noperm(), but the callers must already make sure that
they are drained.

Note that the transaction callbacks still take the lock internally, so
tran_finalize() must be called without the lock held. This is because
bdrv_append() also calls bdrv_attach_child_noperm(), which currently
requires to be called unlocked. Once it changes, the transaction
callbacks can be changed, too.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-10-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Mark bdrv_replace_child_noperm() GRAPH_WRLOCK

Instead of taking the writer lock internally, require callers to already
hold it when calling bdrv_replace_child_noperm(). These callers will
typically already hold the graph lock once the locking work is
completed, which means that they can't call functions that take it
internally.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-9-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block-coroutine-wrapper: Allow arbitrary parameter names

Don't assume specific parameter names like 'bs' or 'blk' in the
generated code, but use the actual name.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-8-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block-coroutine-wrapper: Add no_co_wrapper_bdrv_wrlock functions

Add a new wrapper type for GRAPH_WRLOCK functions that should be called
from coroutine context.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-7-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Introduce bdrv_schedule_unref()

bdrv_unref() is called by a lot of places that need to hold the graph
lock (it naturally happens in the context of operations that change the
graph). However, bdrv_unref() takes the graph writer lock internally, so
it can't actually be called while already holding a graph lock without
causing a deadlock.

bdrv_unref() also can't just become GRAPH_WRLOCK because it drains the
node before closing it, and draining requires that the graph is
unlocked.

The solution is to defer deleting the node until we don't hold the lock
any more and draining is possible again.

Note that keeping images open for longer than necessary can create
problems, too: You can't open an image again before it is really closed
(if image locking didn't prevent it, it would cause corruption).
Reopening an image immediately happens at least during bdrv_open() and
bdrv_co_create().

In order to solve this problem, make sure to run the deferred unref in
bdrv_graph_wrunlock(), i.e. the first possible place where we can drain
again. This is also why bdrv_schedule_unref() is marked GRAPH_WRLOCK.

The output of iotest 051 is updated because the additional polling
changes the order of HMP output, resulting in a new "(qemu)" prompt in
the test output that was previously on a separate line and filtered out.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20230911094620.45040-6-kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Take AioContext lock for bdrv_append() more consistently

The documentation for bdrv_append() says that the caller must hold the
AioContext lock for bs_top. Change all callers to actually adhere to the
contract.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-5-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

preallocate: Don't poll during permission updates

When the permission related BlockDriver callbacks are called, we are in
the middle of an operation traversing the block graph. Polling in such a
place is a very bad idea because the graph could change in unexpected
ways. In the future, callers will also hold the graph lock, which is
likely to turn polling into a deadlock.

So we need to get rid of calls to functions like bdrv_getlength() or
bdrv_truncate() there as these functions poll internally. They are
currently used so that when no parent has write/resize permissions on
the image any more, the preallocate filter drops the extra preallocated
area in the image file and gives up write/resize permissions itself.

In order to achieve this without polling in .bdrv_check_perm, don't
immediately truncate the image, but only schedule a BH to do so. The
filter keeps the write/resize permissions a bit longer now until the BH
has executed.

There is one case in which delaying doesn't work: Reopening the image
read-only. In this case, bs->file will likely be reopened read-only,
too, so keeping write permissions a bit longer on it doesn't work. But
we can already cover this case in preallocate_reopen_prepare() and not
rely on the permission updates for it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-4-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

preallocate: Factor out preallocate_truncate_to_real_size()

It's essentially the same code in preallocate_check_perm() and
preallocate_close(), except that the latter ignores errors.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-3-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Remove unused BlockReopenQueueEntry.perms_checked

This field has been unused since commit 72373e40fbc ('block:
bdrv_reopen_multiple: refresh permissions on updated graph').
Remove it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230911094620.45040-2-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

tests/avocado: Disable MIPS Malta tests due to GitLab issue #1884

Commit 0d58c66068 ("softmmu: Use async_run_on_cpu in tcg_commit")
introduced a regression which is only triggered by the MIPS Malta
machine. Since those tests are gatting and disturb the CI workflow,
disable them until https://gitlab.com/qemu-project/qemu/-/issues/1866
is fixed.

