Merge tag 'block-pull-request' of https://gitlab.com/stefanha/qemu into staging

Pull request

Please include these bug fixes in QEMU 8.1. Thanks!

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEhpWov9P5fNqsNXdanKSrs4Grc8gFAmTCzPUACgkQnKSrs4Gr
# c8g1DAf/fPUQ4zRsCn079pHIyK9TFo4COm23p4kiusxj8otfjt8LH1Zsc9pGWC2+
# bl2RlnPID8JlyJFDRN7b/RCEhj45a83GtCmhDDmqVgy1eO5vwOKm2XyyWeD+pq/U
# Hf2QLPLZZ7tCD8Njpty+gB3Ux4zqthKGXSg8FpJ3w0tl4me2efLvjMa6jHMwtnHT
# aAbyQ3WMpT9w4XHLqRQDHzBqrTSY4od3nl9SrM/DQ2klLIcz8ECTEZVBY9B3pq6m
# QvAg24tfb0QvS14YnZv/PMCfOaVuE87M9G4f93pCynnMxMYze+XczL0sGhIAS9wp
# 03NgGlhGumOix6r2kHjlG6p3xywV8A==
# =jMf8
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 27 Jul 2023 01:00:53 PM PDT
# gpg:                using RSA key 8695A8BFD3F97CDAAC35775A9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>" [full]
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>" [full]

* tag 'block-pull-request' of https://gitlab.com/stefanha/qemu:
  block/blkio: use blkio_set_int("fd") to check fd support
  block/blkio: fall back on using `path` when `fd` setting fails
  block/blkio: retry blkio_connect() if it fails using `fd`
  block/blkio: move blkio_connect() in the drivers functions
  block: Fix pad_request's request restriction
  block/file-posix: fix g_file_get_contents return path
  block/blkio: do not use open flags in qemu_open()
  block/blkio: enable the completion eventfd

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

block/blkio: use blkio_set_int("fd") to check fd support

Setting the `fd` property fails with virtio-blk-* libblkio drivers
that do not support fd passing since
https://gitlab.com/libblkio/libblkio/-/merge_requests/208.

Getting the `fd` property, on the other hand, always succeeds for
virtio-blk-* libblkio drivers even when they don't support fd passing.

This patch switches to setting the `fd` property because it is a
better mechanism for probing fd passing support than getting the `fd`
property.

Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20230727161020.84213-5-sgarzare@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/blkio: fall back on using `path` when `fd` setting fails

qemu_open() fails if called with an unix domain socket in this way:
    -blockdev node-name=drive0,driver=virtio-blk-vhost-user,path=vhost-user-blk.sock,cache.direct=on: Could not open 'vhost-user-blk.sock': No such device or address

Since virtio-blk-vhost-user does not support fd passing, let`s always fall back
on using `path` if we fail the fd passing.

Fixes: cad2ccc395 ("block/blkio: use qemu_open() to support fd passing for virtio-blk")
Reported-by: Qing Wang <qinwang@redhat.com>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20230727161020.84213-4-sgarzare@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/blkio: retry blkio_connect() if it fails using `fd`

libblkio 1.3.0 added support of "fd" property for virtio-blk-vhost-vdpa
driver. In QEMU, starting from commit cad2ccc395 ("block/blkio: use
qemu_open() to support fd passing for virtio-blk") we are using
`blkio_get_int(..., "fd")` to check if the "fd" property is supported
for all the virtio-blk-* driver.

Unfortunately that property is also available for those driver that do
not support it, such as virtio-blk-vhost-user.

So, `blkio_get_int()` is not enough to check whether the driver supports
the `fd` property or not. This is because the virito-blk common libblkio
driver only checks whether or not `fd` is set during `blkio_connect()`
and fails with -EINVAL for those transports that do not support it
(all except vhost-vdpa for now).

So let's handle the `blkio_connect()` failure, retrying it using `path`
directly.

Fixes: cad2ccc395 ("block/blkio: use qemu_open() to support fd passing for virtio-blk")
Suggested-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20230727161020.84213-3-sgarzare@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/blkio: move blkio_connect() in the drivers functions

This is in preparation for the next patch, where for virtio-blk
drivers we need to handle the failure of blkio_connect().

Let's also rename the *_open() functions to *_connect() to make
the code reflect the changes applied.

Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20230727161020.84213-2-sgarzare@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block: Fix pad_request's request restriction

bdrv_pad_request() relies on requests' lengths not to exceed SIZE_MAX,
which bdrv_check_qiov_request() does not guarantee.

bdrv_check_request32() however will guarantee this, and both of
bdrv_pad_request()'s callers (bdrv_co_preadv_part() and
bdrv_co_pwritev_part()) already run it before calling
bdrv_pad_request().  Therefore, bdrv_pad_request() can safely call
bdrv_check_request32() without expecting error, too.

In effect, this patch will not change guest-visible behavior.  It is a
clean-up to tighten a condition to match what is guaranteed by our
callers, and which exists purely to show clearly why the subsequent
assertion (`assert(*bytes <= SIZE_MAX)`) is always true.

Note there is a difference between the interfaces of
bdrv_check_qiov_request() and bdrv_check_request32(): The former takes
an errp, the latter does not, so we can no longer just pass
&error_abort.  Instead, we need to check the returned value.  While we
do expect success (because the callers have already run this function),
an assert(ret == 0) is not much simpler than just to return an error if
it occurs, so let us handle errors by returning them up the stack now.

Reported-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Hanna Czenczek <hreitz@redhat.com>
Message-id: 20230714085938.202730-1-hreitz@redhat.com
Fixes: 18743311b829cafc1737a5f20bc3248d5f91ee2a
       ("block: Collapse padded I/O vecs exceeding IOV_MAX")
Signed-off-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/file-posix: fix g_file_get_contents return path

The g_file_get_contents() function returns a g_boolean. If it fails, the
returned value will be 0 instead of -1. Solve the issue by skipping
assigning ret value.

This issue was found by Matthew Rosato using virtio-blk-{pci,ccw} backed
by an NVMe partition e.g. /dev/nvme0n1p1 on s390x.

Signed-off-by: Sam Li <faithilikerun@gmail.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: 20230727115844.8480-1-faithilikerun@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/blkio: do not use open flags in qemu_open()

qemu_open() in blkio_virtio_blk_common_open() is used to open the
character device (e.g. /dev/vhost-vdpa-0 or /dev/vfio/vfio) or in
the future eventually the unix socket.

In all these cases we cannot open the path in read-only mode,
when the `read-only` option of blockdev is on, because the exchange
of IOCTL commands for example will fail.

In order to open the device read-only, we have to use the `read-only`
property of the libblkio driver as we already do in blkio_file_open().

Fixes: cad2ccc395 ("block/blkio: use qemu_open() to support fd passing for virtio-blk")
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2225439
Reported-by: Qing Wang <qinwang@redhat.com>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: 20230726074807.14041-1-sgarzare@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'pull-qapi-2023-07-26-v2' of https://repo.or.cz/qemu/armbru into staging

QAPI patches patches for 2023-07-26

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEENUvIs9frKmtoZ05fOHC0AOuRhlMFAmTBFvUSHGFybWJydUBy
# ZWRoYXQuY29tAAoJEDhwtADrkYZTML4QAKhHciLnEudtZ6SFSqpOgt80IJnw8a+r
# z1AowVYtgPhlZ8TtQJFXpBtAZtKu8xb/QdFxomm4bdNQnWX6CXCoheF5ZJ9V3Rrz
# A3pA1wt5KTnRif6R9/Rs1dYXEr4cWagg1UNT3g2eOV3fvdDHvJMPOsqK/jWeXuC1
# T94yFMv1bZSLyiLgB7QQNYDZhIWQ06RGU6tZdWaZQReA8N8maXiZN5NnUISK32Rq
# L2X0FtgzyJQ+dLHtbXOw6kIwZdOLNauOM78skZoiZUyFVaH2aDUIg3mnfRw36hN6
# feXGtw68PkTQGexKmonPDljIacfMDApmNBelLwsvB9MTrwVV+hKZPy1ZEwPIFDJ9
# yid63pp2CtQ1TZ3dSjZ1cGbRR+g2NI5X4g1DlcFPAxydMkv9/m5NwQx8OYqVIzqg
# VXeS0++O2BM5+ORjlJxMx3RsyH2O1I8DCfwmifzYSo+3Xg/4nCV3f38czbavjCfJ
# 4T3ooZx0+PRtjlOlfZTkgxV14TMV+XzQr3bsN4wbPdnjnueSE1tyoVGy8MwQ5aXi
# 2oAsjrR8g7iqU6f+6PyRNn5F6D0ge+AYQ7bYS51i3Hyih/y2QUJECpL3XAgOxREb
# /68SEtr4m/GJvmQNdwwwu6e1JFo8LknwMfkfzQAOCK1npAJGsWPmJ6iY7KtWgS8F
# oDwqng/WOhvV
# =mNMX
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 26 Jul 2023 05:52:05 AM PDT
# gpg:                using RSA key 354BC8B3D7EB2A6B68674E5F3870B400EB918653
# gpg:                issuer "armbru@redhat.com"
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>" [undefined]
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* tag 'pull-qapi-2023-07-26-v2' of https://repo.or.cz/qemu/armbru:
  qapi: Reformat recent doc comments to conform to current conventions
  qapi/trace: Tidy up trace-event-get-state, -set-state documentation
  qapi/qdev: Tidy up device_add documentation
  qapi/block: Tidy up block-latency-histogram-set documentation
  qapi/block-core: Tidy up BlockLatencyHistogramInfo documentation

