target/i386: move scalar 0F 38 and 0F 3A instruction to new decoder

Because these are the only VEX instructions that QEMU supports, the
new decoder is entered on the first byte of a valid VEX prefix, and VEX
decoding only needs to be done in decode-new.c.inc.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: validate SSE prefixes directly in the decoding table

Many SSE and AVX instructions are only valid with specific prefixes
(none, 66, F3, F2).  Introduce a direct way to encode this in the
decoding table to avoid using decode groups too much.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: validate VEX prefixes via the instructions' exception classes

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: add AVX_EN hflag

Add a new hflag bit to determine whether AVX instructions are allowed

Signed-off-by: Paul Brook <paul@nowt.org>
Message-Id: <20220424220204.2493824-4-paul@nowt.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: add CPUID feature checks to new decoder

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: add CPUID[EAX=7,ECX=0].ECX to DisasContext

TCG will shortly implement VAES instructions, so add the relevant feature
word to the DisasContext.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: add ALU load/writeback core

Add generic code generation that takes care of preparing operands
around calls to decode.e.gen in a table-driven manner, so that ALU
operations need not take care of that.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: add core of new i386 decoder

The new decoder is based on three principles:

- use mostly table-driven decoding, using tables derived as much as possible
  from the Intel manual.  Centralizing the decode the operands makes it
  more homogeneous, for example all immediates are signed.  All modrm
  handling is in one function, and can be shared between SSE and ALU
  instructions (including XMM<->GPR instructions).  The SSE/AVX decoder
  will also not have duplicated code between the 0F, 0F38 and 0F3A tables.

- keep the code as "non-branchy" as possible.  Generally, the code for
  the new decoder is more verbose, but the control flow is simpler.
  Conditionals are not nested and have small bodies.  All instruction
  groups are resolved even before operands are decoded, and code
  generation is separated as much as possible within small functions
  that only handle one instruction each.

- keep address generation and (for ALU operands) memory loads and writeback
  as much in common code as possible.  All ALU operations for example
  are implemented as T0=f(T0,T1).  For non-ALU instructions,
  read-modify-write memory operations are rare, but registers do not
  have TCGv equivalents: therefore, the common logic sets up pointer
  temporaries with the operands, while load and writeback are handled
  by gvec or by helpers.

These principles make future code review and extensibility simpler, at
the cost of having a relatively large amount of code in the form of this
patch.  Even EVEX should not be _too_ hard to implement (it's just a crazy
large amount of possibilities).

This patch introduces the main decoder flow, and integrates the old
decoder with the new one.  The old decoder takes care of parsing
prefixes and then optionally drops to the new one.  The changes to the
old decoder are minimal and allow it to be replaced incrementally with
the new one.

There is a debugging mechanism through a "LIMIT" environment variable.
In user-mode emulation, the variable is the number of instructions
decoded by the new decoder before permanently switching to the old one.
In system emulation, the variable is the highest opcode that is decoded
by the new decoder (this is less friendly, but it's the best that can
be done without requiring deterministic execution).

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: make rex_w available even in 32-bit mode

REX.W can be used even in 32-bit mode by AVX instructions, where it is retroactively
renamed to VEX.W.  Make the field available even in 32-bit mode but keep the REX_W()
macro as it was; this way, that the handling of dflag does not use it by mistake and
the AVX code more clearly points at the special VEX behavior of the bit.

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

target/i386: make ldo/sto operations consistent with ldq

ldq takes a pointer to the first byte to load the 64-bit word in;
ldo takes a pointer to the first byte of the ZMMReg.  Make them
consistent, which will be useful in the new SSE decoder's
load/writeback routines.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Define XMMReg and access macros, align ZMM registers

This will be used for emission and endian adjustments of gvec operations.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822223722.1697758-2-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Use probe_access_full for final stage2 translation

Rather than recurse directly on mmu_translate, go through the
same softmmu lookup that we did for the page table walk.
This centralizes all knowledge of MMU_NESTED_IDX, with respect
to setup of TranslationParams, to get_physical_address.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-10-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Use atomic operations for pte updates

Use probe_access_full in order to resolve to a host address,
which then lets us use a host cmpxchg to update the pte.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/279
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-9-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Combine 5 sets of variables in mmu_translate

We don't need one variable set per translation level,
which requires copying into pte/pte_addr for huge pages.
Standardize on pte/pte_addr for all levels.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-8-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Use MMU_NESTED_IDX for vmload/vmsave

Use MMU_NESTED_IDX for each memory access, rather than
just a single translation to physical.  Adjust svm_save_seg
and svm_load_seg to pass in mmu_idx.

This removes the last use of get_hphys so remove it.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-7-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Add MMU_PHYS_IDX and MMU_NESTED_IDX

These new mmu indexes will be helpful for improving
paging and code throughout the target.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-6-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Reorg GET_HPHYS

Replace with PTE_HPHYS for the page table walk, and a direct call
to mmu_translate for the final stage2 translation.  Hoist the check
for HF2_NPT_MASK out to get_physical_address, which avoids the
recursive call when stage2 is disabled.

We can now return all the way out to x86_cpu_tlb_fill before raising
an exception, which means probe works.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-5-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Introduce structures for mmu_translate

Create TranslateParams for inputs, TranslateResults for successful
outputs, and TranslateFault for error outputs; return true on success.

Move stage1 error paths from handle_mmu_fault to x86_cpu_tlb_fill;
reorg the rest of handle_mmu_fault into get_physical_address.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-4-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Direct call get_hphys from mmu_translate

Use a boolean to control the call to get_hphys instead
of passing a null function pointer.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-3-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Use MMUAccessType across excp_helper.c

Replace int is_write1 and magic numbers with the proper
MMUAccessType access_type and enumerators.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221002172956.265735-2-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Save and restore pc_save before tcg_remove_ops_after

Restore pc_save while undoing any state change that may have
happened while decoding the instruction.  Leave a TODO about
removing all of that when the table-based decoder is complete.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221016222303.288551-1-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Use device_cold_reset() to reset the APIC

The semantic difference between the deprecated device_legacy_reset()
function and the newer device_cold_reset() function is that the new
function resets both the device itself and any qbuses it owns,
whereas the legacy function resets just the device itself and nothing
else.

The x86_cpu_after_reset() function uses device_legacy_reset() to reset
the APIC; this is an APICCommonState and does not have any qbuses, so
for this purpose the two functions behave identically and we can stop
using the deprecated one.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <20221013171926.1447899-1-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: Avoid using strings binary

When determining the endiandness of the target architecture we're
building for a small program is compiled, which in an obfuscated
way declares two strings. Then, we look which string is in
correct order (using strings binary) and deduct the endiandness.
But using the strings binary is problematic, because it's part of
toolchain (strings is just a symlink to
x86_64-pc-linux-gnu-strings or llvm-strings). And when
(cross-)compiling, it requires users to set the symlink to the
correct toolchain.

Fortunately, we have a better alternative anyways. We can mimic
what compiler.h is already doing: comparing __BYTE_ORDER__
against values for little/big endiandness.

Bug: https://bugs.gentoo.org/876933
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Message-Id: <d6d9c7043cfe6d976d96694f2b4ecf85cf3206f1.1665732504.git.mprivozn@redhat.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hyperv: fix SynIC SINT assertion failure on guest reset

Resetting a guest that has Hyper-V VMBus support enabled triggers a QEMU
assertion failure:
hw/hyperv/hyperv.c:131: synic_reset: Assertion `QLIST_EMPTY(&synic->sint_routes)' failed.