  $ make check-avocado \
      AVOCADO_TAGS='arch:mipsel arch:mips64el' \
      AVOCADO_ALLOW_UNTRUSTED_CODE=1 \
      AVOCADO_TIMEOUT_EXPECTED=1
    AVOCADO tests/avocado
   (04/24) tests/avocado/boot_linux_console.py:BootLinuxConsole.test_mips_malta32el_nanomips_4k: INTERRUPTED: Test interrupted by SIGTERM\nRunner error occurred: Timeout reached\nOriginal status: ERROR\n... (90.39 s)
   (05/24) tests/avocado/boot_linux_console.py:BootLinuxConsole.test_mips_malta32el_nanomips_16k_up: INTERRUPTED: Test interrupted by SIGTERM\nRunner error occurred: Timeout reached\nOriginal status: ERROR\n... (90.29 s)
   (06/24) tests/avocado/boot_linux_console.py:BootLinuxConsole.test_mips_malta32el_nanomips_64k_dbg: INTERRUPTED: Test interrupted by SIGTERM\nRunner error occurred: Timeout reached\nOriginal status: ERROR\n... (92.53 s)
   (11/24) tests/avocado/machine_mips_malta.py:MaltaMachineFramebuffer.test_mips_malta_i6400_framebuffer_logo_1core: INTERRUPTED: Test interrupted by SIGTERM\nRunner error occurred: Timeout reached\nOriginal status: ERROR\n... (25.78 s)
  RESULTS    : PASS 8 | ERROR 0 | FAIL 0 | SKIP 7 | WARN 2 | INTERRUPT 5 | CANCEL 2
  JOB TIME   : 525.60 s                                      ^^^^^^^^^^^

Reported-by: Thomas Huth <thuth@redhat.com>
Suggested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230913135339.9128-1-philmd@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230914155422.426639-10-alex.bennee@linaro.org>

tests/avocado: Fix console data loss

Occasionally some avocado tests will fail waiting for console line
despite the machine running correctly. Console data goes missing, as can
be seen in the console log. This is due to _console_interaction calling
makefile() on the console socket each time it is invoked, which must be
losing old buffer contents when going out of scope.

It is not enough to makefile() with buffered=0. That helps significantly
but data loss is still possible. My guess is that readline() has a line
buffer even when the file is in unbuffered mode, that can eat data.

Fix this by providing a console file that persists for the life of the
console.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: "Daniel P. Berrangé" <berrange@redhat.com>
Message-Id: <20230912131340.405619-1-npiggin@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: John Snow <jsnow@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230914155422.426639-9-alex.bennee@linaro.org>

gitlab: make Cirrus CI jobs gating

The Cirrus CI jobs have been non-gating for a while to let us build
confidence in their reliability. Aside from periodic dependancy
problems when FreeBSD Ports switches to be based on a new FreeBSD
image version, the jobs have been reliable. It is thus worth making
them gating to prevent build failures being missed during merges.

Signed-off-by: "Daniel P. Berrangé" <berrange@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230912184130.3056054-5-berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230914155422.426639-8-alex.bennee@linaro.org>

gitlab: make Cirrus CI timeout explicit

On the GitLab side we're invoking the Cirrus CI job using the
cirrus-run tool which speaks to the Cirrus REST API. Cirrus
sometimes tasks 5-10 minutes to actually schedule the task,
and thus the execution time of 'cirrus-run' inside GitLab will
be slightly longer than the execution time of the Cirrus CI
task.

Setting the timeout in the GitLab CI job should thus be done
in relation to the timeout set for the Cirrus CI job. While
Cirrus CI defaults to 60 minutes, it is better to set this
explicitly, and make the relationship between the jobs
explicit

Signed-off-by: "Daniel P. Berrangé" <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230912184130.3056054-4-berrange@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230914155422.426639-7-alex.bennee@linaro.org>

qtest: kill orphaned qtest QEMU processes on FreeBSD

On Linux we use PR_SET_PDEATHSIG to kill orphaned QEMU processes
if we fail to call qtest_quit(), or the test program aborts/segvs.
This prevents meson from hanging forever due to the orphaned
process keeping stdout open.

On FreeBSD we can achieve the same using PROC_PDEATHSIG_CTL, which
gives us the equivalent protection against hangs.

Signed-off-by: "Daniel P. Berrangé" <berrange@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230912184130.3056054-3-berrange@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230914155422.426639-6-alex.bennee@linaro.org>

microbit: add missing qtest_quit() call

Without this call, the QEMU process is being left running which on
FreeBSD 13.2 at least, makes meson think the test is still running,
and thus execution of "make check" continues forever.

