Merge remote-tracking branch 'remotes/thuth-gitlab/tags/pull-request-2022-03-07' into staging

* Fixes for s390x TCG tests
* Update Haiku VM to a usable level
* Some other miscellaneous small fixes

# gpg: Signature made Mon 07 Mar 2022 18:07:00 GMT
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [full]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [full]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [full]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* remotes/thuth-gitlab/tags/pull-request-2022-03-07:
  Check and report for incomplete 'global' option format
  tests/vm: Update haiku test vm to R1/Beta3
  tests/avocado: Cancel BootLinux tests in case there is no free port
  MAINTAINERS: Update the files in the FreeBSD section
  tests/tcg/s390x: Cleanup of mie3 tests.
  tests/tcg/s390x: Fix the exrl-trt* tests with Clang
  tests/tcg/s390x: Fix mvc, mvo and pack tests with Clang

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

Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20220307' into staging

target-arm queue:
 * cleanups of qemu_oom_check() and qemu_memalign()
 * target/arm/translate-neon: UNDEF if VLD1/VST1 stride bits are non-zero
 * target/arm/translate-neon: Simplify align field check for VLD3
 * GICv3 ITS: add more trace events
 * GICv3 ITS: implement 8-byte accesses properly
 * GICv3: fix minor issues with some trace/log messages
 * ui/cocoa: Use the standard about panel
 * target/arm: Provide cpu property for controling FEAT_LPA2
 * hw/arm/virt: Disable LPA2 for -machine virt-6.2

# gpg: Signature made Mon 07 Mar 2022 16:46:06 GMT
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [ultimate]
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* remotes/pmaydell/tags/pull-target-arm-20220307:
  hw/arm/virt: Disable LPA2 for -machine virt-6.2
  target/arm: Provide cpu property for controling FEAT_LPA2
  ui/cocoa: Use the standard about panel
  hw/intc/arm_gicv3_cpuif: Fix register names in ICV_HPPIR read trace event
  hw/intc/arm_gicv3: Fix missing spaces in error log messages
  hw/intc/arm_gicv3: Specify valid and impl in MemoryRegionOps
  hw/intc/arm_gicv3_its: Add trace events for table reads and writes
  hw/intc/arm_gicv3_its: Add trace events for commands
  target/arm/translate-neon: Simplify align field check for VLD3
  target/arm/translate-neon: UNDEF if VLD1/VST1 stride bits are non-zero
  osdep: Move memalign-related functions to their own header
  util: Put qemu_vfree() in memalign.c
  util: Use meson checks for valloc() and memalign() presence
  util: Share qemu_try_memalign() implementation between POSIX and Windows
  meson.build: Don't misdetect posix_memalign() on Windows
  util: Return valid allocation for qemu_try_memalign() with zero size
  util: Unify implementations of qemu_memalign()
  util: Make qemu_oom_check() a static function

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

Merge remote-tracking branch 'remotes/kraxel/tags/seabios-20220307-pull-request' into staging

update seabios to 1.16.0

# gpg: Signature made Mon 07 Mar 2022 11:02:40 GMT
# 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

* remotes/kraxel/tags/seabios-20220307-pull-request:
  update seabios binaries to 1.16.0
  update seabios submodule to 1.16.0

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

Merge remote-tracking branch 'remotes/cschoenebeck/tags/pull-9p-20220307' into staging

9pfs: introduce macOS host support and cleanup

* Add support for Darwin (a.k.a. macOS) hosts.

* Code cleanup (move qemu_dirent_dup() from osdep -> 9p-util).

* API doc cleanup (convert Doxygen -> kerneldoc format).

# gpg: Signature made Mon 07 Mar 2022 11:14:45 GMT
# gpg:                using RSA key 96D8D110CF7AF8084F88590134C2B58765A47395
# gpg:                issuer "qemu_oss@crudebyte.com"
# gpg: Good signature from "Christian Schoenebeck <qemu_oss@crudebyte.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: ECAB 1A45 4014 1413 BA38  4926 30DB 47C3 A012 D5F4
#      Subkey fingerprint: 96D8 D110 CF7A F808 4F88  5901 34C2 B587 65A4 7395

* remotes/cschoenebeck/tags/pull-9p-20220307:
  fsdev/p9array.h: convert Doxygen -> kerneldoc format
  9pfs/coth.h: drop Doxygen format on v9fs_co_run_in_worker()
  9pfs/9p-util.h: convert Doxygen -> kerneldoc format
  9pfs/9p.c: convert Doxygen -> kerneldoc format
  9pfs/codir.c: convert Doxygen -> kerneldoc format
  9pfs/9p.h: convert Doxygen -> kerneldoc format
  9pfs: drop Doxygen format from qemu_dirent_dup() API comment
  9pfs: move qemu_dirent_dup() from osdep -> 9p-util
  9p: darwin: meson: Allow VirtFS on Darwin
  9p: darwin: Adjust assumption on virtio-9p-test
  9p: darwin: Implement compatibility for mknodat
  9p: darwin: Compatibility for f/l*xattr
  9p: darwin: *xattr_nofollow implementations
  9p: darwin: Move XATTR_SIZE_MAX->P9_XATTR_SIZE_MAX
  9p: darwin: Ignore O_{NOATIME, DIRECT}
  9p: darwin: Handle struct dirent differences
  9p: darwin: Handle struct stat(fs) differences
  9p: Rename 9p-util -> 9p-util-linux
  9p: linux: Fix a couple Linux assumptions

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

Merge remote-tracking branch 'remotes/philmd/tags/abstract-arch-cpu-20220307' into staging

- Re-org accel/ and softmmu/ to have more target-agnostic objects.

