nbd: Use enum for various negotiation modes

Deciphering the hard-coded list of integer return values from
nbd_start_negotiate() will only get more confusing when adding support
for 64-bit extended headers.  Better is to name things in an enum.
Although the function in question is private to client.c, putting the
enum in a public header and including an enum-to-string conversion
will allow its use in more places in upcoming patches.

The enum is intentionally laid out so that operators like <= can be
used to group multiple modes with similar characteristics, and where
the least powerful mode has value 0, even though this patch does not
exploit that.  No semantic change intended.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20230608135653.2918540-9-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

nbd/client: Add safety check on chunk payload length

Our existing use of structured replies either reads into a qiov capped
at 32M (NBD_CMD_READ) or caps allocation to 1000 bytes (see
NBD_MAX_MALLOC_PAYLOAD in block/nbd.c).  But the existing length
checks are rather late; if we encounter a buggy (or malicious) server
that sends a super-large payload length, we should drop the connection
right then rather than assuming the layer on top will be careful.
This becomes more important when we permit 64-bit lengths which are
even more likely to have the potential for attempted denial of service
abuse.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230608135653.2918540-8-eblake@redhat.com>

nbd/client: Simplify cookie vs. index computation

Our code relies on a sentinel cookie value of zero for deciding when a
packet has been handled, as well as relying on array indices between 0
and MAX_NBD_REQUESTS-1 for dereferencing purposes.  As long as we can
symmetrically convert between two forms, there is no reason to go with
the odd choice of using XOR with a random pointer, when we can instead
simplify the mappings with a mere offset of 1.

Using ((uint64_t)-1) as the sentinel instead of NULL such that the two
macros could be entirely eliminated might also be possible, but would
require a more careful audit to find places where we currently rely on
zero-initialization to be interpreted as the sentinel value, so I did
not pursue that course.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20230608135653.2918540-7-eblake@redhat.com>
[eblake: enhance commit message]
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

nbd: s/handle/cookie/ to match NBD spec

Externally, libnbd exposed the 64-bit opaque marker for each client
NBD packet as the "cookie", because it was less confusing when
contrasted with 'struct nbd_handle *' holding all libnbd state.  It
also avoids confusion between the noun 'handle' as a way to identify a
packet and the verb 'handle' for reacting to things like signals.
Upstream NBD changed their spec to favor the name "cookie" based on
libnbd's recommendations[1], so we can do likewise.

[1] https://github.com/NetworkBlockDevice/nbd/commit/ca4392eb2b

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20230608135653.2918540-6-eblake@redhat.com>
[eblake: typo fix]
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

nbd/server: Refactor to pass full request around

Part of NBD's 64-bit headers extension involves passing the client's
requested offset back as part of the reply header (one reason it
stated for this change: converting absolute offsets stored in
NBD_REPLY_TYPE_OFFSET_DATA to relative offsets within the buffer is
easier if the absolute offset of the buffer is also available).  This
is a refactoring patch to pass the full request around the reply
stack, rather than just the handle, so that later patches can then
access request->from when extended headers are active.  Meanwhile,
this patch enables us to now assert that simple replies are only
attempted when appropriate, and otherwise has no semantic change.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230608135653.2918540-5-eblake@redhat.com>

nbd/server: Prepare for alternate-size headers

Upstream NBD now documents[1] an extension that supports 64-bit effect
lengths in requests.  As part of that extension, the size of the reply
headers will change in order to permit a 64-bit length in the reply
for symmetry[2].  Additionally, where the reply header is currently 16
bytes for simple reply, and 20 bytes for structured reply; with the
extension enabled, there will only be one extended reply header, of 32
bytes, with both structured and extended modes sending identical
payloads for chunked replies.

Since we are already wired up to use iovecs, it is easiest to allow
for this change in header size by splitting each structured reply
across multiple iovecs, one for the header (which will become wider in
a future patch according to client negotiation), and the other(s) for
the chunk payload, and removing the header from the payload struct
definitions.  Rename the affected functions with s/structured/chunk/
to make it obvious that the code will be reused in extended mode.

Interestingly, the client side code never utilized the packed types,
so only the server code needs to be updated.

[1] https://github.com/NetworkBlockDevice/nbd/blob/extension-ext-header/doc/proto.md
as of NBD commit e6f3b94a934

[2] Note that on the surface, this is because some future server might
permit a 4G+ NBD_CMD_READ and need to reply with that much data in one
transaction.  But even though the extended reply length is widened to
64 bits, for now the NBD spec is clear that servers will not reply
with more than a maximum payload bounded by the 32-bit
NBD_INFO_BLOCK_SIZE field; allowing a client and server to mutually
agree to transactions larger than 4G would require yet another
extension.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20230608135653.2918540-4-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

nbd: Consistent typedef usage in header

We had a mix of struct declarations followed by typedefs, and direct
struct definitions as part of a typedef.  Pick a single style.  Also
float forward declarations of opaque types to the top of the file,
rather than interspersed with function declarations, which will help a
future patch that wants to expose yet another opaque type that will be
referenced in NBDRequest.  No semantic impact.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20230608135653.2918540-3-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
[eblake: alter patch per mailing list feedback]
Signed-off-by: Eric Blake <eblake@redhat.com>

nbd/client: Use smarter assert

Assigning strlen() to a uint32_t and then asserting that it isn't too
large doesn't catch the case of an input string 4G in length.
Thankfully, the incoming strings can never be that large: if the
export name or query is reflecting a string the client got from the
server, we already guarantee that we dropped the NBD connection if the
server sent more than 32M in a single reply to our NBD_OPT_* request;
if the export name is coming from qemu, nbd_receive_negotiate()
asserted that strlen(info->name) <= NBD_MAX_STRING_SIZE; and
similarly, a query string via x->dirty_bitmap coming from the user was
bounds-checked in either qemu-nbd or by the limitations of QMP.
Still, it doesn't hurt to be more explicit in how we write our
assertions to not have to analyze whether inadvertent wraparound is
possible.

Fixes: 93676c88 ("nbd: Don't send oversize strings", v4.2.0)
Reported-by: Dr. David Alan Gilbert <dave@treblig.org>
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230608135653.2918540-2-eblake@redhat.com>

qemu-nbd: make verbose bool and local variable in main()

Pass 'verbose' to nbd_client_thread() inside NbdClientOpts which looks
a little bit cleaner and make it bool as it is used as bool actually.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230717202520.236999-1-den@openvz.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

qemu-nbd: handle dup2() error when qemu-nbd finished setup process

Fail on error, we are in trouble.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230717145544.194786-6-den@openvz.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
[eblake: avoid intermediate variable]
Signed-off-by: Eric Blake <eblake@redhat.com>

qemu-nbd: properly report error on error in dup2() after qemu_daemon()

We are trying to temporarily redirect stderr of daemonized process to
a pipe to report a error and get failed. In that case we could not
use error_report() helper, but should write the message directly into
the problematic pipe.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230717145544.194786-4-den@openvz.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
[eblake: rearrange patch series, fix typo]
Signed-off-by: Eric Blake <eblake@redhat.com>

qemu-nbd: properly report error if qemu_daemon() is failed

errno has been overwritten by dup2() just below qemu_daemon() and thus
improperly returned to the caller. Fix accordingly.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230717145544.194786-5-den@openvz.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
[eblake: reorder patch series]
Signed-off-by: Eric Blake <eblake@redhat.com>

qemu-nbd: fix regression with qemu-nbd --fork run over ssh

Commit e6df58a5578fee7a50bbf36f4a50a2781cff855d
    Author: Hanna Reitz <hreitz@redhat.com>
    Date:   Wed May 8 23:18:18 2019 +0200
    qemu-nbd: Do not close stderr

has introduced an interesting regression. Original behavior of
    ssh somehost qemu-nbd /home/den/tmp/file -f raw --fork
was the following:
 * qemu-nbd was started as a daemon
 * the command execution is done and ssh exited with success

The patch has changed this behavior and 'ssh' command now hangs forever.