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

Merge tag 'migration-20230726-pull-request' of https://gitlab.com/juan.quintela/qemu into staging

Migration Pull request

Hi

This is the migration PULL request.  It is the same than yesterday with proper PULL headers.
It pass CI. It contains:
- Fabiano rosas trheadinfo cleanups
- Hyman Huang dirtylimit changes
- Part of my changes
- Peter Xu documentation
- Tejus updato to migration descriptions
- Wei want improvements for postocpy and multifd setup

Please apply.

Thanks, Juan.

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEGJn/jt6/WMzuA0uC9IfvGFhy1yMFAmTBCrgACgkQ9IfvGFhy
# 1yPCphAAvZr6HqULECPv/g6gYIiNjl2WQxSgaOnJPnxSV3aaDMl4+rn3GowXbj1a
# V7xQIxxyYR+4BOBPHc1Ey9z2huB6tr5YhzbHhdpOPOfTdGP4LzQogyBCM9elIGbg
# GVnBX4k1yT2bE3qoKkD7FZ8GhQdFTq9NFXg/prAJm5fUnoUVVGhz4YSlWVXcpC19
# XJIAC4QA5LtQYKe9TAlLqECNHeOiMDIFa1QHtrz+52OUWgh8WOvAPtj1CK0pm9Qa
# AsvN8HvKJ2PlCBct7c+E17O/xVihKVciEgu3KXjGHurUipUSD3XCHXOURlS1IrLK
# ShegHFmMQjmS0m9mUy1+2K7DQ+ZcfScqSQCEuEOtTdnzs2him4c6p9VEGyQXa5bc
# PChjihbYmxuz1GwrprtjUGyXgqhjnwGi1yRDl9L3mZc41vfO4m2sHnMZpdJZc+dt
# 5f5oi69cXVmtzSNJqT/4nCa7g5PuaPLg34NdwpbZv7Dt0Hq1yzlkNgUNb9R0XGET
# /BIpIuYYcNdmBUEVebMydndrzY8UDq0KC+e35OADSGkg6B6ZNwYaoungCb2gy6hM
# WCcv+3UATb/oF7HoPmh1+f1MzUZENAdmDtddXOCvWBZQReByKR7eFZLUHR+yBODH
# dVP9zOkPfrm8XVG4fSYhb/4BPK4XhBlibFsxxwOohTttTNHA5ew=
# =J74B
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 26 Jul 2023 04:59:52 AM PDT
# gpg:                using RSA key 1899FF8EDEBF58CCEE034B82F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>" [undefined]
# gpg:                 aka "Juan Quintela <quintela@trasno.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03  4B82 F487 EF18 5872 D723

* tag 'migration-20230726-pull-request' of https://gitlab.com/juan.quintela/qemu: (25 commits)
  migration/rdma: Split qemu_fopen_rdma() into input/output functions
  qemu-file: Make qemu_file_get_error_obj() static
  qemu-file: Simplify qemu_file_shutdown()
  qemu_file: Make qemu_file_is_writable() static
  migration: Change qemu_file_transferred to noflush
  qemu-file: Rename qemu_file_transferred_ fast -> noflush
  qtest/migration-tests.c: use "-incoming defer" for postcopy tests
  migration: enforce multifd and postcopy preempt to be set before incoming
  migration: Update error description whenever migration fails
  docs/migration: Update postcopy bits
  migration: skipped field is really obsolete.
  migration-test: machine_opts is really arch specific
  migration-test: Create arch_opts
  migration-test: Make machine_opts regular with other options
  migration-test: Be consistent for ppc
  migration: Extend query-migrate to provide dirty page limit info
  migration: Implement dirty-limit convergence algo
  migration: Put the detection logic before auto-converge checking
  migration: Refactor auto-converge capability logic
  migration: Introduce dirty-limit capability
  ...

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

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

Miscellaneous fixes

 * Switch canokey license from Apache to GPLv2+
 * Fix uninitialized variable in LUKS driver

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE2vOm/bJrYpEtDo4/vobrtBUQT98FAmS/91MACgkQvobrtBUQ
# T9+WjA/9Gx02s4aZvLJ1gSpzPguIEjwEulVOBCTaxQ1Fuu/5RawWXmFMhQ/iwAbi
# EnbeDpghG+Qk+4DCfQDMq0F8zkozvZOLW8NTZJW66dpV9PSwji39eIpVgvin2GXA
# bGZBz6ZwXoTozplfY8LTzLIGyZNzGNjSO4ND1zsyXq57LXbWXhAdHvsxwi1h1rOc
# FbNMeSPFlwPtCnpQgBDQmRmQ5UzwZiJOCp3zi9njMM/D6AfU/n275QzLvd/3ydBO
# JW4q1IHyDs13g+SCnI4a2rqI7+Uf+Z7h2DfkwhoaGoGuTpZ6llTgM4asjUOFri66
# RzVWz6UK+uCUogq2wgfYJ5jyNwerU8DtyjSW3kxhLcaTTRUGG/+nQu9PV+aPy1xD
# DZWo74KBtiRDFVS1XTLoDd+tNDqzNRdCmWqlc0CWgjUU68b61+GCDnkr+F0rJ59t
# rL1Q+bgKDVnYVxbTVJQs9V6zdeu6o7x94moK2UCAUbGlaCcpkl/sZXqF586dMQAj
# SvaGRYBxMvZvDVeIaINV/sW+vssoSdi7MKaUHAiHydnph/NFzC501bszh7RMyfAd
# 4/PLsm4ezmSFBZ0BS6+zjMBwWEQYiJbl6DDZZI631qSC4G5yOm9TCW2I7ZPNdpRu
# CveFHf8/dREd5o5iE6Vl5mWZF++dOcil64PnevqEv5/wjQcyHJs=
# =+YRm
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 25 Jul 2023 09:24:51 AM PDT
# 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]

* tag 'misc-next-pull-request' of https://gitlab.com/berrange/qemu:
  hw/usb/canokey: change license to GPLv2+
  crypto: Always initialize splitkeylen

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

qapi: Reformat recent doc comments to conform to current conventions

Since commit a937b6aa739 (qapi: Reformat doc comments to conform to
current conventions), a number of comments not conforming to the
current formatting conventions were added.  No problem, just sweep
the entire documentation once more.

To check the generated documentation does not change, I compared the
generated HTML before and after this commit with "wdiff -3".  Finds no
differences.  Comparing with diff is not useful, as the reflown
paragraphs are visible there.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20230720071610.1096458-7-armbru@redhat.com>

qapi/trace: Tidy up trace-event-get-state, -set-state documentation

trace-event-set-state's explanation of how events are selected is
under "Features".  Doesn't belong there.  Simply delete it, as it
feels redundant with documentation of member @name.

trace-event-get-state's explanation is under "Returns".  Tolerable,
but similarly redundant.  Delete it, too.

Cc: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20230720071610.1096458-5-armbru@redhat.com>

qapi/qdev: Tidy up device_add documentation

The notes section comes out like this:

    Notes

    Additional arguments depend on the type.