This fixes the regression introduced in:

  commit a9c9bbee855877293683012942d3485d50f286af
  Author: Chris Laplante <chris@laplante.io>
  Date:   Tue Aug 22 17:31:02 2023 +0100

    qtest: microbit-test: add tests for nRF51 DETECT

Fixes: https://gitlab.com/qemu-project/qemu/-/issues/1882
Signed-off-by: "Daniel P. Berrangé" <berrange@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230912184130.3056054-2-berrange@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230914155422.426639-5-alex.bennee@linaro.org>

tests/docker: Update docker-loongarch-cross toolchain

Update from clfs 5.0 to clfs 8.1, which includes updates
to binutils 2.41, gcc 13.2, and glibc 2.38.

See https://github.com/loongson/build-tools

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230829220228.928506-2-richard.henderson@linaro.org>
Message-Id: <20230914155422.426639-4-alex.bennee@linaro.org>

gitlab: fix typo/spelling in comments

Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230914155422.426639-3-alex.bennee@linaro.org>

tests: update most Debian images to Bookworm

Bookworm has been out a while now. Time to update our containers to
the current stable. This requires the latest lcitool repo so update
the sub-module too.

For some reason the MIPs containers won't build so skip those for now.
We also have to skip the armel builds due to a stuck libc update.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230914155422.426639-2-alex.bennee@linaro.org>

target/loongarch: CPUCFG support LASX

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-58-gaosong@loongson.cn>

target/loongarch: Move simply DO_XX marcos togther

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-57-gaosong@loongson.cn>

target/loongarch: Implement xvld xvst

This patch includes:
- XVLD[X], XVST[X];
- XVLDREPL.{B/H/W/D};
- XVSTELM.{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-56-gaosong@loongson.cn>

target/loongarch: Implement xvshuf xvperm{i} xvshuf4i

This patch includes:
- XVSHUF.{B/H/W/D};
- XVPERM.W;
- XVSHUF4i.{B/H/W/D};
- XVPERMI.{W/D/Q};
- XVEXTRINS.{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-55-gaosong@loongson.cn>

target/loongarch: Implement xvpack xvpick xvilv{l/h}

This patch includes:
- XVPACK{EV/OD}.{B/H/W/D};
- XVPICK{EV/OD}.{B/H/W/D};
- XVILV{L/H}.{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-54-gaosong@loongson.cn>

target/loongarch: Implement xvreplve xvinsve0 xvpickve

This patch includes:
- XVREPLVE.{B/H/W/D};
- XVREPL128VEI.{B/H/W/D};
- XVREPLVE0.{B/H/W/D/Q};
- XVINSVE0.{W/D};
- XVPICKVE.{W/D};
- XVBSLL.V, XVBSRL.V.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-53-gaosong@loongson.cn>

target/loongarch: Implement xvinsgr2vr xvpickve2gr

This patch includes:
- XVINSGR2VR.{W/D};
- XVPICKVE2GR.{W/D}[U].

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-52-gaosong@loongson.cn>

target/loongarch: Implement xvbitsel xvset

This patch includes:
- XVBITSEL.V;
- XVBITSELI.B;
- XVSET{EQZ/NEZ}.V;
- XVSETANYEQZ.{B/H/W/D};
- XVSETALLNEZ.{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-51-gaosong@loongson.cn>

target/loongarch: Implement xvfcmp

This patch includes:
- XVFCMP.cond.{S/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-50-gaosong@loongson.cn>

target/loongarch: Implement xvseq xvsle xvslt

This patch includes:
- XVSEQ[I].{B/H/W/D};
- XVSLE[I].{B/H/W/D}[U];
- XVSLT[I].{B/H/W/D/}[U].