According to the normal specification of the daemon() call, we should
endup with STDERR pointing to /dev/null. That should be done at the
very end of the successful startup sequence when the pipe to the
bootstrap process (used for diagnostics) is no longer needed.

This could be achived in the same way as done for 'qemu-nbd -c' case.
That was commit 0eaf453e, also fixing up e6df58a5. STDOUT copying to
STDERR does the trick.

This also leads to proper 'ssh' connection closing which fixes my
original problem.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
CC: Hanna Reitz <hreitz@redhat.com>
CC: <qemu-stable@nongnu.org>
Message-ID: <20230717145544.194786-3-den@openvz.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

qemu-nbd: pass structure into nbd_client_thread instead of plain char*

We are going to pass additional flag inside next patch.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
CC: <qemu-stable@nongnu.org>
Message-ID: <20230717145544.194786-2-den@openvz.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

Update version for v8.1.0-rc0 release

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

Merge tag 'pull-riscv-to-apply-20230719-1' of https://github.com/alistair23/qemu into staging

Fourth RISC-V PR for 8.1

* Fix LMUL check to use VLEN
* Fix typo field in NUMA error_report
* check priv_ver before auto-enable zca/zcd/zcf
* Fix disas output of upper immediates
* tidy CPU firmware section

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEaukCtqfKh31tZZKWr3yVEwxTgBMFAmS3akMACgkQr3yVEwxT
# gBPQ/BAArrieEkrRco3tIQJFZqTLfII28M0cYdwN+gjMAkL6RlauCh5yKkc+gsGy
# bhhpr0AE+EzrjKfJgdyMQe2ZH08WEpoAfJHAmLTSm2ktgIlnDAjyJtVksZ3FSwfG
# MRK3v0CChyOav3EfDZzK9jcaXeaSSfjCIG8JW3enoZxf2TnpoXlsCIQdRTnMw7Um
# C73BWoOGOfixFehywHBnkkAPo/nkQPofELrRKNTlefAIsH1RcgYw+s3IgCIuYxJN
# zCjM1y6ye1aiaQhKcNJiLoiP4Eq2R6vUuL8RKWkXqTP3QBZUqKMPnRVgI+W0qRAj
# 9DS+l37zMdxytovQ4gmIqnENT8ty9bholOtWM8nI54subJBplQhkRednG3RBFYjH
# hqbsakcHfE1lyyNI7WoBpO8UMtnOad6eBNmMOM48VduSdNuBZN3ksoRVomnJTlCY
# nq1ZdteywHEZ3uBqk3k/4yzKH+jLj0McPz5FswxsMIGScVjd6H8rMYmM95r1He4k
# YTJ8GwnOTBs1tFxOz5DaM3BVfq5hrzB0SbpDHMOdQHNXnqkyfvSd/QWeXfnY09Ux
# kbNvSpzjn7wWRSP7s4KMcTmas4oGtPS2dheREB/gmoC1ubrfuhbzduDNXJt+omuC
# GDcn9cpouyE/Vp/358PuEe1gW9GFMH0CbYBJ66P0hI/76iPfwLY=
# =MOsI
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 19 Jul 2023 05:44:51 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-20230719-1' of https://github.com/alistair23/qemu:
  target/riscv: Fix LMUL check to use VLEN
  hw/riscv: Fix typo field in error_report
  target/riscv/cpu.c: check priv_ver before auto-enable zca/zcd/zcf
  riscv/disas: Fix disas output of upper immediates
  docs/system/target-riscv.rst: tidy CPU firmware section

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

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

hw/nvme fixes

* fix shadow doorbell endian issue

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEEUigzqnXi3OaiR2bATeGvMW1PDekFAmS3kkAACgkQTeGvMW1P
# DenG1ggArIHi1dQQBIG1ubzHx/C+93cybpKwT73/5wfO7BT8CCh1v+qrH/6SsYUT
# 5O7y1MaCLDV4ocf5dRQseXFK0tpjo7EqDnr25UhcSunQ+d2Tn7MAIuubQOFD+Axh
# 5gIwOEJbKqw9apJgnVWnInTBd//ManOgh6OyC1uJ+DEJE7ISJzLlJeWaBekiWpAA
# hNL1zsR5+eTcwnewDRmMs4FlKBlSfgcNgNYnz8tfpnW0DzXKuiY4ITnk6kX9eMAM
# kDlbjFjlgoTPZ8IsYcyhVCJMcH8jqY/LuZcaF7XHHsdX7fa5p17C6rR1hxVyDs+E
# rydOtWetQDhXlyakE+Jp2RB3HLcSmg==
# =j1TL
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 19 Jul 2023 08:35:28 BST
# gpg:                using RSA key 522833AA75E2DCE6A24766C04DE1AF316D4F0DE9
# gpg: Good signature from "Klaus Jensen <its@irrelevant.dk>" [full]
# gpg:                 aka "Klaus Jensen <k.jensen@samsung.com>" [full]
# Primary key fingerprint: DDCA 4D9C 9EF9 31CC 3468  4272 63D5 6FC5 E55D A838
#      Subkey fingerprint: 5228 33AA 75E2 DCE6 A247  66C0 4DE1 AF31 6D4F 0DE9

* tag 'nvme-next-pull-request' of https://gitlab.com/birkelund/qemu:
  hw/nvme: fix endianness issue for shadow doorbells

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

Merge tag 'pull-request-2023-07-18' of https://gitlab.com/thuth/qemu into staging

* Fix s390x KVM guests when compiling with --without-default-devices
* Fix /proc/cpuinfo features list in s390x linux-user emulation
* Generate FreeBSD VM package list via lcitool
* Disable the flaky test_sbsaref_edk2_firmware avocado test by default