This happens both on normal guest reboot or when using "system_reset" HMP
command.

The failing assertion was introduced by commit 64ddecc88bcf ("hyperv: SControl is optional to enable SynIc")
to catch dangling SINT routes on SynIC reset.

The root cause of this problem is that the SynIC itself is reset before
devices using SINT routes have chance to clean up these routes.

Since there seems to be no existing mechanism to force reset callbacks (or
methods) to be executed in specific order let's use a similar method that
is already used to reset another interrupt controller (APIC) after devices
have been reset - by invoking the SynIC reset from the machine reset
handler via a new x86_cpu_after_reset() function co-located with
the existing x86_cpu_reset() in target/i386/cpu.c.
Opportunistically move the APIC reset handler there, too.

Fixes: 64ddecc88bcf ("hyperv: SControl is optional to enable SynIc") # exposed the bug
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Message-Id: <cb57cee2e29b20d06f81dce054cbcea8b5d497e8.1664552976.git.maciej.szmigiero@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/scsi/vmw_pvscsi.c: Use device_cold_reset() to reset SCSI devices

Currently the vwm_pvscsi controller resets individual SCSI devices
with the device_legacy_reset() function.  The only difference between
this and device_cold_reset() is that device_legacy_reset() resets the
device but not any child qbuses it might have.

In this case, no SCSI device has a child qbus, so the functions have
the same behaviour.  Switch to device_cold_reset() to move away from
using the deprecated function, and bring this SCSI controller in to
line with what all the others do.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20221013160623.1296109-3-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

scsi: Use device_cold_reset() and bus_cold_reset()

In the SCSI subsystem we currently use the legacy functions
qdev_reset_all() and qbus_reset_all().  These perform a recursive
reset, starting from either a qbus or a qdev.  However they do not
permit any of the devices in the tree to use three-phase reset,
because device reset goes through the device_legacy_reset() function
that only calls the single DeviceClass::reset method.

Switch to using the device_cold_reset() and bus_cold_reset()
functions.  These also perform a recursive reset, where first the
children are reset and then finally the parent, but they use the new
(...in 2020...) Resettable mechanism, which supports both the old
style single-reset method and also the new 3-phase reset handling.

Since no devices attached to SCSI buses currently try to use 3-phase
reset, this should be a no-behaviour-change commit which just reduces
the use of a deprecated API.

Commit created with:
  sed -i -e 's/qdev_reset_all/device_cold_reset/g;s/qbus_reset_all/bus_cold_reset/g' hw/scsi/*.c

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20221013160623.1296109-2-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: don't enable firmware for targets that are not built

This avoids the unfortunate effect of building pc-bios blobs
even for targets the user isn't interested in.

Due to the bi-arch nature of x86 and PPC firmware, check for the
desired target by hand, and don't just look for the compilation target
in $target_list.

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

virtio-scsi: Send "REPORTED LUNS CHANGED" sense data upon disk hotplug events

Section 5.6.6.3 of VirtIO specification states, "Events will also
be reported via sense codes..." However, no sense data is sent when
VIRTIO_SCSI_EVT_RESET_RESCAN or VIRTIO_SCSI_EVT_RESET_REMOVED events
are reported (when disk hotplug/hotunplug events occur). SCSI layer
on Solaris depends on this sense data, and hence does not handle disk
hotplug/hotunplug events.

When the disk inventory changes, use the bus unit attention mechanism
to return a CHECK_CONDITION status with sense data of 0x06/0x3F/0x0E
(sense code REPORTED_LUNS_CHANGED).  The first device on the bus to
execute a command successfully will report and consume the unit
attention status.

Signed-off-by: Venu Busireddy <venu.busireddy@oracle.com>
Message-Id: <20221006194946.24134-1-venu.busireddy@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge tag 'kraxel-20221013-pull-request' of https://gitlab.com/kraxel/qemu into staging

pci: cleanup virtio ids.
audio: bugfixes and latency improvements.
misc fixes for hw/display and ui

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCgAdFiEEoDKM/7k6F6eZAf59TLbY7tPocTgFAmNHtYsACgkQTLbY7tPo
# cTjHhg//RDkHbqVSExe+Odw5ISuLu/EXZSHAVjo3KOCUvaj7O2cXi8N7DVfEy5a5
# T3+WSv0v4X6TYSV0PoMb36a11rCuOKzeLZrtEOQeYfG3D1WCVc9gIWMt6omzBC7A
# YQ59P+u19qHD7xD2PP3WRtdcqmsceg1RG+47adX2EnsRZmmu/yJxD72w/Q1kXMuB
# jIzuJU2ZVorYX9y11hnIU3M5pvoX/vjFA+Ib2UGZZdlE3KlUKtJeAtLiZkHfoyd1
# 5janU+PtSU6Z1yVirE7RVz3+IBbfqqEFTkDtMXJucJW/Eod0NHCyo4Q6D64HoiZe
# +JZKkHmuvn8ZUgXMtIOZdH+aOHlaIJzA5SoA2IFxCBVuxn7p4NtPbCRoHHg7gkDh
# BDsq+p/wsdOY06u1txFw9dYy+4tKvWS7+Dxhyme7GT2YUQHrEEG3pzGFmk3PE0Vi
# tEAhmfNRxWzUgIcynQiN/3SnShAI8lANq0SEiiTvqcX7h1TK+cjEYjOTMsjK43nL
# 2W/pgQxJpEPcSs3jgFLnBLk9rUHRNRC+GtMBlwN+Wdc1y17leZHiIinqhHjXuts3
# cJTdv4veeGuJENPIl2rk5JOdvpVtzduDkz+Rzx0mGb+LnAYdK2lBUV5LY9FfdwaK
# 2Bgg02ZYNBz7K2zzFeeV+7b7K/LYOuWkGdzGvKbpqjbefopZmTM=
# =6d/F
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 13 Oct 2022 02:51:55 EDT
# gpg:                using RSA key A0328CFFB93A17A79901FE7D4CB6D8EED3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>" [full]
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>" [full]
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>" [full]
# Primary key fingerprint: A032 8CFF B93A 17A7 9901  FE7D 4CB6 D8EE D3E8 7138

* tag 'kraxel-20221013-pull-request' of https://gitlab.com/kraxel/qemu: (26 commits)
  audio: improve out.voices test
  audio: fix in.voices test
  gtk: Add show_menubar=on|off command line option.
  qemu-edid: Restrict input parameter -d to avoid division by zero
  ui/gtk: Fix the implicit mouse ungrabbing logic
  pci-ids: document modern virtio-pci ids in pci.h too
  pci-ids: drop list of modern virtio devices
  pci-ids: drop PCI_DEVICE_ID_VIRTIO_PMEM
  pci-ids: drop PCI_DEVICE_ID_VIRTIO_MEM
  pci-ids: drop PCI_DEVICE_ID_VIRTIO_IOMMU
  docs: add firmware feature flags
  cirrus_vga: fix potential memory overflow
  ui/gtk-egl: egl context needs to be unbound in the end of gd_egl_switch
  ui/vnc-clipboard: fix integer underflow in vnc_client_cut_text_ext
  audio: prevent an integer overflow in resampling code
  audio: fix sw->buf size for audio recording
  audio: refactor audio_get_avail()
  audio: rename audio_sw_bytes_free()
  audio: swap audio_rate_get_bytes() function parameters
  spiceaudio: update comment
  ...