    1. For detailed information about this command, please refer to the
       ‘docs/qdev-device-use.txt’ file.

    2. It’s possible to list device properties by running QEMU with the
       “-device DEVICE,help” command-line argument, where DEVICE is the
       device’s name

The first item isn't numbered.  Fix that:

    1. Additional arguments depend on the type.

    2. For detailed information about this command, please refer to the
       ‘docs/qdev-device-use.txt’ file.

    3. It’s possible to list device properties by running QEMU with the
       “-device DEVICE,help” command-line argument, where DEVICE is the
       device’s name

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20230720071610.1096458-4-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

qapi/block: Tidy up block-latency-histogram-set documentation

Examples come out like

    Example

       set new histograms for all io types with intervals [0, 10), [10,
       50), [50, 100), [100, +inf):

The sentence "set new histograms ..." starts with a lower case letter.
Capitalize it.  Same for the other examples.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20230720071610.1096458-3-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

qapi/block-core: Tidy up BlockLatencyHistogramInfo documentation

Documentation for member @bin comes out like

    list of io request counts corresponding to histogram intervals.
    len("bins") = len("boundaries") + 1 For the example above, "bins"
    may be something like [3, 1, 5, 2], and corresponding histogram
    looks like:

Note how the equation and the sentence following it run together.
Replace the equation:

    list of io request counts corresponding to histogram intervals,
    one more element than "boundaries" has.  For the example above,
    "bins" may be something like [3, 1, 5, 2], and corresponding
    histogram looks like:

Cc: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20230720071610.1096458-2-armbru@redhat.com>
[Off by one fixed]

migration/rdma: Split qemu_fopen_rdma() into input/output functions

This is how everything else in QEMUFile is structured.
As a bonus they are three less lines of code.

Reviewed-by: Peter Xu <peterx@redhat.com>
Message-ID: <20230530183941.7223-17-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

qemu-file: Make qemu_file_get_error_obj() static

It was not used outside of qemu_file.c anyways.

Reviewed-by: Peter Xu <peterx@redhat.com>
Message-ID: <20230530183941.7223-21-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

qemu-file: Simplify qemu_file_shutdown()

Reviewed-by: Peter Xu <peterx@redhat.com>
Message-ID: <20230530183941.7223-20-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

qemu_file: Make qemu_file_is_writable() static

It is not used outside of qemu_file, and it shouldn't.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230530183941.7223-19-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Change qemu_file_transferred to noflush

We do a qemu_fclose() just after that, that also does a qemu_fflush(),
so remove one qemu_fflush().

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230530183941.7223-3-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

qemu-file: Rename qemu_file_transferred_ fast -> noflush

Fast don't say much.  Noflush indicates more clearly that it is like
qemu_file_transferred but without the flush.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230530183941.7223-2-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

qtest/migration-tests.c: use "-incoming defer" for postcopy tests

The Postcopy preempt capability is expected to be set before incoming
starts, so change the postcopy tests to start with deferred incoming and
call migrate-incoming after the cap has been set.

Why the existing tests (without this patch) didn't fail?
There could be two reasons:
1) "backlog" specifies the number of pending connections. As long as the
   server accepts the connections faster than the clients side connecting,
   connection will succeed. For the preempt test, it uses only 2 channels,
   so very likely to not have pending connections.
2) per my tests (on kernel 6.2), the number of pending connections allowed
   is actually "backlog + 1", which is 2 in this case.
That said, the implementation of socket_start_incoming_migration_internal
expects "migrate defer" to be used, and for safety, change the test to
work with the expected usage.

Signed-off-by: Wei Wang <wei.w.wang@intel.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230606101910.20456-3-wei.w.wang@intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: enforce multifd and postcopy preempt to be set before incoming

qemu_start_incoming_migration needs to check the number of multifd
channels or postcopy ram channels to configure the backlog parameter (i.e.
the maximum length to which the queue of pending connections for sockfd
may grow) of listen(). So enforce the usage of postcopy-preempt and
multifd as below:
- need to use "-incoming defer" on the destination; and
- set_capability and set_parameter need to be done before migrate_incoming

Otherwise, disable the use of the features and report error messages to
remind users to adjust the commands.

Signed-off-by: Wei Wang <wei.w.wang@intel.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230606101910.20456-2-wei.w.wang@intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Acked-by: Juan Quintela <quintela@redhat.com>

migration: Update error description whenever migration fails

There are places in migration.c where the migration is marked failed with
MIGRATION_STATUS_FAILED, but the failure reason is never updated. Hence
libvirt doesn't know why the migration failed when it queries for it.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Tejus GK <tejus.gk@nutanix.com>
Message-ID: <20230621130940.178659-2-tejus.gk@nutanix.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

docs/migration: Update postcopy bits

We have postcopy recovery but not reflected in the document, do an update
for that.

Add a very small section on postcopy preempt.

Touch up the pagemap section, dropping the unsent map because it's already
been dropped in the source code in commit 1e7cf8c323 ("migration/postcopy:
unsentmap is not necessary for postcopy").

Touch up the postcopy section to remove "network connection" failures as
downside, because now it's not fatal and can be recovered.  Suggested by
Laszlo.

Acked-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230706115611.371048-1-peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: skipped field is really obsolete.

Has return zero for more than 10 years.

Specifically we introduced the field in 1.5.0

commit f1c72795af573b24a7da5eb52375c9aba8a37972
Author: Peter Lieven <pl@kamp.de>
Date:   Tue Mar 26 10:58:37 2013 +0100

    migration: do not sent zero pages in bulk stage

    during bulk stage of ram migration if a page is a
    zero page do not send it at all.
    the memory at the destination reads as zero anyway.

    even if there is an madvise with QEMU_MADV_DONTNEED
    at the target upon receipt of a zero page I have observed
    that the target starts swapping if the memory is overcommitted.
    it seems that the pages are dropped asynchronously.

    this patch also updates QMP to return the number of
    skipped pages in MigrationStats.

but removed its usage in 1.5.3

commit 9ef051e5536b6368a1076046ec6c4ec4ac12b5c6
Author: Peter Lieven <pl@kamp.de>
Date:   Mon Jun 10 12:14:19 2013 +0200

    Revert "migration: do not sent zero pages in bulk stage"

    Not sending zero pages breaks migration if a page is zero
    at the source but not at the destination. This can e.g. happen
    if different BIOS versions are used at source and destination.
    It has also been reported that migration on pseries is completely
    broken with this patch.

    This effectively reverts commit f1c72795af573b24a7da5eb52375c9aba8a37972.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20230612193344.3796-2-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration-test: machine_opts is really arch specific

And it needs to be in both source and target, so put it on arch_opts.

Reviewed-by: Peter Xu <peterx@redhat.com>
Message-ID: <20230608224943.3877-7-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration-test: Create arch_opts

This will contain the options needed for both source and target.

Reviewed-by: Peter Xu <peterx@redhat.com>
Message-ID: <20230608224943.3877-6-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration-test: Make machine_opts regular with other options

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20230608224943.3877-5-quintela@redhat.com>

migration-test: Be consistent for ppc

It makes no sense that we don't have the same configuration on both sides.

Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Message-ID: <20230608224943.3877-2-quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Extend query-migrate to provide dirty page limit info

Extend query-migrate to provide throttle time and estimated
ring full time with dirty-limit capability enabled, through which
we can observe if dirty limit take effect during live migration.

Signed-off-by: Hyman Huang(黄勇) <yong.huang@smartx.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-ID: <168733225273.5845.15871826788879741674-8@git.sr.ht>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Implement dirty-limit convergence algo

Implement dirty-limit convergence algo for live migration,
which is kind of like auto-converge algo but using dirty-limit
instead of cpu throttle to make migration convergent.

Enable dirty page limit if dirty_rate_high_cnt greater than 2
when dirty-limit capability enabled, Disable dirty-limit if
migration be canceled.

Note that "set_vcpu_dirty_limit", "cancel_vcpu_dirty_limit"
commands are not allowed during dirty-limit live migration.

Signed-off-by: Hyman Huang(黄勇) <yong.huang@smartx.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <168733225273.5845.15871826788879741674-7@git.sr.ht>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Put the detection logic before auto-converge checking

This commit is prepared for the implementation of dirty-limit
convergence algo.

The detection logic of throttling condition can apply to both
auto-converge and dirty-limit algo, putting it's position
before the checking logic for auto-converge feature.

Signed-off-by: Hyman Huang(黄勇) <yong.huang@smartx.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-ID: <168733225273.5845.15871826788879741674-6@git.sr.ht>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Refactor auto-converge capability logic

Check if block migration is running before throttling
guest down in auto-converge way.

Note that this modification is kind of like code clean,
because block migration does not depend on auto-converge
capability, so the order of checks can be adjusted.

Signed-off-by: Hyman Huang(黄勇) <yong.huang@smartx.com>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <168618975839.6361.17407633874747688653-5@git.sr.ht>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Introduce dirty-limit capability

Introduce migration dirty-limit capability, which can
be turned on before live migration and limit dirty
page rate durty live migration.

Introduce migrate_dirty_limit function to help check
if dirty-limit capability enabled during live migration.

Meanwhile, refactor vcpu_dirty_rate_stat_collect
so that period can be configured instead of hardcoded.

dirty-limit capability is kind of like auto-converge
but using dirty limit instead of traditional cpu-throttle
to throttle guest down. To enable this feature, turn on
the dirty-limit capability before live migration using
migrate-set-capabilities, and set the parameters
"x-vcpu-dirty-limit-period", "vcpu-dirty-limit" suitably
to speed up convergence.