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmS2W14RHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbW25g/7B68fy5dMzrrYlESm/PSYKa+POZtXt7IO
# EuDNgqIz8bBtlK/4HgMiY0huoT+5cWJ73qZeSzoiDjte8CqOAfjy2TeOWEpd+1qZ
# xHZ7W1t1fwN2rNQlOlGw4jASkrwdYeiBbD8PopixKeevwMOOAxroYNHSvzPNG+Qa
# rTJAlzq1Q80z8DP5zl6ufsxQNiWQsCdCO3hdTZQsRi+RCxl+3SMx2xKA2bBtw9VL
# skwItYqvvw21LtR7yYhnb9bMtikh+nVYuyAW6CJl9vrG2zwf6D/tURIk4LLaIxlS
# XfDVRul6BJHWWdWAZ42Q7Xwe0olrzOfl57wYb/gvh09E6NEjyJcyIr/mor+yz1/q
# Qomb/Stt5K8pphiSSwkKmZUGA0s96MjOYe/xXbJegu+6X6HNgYVobZDMMq/BABy2
# XeESwUE/Hfz+VW14oqCk4edZb6+jQbYaKO/YDPd7uyXzfas079koWYfWx9Ew08U8
# f5h7achjECr3R4DJ9VgYIOKJuv37LcnwbePspXdRK5pjRrqr9FD9XsEg4gF7rbIe
# A/g/UUx1+AXeie4z1MrUwFhOrqV1g5t34cBR9iqZtISd0UiNk6ExoiX8i+ELaw5+
# ADirr46HJFS9jVAKKw9zym932qC92AQNH0d0iOYv3AHzvKvTDiQOXHyl1bLTMeYd
# yUJNn+Jkdns=
# =dw94
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 18 Jul 2023 10:29:02 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-2023-07-18' of https://gitlab.com/thuth/qemu:
  tests/avocado: Disable the test_sbsaref_edk2_firmware by default
  tests/vm/freebsd: Get up-to-date package list from lcitool vars file
  tests/vm: Introduce get_qemu_packages_from_lcitool_json() helper
  tests/lcitool: Refresh generated files
  tests/lcitool: Generate distribution packages list in JSON format
  tests/qtest: Fix typo in multifd cancel test
  linux-user/elfload: Fix /proc/cpuinfo features: on s390x
  s390x: Fix QEMU abort by selecting S390_FLIC_KVM

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

hw/nvme: fix endianness issue for shadow doorbells

In commit 2fda0726e514 ("hw/nvme: fix missing endian conversions for
doorbell buffers"), we fixed shadow doorbells for big-endian guests
running on little endian hosts. But I did not fix little-endian guests
on big-endian hosts. Fix this.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1765
Fixes: 3f7fe8de3d49 ("hw/nvme: Implement shadow doorbell buffer support")
Cc: qemu-stable@nongnu.org
Reported-by: Thomas Huth <thuth@redhat.com>
Tested-by: Cédric Le Goater <clg@redhat.com>
Tested-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>

target/riscv: Fix LMUL check to use VLEN

The previous check was failing with:

VLEN=128 ELEN = 64 SEW = 16 and LMUL = 1/8 which is a
valid combination.

Fix the check to allow valid combinations when VLEN is a multiple of
ELEN.

From the specification:

"In general, the requirement is to support LMUL ≥ SEWMIN/ELEN, where
SEWMIN is the narrowest supported SEW value and ELEN is the widest
supported SEW value. In the standard extensions, SEWMIN=8. For standard
vector extensions with ELEN=32, fractional LMULs of 1/2 and 1/4 must be
supported. For standard vector extensions with ELEN=64, fractional LMULs
of 1/2, 1/4, and 1/8 must be supported." Elsewhere in the specification
it makes clear that VLEN>=ELEN.

From inspection this new check allows:

VLEN=ELEN=64 1/2, 1/4, 1/8 for SEW >=8
VLEN=ELEN=32 1/2, 1/4 for SEW >=8

Fixes: d9b7609a1fb2 ("target/riscv: rvv-1.0: configure instructions")
Signed-off-by: Rob Bradford <rbradford@rivosinc.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Message-Id: <20230718131316.12283-2-rbradford@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: Fix typo field in error_report

"smp.cpus" means the number of online CPUs and "smp.max_cpus" means the
total number of CPUs.

riscv_numa_get_default_cpu_node_id() checks "smp.cpus" and the
"available CPUs" description in the next error message also indicates
online CPUs.

So report "smp.cpus" in error_report() instand of "smp.max_cpus".

Since "smp.cpus" is "unsigned int", use "%u".

Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230718080712.503333-1-zhao1.liu@linux.intel.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv/cpu.c: check priv_ver before auto-enable zca/zcd/zcf

Commit bd30559568 made changes in how we're checking and disabling
extensions based on env->priv_ver. One of the changes was to move the
extension disablement code to the end of realize(), being able to
disable extensions after we've auto-enabled some of them.

An unfortunate side effect of this change started to happen with CPUs
that has an older priv version, like sifive-u54. Starting on commit
2288a5ce43e5 we're auto-enabling zca, zcd and zcf if RVC is enabled,
but these extensions are priv version 1.12.0. When running a cpu that
has an older priv ver (like sifive-u54) the user is spammed with
warnings like these:

qemu-system-riscv64: warning: disabling zca extension for hart 0x0000000000000000 because privilege spec version does not match
qemu-system-riscv64: warning: disabling zcd extension for hart 0x0000000000000000 because privilege spec version does not match

The warnings are part of the code that disables the extension, but in this
case we're throwing user warnings for stuff that we enabled on our own,
without user intervention. Users are left wondering what they did wrong.

A quick 8.1 fix for this nuisance is to check the CPU priv spec before
auto-enabling zca/zcd/zcf. A more appropriate fix will include a more
robust framework that will account for both priv_ver and user choice
when auto-enabling/disabling extensions, but for 8.1 we'll make it do
with this simple check.

It's also worth noticing that this is the only case where we're
auto-enabling extensions based on a criteria (in this case RVC) that
doesn't match the priv spec of the extensions we're enabling. There's no
need for more 8.1 band-aids.

Cc: Conor Dooley <conor@kernel.org>
Fixes: 2288a5ce43e5 ("target/riscv: add cfg properties for Zc* extension")
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Message-Id: <20230717154141.60898-1-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

riscv/disas: Fix disas output of upper immediates

The GNU assembler produces the following output for instructions
with upper immediates:
    00002597                auipc   a1,0x2
    000024b7                lui     s1,0x2
    6409                    lui     s0,0x2 # c.lui

The immediate operands of upper immediates are not shifted.

However, the QEMU disassembler prints them shifted:
    00002597          auipc                   a1,8192
    000024b7          lui                     s1,8192
    6409              lui                     s0,8192 # c.lui

The current implementation extracts the immediate bits and shifts the by 12,
so the internal representation of the immediate is the actual immediate.
However, the immediates are later printed using rv_fmt_rd_imm or
rv_fmt_rd_offset, which don't undo the shift.

Let's fix this by using specific output formats for instructions
with upper immediates, that take care of the shift.

Signed-off-by: Christoph Müllner <christoph.muellner@vrull.eu>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230711075051.1531007-1-christoph.muellner@vrull.eu>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

docs/system/target-riscv.rst: tidy CPU firmware section

This is how the content of the "RISC-V CPU firmware" section is
displayed after the html is generated:

"When using the sifive_u or virt machine there are three different
firmware boot options: 1. -bios default - This is the default behaviour
if no -bios option is included. (...) 3. -bios <file> - Tells QEMU to
load the specified file as the firmware."

It's all in the same paragraph, in a numbered list, and no special
formatting for the options.

Tidy it a bit by adding line breaks between items and its description.
Remove the numbered list. And apply formatting for the options cited in
the middle of the text.

Cc: qemu-trivial@nongnu.org
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230712143728.383528-1-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

tests/avocado: Disable the test_sbsaref_edk2_firmware by default

The test fails occasionally, see e.g.:

 https://gitlab.com/thuth/qemu/-/jobs/4196177756#L489
 https://gitlab.com/thuth/qemu/-/jobs/4623296271#L290

It also fails on my laptop in ca. 50% of all runs. Thus disable it by
default by using the QEMU_TEST_FLAKY_TESTS environment variable to fence
it (which we also already use in flaky qtests). While we're at it, also
document this variable in docs/devel/testing.rst.