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

Merge tag 'win32-pull-request' of https://gitlab.com/marcandre.lureau/qemu into staging

win32-related misc patches

# -----BEGIN PGP SIGNATURE-----
#
# iQJPBAABCAA6FiEEh6m9kz+HxgbSdvYt2ujhCXWWnOUFAmNG488cHG1hcmNhbmRy
# ZS5sdXJlYXVAcmVkaGF0LmNvbQAKCRDa6OEJdZac5eQTD/j/rEcONwL4gZn/Rcp8
# aJlr39GEHo0JxBAF3eoxCLJlebPcdaUQ4pu/FTegS1A4abPaajDH7rdtcA58ciAG
# rCQjUOrobHzxmI9XaTIPT4PQh3DA4HB58rTpAvb/6P/UDRc0MpkcvaOkGlJVhi+7
# WB63+gnQOBEjcieNcQtmRwYRkx7K5/9G4qEESl0i2E+SE4DM+/vcVa7lfqEZ+6HS
# bsDy2BslxtPFmHj1UElwXjTbCs4Y7pfTFd+9z8ySsGL1Komf45MZs0iS4FmZLqL/
# 7Cuj+xRWibnPN9jnAc+Sdua3FAFZbqmfPQaH6DN6SICZ6Txf2hxFkAgTahagcxYX
# 9EiKGHZzI4L3l/YAxFg9RfK+AsF44ZLPId58AVvUnG1jWwxl3nRaTmvtvHaEwJuZ
# PgnbAdsNzQAJjLnk8ndpTq4mQFM+9/mrQo+iaOCwmB5s07woyEq+L+KJHMUgyk2D
# lECn3vlqVGGb6GA6MS5gSXh0TDRxPxLyr9ofIG5i5YaTo4nH56S80tHrzZMUYNKD
# xe2yUrEZ7UjeV4/6M19xdw3haPOdrG3BoBshb61vI1bF/4iQxYNo8AxptCRhzNNM
# 5Jrn/gyt47SEgMYpGIvHa/qo1lQiLsQAVKAK3O2QWd5T58V6J1a804zhTuT7T45O
# kZS2c8XEdAiBtUAkYNgFxwGM
# =Lpqm
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 12 Oct 2022 11:57:03 EDT
# 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]
# Primary key fingerprint: 87A9 BD93 3F87 C606 D276  F62D DAE8 E109 7596 9CE5

* tag 'win32-pull-request' of https://gitlab.com/marcandre.lureau/qemu:
  tests/unit: make test-io-channel-command work on win32
  io/command: implement support for win32
  io/command: use glib GSpawn, instead of open-coding fork/exec
  tests/channel-helper: set blocking in main thread
  util: make do_send_recv work with partial send/recv
  osdep: make readv_writev() work with partial read/write
  win32: set threads name

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

Merge tag 'pull-request-2022-10-12' of https://gitlab.com/thuth/qemu into staging

* Rework of qtests to support hot plugging tests on q35
* New VNC qtest
* Fixes related to temporary file handling in the tests
* Use signal() instead of sigaction() since the latter does not work on Windows
* Some other small clean-ups

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmNGz1URHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbXwxxAAoh9CeYKLyPFPe71XGsSWW4mQDkny+4CA
# AFBfcEchPSyOoTJwuSHS5cL27KkL1Wy7wTSlWdcQ2Z4LA8hexopfPtCVlizH8vQh
# hHVP5KLDY7WA7bsuRznrjUjXtjLL9mExuPm3QZ+YxH/3rNhISvq1GfC9Z5PQ/zNp
# GagCW+WiFyEYyhRHJnVVTJfZV0NCbJp9KCCJ6n/Rag80JRQiouGQsKLfRoRhvkGC
# 13dWq2NeDq9xE1k1ThUZdIpUxJO4cA+9bh6LzGOIcHX8akfkryh3NGCWuojX3XDF
# 2CRyyblfjMIp5HUcKNOrQzY1oGRn8y5537ycUkEsropSMdAccIqNwlcWO7whLEGn
# 7nbogrPt8Orezytl5ZM3WopWlge5KA2vxOUURx487LnZC2/g7l2unrCDv0c3PHDb
# U2fD5do7ZtPOl9Uz8QkYPo5CzEDnKPmPv1txOMBRcBZsvmXPZMxRWwsYavB/RloG
# nI9IvG3ZW6ivKwoPPceJberL8Ndq+pqn7y4CNNAAunHSjJ2H0YS1m1F/HfVWKJYJ
# uStXy3IZ5Qo8E+IRbPYyM68yBgrnaShuAvTSax8tWjhHOR8HetrzPABMrWylCrdo
# gaeSCEZmqf2zIYHQ2q8c7U/YrHgFNHlLz9yCyM6rZtU9dOAe52PG2TlRQomlGP76
# NKnti3RJWG4=
# =4f+b
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 12 Oct 2022 10:29:41 EDT
# 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-2022-10-12' of https://gitlab.com/thuth/qemu:
  tests/unit/test-image-locking: Fix handling of temporary files
  tests/qtest: libqtest: Install signal handler via signal()
  tests/qtest: migration-test: Avoid using hardcoded /tmp
  qtest: start a VNC test
  tests/avocado: Add missing require_netdev('user') checks
  tests/x86: Add 'q35' machine type to ivshmem-test
  tests/x86: Add 'q35' machine type to drive_del-test
  tests/x86: replace snprint() by g_strdup_printf() in drive_del-test
  tests/x86: Fix comment typo in drive_del-test
  tests/x86: Add 'q35' machine type to hotplug hd-geo-test
  tests/x86: Add 'q35' machine type to override-tests in hd-geo-test
  tests/x86: Refactor hot unplug hd-geo-test
  tests/x86: Add subtest with 'q35' machine type to device-plug-test
  tests/x86: add helper qtest_qmp_device_del_send()
  tests/migration: remove the unused local variable
  qtest: "-display none" is set in qtest_init()

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

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

* scsi-disk: support setting CD-ROM block size via device options
* target/i386: Implement MSR_CORE_THREAD_COUNT MSR
* target/i386: notify VM exit support
* target/i386: PC-relative translation block support
* target/i386: support for XSAVE state in signal frames (linux-user)