Signed-off-by: Hyman Huang(黄勇) <yong.huang@smartx.com>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <168618975839.6361.17407633874747688653-4@git.sr.ht>
Signed-off-by: Juan Quintela <quintela@redhat.com>

qapi/migration: Introduce vcpu-dirty-limit parameters

Introduce "vcpu-dirty-limit" migration parameter used
to limit dirty page rate during live migration.

"vcpu-dirty-limit" and "x-vcpu-dirty-limit-period" are
two dirty-limit-related migration parameters, which can
be set before and during live migration by qmp
migrate-set-parameters.

This two parameters are used to help implement the dirty
page rate limit algo of migration.

Signed-off-by: Hyman Huang(黄勇) <yong.huang@smartx.com>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <168618975839.6361.17407633874747688653-3@git.sr.ht>
Signed-off-by: Juan Quintela <quintela@redhat.com>

qapi/migration: Introduce x-vcpu-dirty-limit-period parameter

Introduce "x-vcpu-dirty-limit-period" migration experimental
parameter, which is in the range of 1 to 1000ms and used to
make dirtyrate calculation period configurable.

Currently with the "x-vcpu-dirty-limit-period" varies, the
total time of live migration changes, test results show the
optimal value of "x-vcpu-dirty-limit-period" ranges from
500ms to 1000 ms. "x-vcpu-dirty-limit-period" should be made
stable once it proves best value can not be determined with
developer's experiments.

Signed-off-by: Hyman Huang(黄勇) <yong.huang@smartx.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <168618975839.6361.17407633874747688653-2@git.sr.ht>
Signed-off-by: Juan Quintela <quintela@redhat.com>

softmmu/dirtylimit: Add parameter check for hmp "set_vcpu_dirty_limit"

dirty_rate paraemter of hmp command "set_vcpu_dirty_limit" is invalid
if less than 0, so add parameter check for it.

Note that this patch also delete the unsolicited help message and
clean up the code.

Signed-off-by: Hyman Huang(黄勇) <yong.huang@smartx.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <168618975839.6361.17407633874747688653-1@git.sr.ht>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration/multifd: Protect accesses to migration_threads

This doubly linked list is common for all the multifd and migration
threads so we need to avoid concurrent access.

Add a mutex to protect the data from concurrent access. This fixes a
crash when removing two MigrationThread objects from the list at the
same time during cleanup of multifd threads.

Fixes: 671326201d ("migration: Introduce interface query-migrationthreads")
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20230607161306.31425-3-farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration/multifd: Rename threadinfo.c functions

We're about to add more functions to this file so make it use the same
coding style as the rest of the code.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <20230607161306.31425-2-farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>

block/blkio: enable the completion eventfd

Until libblkio 1.3.0, virtio-blk drivers had completion eventfd
notifications enabled from the start, but from the next releases
this is no longer the case, so we have to explicitly enable them.

In fact, the libblkio documentation says they could be disabled,
so we should always enable them at the start if we want to be
sure to get completion eventfd notifications:

    By default, the driver might not generate completion events for
    requests so it is necessary to explicitly enable the completion
    file descriptor before use:

    void blkioq_set_completion_fd_enabled(struct blkioq *q, bool enable);

I discovered this while trying a development version of libblkio:
the guest kernel hangs during boot, while probing the device.

Fixes: fd66dbd424f5 ("blkio: add libblkio block driver")
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20230725103744.77343-1-sgarzare@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Update version for v8.1.0-rc1 release

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

hw/usb/canokey: change license to GPLv2+

Apache license is considered by some to be not compatible
with GPLv2+. Since QEMU as combined work is GPLv2-only,
these two files should be made compatible.

Reported-by: "Daniel P. Berrangé" <berrange@redhat.com>
Link: https://lore.kernel.org/qemu-devel/ZEpKXncC%2Fe6FKRe9@redhat.com/
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-By: canokeys.org (http://canokeys.org) <contact@canokeys.org>
Acked-by: YuanYang Meng <mkfssion@mkfssion.com>
Signed-off-by: Hongren (Zenithal) Zheng <i@zenithal.me>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

crypto: Always initialize splitkeylen

When _FORTIFY_SOURCE=2, glibc version is 2.35, and GCC version is
12.1.0, the compiler complains as follows:

In file included from /usr/include/string.h:535,
                 from /home/alarm/q/var/qemu/include/qemu/osdep.h:99,
                 from ../crypto/block-luks.c:21:
In function 'memset',
    inlined from 'qcrypto_block_luks_store_key' at ../crypto/block-luks.c:843:9:
/usr/include/bits/string_fortified.h:59:10: error: 'splitkeylen' may be used uninitialized [-Werror=maybe-uninitialized]
   59 |   return __builtin___memset_chk (__dest, __ch, __len,
      |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   60 |                                  __glibc_objsize0 (__dest));
      |                                  ~~~~~~~~~~~~~~~~~~~~~~~~~~
../crypto/block-luks.c: In function 'qcrypto_block_luks_store_key':
../crypto/block-luks.c:699:12: note: 'splitkeylen' was declared here
  699 |     size_t splitkeylen;
      |            ^~~~~~~~~~~

It seems the compiler cannot see that splitkeylen will not be used
when splitkey is NULL. Suppress the warning by initializing splitkeylen
even when splitkey stays NULL.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

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

trivial-patches 25-07-2023

# -----BEGIN PGP SIGNATURE-----
#
# iQFDBAABCAAtFiEEe3O61ovnosKJMUsicBtPaxppPlkFAmS/2vgPHG1qdEB0bHMu
# bXNrLnJ1AAoJEHAbT2saaT5ZT6MH/j5L3P9yLV6TqW+DkhCppbmBygqxz2SbQjwl
# dVVfSLpJNbtpvLfEnvpb+ms+ZdaOCGj8IofAVf9w0VaIYJFP1srFphY/1x+RYVnw
# kDjCLzuLNSCAdCV2HPqsrMKzdFctZ/MfK+QzfcGik9IvmCNPYWOhpmevs+xAIEJd
# b0xk152zy2fIIC3vKK+3KcM7MFkqZWJ6z0pzUZAyEBS+aQyuZNPJ/cO8xMXotbP2
# jqv12SNGV2GLH1acvsd8GQwDB9MamstB4r8NWpSpT/AyPwOgmMR+j5B8a/WEBJCs
# OcEW/pEyrumSygqf9z01YoNJQUCSvSpg5aq4+S2cRDslmUgFDmw=
# =wCoQ
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 25 Jul 2023 15:23:52 BST
# 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:
  qapi: Correct "eg." to "e.g." in documentation
  hw/pci: add comment to explain checking for available function 0 in pci hotplug
  target/tricore: Rename tricore_feature
  hw/9pfs: spelling fixes
  other architectures: spelling fixes
  arm: spelling fixes
  s390x: spelling fixes
  migration: spelling fixes

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

Merge tag 'misc-fixes-20230725' of https://github.com/philmd/qemu into staging

Misc patches queue

hw/sd/sdhci: Default I/O ops to little endian
hw/mips/loongson3-virt: Only use default USB if available
hw/char/escc: Implement loopback mode to allow self-testing
target/mips: Avoid overruns and shifts by negative number
target/sparc: Handle FPRS correctly on big-endian hosts
target/tricore: Rename tricore_feature to avoid clash with libcapstone

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE+qvnXhKRciHc/Wuy4+MsLN6twN4FAmS/4ksACgkQ4+MsLN6t
# wN6OSg//cZY9C6fRXNNaIqkmnhjbaV6KLtjE7mOKp0RUyh3aN0dtTwWIjdJc0O5C
# iipHESYhcbHTiN/TxK0zXg4KgtKmtwqGsa3QTXGdTlSkTY/dMNioSpb7p82becu0
# fhCvGRLJ97j7/mhebiBNT/urrcG5h3n7CjA5IoFMMA4f+cajsGZHwmq5TTzc2ehy
# 4FuchjFUw+cgqU1peNYoqt2dfnxFg0EgKBSRikl8MyPf9lFzTlXOKbgd+qppG6hI
# 2fAUHyMqBkU22sAoK0eB0077LjgjPPQfmn8UPGkpGD5QZQcvBRNArg4fyHxCKTS7
# zOsO1Qc+4D2l2RJlIHgct2pmcHdT29TlTn2T4Lg900Hm09KelZh1XF+1BemCC13z
# cGWjPcYozvGFFiHlhazINtbGpB6XaP/Z3OwroRHRn+Mn3ss+FaU+j/p+4YlEVyFi
# 4yoEyjhNma6/hssmstifSQsaOf6XthzpS+XdKNB6G1b2WuRSc1Z59b2gcPBTwbXY
# B52lfI61nzSrP9pLuS8c/6hQXQvADIEndeWEcWZ50h3WW2Cemj9jTDVgfjWC4Vg9
# wV2U6NeTr+g54cSU5vcKiZrqsQHUoLiKbZFRJkXF7EEMbOErIQnyIS5l8xf71Pay
# YPxuPf1VprRiR07d+ZaA+wmEaBxLCUPEl1CEuu5NPVA9S4yIIWE=
# =F+Wb
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 25 Jul 2023 15:55:07 BST
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.org>" [full]
# Primary key fingerprint: FAAB E75E 1291 7221 DCFD  6BB2 E3E3 2C2C DEAD C0DE