Message-Id: <20230710170155.7192-1-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/vm/freebsd: Get up-to-date package list from lcitool vars file

Get an up-to-date package list from lcitool, that way we
don't need to manually keep this array in sync.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Inspired-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230711144922.67491-5-philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/vm: Introduce get_qemu_packages_from_lcitool_json() helper

Add the get_qemu_packages_from_lcitool_json() helper which return
such package list from a lcitool env var file in JSON format.

Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230711144922.67491-4-philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/lcitool: Refresh generated files

Refresh the generated files by running:

  $ make lcitool-refresh

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230711144922.67491-3-philmd@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[thuth: Drop changes to libpmem-dev and libxen-dev]
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/lcitool: Generate distribution packages list in JSON format

Add the generate_pkglist() helper to generate a list of packages
required by a distribution to build QEMU.

Since we can not add a "THIS FILE WAS AUTO-GENERATED" comment in
JSON, create the files under tests/vm/generated/ sub-directory;
add a README mentioning the files are generated.

Suggested-by: Erik Skultety <eskultet@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Message-Id: <20230711144922.67491-2-philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Fix typo in multifd cancel test

This wasn't noticed because the test is currently disabled.

Fixes: 02f56e3de ("tests/qtest: massively speed up migration-test")
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20230711212131.2370-1-farosas@suse.de>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

linux-user/elfload: Fix /proc/cpuinfo features: on s390x

elf_hwcap_str() takes a bit number, but compares it for equality with
the HWCAP_S390_* masks. This causes /proc/cpuinfo to display incorrect
hwcaps.

Fix by introducing the HWCAP_S390_NR_* constants and using them in
elf_hwcap_str() instead of the HWCAP_S390_*. While at it, add the
missing nnpa, pcimio and sie hwcaps from the latest kernel.

Output before:

features : esan3 zarch stfle msa

Output after:

features : esan3 zarch stfle msa ldisp eimm etf3eh highgprs vx vxe

Fixes: e19807bee357 ("linux-user/elfload: Introduce elf_hwcap_str() on s390x")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230627151356.273259-1-iii@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

s390x: Fix QEMU abort by selecting S390_FLIC_KVM

If QEMU is built with --without-default-devices, the s390-flic-kvm
device is missing and QEMU aborts when started with the KVM accelerator.
Make sure it's available by selecting S390_FLIC_KVM in Kconfig.

Consequently, this also fixes an abort in tests/qtest/migration-test.

Signed-off-by: Cédric Le Goater <clg@redhat.com>
Message-Id: <20230711151440.716822-1-clg@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

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

Pull request

Fix the hang in the nvme:// block driver during startup.

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEhpWov9P5fNqsNXdanKSrs4Grc8gFAmS1QFoACgkQnKSrs4Gr
# c8i1KQgAqTW/rH+Mq9gC+MabLKXL8BW3KAp/lJqcaTrWRX+9O968E5RjF8fk5KzY
# CZGKGcFd2+r6poUaRrp2CcnDe9AU9r3NBfQIoVAnyWL8fwRBl5ERPZ3q6dmW63vX
# K8ckI9c0tHUNRuKpWiFCFlcEh7GGgl2pZhY1U50J1iaFTtvmDKl4Ls/2GqVUSiSY
# 4HbJ99qAYyZ/EUzbFhDPq4XFneqJzU0Gu4uL3bbVY+KlmGOp36W8wwPaKaD6yXWv
# mnv4DkoO3qlZ8tGegwk4SwePWN2Coec+KssMgYR1GWv0pbF8G4sX+UTeQin56Xny
# EY3lJy2z793lMjFVMDd7N8+uKAczsQ==
# =xvCy
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 17 Jul 2023 02:21:30 PM BST
# 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/nvme: invoke blk_io_plug_call() outside q->lock

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

Merge tag 'pull-target-arm-20230717' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * hw/arm/sbsa-ref: set 'slots' property of xhci
 * linux-user: Remove pointless NULL check in clock_adjtime handling
 * ptw: Fix S1_ptw_translate() debug path
 * ptw: Account for FEAT_RME when applying {N}SW, SA bits
 * accel/tcg: Zero-pad PC in TCG CPU exec trace lines
 * hw/nvram: Avoid unnecessary Xilinx eFuse backstore write

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmS1OEUZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3syoEACBj2B+btKASbWs6c7iUF9R
# bsMhVVZbeNrW7try7fIjAcvRQV2X7cdqHMGeX0yP9M5EcqBfz4ptxDbxcmEsgm0h
# kZJudG8RuZ/gnw7wbwQ1TfJf4KgsBh49yZjlom2s8CgVStpbuFO4xz7ZucR65uhl
# PwLCgW0/DJR4SQTvDLnCOTTNbY/cuWCKK1CmuLMOE9IgozMNOxxW5wkryrvdllKs
# hYSCWM1jy9fJ4TRlhDJy8JI7+t4TEZN9ESwYGE6QDly8r3GoGMFj5Z9okUbGp3/V
# MYfkbz7l2/C5QxcpY5d0mJUR1HlP7McO7rWhtQjqmCPGpDVqMUu4/DClu6Q/2Ob3
# GRQcgztZ8a9wgVa6b4g1UBkqCnloT7WtU3wLVVmZGF3DO4k+oz53XPHb2zFtI3Xx
# pQ9LyABIoKCM5ql+/WaA3thtTC1qH6lZnjMBqVBx8+d0zKYWSG4wlnbihy70GOpw
# V5n0fQlTXr5WV4tZT/euP17odvnkictH7Vmj6zHUFkHdqHxwFwG0OCw1ZjBrMbzl
# 7kY9DxGA+5iKEZoTwHpxXYny70MnpdRIrUhpZ/4PNq68hzIAQ5Dqm29DtKjodM60
# M49CIo+O9E3+0xpcGPDtcuJ7bVPd/95o3usVjapDdBREGWcJsPS6PHK3MuAxgkHo
# B0y1egitacJYp3x91gYIRA==
# =JPpH
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 17 Jul 2023 01:47:01 PM BST
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [full]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [full]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [full]
# gpg:                 aka "Peter Maydell <peter@archaic.org.uk>" [unknown]

* tag 'pull-target-arm-20230717' of https://git.linaro.org/people/pmaydell/qemu-arm:
  hw/nvram: Avoid unnecessary Xilinx eFuse backstore write
  accel/tcg: Zero-pad PC in TCG CPU exec trace lines
  target/arm/ptw.c: Account for FEAT_RME when applying {N}SW, SA bits
  target/arm: Fix S1_ptw_translate() debug path
  target/arm/ptw.c: Add comments to S1Translate struct fields
  linux-user: Remove pointless NULL check in clock_adjtime handling
  hw/arm/sbsa-ref: set 'slots' property of xhci