- Use CPUArchState as an abstract type, defined by each target
  (CPUState is our interface with generic code, CPUArchState is
   our interface with target-specific code).

# gpg: Signature made Sun 06 Mar 2022 23:23:19 GMT
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.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: FAAB E75E 1291 7221 DCFD  6BB2 E3E3 2C2C DEAD C0DE

* remotes/philmd/tags/abstract-arch-cpu-20220307: (33 commits)
  accel/tcg: Remove pointless CPUArchState casts
  target/i386: Remove pointless CPUArchState casts
  target: Use ArchCPU as interface to target CPU
  target: Introduce and use OBJECT_DECLARE_CPU_TYPE() macro
  target: Use CPUArchState as interface to target-specific CPU state
  target: Use forward declared type instead of structure type
  target/hexagon: Add missing 'hw/core/cpu.h' include
  target: Include missing 'cpu.h'
  Hexagon (target/hexagon) convert to OBJECT_DECLARE_TYPE
  target/i386/tcg/sysemu: Include missing 'exec/exec-all.h' header
  cpu: Add missing 'exec/exec-all.h' and 'qemu/accel.h' headers
  exec/cpu_ldst: Include 'cpu.h' to get target_ulong definition
  meson: Display libfdt as disabled when system emulation is disabled
  softmmu: Build target-agnostic objects once
  softmmu: Add qemu_init_arch_modules()
  exec/cpu: Make address_space_init/reloading_memory_map target agnostic
  exec/gdbstub: Make gdb_exit() / gdb_set_stop_cpu() target agnostic
  misc: Add missing "sysemu/cpu-timers.h" include
  misc: Remove unnecessary "sysemu/cpu-timers.h" include
  softmmu/cpu-timers: Remove unused 'exec/exec-all.h' header
  ...

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

Check and report for incomplete 'global' option format

Qemu might crash when provided incomplete '-global' option.
For example:
     qemu-system-x86_64 -global driver=isa-fdc
     qemu-system-x86_64: ../../devel/qemu/qapi/string-input-visitor.c:394:
     string_input_visitor_new: Assertion `str' failed.
     Aborted (core dumped)

Fixes: 3751d7c43f795b ("vl: allow full-blown QemuOpts syntax for -global")
Signed-off-by: Rohit Kumar <rohit.kumar3@nutanix.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/604
Message-Id: <20220216071508.412974-1-rohit.kumar3@nutanix.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/vm: Update haiku test vm to R1/Beta3

The old image did not have python3 yet, and thus was not usable
for compiling QEMU anymore.

Suggested-by: Alexander von Gluck IV <kallisti5@unixzen.com>
Message-Id: <20220216154208.2985103-1-kallisti5@unixzen.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/avocado: Cancel BootLinux tests in case there is no free port

The BootLinux tests are currently failing with an ugly python
stack trace on my RHEL8 system since they cannot get a free port
(likely due to the firewall settings on my system). Let's properly
check the return value of find_free_port() instead and cancel the
test gracefully if it cannot get a free port.

Message-Id: <20220228114325.818294-1-thuth@redhat.com>
Reviewed-by: Beraldo Leal <bleal@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Merge remote-tracking branch 'remotes/hreitz-gitlab/tags/pull-block-2022-03-07' into staging

Block patches for 7.0-rc0:
- New fleecing backup scheme
- iotest fixes
- Fixes for the curl block driver
- Fix for the preallocate block driver
- IDE fix for zero-length TRIM requests

# gpg: Signature made Mon 07 Mar 2022 10:33:31 GMT
# gpg:                using RSA key CB62D7A0EE3829E45F004D34A1FA40D098019CDF
# gpg:                issuer "hreitz@redhat.com"
# gpg: Good signature from "Hanna Reitz <hreitz@redhat.com>" [marginal]
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg:          It is not certain that the signature belongs to the owner.
# Primary key fingerprint: CB62 D7A0 EE38 29E4 5F00  4D34 A1FA 40D0 9801 9CDF

* remotes/hreitz-gitlab/tags/pull-block-2022-03-07: (23 commits)
  iotests/image-fleecing: test push backup with fleecing
  iotests/image-fleecing: add test case with bitmap
  iotests.py: add qemu_io_pipe_and_status()
  iotests/image-fleecing: add test-case for fleecing format node
  block: copy-before-write: realize snapshot-access API
  block: introduce snapshot-access block driver
  block/io: introduce block driver snapshot-access API
  block/reqlist: add reqlist_wait_all()
  block/dirty-bitmap: introduce bdrv_dirty_bitmap_status()
  block/reqlist: reqlist_find_conflict(): use ranges_overlap()
  block: intoduce reqlist
  block/block-copy: add block_copy_reset()
  block/copy-before-write: add bitmap open parameter
  block/block-copy: block_copy_state_new(): add bitmap parameter
  block/dirty-bitmap: bdrv_merge_dirty_bitmap(): add return value
  block/block-copy: move copy_bitmap initialization to block_copy_state_new()
  iotests: Write test output to TEST_DIR
  tests/qemu-iotests/testrunner: Quote "case not run" lines in TAP mode
  tests/qemu-iotests/040: Skip TestCommitWithFilters without 'throttle'
  block: fix preallocate filter: don't do unaligned preallocate requests
  ...