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmNFKP4UHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroNJnwgAgCcOOxmY4Qem0Gd1L+SJKpEtGMOd
# 4LY7443vT36pMpvqFNSfp5GBjDT1MgTD8BIY28miLMq959LT89LyM9g/H7IKOT82
# uyCsW3jW+6F19EZVkNvzTt+3USn/kaHn50zA4Ss9kvdNZr31b2LYqtglVCznfZwH
# oI1rDhvsXubq8oWvwkqH7IwduK8mw+EB5Yz7AjYQ6eiYjenTrQBObpwQNbb4rlUf
# oRm8dk/YJ2gfI2HQkoznGEbgpngy2tIU1vHNEpIk5NpwXxrulOyui3+sWaG4pH8f
# oAOrSDC23M5A6jBJJAzDJ1q6M677U/kwJypyGQ7IyvyhECXE3tR+lHX1eA==
# =tqeJ
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 11 Oct 2022 04:27:42 EDT
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [full]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>" [full]
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* tag 'for-upstream' of https://gitlab.com/bonzini/qemu: (37 commits)
  linux-user: i386/signal: support XSAVE/XRSTOR for signal frame fpstate
  linux-user: i386/signal: support FXSAVE fpstate on 32-bit emulation
  linux-user: i386/signal: move fpstate at the end of the 32-bit frames
  KVM: x86: Implement MSR_CORE_THREAD_COUNT MSR
  i386: kvm: Add support for MSR filtering
  x86: Implement MSR_CORE_THREAD_COUNT MSR
  target/i386: Enable TARGET_TB_PCREL
  target/i386: Inline gen_jmp_im
  target/i386: Add cpu_eip
  target/i386: Create eip_cur_tl
  target/i386: Merge gen_jmp_tb and gen_goto_tb into gen_jmp_rel
  target/i386: Remove MemOp argument to gen_op_j*_ecx
  target/i386: Use gen_jmp_rel for DISAS_TOO_MANY
  target/i386: Use gen_jmp_rel for gen_jcc
  target/i386: Use gen_jmp_rel for loop, repz, jecxz insns
  target/i386: Create gen_jmp_rel
  target/i386: Use DISAS_TOO_MANY to exit after gen_io_start
  target/i386: Create eip_next_*
  target/i386: Truncate values for lcall_real to i32
  target/i386: Introduce DISAS_JUMP
  ...

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

Merge tag 'for_upstream' of https://git./virt/kvm/mst/qemu into staging

pc,virtio: features, tests, fixes, cleanups

virtio introspection
new serial number opton for cxl
vhost user blk dynamic config size
virtio-gpio vhost user backend

Tests fixes cleanups all over the place

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# -----BEGIN PGP SIGNATURE-----
#
# iQFDBAABCAAtFiEEXQn9CHHI+FuUyooNKB8NuNKNVGkFAmNEVeoPHG1zdEByZWRo
# YXQuY29tAAoJECgfDbjSjVRph8YH+gMWpb5IglE0Q+H2JiQPBwB/Ghy1ohRKnOvZ
# lChB7+oy18o2xXRFTOXwG9Ijqsbdn0QMbU/r3NWxBuMzxDow012xiMDniJlJmcXw
# /4POOCSTKrIfzVBhsEErVSA9NwSE5cQKr1oiRBGIa9UdZfZ//v7s6SoP4vtyj8RZ
# UJVYVnMDtq/0PaN92IMs06lhqo/LkegE7gTGHMBf8Nvw4SgQoZgfPyp1eR+dKOhz
# lXNqqvTds9yt8yS65UWbuSrZ9d7GpCQf8nuyLaLaENHd6FQUVfmTTT37l2EKziwp
# PK0EwWMHeGkj7LHrylztradhE9xBlIW23ROP8wPdGZHmgLNHbC0=
# =20Zb
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 10 Oct 2022 13:27:06 EDT
# gpg:                using RSA key 5D09FD0871C8F85B94CA8A0D281F0DB8D28D5469
# gpg:                issuer "mst@redhat.com"
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>" [full]
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>" [full]
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* tag 'for_upstream' of https://git.kernel.org/pub/scm/virt/kvm/mst/qemu: (55 commits)
  x86: pci: acpi: consolidate PCI slots creation
  tests: acpi: update expected blobs
  x86: pci: acpi: reorder Device's _DSM method
  tests: acpi: whitelist pc/q35 DSDT before moving _ADR field
  tests: acpi: update expected blobs
  x86: pci: acpi: reorder Device's _ADR and _SUN fields
  tests: acpi: whitelist pc/q35 DSDT before moving _ADR field
  tests: acpi: update expected blobs
  x86: acpi: cleanup PCI device _DSM duplication
  tests: acpi: whitelist pc/q35 DSDT before switching _DSM to use ASUN
  tests: acpi: update expected blobs
  x86: acpi: _DSM: use Package to pass parameters
  acpi: x86: refactor PDSM method to reduce nesting
  tests: acpi: whitelist pc/q35 DSDT due to HPET AML move
  tests: acpi: update expected blobs after HPET move
  acpi: x86: deduplicate HPET AML building
  tests: acpi: whitelist pc/q35 DSDT due to HPET AML move
  hw/smbios: support for type 8 (port connector)
  pci: Sanity check mask argument to pci_set_*_by_mask()
  pci: Remove unused pci_get_*_by_mask() functions
  ...

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

Merge tag 'pull-target-arm-20221010' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * Retry KVM_CREATE_VM call if it fails EINTR
 * allow setting SCR_EL3.EnTP2 when FEAT_SME is implemented
 * docs/nuvoton: Update URL for images
 * refactoring of page table walk code
 * hw/arm/boot: set CPTR_EL3.ESM and SCR_EL3.EnTP2 when booting Linux with EL3
 * Don't allow guest to use unimplemented granule sizes
 * Report FEAT_GTG support

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmNEK54ZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3kzHD/9StYmulAf0iwe1ZNp6NavK
# CioOgZi6XyZl4rS2DrCf6/IO5XRFJP68byZd4Po554r2jcPc149yTuQAn4wb7d5e
# kejMZRQeWsXdxschhoVzDp9fgfzyZBn9X+gbdEZFFPWzOHMyWuu4cTok0dAKQvQY
# tZDLGmKeTv4MRUFJCri0310Sq0T0v/nAX/AyFtpvIr2SBx7DVCWYY02s5R4Yy5+M
# ntDWb0j12r78/bPwI1ll+g19JXUV5Tfh9AsbcYjKv45kdftz/Xc8fBiSiEpxyMrF
# mnVrr3kesZHOYAnOr2K1MnwsF0vU41kRg7kMRqSnu7pZXlI/8tmRyXoPR3c2aDbW
# Q5HWtsA48j2h0CJ0ESzl5SQnl3TSPa94m/HmpRSBFrYkU727QgnWDhUmBb4n54xs
# 9iBJDhcKGZLq68CB2+j6ENdRNTndolr14OwwEns0lbkoiCKUOQY3AigtZJQGRBGM
# J5r3ED7jfTWpvP6vpp5X484fK6KVprSMxsRFDkmiwhbb3J+WtKLxbSlgsWIrkZ7s
# +JgTGfGB8sD9hJVuFZYyPQb/XWP8Bb8jfgsLsTu1vW9Xs1ASrLimFYdRO3hhwSg3
# c5yubz6Vu9GB/JYh7hGprlMD5Yv48AA3if70hOu2d4P8A4OitavT7o+4Thwqjhds
# cSV1RsBJ8ha6L3CziZaKrQ==
# =s+1f
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 10 Oct 2022 10:26:38 EDT
# 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]
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* tag 'pull-target-arm-20221010' of https://git.linaro.org/people/pmaydell/qemu-arm: (28 commits)
  docs/system/arm/emulation.rst: Report FEAT_GTG support
  target/arm: Use ARMGranuleSize in ARMVAParameters
  target/arm: Don't allow guest to use unimplemented granule sizes
  hw/arm/boot: set CPTR_EL3.ESM and SCR_EL3.EnTP2 when booting Linux with EL3
  target/arm: Use tlb_set_page_full
  target/arm: Fix cacheattr in get_phys_addr_disabled
  target/arm: Split out get_phys_addr_disabled
  target/arm: Fix ATS12NSO* from S PL1
  target/arm: Pass HCR to attribute subroutines.
  target/arm: Remove env argument from combined_attrs_fwb
  target/arm: Hoist read of *is_secure in S1_ptw_translate
  target/arm: Introduce arm_hcr_el2_eff_secstate
  target/arm: Drop secure check for HCR.TGE vs SCTLR_EL1.M
  target/arm: Reorg regime_translation_disabled
  target/arm: Fold secure and non-secure a-profile mmu indexes
  target/arm: Add is_secure parameter to do_ats_write
  target/arm: Merge regime_is_secure into get_phys_addr
  target/arm: Add TBFLAG_M32.SECURE
  target/arm: Add is_secure parameter to v7m_read_half_insn
  target/arm: Split out get_phys_addr_with_secure
  ...