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

Merge tag 'ui-pull-request' of https://gitlab.com/marcandre.lureau/qemu into staging

ui & audio fixes

# -----BEGIN PGP SIGNATURE-----
#
# iQJQBAABCAA6FiEEh6m9kz+HxgbSdvYt2ujhCXWWnOUFAmS1N6IcHG1hcmNhbmRy
# ZS5sdXJlYXVAcmVkaGF0LmNvbQAKCRDa6OEJdZac5SYsD/44+FIoik9v478pZDTp
# CpaezX+DfsW1zee4Ana5eKJkrVld/xEa6i9/msfUHy12bha+kiJ4a6wLu3H4KRZc
# vX/t6sehG2wNcsV5wLhfcjsKzaNUkYpnxLhIZ0fOYXKA0fSBuM/Bsj6zzGTG6kQA
# nt/cK58r1wy63V7werZbA7BI8PF0opDUw5SrZqN0GeoN5clbdyLdcXvD50ibvkDf
# eOVjNQ3QH8IbihmgBVm1wUV8hTuvYRpBmeLJyk7NeR4bnPl3XGIAgtAY8hJL5LdY
# Bm+I3AuxMSskVcag/22QR8mGR0HhDbf3NZauw4ND3LhSctvNN5syaKHVnY5a9aGe
# QLVEV9pxXGfqzWQcsD2HmbupRoBihmp6+WsIpV8ZtuSfeD6slyObw+lqarSQL9b5
# 2C4UFmGCsCOk8rrczZRDp9IWbm23toc/QcQZtg/LhdlCr8nM+7m0XtyEY5WtT3U1
# 8rJEmjOHHqlD4cVBathc8+ZRjKr8HFRRo1ed6WKMoP6voTsw2fiR7I3Vdc7jO7h9
# A1lMiMoLdAXi0Q2VqbmBdLMgb4fXtLzYl2mcbzW0aEUm8uyUfDy2bkVIIUopu40M
# pROmLjaUzUVE3CruckBUCvoYZtJ5hBtvy3W2k8drBNylnP5B8tEqpxpPb+tSFk82
# xgT6oLp8En8asE293eaACbswuw==
# =W2Xa
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 17 Jul 2023 01:44:18 PM BST
# gpg:                using RSA key 87A9BD933F87C606D276F62DDAE8E10975969CE5
# gpg:                issuer "marcandre.lureau@redhat.com"
# gpg: Good signature from "Marc-André Lureau <marcandre.lureau@redhat.com>" [full]
# gpg:                 aka "Marc-André Lureau <marcandre.lureau@gmail.com>" [full]

* tag 'ui-pull-request' of https://gitlab.com/marcandre.lureau/qemu:
  audio/pw: improve channel position code
  audio/pw: remove wrong comment
  audio/pw: simplify error reporting in stream creation
  audio/pw: add more error reporting
  audio/pw: factorize some common code
  audio/pw: add more details on error
  audio/pw: trace during init before calling pipewire API
  audio/pw: needless check for NULL
  audio/pw: drop needless case statement
  audio/pw: Pipewire->PipeWire case fix for user-visible text
  tests/lcitool: add pipewire
  libvirt-ci: update submodule to cover pipewire
  ui/gtk: skip refresh if new dmabuf has been submitted
  ui/gtk: set scanout-mode right before scheduling draw
  virtio-gpu-udmabuf: correct naming of QemuDmaBuf size properties
  virtio-gpu: replace the surface with null surface when resetting
  ui/gtk: Make sure the right EGL context is currently bound
  ui/vnc-clipboard: fix infinite loop in inflate_buffer (CVE-2023-3255)
  virtio-gpu: fix potential divide-by-zero regression

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

block/nvme: invoke blk_io_plug_call() outside q->lock

blk_io_plug_call() is invoked outside a blk_io_plug()/blk_io_unplug()
section while opening the NVMe drive from:

  nvme_file_open() ->
  nvme_init() ->
  nvme_identify() ->
  nvme_admin_cmd_sync() ->
  nvme_submit_command() ->
  blk_io_plug_call()

blk_io_plug_call() immediately invokes the given callback when the
current thread is not plugged, as is the case during nvme_file_open().

Unfortunately, nvme_submit_command() calls blk_io_plug_call() with
q->lock still held:

    ...
    q->sq.tail = (q->sq.tail + 1) % NVME_QUEUE_SIZE;
    q->need_kick++;
    blk_io_plug_call(nvme_unplug_fn, q);
    qemu_mutex_unlock(&q->lock);
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^

nvme_unplug_fn() deadlocks trying to acquire q->lock because the lock is
already acquired by the same thread. The symptom is that QEMU hangs
during startup while opening the NVMe drive.

Fix this by moving the blk_io_plug_call() outside q->lock. This is safe
because no other thread runs code related to this queue and
blk_io_plug_call()'s internal state is immune to thread safety issues
since it is thread-local.

Reported-by: Lukáš Doktor <ldoktor@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Tested-by: Lukas Doktor <ldoktor@redhat.com>
Message-id: 20230712191628.252806-1-stefanha@redhat.com
Fixes: f2e590002bd6 ("block/nvme: convert to blk_io_plug_call() API")
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

audio/pw: improve channel position code

Follow PulseAudio backend comment and code, and only implement the
channels QEMU actually supports at this point, and add the same comment
about limits and future mappings. Simplify a bit the code.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-13-marcandre.lureau@redhat.com>

audio/pw: remove wrong comment

The stream is actually created connected.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-12-marcandre.lureau@redhat.com>

audio/pw: simplify error reporting in stream creation

create_stream() now reports on all error paths.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-11-marcandre.lureau@redhat.com>

audio/pw: add more error reporting

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-10-marcandre.lureau@redhat.com>

audio/pw: factorize some common code

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-9-marcandre.lureau@redhat.com>

audio/pw: add more details on error

PipeWire uses errno to report error details.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-8-marcandre.lureau@redhat.com>

audio/pw: trace during init before calling pipewire API

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-7-marcandre.lureau@redhat.com>

audio/pw: needless check for NULL

g_clear_pointer() already checks for NULL.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-6-marcandre.lureau@redhat.com>

audio/pw: drop needless case statement

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-5-marcandre.lureau@redhat.com>

audio/pw: Pipewire->PipeWire case fix for user-visible text

"PipeWire" is the correct case.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-4-marcandre.lureau@redhat.com>

tests/lcitool: add pipewire

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230506163735.3481387-3-marcandre.lureau@redhat.com>

libvirt-ci: update submodule to cover pipewire

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230506163735.3481387-2-marcandre.lureau@redhat.com>

ui/gtk: skip refresh if new dmabuf has been submitted

Skip refresh if a new dmabuf (guest scanout frame) has already been
submitted and ready to be drawn because the scanout will be updated
with new frame anyway.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Vivek Kasireddy <vivek.kasireddy@intel.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20230706183355.29361-2-dongwon.kim@intel.com>

ui/gtk: set scanout-mode right before scheduling draw

Setting scanout mode is better to be done very last minute
right because the mode can be reset anytime after it is set in
dpy_gl_scanout_texture by any asynchronouse dpy_refresh call,
which eventually cancels drawing of the guest scanout texture.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Vivek Kasireddy <vivek.kasireddy@intel.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20230706183355.29361-1-dongwon.kim@intel.com>

virtio-gpu-udmabuf: correct naming of QemuDmaBuf size properties

Replace 'width' and 'height' in QemuDmaBuf with 'backing_widht'
and 'backing_height' as these commonly indicate the size of the
whole surface (e.g. guest's Xorg extended display). Then use
'width' and 'height' for sub region in there (e.g. guest's
scanouts).

Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Vivek Kasireddy <vivek.kasireddy@intel.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20230713040444.32267-1-dongwon.kim@intel.com>

virtio-gpu: replace the surface with null surface when resetting

The primary guest scanout shows the booting screen right after reboot
but additional guest displays (i.e. max_ouptuts > 1) will keep displaying
the old frames until the guest virtio gpu driver gets initialized, which
could cause some confusion. A better way is to to replace the surface with
a place holder that tells the display is not active during the reset of
virtio-gpu device.