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

hw/arm/virt: Disable LPA2 for -machine virt-6.2

There is a Linux kernel bug present until v5.12 that prevents
booting with FEAT_LPA2 enabled.  As a workaround for TCG,
disable this feature for machine versions prior to 7.0.

Cc: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Provide cpu property for controling FEAT_LPA2

There is a Linux kernel bug present until v5.12 that prevents
booting with FEAT_LPA2 enabled.  As a workaround for TCG, allow
the feature to be disabled from -cpu max.

Since this kernel bug is present in the Fedora 31 image that
we test in avocado, disable lpa2 on the command-line.

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

ui/cocoa: Use the standard about panel

This provides standard look and feel for the about panel and reduces
code.

Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
Message-id: 20220227042241.1543-1-akihiko.odaki@gmail.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/intc/arm_gicv3_cpuif: Fix register names in ICV_HPPIR read trace event

The trace_gicv3_icv_hppir_read trace event takes an integer value
which it uses to form the register name, which should be either
ICV_HPPIR0 or ICV_HPPIR1.  We were passing in the 'grp' variable for
this, but that is either GICV3_G0 or GICV3_G1NS, which happen to be 0
and 2, which meant that tracing for the ICV_HPPIR1 register was
incorrectly printed as ICV_HPPIR2.

Use the same approach we do for all the other similar trace events,
and pass in 'ri->crm == 8 ?  0 : 1', deriving the index value
directly from the ARMCPRegInfo struct.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220303202341.2232284-6-peter.maydell@linaro.org

hw/intc/arm_gicv3: Fix missing spaces in error log messages

We forgot a space in some log messages, so the output ended
up looking like
gicv3_dist_write: invalid guest write at offset 0000000000008000size 8

with a missing space before "size". Add the missing spaces.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220303202341.2232284-5-peter.maydell@linaro.org

hw/intc/arm_gicv3: Specify valid and impl in MemoryRegionOps

The GICv3 has some registers that support byte accesses, and some
that support 8-byte accesses.  Our TCG implementation implements all
of this, switching on the 'size' argument and handling the registers
that must support reads of that size while logging an error for
attempted accesses to registers that do not support that size access.
However we forgot to tell the core memory subsystem about this by
specifying the .impl and .valid fields in the MemoryRegionOps struct,
so the core was happily simulating 8 byte accesses by combining two 4
byte accesses.  This doesn't have much guest-visible effect, since
there aren't many 8 byte registers and they all support being written
in two 4 byte parts.

Set the .impl and .valid fields to say that all sizes from 1 to 8
bytes are both valid and implemented by the device.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220303202341.2232284-4-peter.maydell@linaro.org

hw/intc/arm_gicv3_its: Add trace events for table reads and writes

For debugging guest use of the ITS, it can be helpful to trace
when the ITS reads and writes the in-memory tables.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220303202341.2232284-3-peter.maydell@linaro.org

hw/intc/arm_gicv3_its: Add trace events for commands

When debugging code that's using the ITS, it's helpful to
see tracing of the ITS commands that the guest executes. Add
suitable trace events.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220303202341.2232284-2-peter.maydell@linaro.org

target/arm/translate-neon: Simplify align field check for VLD3

For VLD3 (single 3-element structure to one lane), there is no
alignment specification and the alignment bits in the instruction
must be zero.  This is bit [4] for the size=0 and size=1 cases, and
bits [5:4] for the size=2 case.  We do this check correctly in
VLDST_single(), but we write it a bit oddly: in the 'case 3' code we
check for bit 0 of a->align (bit [4] of the insn), and then we fall
through to the 'case 2' code which checks bit 1 of a->align (bit [5]
of the insn) in the size 2 case.  Replace this with just checking "is
a->align non-zero" for VLD3, which lets us drop the fall-through and
put the cases in this switch in numerical order.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220303113741.2156877-3-peter.maydell@linaro.org

target/arm/translate-neon: UNDEF if VLD1/VST1 stride bits are non-zero

For VLD1/VST1 (single element to one lane) we are only accessing one
register, and so the 'stride' is meaningless.  The bits that would
specify stride (insn bit [4] for size=1, bit [6] for size=2) are
specified to be zero in the encoding (which would correspond to a
stride of 1 for VLD2/VLD3/VLD4 etc), and we must UNDEF if they are
not.