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

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

Block layer patches

- job: replace AioContext lock with job_mutex
- Fixes to make coroutine_fn annotations more accurate
- QAPI schema: Fix incorrect example
- Code cleanup

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEE3D3rFZqa+V09dFb+fwmycsiPL9YFAmNAAz8RHGt3b2xmQHJl
# ZGhhdC5jb20ACgkQfwmycsiPL9a6zg//QYLx+FYMStb50lS+6VBio8AKOVbwn5zp
# ZANoXinMknnxI5wTldjkkM1cBRg27BVjpOHz4XemBtQgT5nBqWq8+Ov31lwASVID
# na/L9o4Pa0xmywM777K+edceWk0fpJTLmnFf1Qxan9qB/VSjNFtk+fjwFopoatKg
# XbHd6maQtrY8bIOyBsBoZozNaS39E/uPqkP67V6GF09re17f0PBctGHKFkTKZr8w
# 2HfyMt8/UIhFet++NFgxppTcvIKfZ20pk4AQ+yYsL+FxWr/cs4leKWl5BSc7thtP
# Sm/y0WiEB4nPNo4CSf9sA1Vo8EIGYzBhUVteqYQUF2vSXSzFmZb191fLJRYwp5bQ
# QxEmHzPVGqcUHr+jkfXI0yLolWduiKV1ATZ0zW3N41VfzGLYZdSgI2ZhbHJ0/yKO
# ZhyC63gye9V6TXxviYIz2V6iOD8QuwJ8X1P0E3yRsGploF1UY/N1lwbmek1XhFn/
# +xn/mrTeV0lu4wKuWRpUfY2C/7SR0Za6MB2GqduRWnbcAonLH3/syAxXSfu2611N
# Z1Cf9Wu8Mm0IQz0LbbVvEJZ4yoEPkg/tGH8q6dpau2uTfCb6sSylRxLcXEa5R0UQ
# W+wX5GSoTDe4DQKOSaJE7jWV/QwY5diTLHBIvSF8uKAfeCenkDDLowrMvbWafL0X
# XTFzpZ/1aA8=
# =jMFT
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 07 Oct 2022 06:45:19 EDT
# gpg:                using RSA key DC3DEB159A9AF95D3D7456FE7F09B272C88F2FD6
# gpg:                issuer "kwolf@redhat.com"
# gpg: Good signature from "Kevin Wolf <kwolf@redhat.com>" [full]
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74  56FE 7F09 B272 C88F 2FD6

* tag 'for-upstream' of git://repo.or.cz/qemu/kevin: (50 commits)
  file-posix: Remove unused s->discard_zeroes
  job: remove unused functions
  blockjob: remove unused functions
  block_job_query: remove atomic read
  job.c: enable job lock/unlock and remove Aiocontext locks
  job.h: categorize JobDriver callbacks that need the AioContext lock
  blockjob: protect iostatus field in BlockJob struct
  blockjob: rename notifier callbacks as _locked
  blockjob.h: categorize fields in struct BlockJob
  jobs: protect job.aio_context with BQL and job_mutex
  job: detect change of aiocontext within job coroutine
  jobs: group together API calls under the same job lock
  block/mirror.c: use of job helpers in drivers
  jobs: use job locks also in the unit tests
  jobs: add job lock in find_* functions
  blockjob: introduce block_job _locked() APIs
  job: move and update comments from blockjob.c
  job.c: add job_lock/unlock while keeping job.h intact
  aio-wait.h: introduce AIO_WAIT_WHILE_UNLOCKED
  job.c: API functions not used outside should be static
  ...

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

audio: improve out.voices test

Improve readability of audio out.voices test:
If 1 is logged and set after positive test, 1 should be tested.

Signed-off-by: Helge Konetzka <hk@zapateado.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221012114925.5084-3-hk@zapateado.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: fix in.voices test

Calling qemu with valid -audiodev ...,in.voices=0 results in an obsolete
warning:
  audio: Bogus number of capture voices 0, setting to 0
This patch fixes the in.voices test.

Signed-off-by: Helge Konetzka <hk@zapateado.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221012114925.5084-2-hk@zapateado.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

tests/unit: make test-io-channel-command work on win32

This has been tested under msys2 & windows 11. I haven't tried to make
it work with other environments yet, but that should be enough to
validate the channel-command implementation anyway.

Here are the changes:
- drop tests/ from fifo/pipe path, to avoid directory issues
- use g_find_program() to lookup the socat executable (otherwise we
would need to change ChanneCommand to use G_SPAWN_SEARCH_PATH, and deal
with missing socat differently)
- skip the "echo" test when socat is missing as well

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20221006113657.2656108-7-marcandre.lureau@redhat.com>

io/command: implement support for win32

The initial implementation was changing the pipe state created by GLib
to PIPE_NOWAIT, but it turns out it doesn't work (read/write returns an
error). Since reading may return less than the requested amount, it
seems to be non-blocking already. However, the IO operation may block
until the FD is ready, I can't find good sources of information, to be
safe we can just poll for readiness before.

Alternatively, we could setup the FDs ourself, and use UNIX sockets on
Windows, which can be used in blocking/non-blocking mode. I haven't
tried it, as I am not sure it is necessary.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20221006113657.2656108-6-marcandre.lureau@redhat.com>

io/command: use glib GSpawn, instead of open-coding fork/exec

Simplify qio_channel_command_new_spawn() with GSpawn API. This will
allow to build for WIN32 in the following patches.

As pointed out by Daniel Berrangé: there is a change in semantics here
too. The current code only touches stdin/stdout/stderr. Any other FDs
which do NOT have O_CLOEXEC set will be inherited. With the new code,
all FDs except stdin/out/err will be explicitly closed, because we don't
set the flag G_SPAWN_LEAVE_DESCRIPTORS_OPEN. The only place we use
QIOChannelCommand today is the migration exec: protocol, and that is
only declared to use stdin/stdout.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221006113657.2656108-5-marcandre.lureau@redhat.com>

tests/channel-helper: set blocking in main thread

The /io/channel/command/echo tests run the reader side and the writer
side with the same underlying command channel. Setting the blocking mode
of the fd/handles while the other end is already reading/writing may
create issues (deadlock in win32 when earlier attempt of this series
were using SetNamedPipeHandleState). Let's just do it before spawning
the threads to avoid further concurrency issues.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20221006113657.2656108-4-marcandre.lureau@redhat.com>

util: make do_send_recv work with partial send/recv

According to msdn documentation and Linux man pages, send() should try
to send as much as possible in blocking mode, while recv() may return
earlier with a smaller available amount, we should try to continue
send/recv from there.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20221006113657.2656108-3-marcandre.lureau@redhat.com>

osdep: make readv_writev() work with partial read/write

With a pipe or other reasons, read/write may return less than the
requested bytes. This happens with the test-io-channel-command test on
Windows. glib spawn code uses a binary pipe of 4096 bytes, and the first
read returns that much (although more are requested), for some unclear
reason...