* tag 'misc-fixes-20230725' of https://github.com/philmd/qemu:
  target/tricore: Rename tricore_feature
  target/sparc: Handle FPRS correctly on big-endian hosts
  target/mips: Avoid shift by negative number in page_table_walk_refill()
  target/mips: Pass directory/leaf shift values to walk_directory()
  target/mips/mxu: Avoid overrun in gen_mxu_q8adde()
  target/mips/mxu: Avoid overrun in gen_mxu_S32SLT()
  target/mips/mxu: Replace magic array size by its definition
  hw/char/escc: Implement loopback mode
  hw/mips: Improve the default USB settings in the loongson3-virt machine
  hw/sd/sdhci: Do not force sdhci_mmio_*_ops onto all SD controllers

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

qapi: Correct "eg." to "e.g." in documentation

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/pci: add comment to explain checking for available function 0 in pci hotplug

This change is cosmetic. A comment is added explaining why we need to check for
the availability of function 0 when we hotplug a device.

CC: mst@redhat.com
CC: mjt@tls.msk.ru
Signed-off-by: Ani Sinha <anisinha@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

target/tricore: Rename tricore_feature

this name is used by capstone and will lead to a build failure of QEMU,
when capstone is enabled. So we rename it to tricore_has_feature(), to
match has_feature() in translate.c.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1774
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

hw/9pfs: spelling fixes

Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

other architectures: spelling fixes

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

arm: spelling fixes

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

s390x: spelling fixes

Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Thomas Huth <thuth@redhat.com>

migration: spelling fixes

Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Fabiano Rosas <farosas@suse.de>

target/tricore: Rename tricore_feature

this name is used by capstone and will lead to a build failure of QEMU,
when capstone is enabled. So we rename it to tricore_has_feature(), to
match has_feature() in translate.c.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1774
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230721060605.76636-1-kbastian@mail.uni-paderborn.de>

target/sparc: Handle FPRS correctly on big-endian hosts

In CPUSparcState we define the fprs field as uint64_t.  However we
then refer to it in translate.c via a TCGv_i32 which we set up with
tcg_global_mem_new_ptr().  This means that on a big-endian host when
the guest does something to writo te the FPRS register this value
ends up in the wrong half of the uint64_t, and the QEMU C code that
refers to env->fprs sees the wrong value.  The effect of this is that
guest code that enables the FPU crashes with spurious FPU Disabled
exceptions.  In particular, this is why
 tests/avocado/machine_sparc64_sun4u.py:Sun4uMachine.test_sparc64_sun4u
times out on an s390 host.

There are multiple ways we could fix this; since there are actually
only three bits in the FPRS register and the code in translate.c
would be a bit painful to convert to dealing with a TCGv_i64, change
the type of the CPU state struct field to match what translate.c is
expecting.

(None of the other fields referenced by the r32[] array in
sparc_tcg_init() have the wrong type.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Message-Id: <20230717103544.637453-1-peter.maydell@linaro.org>

target/mips: Avoid shift by negative number in page_table_walk_refill()

Coverity points out that in page_table_walk_refill() we can
shift by a negative number, which is undefined behaviour
(CID 1452918, 1452920, 1452922).  We already catch the
negative directory_shift and leaf_shift as being a "bail
out early" case, but not until we've already used them to
calculated some offset values.

The shifts can be negative only if ptew > 1, so make the
bail-out-early check look directly at that, and only
calculate the shift amounts and the offsets based on them
after we have done that check. This allows
us to simplify the expressions used to calculate the
shift amounts, use an unsigned type, and avoids the
undefined behaviour.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
[PMD: Check for ptew > 1, use unsigned type]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20230717213504.24777-3-philmd@linaro.org>

target/mips: Pass directory/leaf shift values to walk_directory()

We already evaluated directory_shift and leaf_shift in
page_table_walk_refill(), no need to do that again: pass
as argument.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20230717213504.24777-2-philmd@linaro.org>

target/mips/mxu: Avoid overrun in gen_mxu_q8adde()

Coverity reports a potential overruns (CID 1517770):

  Overrunning array "mxu_gpr" of 15 8-byte elements at
  element index 4294967295 (byte offset 34359738367)
  using index "XRb - 1U" (which evaluates to 4294967295).

Add a gen_extract_mxu_gpr() helper similar to
gen_load_mxu_gpr() to safely extract MXU registers.

Fixes: eb79951ab6 ("target/mips/mxu: Add Q8ADDE ... insns")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230712060806.82323-4-philmd@linaro.org>

target/mips/mxu: Avoid overrun in gen_mxu_S32SLT()

Coverity reports a potential overrun (CID 1517769):

  Overrunning array "mxu_gpr" of 15 8-byte elements at
  element index 4294967295 (byte offset 34359738367)
  using index "XRb - 1U" (which evaluates to 4294967295).

Use gen_load_mxu_gpr() to safely load MXU registers.

Fixes: ff7936f009 ("target/mips/mxu: Add S32SLT ... insns")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230712060806.82323-3-philmd@linaro.org>

target/mips/mxu: Replace magic array size by its definition

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230712060806.82323-2-philmd@linaro.org>

hw/char/escc: Implement loopback mode

The firmware of the m68k next-cube machine uses the loopback mode
for self-testing the hardware and currently fails during this step.
By implementing the loopback mode, we can make the firmware pass
to the next step.

Signed-off-by: Thomas Huth <huth@tuxfamily.org>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230716153519.31722-1-huth@tuxfamily.org>

hw/mips: Improve the default USB settings in the loongson3-virt machine

It's possible to compile QEMU without the USB devices (e.g. when using
"--without-default-devices" as option for the "configure" script).
To be still able to run the loongson3-virt machine in default mode with
such a QEMU binary, we have to check here for the availability of the
OHCI controller first before instantiating the USB devices.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230714104903.284845-1-thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/sd/sdhci: Do not force sdhci_mmio_*_ops onto all SD controllers

Since commit c0a55a0c9da2 "hw/sd/sdhci: Support big endian SD host controller
interfaces" sdhci_common_realize() forces all SD card controllers to use either
sdhci_mmio_le_ops or sdhci_mmio_be_ops, depending on the "endianness" property.
However, there are device models which use different MMIO ops: TYPE_IMX_USDHC
uses usdhc_mmio_ops and TYPE_S3C_SDHCI uses sdhci_s3c_mmio_ops.

Forcing sdhci_mmio_le_ops breaks SD card handling on the "sabrelite" board, for
example. Fix this by defaulting the io_ops to little endian and switch to big
endian in sdhci_common_realize() only if there is a matchig big endian variant
available.

Fixes: c0a55a0c9da2 ("hw/sd/sdhci: Support big endian SD host controller
interfaces")

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Message-Id: <20230709080950.92489-1-shentey@gmail.com>

Merge tag 'pull-target-arm-20230725' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * tests/decode: Suppress "error: " string for expected-failure tests
 * ui/curses: For curses display, recognize a few more control keys
 * target/arm: Special case M-profile in debug_helper.c code
 * scripts/git-submodule.sh: Don't rely on non-POSIX 'read' behaviour
 * hw/arm/smmu: Handle big-endian hosts correctly