We failed to make this check, which meant that we would incorrectly
handle some instruction patterns as loads or stores instead of
UNDEFing them. Enforce that stride == 1 for the nregs == 1 case.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/890
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220303113741.2156877-2-peter.maydell@linaro.org

osdep: Move memalign-related functions to their own header

Move the various memalign-related functions out of osdep.h and into
their own header, which we include only where they are used.
While we're doing this, add some brief documentation comments.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20220226180723.1706285-10-peter.maydell@linaro.org

util: Put qemu_vfree() in memalign.c

qemu_vfree() is the companion free function to qemu_memalign(); put
it in memalign.c so the allocation and free functions are together.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220226180723.1706285-9-peter.maydell@linaro.org
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

util: Use meson checks for valloc() and memalign() presence

Instead of assuming that all CONFIG_BSD have valloc() and anything
else is memalign(), explicitly check for those functions in
meson.build and use the "is the function present" define.  Tests for
specific functionality are better than which-OS checks; this also
lets us give a helpful error message if somehow there's no usable
function present.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20220226180723.1706285-8-peter.maydell@linaro.org

util: Share qemu_try_memalign() implementation between POSIX and Windows

The qemu_try_memalign() functions for POSIX and Windows used to be
significantly different, but these days they are identical except for
the actual allocation function called, and the POSIX version already
has to have ifdeffery for different allocation functions.

Move to a single implementation in memalign.c, which uses the Windows
_aligned_malloc if we detect that function in meson.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220226180723.1706285-7-peter.maydell@linaro.org
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

meson.build: Don't misdetect posix_memalign() on Windows

Currently we incorrectly think that posix_memalign() exists on
Windows.  This is because of a combination of:

 * the msys2/mingw toolchain/libc claim to have a
   __builtin_posix_memalign when there isn't a builtin of that name
 * meson will assume that if you have a __builtin_foo that
   counts for has_function('foo')

Specifying a specific include file via prefix: causes meson to not
treat builtins as sufficient and actually look for the function
itself; see this meson pull request which added that as the official
way to get the right answer:
  https://github.com/mesonbuild/meson/pull/1150

Currently this misdectection doesn't cause problems because we only
use CONFIG_POSIX_MEMALIGN in oslib-posix.c; however that will change
in a following commit.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220226180723.1706285-6-peter.maydell@linaro.org
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

util: Return valid allocation for qemu_try_memalign() with zero size

Currently qemu_try_memalign()'s behaviour if asked to allocate
0 bytes is rather variable:
 * on Windows, we will assert
 * on POSIX platforms, we get the underlying behaviour of
   the posix_memalign() or equivalent function, which may be
   either "return a valid non-NULL pointer" or "return NULL"

Explictly check for 0 byte allocations, so we get consistent
behaviour across platforms.  We handle them by incrementing the size
so that we return a valid non-NULL pointer that can later be passed
to qemu_vfree().  This is permitted behaviour for the
posix_memalign() API and is the most usual way that underlying
malloc() etc implementations handle a zero-sized allocation request,
because it won't trip up calling code that assumes NULL means an
error.  (This includes our own qemu_memalign(), which will abort on
NULL.)

This change is a preparation for sharing the qemu_try_memalign() code
between Windows and POSIX.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

util: Unify implementations of qemu_memalign()

We implement qemu_memalign() in both oslib-posix.c and oslib-win32.c,
but the two versions are essentially the same: they call
qemu_try_memalign(), and abort() after printing an error message if
it fails.  The only difference is that the win32 version prints the
GetLastError() value whereas the POSIX version prints
strerror(errno).  However, this is a bug in the win32 version: in
commit dfbd0b873a85021 in 2020 we changed the implementation of
qemu_try_memalign() from using VirtualAlloc() (which sets the
GetLastError() value) to using _aligned_malloc() (which sets errno),
but didn't update the error message to match.

Replace the two separate functions with a single version in a
new memalign.c file, which drops the unnecessary extra qemu_oom_check()
function and instead prints a more useful message including the
requested size and alignment as well as the errno string.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220226180723.1706285-4-peter.maydell@linaro.org
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

util: Make qemu_oom_check() a static function

The qemu_oom_check() function, which we define in both oslib-posix.c
and oslib-win32.c, is now used only locally in that file; make it
static.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20220226180723.1706285-3-peter.maydell@linaro.org

fsdev/p9array.h: convert Doxygen -> kerneldoc format

API doc comments in QEMU are supposed to be in kerneldoc format, so
convert API doc comments from Doxygen format to kerneldoc format.

Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <2e2d46a402560f155de322d95789ba107d728885.1646314856.git.qemu_oss@crudebyte.com>

9pfs/coth.h: drop Doxygen format on v9fs_co_run_in_worker()

API doc comments in QEMU are supposed to be in kerneldoc format, so
drop Doxygen format used on v9fs_co_run_in_worker() macro.

Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <a8fdf0290d1e40a68f5577f29aeae12298b70733.1646314856.git.qemu_oss@crudebyte.com>

9pfs/9p-util.h: convert Doxygen -> kerneldoc format

API doc comments in QEMU are supposed to be in kerneldoc format, so
convert API doc comments from Doxygen format to kerneldoc format.

Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <dc1c4a85e233f5884ee5f6ec96b87db286083df7.1646314856.git.qemu_oss@crudebyte.com>

9pfs/9p.c: convert Doxygen -> kerneldoc format

API doc comments in QEMU are supposed to be in kerneldoc format, so
convert API doc comments from Doxygen format to kerneldoc format.

Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <4ece6ffa4465c271c6a7c42a3040f42780fcce87.1646314856.git.qemu_oss@crudebyte.com>

9pfs/codir.c: convert Doxygen -> kerneldoc format

API doc comments in QEMU are supposed to be in kerneldoc format, so
convert API doc comments from Doxygen format to kerneldoc format.

Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <c76be7d38ea448c6417b2ffb5ccd6b711519a878.1646314856.git.qemu_oss@crudebyte.com>

9pfs/9p.h: convert Doxygen -> kerneldoc format

API doc comments in QEMU are supposed to be in kerneldoc format, so
convert API doc comments from Doxygen format to kerneldoc format.

Based-on: <E1nPTwO-0006pl-Np@lizzy.crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <2b8f91de7bac3d3bc85d60eb08830a35a394be75.1646314856.git.qemu_oss@crudebyte.com>

9pfs: drop Doxygen format from qemu_dirent_dup() API comment

API doc comments in QEMU are supposed to be in kerneldoc format, so drop
occurrences of "@c" which is Doxygen format for fixed-width text.

Link: https://lore.kernel.org/qemu-devel/CAFEAcA89+ENOM6x19OEF53Kd2DWkhN5SN21Va0D7yepJSa3Jyg@mail.gmail.com/
Based-on: <E1nP9Oz-00043L-KJ@lizzy.crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <E1nPTwO-0006pl-Np@lizzy.crudebyte.com>

9pfs: move qemu_dirent_dup() from osdep -> 9p-util

Function qemu_dirent_dup() is currently only used by 9pfs server, so move
it from project global header osdep.h to 9pfs specific header 9p-util.h.

Link: https://lore.kernel.org/qemu-devel/CAFEAcA_=HAUNomKD2wurSVaAHa5mrk22A1oHKLWUDjk7v6Khmg@mail.gmail.com/
Based-on: <20220227223522.91937-12-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <E1nP9Oz-00043L-KJ@lizzy.crudebyte.com>

9p: darwin: meson: Allow VirtFS on Darwin

To allow VirtFS on darwin, we need to check that pthread_fchdir_np is
available, which has only been available since macOS 10.12.

Additionally, virtfs_proxy_helper is disabled on Darwin. This patch
series does not currently provide an implementation of the proxy-helper,
but this functionality could be implemented later on.

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
[Michael Roitzsch: - Rebase for NixOS]
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
[Will Cohen: - Rebase to master]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
[Will Cohen: - Add check for pthread_fchdir_np to virtfs
             - Add comments to patch commit
             - Note that virtfs_proxy_helper does not work
               on macOS
             - Fully adjust meson virtfs error note to specify
               macOS
             - Rebase to master]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Message-Id: <20220227223522.91937-12-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Acked-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: darwin: Adjust assumption on virtio-9p-test

The previous test depended on the assumption that P9_DOTL_AT_REMOVEDIR
and AT_REMOVEDIR have the same value.

While this is true on Linux, it is not true everywhere, and leads to an
incorrect test failure on unlink_at, noticed when adding 9p to darwin:

Received response 7 (RLERROR) instead of 77 (RUNLINKAT)
Rlerror has errno 22 (Invalid argument)
**

ERROR:../tests/qtest/virtio-9p-test.c:305:v9fs_req_recv: assertion
failed (hdr.id == id): (7 == 77) Bail out!

ERROR:../tests/qtest/virtio-9p-test.c:305:v9fs_req_recv: assertion
failed (hdr.id == id): (7 == 77)

Signed-off-by: Fabian Franz <fabianfranz.oss@gmail.com>
[Will Cohen: - Add explanation of patch and description
               of pre-patch test failure]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Acked-by: Thomas Huth <thuth@redhat.com>
[Will Cohen: - Move this patch before 9p: darwin: meson
               patch to avoid qtest breakage during
               bisecting]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <20220227223522.91937-11-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: darwin: Implement compatibility for mknodat

Darwin does not support mknodat. However, to avoid race conditions
with later setting the permissions, we must avoid using mknod on
the full path instead. We could try to fchdir, but that would cause
problems if multiple threads try to call mknodat at the same time.
However, luckily there is a solution: Darwin includes a function
that sets the cwd for the current thread only.
This should suffice to use mknod safely.

This function (pthread_fchdir_np) is protected by a check in
meson in a patch later in this series.

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
[Will Cohen: - Adjust coding style
             - Replace clang references with gcc
             - Note radar filed with Apple for missing syscall
             - Replace direct syscall with pthread_fchdir_np and
               adjust patch notes accordingly
             - Declare pthread_fchdir_np with
             - __attribute__((weak_import)) to allow checking for
               its presence before usage
             - Move declarations above cplusplus guard
             - Add CONFIG_PTHREAD_FCHDIR_NP to meson and check for
               presence in 9p-util
             - Rebase to apply cleanly on top of the 2022-02-10
               changes to 9pfs
             - Fix line over 90 characters formatting error]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Message-Id: <20220227223522.91937-10-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: darwin: Compatibility for f/l*xattr

On darwin `fgetxattr` takes two extra optional arguments,
and the l* variants are not defined (in favor of an extra
flag to the regular variants.

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
[Michael Roitzsch: - Rebase for NixOS]
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Message-Id: <20220227223522.91937-9-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Acked-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: darwin: *xattr_nofollow implementations

This implements the darwin equivalent of the functions that were
moved to 9p-util(-linux) earlier in this series in the new
9p-util-darwin file.