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20221006113657.2656108-2-marcandre.lureau@redhat.com>

win32: set threads name

As described in:
https://learn.microsoft.com/en-us/visualstudio/debugger/how-to-set-a-thread-name-in-native-code?view=vs-2022

SetThreadDescription() is available since Windows 10, version 1607 and
in some versions only by "Run Time Dynamic Linking". Its declaration is
not yet in mingw, so we lookup the function the same way glib does.

Tested with Visual Studio Community 2022 debugger.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Acked-by: Richard Henderson <richard.henderson@linaro.org>

gtk: Add show_menubar=on|off command line option.

The patch adds "show_menubar" command line option for GTK UI similar to
"show_tabs". This option allows to hide menu bar initially, it still can
be toggled by shortcut and other shortcuts still work.

Signed-off-by: Bryce Mills <brycemills@proton.me>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <NWO_zx1CT5Aj9vAXsRlqBppXd63gcKwL9V1qM1Meh36M_9tCw-EsCnfpvONXhHjmtKIUoSuCy9OO6cHS7M8b0oHBOCZG6f1jZ4Q2tqgI2Qo=@proton.me>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

qemu-edid: Restrict input parameter -d to avoid division by zero

A zero value for dpi will lead to a division by zero in qemu_edid_dpi_to_mm().
Tested by runnig qemu-edid -dX, X = 0, 100.

Resolves: qemu-project/qemu#1249

Suggested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Sebastian Mitterle <smitterl@redhat.com>
Message-Id: <20221011151216.64897-1-smitterl@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

tests/unit/test-image-locking: Fix handling of temporary files

test-image-locking leaves some temporary files around - clean
them up. While we're at it, test-image-locking is a unit test,
so it should not use "qtest.*" for temporary file names. Give
them better names instead, so that it clear where the temporary
files come from.

Message-Id: <20221012085932.799221-1-thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

ui/gtk: Fix the implicit mouse ungrabbing logic

Although the grab menu item represents the tabbed displays, the old
implicit mouse ungrabbing logic changes the grab menu item even for
an untabbed display.

Leave the grab menu item when implicitly ungrabbing mouse for an
untabbed display. The new ungrabbing logic introduced in
gd_mouse_mode_change() strictly follows the corresponding grabbing
logic found in gd_button_event().

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20221008140116.11473-1-akihiko.odaki@daynix.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

pci-ids: document modern virtio-pci ids in pci.h too

While being at it add a #define for the magic 0x1040 number.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20221004112100.301935-6-kraxel@redhat.com>

pci-ids: drop list of modern virtio devices

Drop the list of modern virtio devices and explain how they
are calculated instead.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20221004112100.301935-5-kraxel@redhat.com>

pci-ids: drop PCI_DEVICE_ID_VIRTIO_PMEM

Not needed for a virtio 1.0 device.  virtio_pci_device_plugged()
overrides them anyway (so no functional change).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Pankaj Gupta <pankaj.gupta@amd.com>
Tested-by: Pankaj Gupta <pankaj.gupta@amd.com>
Message-Id: <20221004112100.301935-4-kraxel@redhat.com>

pci-ids: drop PCI_DEVICE_ID_VIRTIO_MEM

Not needed for a virtio 1.0 device.  virtio_pci_device_plugged()
overrides them anyway (so no functional change).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20221004112100.301935-3-kraxel@redhat.com>

pci-ids: drop PCI_DEVICE_ID_VIRTIO_IOMMU

Not needed for a virtio 1.0 device.  virtio_pci_device_plugged()
overrides them anyway (so no functional change).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Tested-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20221004112100.301935-2-kraxel@redhat.com>

docs: add firmware feature flags

Add new firmware feature flags for the recently added confidential
computing operating modes by amd and intel.

While being at it also fix the path to the amd sev documentation.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Kashyap Chamarthy <kchamart@redhat.com>
Message-Id: <20220930133220.1771336-1-kraxel@redhat.com>

cirrus_vga: fix potential memory overflow

Signed-off-by: lu zhipeng <luzhipeng@cestc.cn>
Message-Id: <20220929122352.1891-1-luzhipeng@cestc.cn>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

tests/qtest: libqtest: Install signal handler via signal()

At present the codes uses sigaction() to install signal handler with
a flag SA_RESETHAND. Such usage can be covered by the signal() API
that is a simplified interface to the general sigaction() facility.

Update to use signal() to install the signal handler, as it is
available on Windows which we are going to support.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221006151927.2079583-11-bmeng.cn@gmail.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: migration-test: Avoid using hardcoded /tmp

This case was written to use hardcoded /tmp directory for temporary
files. Update to use g_dir_make_tmp() for a portable implementation.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221006151927.2079583-5-bmeng.cn@gmail.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

qtest: start a VNC test

This is some of the simplest test we could perform, it simply connects
to the VNC server via passed-in socket FDs and checks the connection can
be established.

Another series will make this test work on Windows as well.

As always, more tests can be added later! :)

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221006130513.2683873-1-marcandre.lureau@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Merge tag 'pull-testing-gdbstub-plugins-gitdm-111022-1' of https://github.com/stsquad/qemu into staging

testing:

  - revert patch to stop always building pc-bios roms

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEEZoWumedRZ7yvyN81+9DbCVqeKkQFAmNFw8YACgkQ+9DbCVqe
# KkT40wf+JTJSu9yashD38isN26LtW/Ie7+D0lhBCpmHlQy256Qumc4kaMotC6da5
# iU58rfDzN6EUjfbZSPfbLQTVnmH6om5/ejJoEWZIeoYtcgBHdzHYGpHxOf68r/BA
# 1y5GlahyG0rtR80J9vjAIqlQswBgklZIomqQj2VyAy3I7fFcYKIW14FyY6by/cZU
# b7YVn8esZ/+6q6KirrX/zksf04uek8twoOZ4oBRNW1aeApvs73NUXmIXd1IidlgW
# qxDOCm2IGZ1skRoaWH8kOjsYXfE/w4DFt+vOAKmSHnnYLAEYjfA+J4TUfoqmaOQ1
# LhEm/v0rDvteX8hEErBvXAsXdm3HqA==
# =ux3j
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 11 Oct 2022 15:28:06 EDT
# gpg:                using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <alex.bennee@linaro.org>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8  DF35 FBD0 DB09 5A9E 2A44

* tag 'pull-testing-gdbstub-plugins-gitdm-111022-1' of https://github.com/stsquad/qemu:
  Revert "configure: build ROMs with container-based cross compilers"

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

Revert "configure: build ROMs with container-based cross compilers"

This reverts commit 730fe750fba63023e294ff0acf0f874369f1946f.