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmS/ot8ZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3slqEACaLJwIYl1bJBfCda2u53+C
# q20t50SQjkvV2CSW6A9uOHPPahKUcxAXh6K+d54BhzD6Dsrv5g1rpo/2fnNhHDSG
# 7fHlla+fPnywmAOahE2FPUw4pckRX1tpPIM1RDjM9szLYqkJlShKYP28QsLu1Eku
# bnTty6OcId5hAZILag53QLwL9EYsVYoCEe6xRcgY3He0UZcCEisCUdfeCXEN1Uc8
# 57wd+q3KNUTgOScqmDJRAH2NaET0UOYlUvQGVu8/Bh3t0huQCtfyT4gc8z7v/TZ8
# 2PfI6bFb9nei09avxhBMN9Nu7BVD6eHBkAAe4JHDBhkJKCZn+LASDCMUAFPrFD2V
# NeIObNHBMaE9FqIG/SZxf7kEOaFcUwt4GrVfQNguaqiXIwALsfT/jiX4r+jXX4WS
# ii0mdoS2ZuAcRtUhTA7S6x44B3wa47sidSogoe3t2k8ObYB/AZ34F1cSZDgEmIG7
# nobJE2OgzSRMWUHXhCUEzGvn8MMPeI0HQmKr4sOD6CnlqHIzLZDH4Jx0DL4yvoyc
# XLs0D2G4yscUTtWh15R/nTWTJKxjumbs05bqwRKLTMsVj6kpDDY/EqhHMvB6Xm70
# z+xDGki9xsBOTGRO7GdqGlWEKfnwUIPjipwy9crhsjSe121XrP8uwmmDBL1tOLgc
# L+geqtruzJgFmo3rOBGxXA==
# =4paq
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 25 Jul 2023 11:24:31 BST
# 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-20230725' of https://git.linaro.org/people/pmaydell/qemu-arm:
  tests/decode: Suppress "error: " string for expected-failure tests
  For curses display, recognize a few more control keys
  target/arm: Special case M-profile in debug_helper.c code
  scripts/git-submodule.sh: Don't rely on non-POSIX 'read' behaviour
  hw/arm/smmu: Handle big-endian hosts correctly

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

tests/decode: Suppress "error: " string for expected-failure tests

The "expected failure" tests for decodetree result in the
error messages from decodetree ending up in logs and in
V=1 output:

>>> MALLOC_PERTURB_=226 /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/x86/pyvenv/bin/python3 /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/scripts/decodetree.py --output-null --test-for-error /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/x86/../../tests/decode/err_argset1.decode
――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――― ✀  ――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/x86/../../tests/decode/err_argset1.decode:5: error: duplicate argument "a"
―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
 1/44 qemu:decodetree / err_argset1                OK              0.05s

This then produces false positives when scanning the
logfiles for strings like "error: ".

For the expected-failure tests, make decodetree print
"detected:" instead of "error:".

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230720131521.1325905-1-peter.maydell@linaro.org

For curses display, recognize a few more control keys

The curses display handles most control-X keys, and translates
them into their corresponding keycode.  Here we recognize
a few that are missing, Ctrl-@ (null), Ctrl-\ (backslash),
Ctrl-] (right bracket), Ctrl-^ (caret), Ctrl-_ (underscore).

Signed-off-by: Sean Estabrooks <sean.estabrooks@gmail.com>
Message-id: CAHyVn3Bh9CRgDuOmf7G7Ngwamu8d4cVozAcB2i4ymnnggBXNmg@mail.gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Special case M-profile in debug_helper.c code

A lot of the code called from helper_exception_bkpt_insn() is written
assuming A-profile, but we will also call this helper on M-profile
CPUs when they execute a BKPT insn.  This used to work by accident,
but recent changes mean that we will hit an assert when some of this
code calls down into lower level functions that end up calling
arm_security_space_below_el3(), arm_el_is_aa64(), and other functions
that now explicitly assert that the guest CPU is not M-profile.

Handle M-profile directly to avoid the assertions:
 * in arm_debug_target_el(), M-profile debug exceptions always
   go to EL1
 * in arm_debug_exception_fsr(), M-profile always uses the short
   format FSR (compare commit d7fe699be54b2, though in this case
   the code in arm_v7m_cpu_do_interrupt() does not need to
   look at the FSR value at all)

Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1775
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230721143239.1753066-1-peter.maydell@linaro.org

scripts/git-submodule.sh: Don't rely on non-POSIX 'read' behaviour

The POSIX definition of the 'read' utility requires that you
specify the variable name to set; omitting the name and
having it default to 'REPLY' is a bashism. If your system
sh is dash, then it will print an error message during build:

qemu/pc-bios/s390-ccw/../../scripts/git-submodule.sh: 106: read: arg count

Specify the variable name explicitly.

Fixes: fdb8fd8cb915647b ("git-submodule: allow partial update of .git-submodule-status")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230720153038.1587196-1-peter.maydell@linaro.org

hw/arm/smmu: Handle big-endian hosts correctly

The implementation of the SMMUv3 has multiple places where it reads a
data structure from the guest and directly operates on it without
doing a guest-to-host endianness conversion.  Since all SMMU data
structures are little-endian, this means that the SMMU doesn't work
on a big-endian host.  In particular, this causes the Avocado test
  machine_aarch64_virt.py:Aarch64VirtMachine.test_alpine_virt_tcg_gic_max
to fail on an s390x host.

Add appropriate byte-swapping on reads and writes of guest in-memory
data structures so that the device works correctly on big-endian
hosts.

As part of this we constrain queue_read() to operate only on Cmd
structs and queue_write() on Evt structs, because in practice these
are the only data structures the two functions are used with, and we
need to know what the data structure is to be able to byte-swap its
parts correctly.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Tested-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Message-id: 20230717132641.764660-1-peter.maydell@linaro.org
Cc: qemu-stable@nongnu.org

Merge tag 'pull-request-2023-07-24' of https://gitlab.com/thuth/qemu into staging

* Fix emulation of s390x instructions: CKSM, CLM, ICM, MC, CLGEBR(A)
* Remove useless and non-working s390x migration avocado tests
* Fix loongarch CSRRD CPUID instruction when running on s390x hosts
* Disable flaky s390x virtio-gpu test by default

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmS+q4IRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbXB9xAAhF6bV23f1C9SHTWpxoN8oDYqJ3ZRR/D2
# z/4RHzXC5icw4jWPXG/zf19a+TT5UA03HFqvXbFlEtLVLF1/RY2wDnQbHJd2msRX
# YwvqE9SVoxbS3Jy+FS6XQ69WmPNoz01x+BIG5ig/jzXdOPZmylfik+1eAmHNHvcF
# GNCvdFwDzAZnB7N7y973BRnYn73CXBk3ap3UnnX/PaFD9k7DowTW5HB4ej/Uydwl
# z71/NbZhjgKYNpWgPYCI0Mypmz8b870o0VJxEJO9/1pWYLOoUcy/BKdOeGr3vqqO
# cduRL2c78Xd2RaUt4TgYmsZKkLazcG37XnQ5iaewDctTjfjuL5Oss16SO0WTut3p
# 8DtJHZCwD+0GcnbdFxyy7qdFmW2jgqsJtIPtnZs29ruEXnl0iyovjw38/J4kgBrF
# eHY+X2AyvlXI6zNkRMjKp+ltAkVx1vXbedM/bfaEcSBiKAI/OlV0Y1slX7+K14pR
# ITGOmBEG+yytMDnanzZKdN1DlBJMbGFYg9MctL7/yEwtD9vTl7gc66JhqcndkmCg
# VXSvqN8ECXPRbeNyZQZP+weXQydpkk3ndVIlSUrWpWsJtj0GDQOt+iK/KWcQ+AkS
# TXXp2OKM/mC+z6pdODsn/EoLl0toeDjpZFu27t/AvglNkmDFneY72LrtSwKgYHsi
# OJpjY1zCoxM=
# =Q+jQ
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 24 Jul 2023 17:49:06 BST
# 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-2023-07-24' of https://gitlab.com/thuth/qemu:
  tests/avocado/machine_s390_ccw_virtio: Skip the flaky virtio-gpu test by default
  target/loongarch: Fix the CSRRD CPUID instruction on big endian hosts
  tests/avocado/migration: Remove the malfunctioning s390x tests
  tests/tcg/s390x: Test VCKSM
  tests/tcg/s390x: Test STPQ
  tests/tcg/s390x: Test MC
  tests/tcg/s390x: Test ICM
  tests/tcg/s390x: Test CLM
  tests/tcg/s390x: Test CLGEBR and CGEBRA
  tests/tcg/s390x: Test CKSM
  target/s390x: Fix assertion failure in VFMIN/VFMAX with type 13
  target/s390x: Make MC raise specification exception when class >= 16
  target/s390x: Fix ICM with M3=0
  target/s390x: Fix CONVERT TO LOGICAL/FIXED with out-of-range inputs
  target/s390x: Fix CLM with M3=0
  target/s390x: Make CKSM raise an exception if R2 is odd