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
[Michael Roitzsch: - Rebase for NixOS]
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Message-Id: <20220227223522.91937-8-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Acked-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: darwin: Move XATTR_SIZE_MAX->P9_XATTR_SIZE_MAX

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
Because XATTR_SIZE_MAX is not defined on Darwin,
create a cross-platform P9_XATTR_SIZE_MAX instead.

[Will Cohen: - Adjust coding style
             - Lower XATTR_SIZE_MAX to 64k
             - Add explanatory context related to XATTR_SIZE_MAX]
[Fabian Franz: - Move XATTR_SIZE_MAX reference from 9p.c to
                 P9_XATTR_SIZE_MAX in 9p.h]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Signed-off-by: Fabian Franz <fabianfranz.oss@gmail.com>
[Will Cohen: - For P9_XATTR_MAX, ensure that Linux uses
               XATTR_SIZE_MAX, Darwin uses 64k, and error
               out for undefined hosts]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Message-Id: <20220227223522.91937-7-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: darwin: Ignore O_{NOATIME, DIRECT}

Darwin doesn't have either of these flags. Darwin does have
F_NOCACHE, which is similar to O_DIRECT, but has different
enough semantics that other projects don't generally map
them automatically. In any case, we don't support O_DIRECT
on Linux at the moment either.

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
[Michael Roitzsch: - Rebase for NixOS]
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
[Will Cohen: - Adjust coding style]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Message-Id: <20220227223522.91937-6-wwcohen@gmail.com>
[C.S.: - Fix compiler warning "unused label 'again'". ]
Link: https://lore.kernel.org/qemu-devel/11201492.CjeqJxXfGd@silver/
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: darwin: Handle struct dirent differences

On darwin d_seekoff exists, but is optional and does not seem to
be commonly used by file systems. Use `telldir` instead to obtain
the seek offset and inject it into d_seekoff, and create a
qemu_dirent_off helper to call it appropriately when appropriate.

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
[Michael Roitzsch: - Rebase for NixOS]
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
[Will Cohen: - Adjust to pass testing
             - Ensure that d_seekoff is filled using telldir
               on darwin, and create qemu_dirent_off helper
               to decide which to access]
[Fabian Franz: - Add telldir error handling for darwin]
Signed-off-by: Fabian Franz <fabianfranz.oss@gmail.com>
[Will Cohen: - Ensure that telldir error handling uses
               signed int
             - Cleanup of telldir error handling
             - Remove superfluous error handling for
               qemu_dirent_off
             - Adjust formatting
             - Use qemu_dirent_off in codir.c
             - Declare qemu_dirent_off as static to prevent
               linker error
             - Move qemu_dirent_off above the end-of-file
               endif to fix compilation]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Message-Id: <20220227223522.91937-5-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: darwin: Handle struct stat(fs) differences

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
[Will Cohen: - Note lack of f_namelen and f_frsize on Darwin
             - Ensure that tv_sec and tv_nsec are both
               initialized for Darwin and non-Darwin]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Message-Id: <20220227223522.91937-4-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: Rename 9p-util -> 9p-util-linux

The current file only has the Linux versions of these functions.
Rename the file accordingly and update the Makefile to only build
it on Linux. A Darwin version of these will follow later in the
series.

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
[Michael Roitzsch: - Rebase for NixOS]
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220227223522.91937-3-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

9p: linux: Fix a couple Linux assumptions

 - Guard Linux only headers.
 - Add qemu/statfs.h header to abstract over the which
   headers are needed for struct statfs
 - Define `ENOATTR` only if not only defined
   (it's defined in system headers on Darwin).

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
[Michael Roitzsch: - Rebase for NixOS]
Signed-off-by: Michael Roitzsch <reactorcontrol@icloud.com>
While it might at first appear that fsdev/virtfs-proxy-header.c would
need similar adjustment for darwin as file-op-9p here, a later patch in
this series disables virtfs-proxy-helper for non-Linux. Allowing
virtfs-proxy-helper on darwin could potentially be an additional
optimization later.

[Will Cohen: - Fix headers for Alpine
             - Integrate statfs.h back into file-op-9p.h
             - Remove superfluous header guards from file-opt-9p
             - Add note about virtfs-proxy-helper being disabled
               on non-Linux for this patch series]
Signed-off-by: Will Cohen <wwcohen@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <20220227223522.91937-2-wwcohen@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>

update seabios binaries to 1.16.0

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

update seabios submodule to 1.16.0

git shortlog 6a62e0cb0dfe..rel-1.16.0
-------------------------------------

Florian Larysch (1):
      nvme: fix LBA format data structure

Jan Beulich via SeaBIOS (1):
      nvme: avoid use-after-free in nvme_controller_enable()

Kevin O'Connor (9):
      smm: Suppress gcc array-bounds warnings
      nvme: Rework nvme_io_readwrite() to return -1 on error
      nvme: Add nvme_bounce_xfer() helper function
      nvme: Convert nvme_build_prpl() to nvme_prpl_xfer()
      nvme: Pass prp1 and prp2 directly to nvme_io_xfer()
      nvme: Build the page list in the existing dma buffer
      nvme: Only allocate one dma bounce buffer for all nvme drives
      sercon: Fix missing GET_LOW() to access rx_bytes
      docs: Note v1.16.0 release

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

MAINTAINERS: Update the files in the FreeBSD section

The FreeBSD CI definitions now reside in other files than .cirrs.yml.
Update the entry in MAINTAINERS accordingly.