And to immediately update the surface with the place holder image after
the switch, displaychangelistener_gfx_switch needs to be called with
'update == TRUE' in dpy_gfx_replace_surface when the new surface is NULL.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Vivek Kasireddy <vivek.kasireddy@intel.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20230627224451.11739-1-dongwon.kim@intel.com>

ui/gtk: Make sure the right EGL context is currently bound

Observed a wrong context is bound when changing the scanout mode.
To prevent problem, it is needed to make sure to bind the right
context in gtk_egl_set_scanout_mode/gtk_gl_area_set_scanout_mode
as well as unbind one in the end of gd_egl_update/gd_gl_area_update.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Vivek Kasireddy <vivek.kasireddy@intel.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20230628191504.17185-1-dongwon.kim@intel.com>

ui/vnc-clipboard: fix infinite loop in inflate_buffer (CVE-2023-3255)

A wrong exit condition may lead to an infinite loop when inflating a
valid zlib buffer containing some extra bytes in the `inflate_buffer`
function. The bug only occurs post-authentication. Return the buffer
immediately if the end of the compressed data has been reached
(Z_STREAM_END).

Fixes: CVE-2023-3255
Fixes: 0bf41cab ("ui/vnc: clipboard support")
Reported-by: Kevin Denis <kevin.denis@synacktiv.com>
Signed-off-by: Mauro Matteo Cascella <mcascell@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20230704084210.101822-1-mcascell@redhat.com>

virtio-gpu: fix potential divide-by-zero regression

Commit 9462ff4695aa0 ("virtio-gpu/win32: allocate shareable 2d
resources/images") introduces a division, which can lead to crashes when
"height" is 0.

Fixes: https://gitlab.com/qemu-project/qemu/-/issues/1744
Reviewed-by: Alexander Bulekov <alxndr@bu.edu>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>

hw/nvram: Avoid unnecessary Xilinx eFuse backstore write

Add a check in the bit-set operation to write the backstore
only if the affected bit is 0 before.

With this in place, there will be no need for callers to
do the checking in order to avoid unnecessary writes.

Signed-off-by: Tong Ho <tong.ho@amd.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

accel/tcg: Zero-pad PC in TCG CPU exec trace lines

In commit f0a08b0913befbd we changed the type of the PC from
target_ulong to vaddr.  In doing so we inadvertently dropped the
zero-padding on the PC in trace lines (the second item inside the []
in these lines).  They used to look like this on AArch64, for
instance:

Trace 0: 0x7f2260000100 [00000000/0000000040000000/00000061/ff200000]

and now they look like this:
Trace 0: 0x7f4f50000100 [00000000/40000000/00000061/ff200000]

and if the PC happens to be somewhere low like 0x5000
then the field is shown as /5000/.

This is because TARGET_FMT_lx is a "%08x" or "%016x" specifier,
depending on TARGET_LONG_SIZE, whereas VADDR_PRIx is just PRIx64
with no width specifier.

Restore the zero-padding by adding an 016 width specifier to
this tracing and a couple of others that were similarly recently
changed to use VADDR_PRIx without a width specifier.

We can't unfortunately restore the "32-bit guests are padded to
8 hex digits and 64-bit guests to 16 hex digits" behaviour so
easily.

Fixes: f0a08b0913befbd ("accel/tcg/cpu-exec.c: Widen pc to vaddr")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-id: 20230711165434.4123674-1-peter.maydell@linaro.org

target/arm/ptw.c: Account for FEAT_RME when applying {N}SW, SA bits

In get_phys_addr_twostage() the code that applies the effects of
VSTCR.{SA,SW} and VTCR.{NSA,NSW} only updates result->f.attrs.secure.
Now we also have f.attrs.space for FEAT_RME, we need to keep the two
in sync.

These bits only have an effect for Secure space translations, not
for Root, so use the input in_space field to determine whether to
apply them rather than the input is_secure. This doesn't actually
make a difference because Root translations are never two-stage,
but it's a little clearer.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230710152130.3928330-4-peter.maydell@linaro.org

target/arm: Fix S1_ptw_translate() debug path

In commit fe4a5472ccd6 we rearranged the logic in S1_ptw_translate()
so that the debug-access "call get_phys_addr_*" codepath is used both
when S1 is doing ptw reads from stage 2 and when it is doing ptw
reads from physical memory.  However, we didn't update the
calculation of s2ptw->in_space and s2ptw->in_secure to account for
the "ptw reads from physical memory" case.  This meant that debug
accesses when in Secure state broke.

Create a new function S2_security_space() which returns the
correct security space to use for the ptw load, and use it to
determine the correct .in_secure and .in_space fields for the
stage 2 lookup for the ptw load.

Reported-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Tested-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230710152130.3928330-3-peter.maydell@linaro.org
Fixes: fe4a5472ccd6 ("target/arm: Use get_phys_addr_with_struct in S1_ptw_translate")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm/ptw.c: Add comments to S1Translate struct fields

Add comments to the in_* fields in the S1Translate struct
that explain what they're doing.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230710152130.3928330-2-peter.maydell@linaro.org

linux-user: Remove pointless NULL check in clock_adjtime handling

In the code for TARGET_NR_clock_adjtime, we set the pointer phtx to
the address of the local variable htx.  This means it can never be
NULL, but later in the code we check it for NULL anyway.  Coverity
complains about this (CID 1507683) because the NULL check comes after
a call to clock_adjtime() that assumes it is non-NULL.

Since phtx is always &htx, and is used only in three places, it's not
really necessary.  Remove it, bringing the code structure in to line
with that for TARGET_NR_clock_adjtime64, which already uses a simple
'&htx' when it wants a pointer to 'htx'.

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: 20230623144410.1837261-1-peter.maydell@linaro.org

hw/arm/sbsa-ref: set 'slots' property of xhci

This extends the slots of xhci to 64, since the default xhci_sysbus
just supports one slot.

Signed-off-by: Wang Yuquan <wangyuquan1236@phytium.com.cn>
Signed-off-by: Chen Baozi <chenbaozi@phytium.com.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Tested-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Message-id: 20230710063750.473510-2-wangyuquan1236@phytium.com.cn
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge tag 'pull-tpm-2023-07-14-1' of https://github.com/stefanberger/qemu-tpm into staging

Merge tpm 2023/07/14 v1

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEuBi5yt+QicLVzsZrda1lgCoLQhEFAmSxasMACgkQda1lgCoL
# QhHaJwgAzOzXV9xRId1wjNLHI9HF4dkblfr12PgKsECKz7uPvcbpZOsDvu629Z8f
# a2X502Mep+6lBqq5RNUF59a5nE3laoYItaF+P+TFWz74+na7WKyFg107YyoHvCfJ
# GdjpCNyOGaXzDycAcwjqvkX5q532YBD9npedD9NzZ2UgUYWVXOYAZFhek6H9kwHN
# xlewj80pHLw3X8MM/Z1jR5hDFQa4ocCvF8F2Fnr4u0L5b+noga1WzspKTY8mcbVJ
# 3ncOTuvoWReBPweZhvSzsFbYAFb3OEeaimUHQrKVJj1L7pJtSJZeOuS+umbhvpFR
# 9baczGwrqYqpPtiDT48lTjArdJ8Cqw==
# =6Hmf
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 14 Jul 2023 04:33:23 PM BST
# gpg:                using RSA key B818B9CADF9089C2D5CEC66B75AD65802A0B4211
# gpg: Good signature from "Stefan Berger <stefanb@linux.vnet.ibm.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: B818 B9CA DF90 89C2 D5CE  C66B 75AD 6580 2A0B 4211