Unconditionally building all the bios for all arches was a little too
far too fast.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Tested-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20221011113417.794841-4-alex.bennee@linaro.org>

tests/avocado: Add missing require_netdev('user') checks

Some avocado tests fail if QEMU was built without libslirp. Add
require_netdev('user') checks where necessary:

These tests try to ping 10.0.2.2 and expect it to succeed:
  boot_linux_console.py:BootLinuxConsole.test_arm_emcraft_sf2
  boot_linux_console.py:BootLinuxConsole.test_arm_orangepi_sd
  ppc_bamboo.py:BambooMachine.test_ppc_bamboo

These tests run a commandline that includes '-net user':
  machine_aspeed.py:AST2x00Machine.test_arm_ast2500_evb_builroot
  (and others that use the do_test_arm_aspeed_buidroot_start()
  or do_test_arm_aspeed_sdk_start() helper functions)

These changes seem to be sufficient for 'make check-avocado'
to not fail on a --disable-slirp build.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20221001195224.2453581-1-peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

ui/gtk-egl: egl context needs to be unbound in the end of gd_egl_switch

A thread often fails to bind an egl context to itself after guest VM is
rebooted because the context is still owned by another thread. It is not
very clear what condition makes this happen but this can be prevented
by unbinding the context from the thread in the end of gd_egl_switch.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Message-Id: <20220928215805.4661-1-dongwon.kim@intel.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

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

dump patches

Hi

Includes:
- the first patches from "[PATCH v5 00/18] dump: Add arch section and s390x PV dump"
- "[PATCH v2 0/2] Fix dumping in kdump format with non-aligned memory"

# -----BEGIN PGP SIGNATURE-----
#
# iQJQBAABCAA6FiEEh6m9kz+HxgbSdvYt2ujhCXWWnOUFAmM+9UocHG1hcmNhbmRy
# ZS5sdXJlYXVAcmVkaGF0LmNvbQAKCRDa6OEJdZac5aoMD/0eya5IqxX0ke9jjUcB
# wWhRbsVdiO9yt2oN3gsQVkUtuK98p7/JpWSKHWXsQQjd7vIYPCj8RBHXZ4Cp71+S
# n+Db/K22fmWvuP0LtCNzYujq1ZxKdQI8KdKmYwHQdGkgy85QwYO+0SgpVgLjRd/2
# 1IWzHuzIcEmraxkSLvR8N0lmz/Z2CBc7ME5izO1mHSZrs8Ria2tDpXnz5rFxPy+q
# TDqo+LP0GIapoHbbE+6JWGL2u9mLiP0sg9SclZOPZG3CsPQ5XYEStyZcLw1YYIO1
# rruom463SbP4NJHIRspC8pADFI+d4uBamj/eUu8/9CUVqetk0UpKyXWiqgEDvmkO
# /2/yFYS60gEghvd3XVnuZnTNeRSDSE1aUXUmKdGqDjYL4DYcsehIQ9z8ut/tuYIO
# D+4RuiuCmEyznV/DDecnwHhrv9jWftNwdwjW5GAniEjDBp/DtoouAeMArQw9rE1Z
# mXqTa5NaeW69VTtzxzN25GzSAjiEzFT7UFRt3bT8fb0NX+UOkluD/agBqRIM5lzh
# KbUJjqWhqA6TnHtDekbNLlmocDCn+NeBeXxDUIa19C4LICCuuxbFU+MG7cvdjSEg
# sXRB6/qAdWgv4O3zGw1SHff+qmvtHjKVj0ihrNkrrYrL+31O2splqTMeaCqGdq9C
# elE/TAS46CcMUSYuWzxy6mAMmA==
# =O9tr
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 06 Oct 2022 11:33:30 EDT
# 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]
# Primary key fingerprint: 87A9 BD93 3F87 C606 D276  F62D DAE8 E109 7596 9CE5

* tag 'dump-pull-request' of https://gitlab.com/marcandre.lureau/qemu:
  dump: fix kdump to work over non-aligned blocks
  dump: simplify a bit kdump get_next_page()
  dump: Rename write_elf*_phdr_note to prepare_elf*_phdr_note
  dump: Split elf header functions into prepare and write
  dump: Rework dump_calculate_size function
  dump: Rework filter area variables
  dump: Rework get_start_block
  dump: Refactor dump_iterate and introduce dump_filter_memblock_*()
  dump: Rename write_elf_loads to write_elf_phdr_loads
  dump: Replace opaque DumpState pointer with a typed one

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

ui/vnc-clipboard: fix integer underflow in vnc_client_cut_text_ext

Extended ClientCutText messages start with a 4-byte header. If len < 4,
an integer underflow occurs in vnc_client_cut_text_ext. The result is
used to decompress data in a while loop in inflate_buffer, leading to
CPU consumption and denial of service. Prevent this by checking dlen in
protocol_client_msg.

Fixes: CVE-2022-3165
Fixes: 0bf41cab93e5 ("ui/vnc: clipboard support")
Reported-by: TangPeng <tangpeng@qianxin.com>
Signed-off-by: Mauro Matteo Cascella <mcascell@redhat.com>
Message-Id: <20220925204511.1103214-1-mcascell@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

tests/x86: Add 'q35' machine type to ivshmem-test

Configure pci bridge setting to test ivshmem on 'q35'.

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-10-michael.labiuk@virtuozzo.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/x86: Add 'q35' machine type to drive_del-test

Configure pci bridge setting to run tests on 'q35' machine type.

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-9-michael.labiuk@virtuozzo.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/x86: replace snprint() by g_strdup_printf() in drive_del-test

Using g_autofree char* and  g_strdup_printf(...) instead of ugly
snprintf on stack array.

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-8-michael.labiuk@virtuozzo.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/x86: Fix comment typo in drive_del-test

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-7-michael.labiuk@virtuozzo.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/x86: Add 'q35' machine type to hotplug hd-geo-test

Add pci bridge setting to test hotplug.
Duplicate tests for plugging scsi and virtio devices for q35 machine type.

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-6-michael.labiuk@virtuozzo.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/x86: Add 'q35' machine type to override-tests in hd-geo-test

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-5-michael.labiuk@virtuozzo.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/x86: Refactor hot unplug hd-geo-test

Moving common code to function.

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-4-michael.labiuk@virtuozzo.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/x86: Add subtest with 'q35' machine type to device-plug-test

Configure pci bridge setting to plug pci device and unplug.

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-3-michael.labiuk@virtuozzo.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/x86: add helper qtest_qmp_device_del_send()

Move sending 'device_del' command to separate function.
Function can be used in case of addition action is needed to start
actual removing device after sending command.

Signed-off-by: Michael Labiuk <michael.labiuk@virtuozzo.com>
Message-Id: <20220929223547.1429580-2-michael.labiuk@virtuozzo.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
[thuth: Fixed typo]
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/migration: remove the unused local variable

Remove the unused local variable "records".

Signed-off-by: dinglimin <dinglimin@cmss.chinamobile.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Message-Id: <20220928080555.2263-1-dinglimin@cmss.chinamobile.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

qtest: "-display none" is set in qtest_init()

So we don't need to set anywhere else.