Message-Id: <20220217141138.917292-1-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Cleanup of mie3 tests.

Adds clobbers and merges remaining separate asm statements.

Signed-off-by: David Miller <dmiller423@gmail.com>
Message-Id: <20220301214305.2778-1-dmiller423@gmail.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
[thuth: dropped changes to mie3-compl.c, whitespace fixes]
Signed-off-by: Thomas Huth <thuth@redhat.com>

iotests/image-fleecing: test push backup with fleecing

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20220303194349.2304213-17-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests/image-fleecing: add test case with bitmap

Note that reads zero areas (not dirty in the bitmap) fails, that's
correct.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20220303194349.2304213-16-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests.py: add qemu_io_pipe_and_status()

Add helper that returns both status and output, to be used in the
following commit

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20220303194349.2304213-15-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests/image-fleecing: add test-case for fleecing format node

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-14-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block: copy-before-write: realize snapshot-access API

Current scheme of image fleecing looks like this:

[guest]                    [NBD export]
  |                              |
  |root                          | root
  v                              v
[copy-before-write] -----> [temp.qcow2]
  |                 target  |
  |file                     |backing
  v                         |
[active disk] <-------------+

 - On guest writes copy-before-write filter copies old data from active
   disk to temp.qcow2. So fleecing client (NBD export) when reads
   changed regions from temp.qcow2 image and unchanged from active disk
   through backing link.

This patch makes possible new image fleecing scheme:

[guest]                   [NBD export]
   |                            |
   | root                       | root
   v                 file       v
[copy-before-write]<------[snapshot-access]
   |           |
   | file      | target
   v           v
[active-disk] [temp.img]

 - copy-before-write does CBW operations and also provides
   snapshot-access API. The API may be accessed through
   snapshot-access driver.

Benefits of new scheme:

1. Access control: if remote client try to read data that not covered
   by original dirty bitmap used on copy-before-write open, client gets
   -EACCES.

2. Discard support: if remote client do DISCARD, this additionally to
   discarding data in temp.img informs block-copy process to not copy
   these clusters. Next read from discarded area will return -EACCES.
   This is significant thing: when fleecing user reads data that was
   not yet copied to temp.img, we can avoid copying it on further guest
   write.

3. Synchronisation between client reads and block-copy write is more
   efficient. In old scheme we just rely on BDRV_REQ_SERIALISING flag
   used for writes to temp.qcow2. New scheme is less blocking:
     - fleecing reads are never blocked: if data region is untouched or
       in-flight, we just read from active-disk, otherwise we read from
       temp.img
     - writes to temp.img are not blocked by fleecing reads
     - still, guest writes of-course are blocked by in-flight fleecing
       reads, that currently read from active-disk - it's the minimum
       necessary blocking

4. Temporary image may be of any format, as we don't rely on backing
   feature.

5. Permission relation are simplified. With old scheme we have to share
   write permission on target child of copy-before-write, otherwise
   backing link conflicts with copy-before-write file child write
   permissions. With new scheme we don't have backing link, and
   copy-before-write node may have unshared access to temporary node.
   (Not realized in this commit, will be in future).

6. Having control on fleecing reads we'll be able to implement
   alternative behavior on failed copy-before-write operations.
   Currently we just break guest request (that's a historical behavior
   of backup). But in some scenarios it's a bad behavior: better
   is to drop the backup as failed but don't break guest request.
   With new scheme we can simply unset some bits in a bitmap on CBW
   failure and further fleecing reads will -EACCES, or something like
   this. (Not implemented in this commit, will be in future)
   Additional application for this is implementing timeout for CBW
   operations.

Iotest 257 output is updated, as two more bitmaps now live in
copy-before-write filter.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20220303194349.2304213-13-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block: introduce snapshot-access block driver

The new block driver simply utilizes snapshot-access API of underlying
block node.

In further patches we want to use it like this:

[guest]                   [NBD export]
   |                            |
   | root                       | root
   v                 file       v
[copy-before-write]<------[snapshot-access]
   |           |
   | file      | target
   v           v
[active-disk] [temp.img]

This way, NBD client will be able to read snapshotted state of active
disk, when active disk is continued to be written by guest. This is
known as "fleecing", and currently uses another scheme based on qcow2
temporary image which backing file is active-disk. New scheme comes
with benefits - see next commit.

The other possible application is exporting internal snapshots of
qcow2, like this:

[guest]          [NBD export]
   |                  |
   | root             | root
   v       file       v
[qcow2]<---------[snapshot-access]

For this, we'll need to implement snapshot-access API handlers in
qcow2 driver, and improve snapshot-access block driver (and API) to
make it possible to select snapshot by name. Another thing to improve
is size of snapshot. Now for simplicity we just use size of bs->file,
which is OK for backup, but for qcow2 snapshots export we'll need to
imporve snapshot-access API to get size of snapshot.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20220303194349.2304213-12-vsementsov@virtuozzo.com>
[hreitz: Rebased on block GS/IO split]
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/io: introduce block driver snapshot-access API

Add new block driver handlers and corresponding generic wrappers.
It will be used to allow copy-before-write filter to provide
reach fleecing interface in further commit.