* tag 'pull-tpm-2023-07-14-1' of https://github.com/stefanberger/qemu-tpm:
  hw/tpm: TIS on sysbus: Remove unsupport ppi command line option

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

Merge tag 'pull-tcg-20230715' of https://gitlab.com/rth7680/qemu into staging

tcg: Use HAVE_CMPXCHG128 instead of CONFIG_CMPXCHG128
accel/tcg: Introduce page_check_range_empty
accel/tcg: Introduce page_find_range_empty
accel/tcg: Accept more page flags in page_check_range
accel/tcg: Return bool from page_check_range
accel/tcg: Always lock pages before translation
linux-user: Use abi_* types for target structures in syscall_defs.h
linux-user: Fix abi_llong alignment for microblaze and nios2
linux-user: Fix do_shmat type errors
linux-user: Implement execve without execveat
linux-user: Make sure initial brk is aligned
linux-user: Use a mask with strace flags
linux-user: Implement MAP_FIXED_NOREPLACE
linux-user: Widen target_mmap offset argument to off_t
linux-user: Use page_find_range_empty for mmap_find_vma_reserved
linux-user: Use 'last' instead of 'end' in target_mmap and subroutines
linux-user: Remove can_passthrough_madvise
linux-user: Simplify target_madvise
linux-user: Drop uint and ulong types
linux-user/arm: Do not allocate a commpage at all for M-profile CPUs
bsd-user: Use page_check_range_empty for MAP_EXCL
bsd-user: Use page_find_range_empty for mmap_find_vma_reserved

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmSypEYdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV9VzQf/RMRK4SQDJiJEbQ6K
# 5U1i955Rl4MMLT8PrkbT/UDA9soyIlSVjUenW8ThJJg6SLbSvkXZsWn165PFu+yW
# nYkeCYxkJtAjWmmFlZ44J+VLEZZ6LkWrIvPZHvKohelpi6uT/fuQaAZjKuH2prI/
# 7bdP5YdLUMpCztERHYfxmroEX4wJR6knsRpt5rYchADxEfkWk82PanneCw7grQ6V
# VNg1pRGplp0jMkpOOBvMD1ENkmoipklMe9P1gQdCHobg2/kqpozhT1oQp/gHNkP5
# 66Cjzv8o0nnPjJetm74pnP06iNhuMjDesD7f+Vq/DALgMobwjvhDW5GD+Ccto85B
# hqvwHA==
# =vm0t
# -----END PGP SIGNATURE-----
# gpg: Signature made Sat 15 Jul 2023 02:51:02 PM BST
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [ultimate]

* tag 'pull-tcg-20230715' of https://gitlab.com/rth7680/qemu: (47 commits)
  tcg: Use HAVE_CMPXCHG128 instead of CONFIG_CMPXCHG128
  accel/tcg: Always lock pages before translation
  linux-user/arm: Do not allocate a commpage at all for M-profile CPUs
  linux-user: Drop uint and ulong
  linux-user: Simplify target_madvise
  linux-user: Remove can_passthrough_madvise
  accel/tcg: Return bool from page_check_range
  accel/tcg: Accept more page flags in page_check_range
  linux-user: Simplify target_munmap
  linux-user: Rename mmap_reserve to mmap_reserve_or_unmap
  linux-user: Rewrite mmap_reserve
  linux-user: Use 'last' instead of 'end' in target_mmap
  linux-user: Use page_find_range_empty for mmap_find_vma_reserved
  bsd-user: Use page_find_range_empty for mmap_find_vma_reserved
  accel/tcg: Introduce page_find_range_empty
  linux-user: Rewrite mmap_frag
  linux-user: Rewrite target_mprotect
  linux-user: Widen target_mmap offset argument to off_t
  linux-user: Split out target_to_host_prot
  linux-user: Implement MAP_FIXED_NOREPLACE
  ...

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

tcg: Use HAVE_CMPXCHG128 instead of CONFIG_CMPXCHG128

We adjust CONFIG_ATOMIC128 and CONFIG_CMPXCHG128 with
CONFIG_ATOMIC128_OPT in atomic128.h.  It is difficult
to tell when those changes have been applied with the
ifdef we must use with CONFIG_CMPXCHG128.  So instead
use HAVE_CMPXCHG128, which triggers -Werror-undef when
the proper header has not been included.

Improves tcg_gen_atomic_cmpxchg_i128 for s390x host, which
requires CONFIG_ATOMIC128_OPT.  Without this we fall back
to EXCP_ATOMIC to single-step 128-bit atomics, which is
slow enough to cause some tests to time out.

Reported-by: Thomas Huth <thuth@redhat.com>
Tested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Always lock pages before translation

We had done this for user-mode by invoking page_protect
within the translator loop.  Extend this to handle system
mode as well.  Move page locking out of tb_link_page.

Reported-by: Liren Wei <lrwei@bupt.edu.cn>
Reported-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Richard W.M. Jones <rjones@redhat.com>

linux-user/arm: Do not allocate a commpage at all for M-profile CPUs

Since commit fbd3c4cff6 ("linux-user/arm: Mark the commpage
executable") executing bare-metal (linked with rdimon.specs)
cortex-M code fails as:

  $ qemu-arm -cpu cortex-m3 ~/hello.exe.m3
  qemu-arm: ../../accel/tcg/user-exec.c:492: page_set_flags: Assertion `last <= GUEST_ADDR_MAX' failed.
  Aborted (core dumped)

Commit 4f5c67f8df ("linux-user/arm: Take more care allocating
commpage") already took care of not allocating a commpage for
M-profile CPUs, however it had to be reverted as commit 6cda41daa2.

Re-introduce the M-profile fix from commit 4f5c67f8df.

Fixes: fbd3c4cff6 ("linux-user/arm: Mark the commpage executable")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1755
Reported-by: Christophe Lyon <christophe.lyon@linaro.org>
Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230711153408.68389-1-philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

linux-user: Drop uint and ulong

These are types not used anymore anywhere else.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: <20230511085056.13809-1-quintela@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

linux-user: Simplify target_madvise

The trivial length 0 check can be moved up, simplifying some
of the other cases.  The end < start test is handled by
guest_range_valid_untagged.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-27-richard.henderson@linaro.org>

linux-user: Remove can_passthrough_madvise

Use page_check_range instead, which uses the interval tree
instead of checking each page individually.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-26-richard.henderson@linaro.org>

accel/tcg: Return bool from page_check_range

Replace the 0/-1 result with true/false.
Invert the sense of the test of all callers.
Document the function.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-25-richard.henderson@linaro.org>

accel/tcg: Accept more page flags in page_check_range

Only PAGE_WRITE needs special attention, all others can be
handled as we do for PAGE_READ.  Adjust the mask.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230707204054.8792-24-richard.henderson@linaro.org>

linux-user: Simplify target_munmap

All of the guest to host page adjustment is handled by
mmap_reserve_or_unmap; there is no need to duplicate that.
There are no failure modes for munmap after alignment and
guest address range have been validated.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-23-richard.henderson@linaro.org>

linux-user: Rename mmap_reserve to mmap_reserve_or_unmap

If !reserved_va, munmap instead and assert success.
Update all callers.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-22-richard.henderson@linaro.org>

linux-user: Rewrite mmap_reserve

Use 'last' variables instead of 'end' variables; be careful
about avoiding overflow.  Assert that the mmap succeeded.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-21-richard.henderson@linaro.org>

linux-user: Use 'last' instead of 'end' in target_mmap

Complete the transition within the mmap functions to a formulation
that does not overflow at the end of the address space.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230707204054.8792-20-richard.henderson@linaro.org>

linux-user: Use page_find_range_empty for mmap_find_vma_reserved

Use the interval tree to find empty space, rather than
probing each page in turn.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-19-richard.henderson@linaro.org>

bsd-user: Use page_find_range_empty for mmap_find_vma_reserved

Use the interval tree to find empty space, rather than
probing each page in turn.