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

Merge tag 'pull-qapi-2023-07-10' of https://repo.or.cz/qemu/armbru into staging

QAPI patches patches for 2023-07-10

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEENUvIs9frKmtoZ05fOHC0AOuRhlMFAmSr6HsSHGFybWJydUBy
# ZWRoYXQuY29tAAoJEDhwtADrkYZTgnUP/1XvFPJ8NUWBjqe4DgYqkjx7rf5Zym+y
# rluYzLNARWrOZuexvcn6tEiv74MilhSsZHuLvPQaQRF7voLPgD7fbRUBuYWPgodU
# 36+i3Hk76hAdhx0UMY62wHEviv0sWBr/ZiZjPcwrSS5tSEY23iUMY4ZVf/mIfPHH
# XYtF0co95SWKvqp9FSnejoYiNBCWKqZpHyDnJoXcd8RynqDt+cmNaZcU+Id+/WTv
# fLiLGQgHNyKBIYWlljxiDXGMlybnbV88N0dkLZtJ0Z1aJhh6j5grxTp0BRd85nsw
# QQjGO1qot6adQy04xi1RiMp4VZDJH18/9gBhDRLddVul0q49J1CT9LmKv/lYbpPj
# 6duZwrO5ciEUQ2usc8/L8ZtM7xIbAXRGqyg69IpmfwVE906LFrHt6N23WJl14a7F
# UBwD2+uGQNFyjxhFtPZZIYzYIH/49eGA/i6nhSIsd+LCD2r4n3M7FukgF8phuI9t
# xEX++sW4ix8cStqtsRAtFJ7OCFFKK2al1zpPzgHyZQ4mwMZimRKh6blcD+AnOZms
# uhiqONr2VlS9kefLAn5oCyTRUzxjJplnsqK44o8bKTfXxGcWBX2mt2nYMZECSLrQ
# B1HWzr8y4uc8ivYzIErhWMWtIwISa9KQSsuurZXz83vEWnrtVq6hh9B8z6j24hk9
# RJRSRZjHHjt7
# =3XVF
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 10 Jul 2023 12:16:11 BST
# gpg:                using RSA key 354BC8B3D7EB2A6B68674E5F3870B400EB918653
# gpg:                issuer "armbru@redhat.com"
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>" [full]
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>" [full]
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* tag 'pull-qapi-2023-07-10' of https://repo.or.cz/qemu/armbru:
  migration.json: Don't use space before colon
  qapi: better docs for calc-dirty-rate and friends

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

tests/avocado/machine_s390_ccw_virtio: Skip the flaky virtio-gpu test by default

The virtio-gpu test is known to be flaky - that's why we also did
not enable the test_s390x_fedora in the gitlab CI. However, a flaky
test can also be annoying when testing locally, so let's rather skip
this subtest by default and start running the test_s390x_fedora test
in the gitlab CI again (since the other things that are tested here
are quite valuable).

Message-Id: <20230724084851.24251-1-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/loongarch: Fix the CSRRD CPUID instruction on big endian hosts

The test in tests/avocado/machine_loongarch.py is currently failing
on big endian hosts like s390x. By comparing the traces between running
the QEMU_EFI.fd bios on a s390x and on a x86 host, it's quickly obvious
that the CSRRD instruction for the CPUID is behaving differently. And
indeed: The code currently does a long read (i.e. 64 bit) from the
address that points to the CPUState->cpu_index field (with tcg_gen_ld_tl()
in the trans_csrrd() function). But this cpu_index field is only an "int"
(i.e. 32 bit). While this dirty pointer magic works on little endian hosts,
it of course fails on big endian hosts. Fix it by using a proper helper
function instead.

Message-Id: <20230720175307.854460-1-thuth@redhat.com>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/avocado/migration: Remove the malfunctioning s390x tests

The tests from tests/avocado/migration.py do not work at all
on s390x - the bios shuts down immediately when it cannot find
a boot disk, so there is nothing left to migrate here. For doing
a proper migration test, we would need a proper payload, but we
already do such tests in the migration *qtest*, so it is unnecessary
to redo such a test here, thus let's simply remove this test.

Message-Id: <20230721164346.10112-1-thuth@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Test VCKSM

Add a small test to prevent regressions.

Tested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-15-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Test STPQ

Add a small test to prevent regressions.

Tested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-14-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Test MC

Add a small test to prevent regressions.

Tested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-13-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Test ICM

Add a small test to prevent regressions.

Tested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-12-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Test CLM

Add a small test to prevent regressions.

Tested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-11-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Test CLGEBR and CGEBRA

Add a small test to prevent regressions.

Tested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-10-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Test CKSM

Add a small test to prevent regressions.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-9-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Fix assertion failure in VFMIN/VFMAX with type 13

Type 13 is reserved, so using it should result in specification
exception. Due to an off-by-1 error the code triggers an assertion at a
later point in time instead.

Cc: qemu-stable@nongnu.org
Fixes: da4807527f3b ("s390x/tcg: Implement VECTOR FP (MAXIMUM|MINIMUM)")
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-8-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Make MC raise specification exception when class >= 16

MC requires bit positions 8-11 (upper 4 bits of class) to be zeros,
otherwise it must raise a specification exception.

Cc: qemu-stable@nongnu.org
Fixes: 20d143e2cab8 ("s390x/tcg: Implement MONITOR CALL")
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-6-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Fix ICM with M3=0

When the mask is zero, access exceptions should still be recognized for
1 byte at the second-operand address. CC should be set to 0.

Cc: qemu-stable@nongnu.org
Fixes: e023e832d0ac ("s390x: translate engine for s390x CPU")
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-5-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Fix CONVERT TO LOGICAL/FIXED with out-of-range inputs

CONVERT TO LOGICAL/FIXED deviate from IEEE 754 in that they raise an
inexact exception on out-of-range inputs. float_flag_invalid_cvti
aligns nicely with that behavior, so convert it to
S390_IEEE_MASK_INEXACT.

Cc: qemu-stable@nongnu.org
Fixes: defb0e3157af ("s390x: Implement opcode helpers")
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-4-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Fix CLM with M3=0

When the mask is zero, access exceptions should still be recognized for
1 byte at the second-operand address. CC should be set to 0.

Cc: qemu-stable@nongnu.org
Fixes: defb0e3157af ("s390x: Implement opcode helpers")
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-3-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Make CKSM raise an exception if R2 is odd

R2 designates an even-odd register pair; the instruction should raise
a specification exception when R2 is not even.

Cc: qemu-stable@nongnu.org
Fixes: e023e832d0ac ("s390x: translate engine for s390x CPU")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230724082032.66864-2-iii@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Merge tag 'pull-riscv-to-apply-20230723-3' of https://github.com/alistair23/qemu into staging

Fifth RISC-V PR for 8.1

* roms/opensbi: Upgrade from v1.3 to v1.3.1

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEaukCtqfKh31tZZKWr3yVEwxTgBMFAmS88+wACgkQr3yVEwxT
# gBNxwA//ZJxbSN4LR+5Cs12tW1ad4GMfkMyoRHp6CN6ZFA38W3xjvchqEAKMlk9C
# S8GHfoGukk0+dxqZ6QID/GTgaR0aH09WVFkr4SzWCvvFaJFnzU+wJknQv7aLOT/M
# yFflWbpUFM/JJlpouskSqG1eMjcC4P2ZD8e5CiP1OqRgzQ0HyQi99ADVpFMzET6X
# xP9LfFKvgaOrsTUJAGrnJ3EUkJIx9e1yTBm7wt+tREIj7peLZuwUGG6+vPAXnEq2
# JpAnFHlsiDWfOf72bIZt7Gw9AS64f6ej6IvtqhfjF5a7nOhPb0soejilIsvnTVS7
# akp4Ip2TQ8wULb4wehHPkmo882mzacmeHHsxPAzgW+FKbSK+LKiDvesJk0suO+SW
# 4tCL6xo2gFrTgSUxo762myTN6u5JxkPZnLJV7Lw/nfWJ04DYaZWJ4KdZ39HH+34/
# 1jNt1SXK/WF1DlXoRkRnQtzeenhIvmlSOtyhPhpAjSXHnwk5vfnarq/EAcKx2t+B
# OHWDwQlWgnZ/53m0EwBB91IDW4dMMc7CwTw8VPDjUQeRk8JFhrRjnY4TdT/LGBZt
# 87AfKEH8RPo0mIbDou7/bjXwraW647SzlZhrCfyNNyNQ4fo1z3Qo5tO5liloiBQb
# SRdhdZ6UCg6epokVuvaRPH+TMmMGWad6n4GKGqXa1edK1yCIKEE=
# =pNh6
# -----END PGP SIGNATURE-----
# gpg: Signature made Sun 23 Jul 2023 10:33:32 BST
# gpg:                using RSA key 6AE902B6A7CA877D6D659296AF7C95130C538013
# gpg: Good signature from "Alistair Francis <alistair@alistair23.me>" [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: 6AE9 02B6 A7CA 877D 6D65  9296 AF7C 9513 0C53 8013

* tag 'pull-riscv-to-apply-20230723-3' of https://github.com/alistair23/qemu:
  roms/opensbi: Upgrade from v1.3 to v1.3.1