Signed-off-by: Juan Quintela <quintela@redhat.com>
[thuth: Drop changes in tests/qtest/fuzz/ since the fuzzers still need this]
Message-Id: <20220902165126.1482-2-quintela@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

linux-user: i386/signal: support XSAVE/XRSTOR for signal frame fpstate

Add support for saving/restoring extended save states when signals
are delivered.  This allows using AVX, MPX or PKRU registers in
signal handlers.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

audio: prevent an integer overflow in resampling code

There are corner cases where rate->opos can overflow. For
example, if QEMU is started with -audiodev pa,id=audio0,
out.frequency=11025 -device ich9-intel-hda -device hda-duplex,
audiodev=audio0 and the guest plays audio with a sampling
frequency of 44100Hz, rate->opos will overflow after 27.05h
and the audio stream will be silent for a long time.

To prevent a rate->opos and also a rate->ipos overflow, both
are wrapped around after a short time. The wrap around point
rate->ipos >= 0x10001 is an arbitrarily selected value and can
be any small value, 0 and 1 included.

The comment that an ipos overflow will result in an infinite
loop has been removed, because in this case the resampling code
only generates no more output samples and the audio stream stalls.
However, there is no infinite loop.

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20220923183640.8314-12-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: fix sw->buf size for audio recording

The calculation of the buffer size needed to store audio samples
after resampling is wrong for audio recording. For audio recording
sw->ratio is calculated as

sw->ratio = frontend sample rate / backend sample rate.

From this follows

frontend samples = frontend sample rate / backend sample rate
 * backend samples
frontend samples = sw->ratio * backend samples

In 2 of 3 places in the audio recording code where sw->ratio
is used in a calculation to get the number of frontend frames,
the calculation is wrong. Fix this. The 3rd formula in
audio_pcm_sw_read() is correct.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/71
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-11-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: refactor audio_get_avail()

Split out the code in audio_get_avail() that calculates the
buffer size that the audio frontend can read. This is similar
to the code changes in audio_get_free().

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-10-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: rename audio_sw_bytes_free()

Rename and refactor audio_sw_bytes_free(). This function is not
limited to calculate the free audio buffer size. The renamed
function returns the number of frames instead of bytes.

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-9-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: swap audio_rate_get_bytes() function parameters

Swap the rate and info parameters of the audio_rate_get_bytes()
function to align the parameter order with the rest of the
audio_rate_*() functions.

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-8-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

spiceaudio: update comment

Replace a comment with a question with the answer.

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-7-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

spiceaudio: add a pcm_ops buffer_get_free function

It seems there is a demand [1] for low latency playback over
SPICE. Add a pcm_ops buffer_get_free function to reduce the
playback latency. The mixing engine buffer becomes a temporary
buffer.

[1] https://lists.nongnu.org/archive/html/qemu-devel/2022-01/msg01644.html

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-6-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: add more audio rate control functions

The next patch needs two new rate control functions. The first
one returns the bytes needed at call time to maintain the
selected rate. The second one adjusts the bytes actually sent.

Split the audio_rate_get_bytes() function into these two
functions and reintroduce audio_rate_get_bytes().

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-5-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

alsaaudio: reduce playback latency

Change the buffer_get_free pcm_ops function to report the free
ALSA playback buffer. The generic buffer becomes a temporary
buffer and is empty after a call to audio_run_out().

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-4-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: run downstream playback queue unconditionally

Run the downstream playback queue even if the emulated audio
device didn't write new samples. There still may be buffered
audio samples downstream.

This is for the -audiodev out.mixing-engine=off case. Commit
a8a98cfd42 ("audio: run downstream playback queue uncondition-
ally") fixed the out.mixing-engine=on case.

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-3-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: fix GUS audio playback with out.mixing-engine=off

Fix GUS audio playback with out.mixing-engine=off.

The GUS audio device needs to know the amount of samples to
produce in advance.

To reproduce start qemu with
-parallel none -device gus,audiodev=audio0
-audiodev pa,id=audio0,out.mixing-engine=off

and start the cartoon.exe demo in a FreeDOS guest. The demo file
is available on the download page of the GUSemu32 author.

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-2-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

audio: refactor code in audio_run_out()

Refactoring the code in audio_run_out() avoids code duplication
in the next patch. There's no functional change.

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220923183640.8314-1-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

linux-user: i386/signal: support FXSAVE fpstate on 32-bit emulation

Linux can use FXSAVE to save/restore XMM registers even on 32-bit
systems.  This requires some care in order to keep the FXSAVE area
aligned to 16 bytes; for this reason, get_sigframe is changed to
pass the offset into the FXSAVE area rather than the full frame
size.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

linux-user: i386/signal: move fpstate at the end of the 32-bit frames

Recent versions of Linux moved the 32-bit fpstate towards the end of the
frame, so that the variable-sized xsave data does not overwrite the
(ABI-defined) extramask[] field.  Follow suit in QEMU.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: x86: Implement MSR_CORE_THREAD_COUNT MSR

The MSR_CORE_THREAD_COUNT MSR describes CPU package topology, such as number
of threads and cores for a given package. This is information that QEMU has
readily available and can provide through the new user space MSR deflection
interface.

This patch propagates the existing hvf logic from patch 027ac0cb516
("target/i386/hvf: add rdmsr 35H MSR_CORE_THREAD_COUNT") to KVM.

Signed-off-by: Alexander Graf <agraf@csgraf.de>
Message-Id: <20221004225643.65036-4-agraf@csgraf.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

i386: kvm: Add support for MSR filtering

KVM has grown support to deflect arbitrary MSRs to user space since
Linux 5.10. For now we don't expect to make a lot of use of this
feature, so let's expose it the easiest way possible: With up to 16
individually maskable MSRs.

This patch adds a kvm_filter_msr() function that other code can call
to install a hook on KVM MSR reads or writes.

Signed-off-by: Alexander Graf <agraf@csgraf.de>
Message-Id: <20221004225643.65036-3-agraf@csgraf.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

x86: Implement MSR_CORE_THREAD_COUNT MSR

Intel CPUs starting with Haswell-E implement a new MSR called
MSR_CORE_THREAD_COUNT which exposes the number of threads and cores
inside of a package.

This MSR is used by XNU to populate internal data structures and not
implementing it prevents virtual machines with more than 1 vCPU from
booting if the emulated CPU generation is at least Haswell-E.

This patch propagates the existing hvf logic from patch 027ac0cb516
("target/i386/hvf: add rdmsr 35H MSR_CORE_THREAD_COUNT") to TCG.

Signed-off-by: Alexander Graf <agraf@csgraf.de>
Message-Id: <20221004225643.65036-2-agraf@csgraf.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Enable TARGET_TB_PCREL

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20221001140935.465607-27-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Inline gen_jmp_im

Expand this function at each of its callers.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20221001140935.465607-26-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Add cpu_eip

Create a tcg global temp for this, and use it instead of explicit stores.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20221001140935.465607-25-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Create eip_cur_tl

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20221001140935.465607-24-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Merge gen_jmp_tb and gen_goto_tb into gen_jmp_rel

These functions have only one caller, and the logic is more
obvious this way.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20221001140935.465607-23-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Remove MemOp argument to gen_op_j*_ecx

These functions are always passed aflag, so we might as well
read it from DisasContext directly.  While we're at it, use
a common subroutine for these two functions.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20221001140935.465607-22-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: Use gen_jmp_rel for DISAS_TOO_MANY

With gen_jmp_rel, we may chain between two translation blocks
which may only be separated because of TB size limits.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221001140935.465607-21-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>