Cc: Warner Losh <imp@bsdimp.com>
Cc: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-bt: Warner Losh <imp@bsdimp.com>
Message-Id: <20230707204054.8792-18-richard.henderson@linaro.org>

accel/tcg: Introduce page_find_range_empty

Use the interval tree to locate an unused range in the VM.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-17-richard.henderson@linaro.org>

linux-user: Rewrite mmap_frag

Use 'last' variables instead of 'end' variables.
Always zero MAP_ANONYMOUS fragments, which we previously
failed to do if they were not writable; early exit in case
we allocate a new page from the kernel, known zeros.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-16-richard.henderson@linaro.org>

linux-user: Rewrite target_mprotect

Use 'last' variables instead of 'end' variables.
When host page size > guest page size, detect when
adjacent host pages have the same protection and
merge that expanded host range into fewer syscalls.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-15-richard.henderson@linaro.org>

linux-user: Widen target_mmap offset argument to off_t

We build with _FILE_OFFSET_BITS=64, so off_t = off64_t = uint64_t.
With an extra cast, this fixes emulation of mmap2, which could
overflow the computation of the full value of offset.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-14-richard.henderson@linaro.org>

linux-user: Split out target_to_host_prot

Split out from validate_prot_to_pageflags, as there is not
one single host_prot for the entire range.  We need to adjust
prot for every host page that overlaps multiple guest pages.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-13-richard.henderson@linaro.org>

linux-user: Implement MAP_FIXED_NOREPLACE

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-12-richard.henderson@linaro.org>

bsd-user: Use page_check_range_empty for MAP_EXCL

The previous check returned -1 when any page within
[start, start+len) is unmapped, not when all are unmapped.

Cc: Warner Losh <imp@bsdimp.com>
Cc: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Message-Id: <20230707204054.8792-11-richard.henderson@linaro.org>

accel/tcg: Introduce page_check_range_empty

Examine the interval tree to validate that a region
has no existing mappings.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-10-richard.henderson@linaro.org>

linux-user: Populate more bits in mmap_flags_tbl

Fix translation of TARGET_MAP_SHARED and TARGET_MAP_PRIVATE,
which are types not single bits.  Add TARGET_MAP_SHARED_VALIDATE,
TARGET_MAP_SYNC, TARGET_MAP_NONBLOCK, TARGET_MAP_POPULATE,
TARGET_MAP_FIXED_NOREPLACE, and TARGET_MAP_UNINITIALIZED.

Update strace to match.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-9-richard.henderson@linaro.org>

linux-user: Split TARGET_PROT_* out of syscall_defs.h

Move the values into the per-target target_mman.h headers

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-8-richard.henderson@linaro.org>

linux-user: Split TARGET_MAP_* out of syscall_defs.h

Move the values into the per-target target_mman.h headers

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-7-richard.henderson@linaro.org>

linux-user/strace: Expand struct flags to hold a mask

A zero bit value does not make sense -- it must relate to
some field in some way.

Define FLAG_BASIC with a build-time sanity check.
Adjust FLAG_GENERIC and FLAG_TARGET to use it.
Add FLAG_GENERIC_MASK and FLAG_TARGET_MASK.

Fix up the existing flag definitions for build errors.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-6-richard.henderson@linaro.org>

linux-user: Fix formatting of mmap.c

Fix all checkpatch.pl errors within mmap.c.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230707204054.8792-5-richard.henderson@linaro.org>

linux-user: Make sure initial brk(0) is page-aligned

Fixes: 86f04735ac ("linux-user: Fix brk() to release pages")
Signed-off-by: Andreas Schwab <schwab@suse.de>
Message-Id: <mvmpm55qnno.fsf@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg: Fix info_in_idx increment in layout_arg_by_ref

Off by one error, failing to take into account that layout_arg_1
already incremented info_in_idx for the first piece.  We only
need care for the n-1 TCG_CALL_ARG_BY_REF_N pieces here.

Cc: qemu-stable@nongnu.org
Fixes: 313bdea84d2 ("tcg: Add TCG_CALL_{RET,ARG}_BY_REF")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1751
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Peter Maydell <peter.maydell@linaro.org>

accel/tcg: Split out cpu_exec_longjmp_cleanup

Share the setjmp cleanup between cpu_exec_step_atomic
and cpu_exec_setjmp.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

linux-user: Fix do_shmat type errors

The guest address, raddr, should be unsigned, aka abi_ulong.
The host addresses should be cast via *intptr_t not long.
Drop the inline and fix two other whitespace issues.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230626140250.69572-1-richard.henderson@linaro.org>

linux-user/syscall: Implement execve without execveat

Support for execveat syscall was implemented in 55bbe4 and is available
since QEMU 8.0.0. It relies on host execveat, which is widely available
on most of Linux kernels today.

However, this change breaks qemu-user self emulation, if "host" qemu
version is less than 8.0.0. Indeed, it does not implement yet execveat.
This strange use case happens with most of distribution today having
binfmt support.

With a concrete failing example:
$ qemu-x86_64-7.2 qemu-x86_64-8.0 /bin/bash -c /bin/ls
/bin/bash: line 1: /bin/ls: Function not implemented
-> not implemented means execve returned ENOSYS

qemu-user-static 7.2 and 8.0 can be conveniently grabbed from debian
packages qemu-user-static* [1].

One usage of this is running wine-arm64 from linux-x64 (details [2]).
This is by updating qemu embedded in docker image that we ran into this
issue.

The solution to update host qemu is not always possible. Either it's
complicated or ask you to recompile it, or simply is not accessible
(GitLab CI, GitHub Actions). Thus, it could be worth to implement execve
without relying on execveat, which is the goal of this patch.

This patch was tested with example presented in this commit message.

[1] http://ftp.us.debian.org/debian/pool/main/q/qemu/
[1] https://www.linaro.org/blog/emulate-windows-on-arm/

Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Message-Id: <20230705121023.973284-1-pierrick.bouvier@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

include/exec/user: Set ABI_LLONG_ALIGNMENT to 4 for nios2

Based on gcc's nios2.h setting BIGGEST_ALIGNMENT to 32 bits.

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

include/exec/user: Set ABI_LLONG_ALIGNMENT to 4 for microblaze

Based on gcc's microblaze.h setting BIGGEST_ALIGNMENT to 32 bits.

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

linux-user: Use abi_uint not unsigned in syscall_defs.h

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

linux-user: Use abi_short not short in syscall_defs.h

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