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

Merge tag 'pull-revert-armhf-brk-fix' of https://gitlab.com/mjt0k/qemu into staging

Revert "linux-user: Fix qemu-arm to run static armhf binaries"

# -----BEGIN PGP SIGNATURE-----
#
# iQFDBAABCAAtFiEEe3O61ovnosKJMUsicBtPaxppPlkFAmS8Cj4PHG1qdEB0bHMu
# bXNrLnJ1AAoJEHAbT2saaT5ZLdsH/A7uTAODHopQnUGySPpHAT9uPjf6A21eZnnE
# MAcd3w+iEnrleJbpSftaZfKBKEUEN/vPPXg8nfCxhtfvgWUoHlglR3EoeTU+viFG
# cW7YO2LG9EbITL62uWDM6blVJzh7OHq+IjUf9xYuErnpREyXMvmGGAlRRjAN2yUz
# 3TMQ35JBNuSay2hwvC12jj4AqsmCEFBxvUiJkrrQTFeCVa+Gj/A9xoH7NWSrkW5F
# YE1qIfc2lCQFuWag3+Iag1PtTmj4oeW2LmLCJpxc75XnRqcl7MKK2mNhNJ012kun
# YZtySEG35t4x4Pyf11dhydtv4fbe5K+L0OjMf444zzEcW7TJ840=
# =Cp7j
# -----END PGP SIGNATURE-----
# gpg: Signature made Sat 22 Jul 2023 17:56:30 BST
# 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-revert-armhf-brk-fix' of https://gitlab.com/mjt0k/qemu:
  Revert "linux-user: Fix qemu-arm to run static armhf binaries"

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

Merge tag 'pull-tcg-20230724' of https://gitlab.com/rth7680/qemu into staging

accel/tcg: Zero-pad vaddr in tlb debug output
accel/tcg: Fix type of 'last' for pageflags_{find,next}
accel/tcg: Fix sense of read-only probes in ldst_atomicity
accel/tcg: Take mmap_lock in load_atomic*_or_exit
tcg: Add earlyclobber to op_add2 for x86 and s390x
tcg/ppc: Fix race in goto_tb implementation

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmS+O7cdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV8qrAf/VeAFnMbtantUTfM5
# zOcfBlutsDlJrNwA/ajFDrPwUDewP7s5cqxImAYqhXfhqlc2RIB3UiMCgSaQ+q6O
# MBOH0bEj/zbeIlwRX07ZBWhUYVdqJVd7Nxb1W19YwgG9yieWUxa+Xo1i2fhyXMv+
# 20VOFB1dPnxYyUMrzh/bSiHE90JFZktO1WzV10FRD+IpnImY9R+YGdpGTpVzUhor
# ReRHTkMKyYilY6EEUG2gFhotrY/bbSSSFyl9BcQjkZh11603nAN0mNKxtSjPJnNB
# rXhCVEgmbbBvCufsO6szQ03W/7RZ/KCg/DyKqxyCP1Ril4BIOx3tiucROcapXH/K
# 0y/ycA==
# =hdk/
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 24 Jul 2023 09:52:07 BST
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* tag 'pull-tcg-20230724' of https://gitlab.com/rth7680/qemu:
  accel/tcg: Fix type of 'last' for pageflags_{find,next}
  accel/tcg: Zero-pad vaddr in tlb_debug output
  tcg/{i386, s390x}: Add earlyclobber to the op_add2's first output
  accel/tcg: Take mmap_lock in load_atomic*_or_exit
  accel/tcg: Fix sense of read-only probes in ldst_atomicity
  include/exec: Add WITH_MMAP_LOCK_GUARD
  tcg/ppc: Fix race in goto_tb implementation

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

accel/tcg: Fix type of 'last' for pageflags_{find,next}

These should match 'start' as target_ulong, not target_long.

On 32bit targets, the parameter was sign-extended to uint64_t,
so only the first mmap within the upper 2GB memory can succeed.

Signed-off-by: Luca Bonissi <qemu@bonslack.org>
Message-Id: <327460e2-0ebd-9edb-426b-1df80d16c32a@bonslack.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Zero-pad vaddr in tlb_debug output

In replacing target_ulong with vaddr and TARGET_FMT_lx with VADDR_PRIx,
the zero-padding of TARGET_FMT_lx got lost.  Readd 16-wide zero-padding
for logging consistency.

Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230713120746.26897-1-anjo@rev.ng>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/{i386, s390x}: Add earlyclobber to the op_add2's first output

i386 and s390x implementations of op_add2 require an earlyclobber,
which is currently missing. This breaks VCKSM in s390x guests. E.g., on
x86_64 the following op:

    add2_i32 tmp2,tmp3,tmp2,tmp3,tmp3,tmp2   dead: 0 2 3 4 5  pref=none,0xffff

is translated to:

    addl     %ebx, %r12d
    adcl     %r12d, %ebx

Introduce a new C_N1_O1_I4 constraint, and make sure that earlyclobber
of aliased outputs is honored.

Cc: qemu-stable@nongnu.org
Fixes: 82790a870992 ("tcg: Add markup for output requires new register")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230719221310.1968845-7-iii@linux.ibm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Take mmap_lock in load_atomic*_or_exit

For user-only, the probe for page writability may race with another
thread's mprotect.  Take the mmap_lock around the operation.  This
is still faster than the start/end_exclusive fallback.

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

accel/tcg: Fix sense of read-only probes in ldst_atomicity

In the initial commit, cdfac37be0d, the sense of the test is incorrect,
as the -1/0 return was confusing.  In bef6f008b981, we mechanically
invert all callers while changing to false/true return, preserving the
incorrectness of the test.

Now that the return sense is sane, it's easy to see that if !write,
then the page is not modifiable (i.e. most likely read-only, with
PROT_NONE handled via SIGSEGV).

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

include/exec: Add WITH_MMAP_LOCK_GUARD

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

tcg/ppc: Fix race in goto_tb implementation

Commit 20b6643324 ("tcg/ppc: Reorg goto_tb implementation") modified
goto_tb to ensure only a single instruction was patched to prevent
incorrect behavior if a thread was in the middle of multiple
instructions when they were replaced. However this introduced a race
between loading the jmp target into TCG_REG_TB and patching and
executing the direct branch.

The relevant part of the goto_tb implementation:

    ld TCG_REG_TB, TARGET_ADDR_LOCATION(TCG_REG_TB)
  patch_location:
    mtctr TCG_REG_TB
    bctr

tb_target_set_jmp_target() will replace 'patch_location' with a direct
branch if the target is in range. The direct branch now relies on
TCG_REG_TB being set up correctly by the ld. Prior to this commit
multiple instructions were patched in for the direct branch case; these
instructions would initialize TCG_REG_TB to the same value as the branch
target.

Imagine the following sequence:

1) Thread A is executing the goto_tb sequence and loads the jmp
   target into TCG_REG_TB.

2) Thread B updates the jmp target address and calls
   tb_target_set_jmp_target(). This patches a new direct branch into the
   goto_tb sequence.

3) Thread A executes the newly patched direct branch. The value in
   TCG_REG_TB still contains the old jmp target.

TCG_REG_TB MUST contain the translation block's tc.ptr. Execution will
eventually crash after performing memory accesses generated from a
faulty value in TCG_REG_TB.

This presents as segfaults or illegal instruction exceptions.

Do not revert commit 20b6643324 as it did fix a different race
condition. Instead remove the direct branch optimization and always use
indirect branches.

The direct branch optimization can be re-added later with a race free
sequence.

Fixes: 20b6643324 ("tcg/ppc: Reorg goto_tb implementation")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1726
Reported-by: Anushree Mathur <anushree.mathur@linux.vnet.ibm.com>
Tested-by: Anushree Mathur <anushree.mathur@linux.vnet.ibm.com>
Tested-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Co-developed-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Jordan Niethe <jniethe5@gmail.com>
Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Message-Id: <20230717093001.13167-1-jniethe5@gmail.com>

roms/opensbi: Upgrade from v1.3 to v1.3.1

Upgrade OpenSBI from v1.3 to v1.3.1 and the pre-built bios images
which fixes the boot failure seen when using QEMU to do a direct
kernel boot with Microchip Icicle Kit board machine.

The v1.3.1 release includes the following commits:

0907de3 lib: sbi: fix comment indent
eb736a5 lib: sbi_pmu: Avoid out of bounds access
7828eeb gpio/desginware: add Synopsys DesignWare APB GPIO support
c6a3573 lib: utils: Fix sbi_hartid_to_scratch() usage in ACLINT drivers
057eb10 lib: utils/gpio: Fix RV32 compile error for designware GPIO driver

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Message-Id: <20230719165817.889465-1-bmeng@tinylab.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>