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-49-gaosong@loongson.cn>

target/loongarch: Implement LASX fpu fcvt instructions

This patch includes:
- XVFCVT{L/H}.{S.H/D.S};
- XVFCVT.{H.S/S.D};
- XVFRINT[{RNE/RZ/RP/RM}].{S/D};
- XVFTINT[{RNE/RZ/RP/RM}].{W.S/L.D};
- XVFTINT[RZ].{WU.S/LU.D};
- XVFTINT[{RNE/RZ/RP/RM}].W.D;
- XVFTINT[{RNE/RZ/RP/RM}]{L/H}.L.S;
- XVFFINT.{S.W/D.L}[U];
- X[CVFFINT.S.L, VFFINT{L/H}.D.W.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-48-gaosong@loongson.cn>

target/loongarch: Implement LASX fpu arith instructions

This patch includes:
- XVF{ADD/SUB/MUL/DIV}.{S/D};
- XVF{MADD/MSUB/NMADD/NMSUB}.{S/D};
- XVF{MAX/MIN}.{S/D};
- XVF{MAXA/MINA}.{S/D};
- XVFLOGB.{S/D};
- XVFCLASS.{S/D};
- XVF{SQRT/RECIP/RSQRT}.{S/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-47-gaosong@loongson.cn>

target/loongarch: Implement xvfrstp

This patch includes:
- XVFRSTP[I].{B/H}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-46-gaosong@loongson.cn>

target/loongarch: Implement xvbitclr xvbitset xvbitrev

This patch includes:
- XVBITCLR[I].{B/H/W/D};
- XVBITSET[I].{B/H/W/D};
- XVBITREV[I].{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-45-gaosong@loongson.cn>

target/loongarch: Implement xvpcnt

This patch includes:
- VPCNT.{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-44-gaosong@loongson.cn>

target/loongarch: Implement xvclo xvclz

This patch includes:
- XVCLO.{B/H/W/D};
- XVCLZ.{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-43-gaosong@loongson.cn>

target/loongarch: Implement xvssrlrn xvssrarn

This patch includes:
- XVSSRLRN.{B.H/H.W/W.D};
- XVSSRARN.{B.H/H.W/W.D};
- XVSSRLRN.{BU.H/HU.W/WU.D};
- XVSSRARN.{BU.H/HU.W/WU.D};
- XVSSRLRNI.{B.H/H.W/W.D/D.Q};
- XVSSRARNI.{B.H/H.W/W.D/D.Q};
- XVSSRLRNI.{BU.H/HU.W/WU.D/DU.Q};
- XVSSRARNI.{BU.H/HU.W/WU.D/DU.Q}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-42-gaosong@loongson.cn>

target/loongarch: Implement xvssrln xvssran

This patch includes:
- XVSSRLN.{B.H/H.W/W.D};
- XVSSRAN.{B.H/H.W/W.D};
- XVSSRLN.{BU.H/HU.W/WU.D};
- XVSSRAN.{BU.H/HU.W/WU.D};
- XVSSRLNI.{B.H/H.W/W.D/D.Q};
- XVSSRANI.{B.H/H.W/W.D/D.Q};
- XVSSRLNI.{BU.H/HU.W/WU.D/DU.Q};
- XVSSRANI.{BU.H/HU.W/WU.D/DU.Q}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-41-gaosong@loongson.cn>

target/loongarch: Implement xvsrlrn xvsrarn

This patch includes:
- XVSRLRN.{B.H/H.W/W.D};
- XVSRARN.{B.H/H.W/W.D};
- XVSRLRNI.{B.H/H.W/W.D/D.Q};
- XVSRARNI.{B.H/H.W/W.D/D.Q}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-40-gaosong@loongson.cn>

target/loongarch: Implement xvsrln xvsran

This patch includes:
- XVSRLN.{B.H/H.W/W.D};
- XVSRAN.{B.H/H.W/W.D};
- XVSRLNI.{B.H/H.W/W.D/D.Q};
- XVSRANI.{B.H/H.W/W.D/D.Q}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-39-gaosong@loongson.cn>

target/loongarch: Implement xvsrlr xvsrar

This patch includes:
- XVSRLR[I].{B/H/W/D};
- XVSRAR[I].{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-38-gaosong@loongson.cn>

target/loongarch: Implement xvsllwil xvextl

This patch includes:
- XVSLLWIL.{H.B/W.H/D.W};
- XVSLLWIL.{HU.BU/WU.HU/DU.WU};
- XVEXTL.Q.D, VEXTL.QU.DU.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-37-gaosong@loongson.cn>

target/loongarch: Implement xvsll xvsrl xvsra xvrotr

This patch includes:
- XVSLL[I].{B/H/W/D};
- XVSRL[I].{B/H/W/D};
- XVSRA[I].{B/H/W/D};
- XVROTR[I].{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-36-gaosong@loongson.cn>