In future this approach may be used to allow reading qcow2 internal
snapshots, for example to export them through NBD.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-11-vsementsov@virtuozzo.com>
[hreitz: Rebased on block GS/IO split]
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/reqlist: add reqlist_wait_all()

Add function to wait for all intersecting requests.
To be used in the further commit.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Nikita Lapshin <nikita.lapshin@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-10-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/dirty-bitmap: introduce bdrv_dirty_bitmap_status()

Add a convenient function similar with bdrv_block_status() to get
status of dirty bitmap.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-9-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/reqlist: reqlist_find_conflict(): use ranges_overlap()

Let's reuse convenient helper.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-8-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block: intoduce reqlist

Split intersecting-requests functionality out of block-copy to be
reused in copy-before-write filter.

Note: while being here, fix tiny typo in MAINTAINERS.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-7-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/block-copy: add block_copy_reset()

Split block_copy_reset() out of block_copy_reset_unallocated() to be
used separately later.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-6-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/copy-before-write: add bitmap open parameter

This brings "incremental" mode to copy-before-write filter: user can
specify bitmap so that filter will copy only "dirty" areas.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20220303194349.2304213-5-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/block-copy: block_copy_state_new(): add bitmap parameter

This will be used in the following commit to bring "incremental" mode
to copy-before-write filter.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-4-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/dirty-bitmap: bdrv_merge_dirty_bitmap(): add return value

That simplifies handling failure in existing code and in further new
usage of bdrv_merge_dirty_bitmap().

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-3-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/block-copy: move copy_bitmap initialization to block_copy_state_new()

We are going to complicate bitmap initialization in the further
commit. And in future, backup job will be able to work without filter
(when source is immutable), so we'll need same bitmap initialization in
copy-before-write filter and in backup job. So, it's reasonable to do
it in block-copy.

Note that for now cbw_open() is the only caller of
block_copy_state_new().

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220303194349.2304213-2-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests: Write test output to TEST_DIR

Drop the use of OUTPUT_DIR (test/qemu-iotests under the build
directory), and instead write test output files (.out.bad, .notrun, and
.casenotrun) to TEST_DIR.

With this, the same test can be run concurrently without the separate
instances interfering, because they will need separate TEST_DIRs anyway.
Running the same test separately is useful when running the iotests with
various format/protocol combinations in parallel, or when you just want
to aggressively exercise a single test (e.g. when it fails only
sporadically).

Putting this output into TEST_DIR means that it will stick around for
inspection after the test run is done (though running the same test in
the same TEST_DIR will overwrite it, just as it used to be); but given
that TEST_DIR is a scratch directory, it should be clear that users can
delete all of its content at any point.  (And if TEST_DIR is on tmpfs,
it will just disappear on shutdown.)  Contrarily, alternative approaches
that would put these output files into OUTPUT_DIR with some prefix to
differentiate between separate test runs might easily lead to cluttering
OUTPUT_DIR.

(This change means OUTPUT_DIR is no longer written to by the iotests, so
we can drop its usage altogether.)

Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220221172909.762858-1-hreitz@redhat.com>
[hreitz: Simplified `Path(os.path.join(x, y))` to `Path(x, y)`, as
         suggested by Vladimir; and rebased on 9086c7639822b6
         ("tests/qemu-iotests: Rework the checks and spots using GNU
         sed")]
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

tests/qemu-iotests/testrunner: Quote "case not run" lines in TAP mode

In TAP mode, the stdout is reserved for the TAP protocol, so we
have to make sure to mark other lines with a comment '#' character
at the beginning to avoid that the TAP parser at the other end
gets confused.

To test this condition, run "configure" for example with:

 --block-drv-rw-whitelist=copy-before-write,qcow2,raw,file,host_device,blkdebug,null-co,copy-on-read

so that iotest 041 will report that some tests are not run due to
the missing "quorum" driver. Without this change, "make check-block"
fails since the meson tap parser gets confused by these messages.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220223124353.3273898-1-thuth@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

tests/qemu-iotests/040: Skip TestCommitWithFilters without 'throttle'

iotest 040 already has some checks for the availability of the 'throttle'
driver, but some new code has been added in the course of time that
depends on 'throttle' but does not check for its availability. Add
a check to the TestCommitWithFilters class so that this iotest now
also passes again if 'throttle' has not been enabled in the QEMU
binaries.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220223123127.3206042-1-thuth@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block: fix preallocate filter: don't do unaligned preallocate requests

There is a bug in handling BDRV_REQ_NO_WAIT flag: we still may wait in
wait_serialising_requests() if request is unaligned. And this is
possible for the only user of this flag (preallocate filter) if
underlying file is unaligned to its request_alignment on start.

So, we have to fix preallocate filter to do only aligned preallocate
requests.

Next, we should fix generic block/io.c somehow. Keeping in mind that
preallocate is the only user of BDRV_REQ_NO_WAIT and that we have to
fix its behavior now, it seems more safe to just assert that we never
use BDRV_REQ_NO_WAIT with unaligned requests and add corresponding
comment. Let's do so.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Message-Id: <20220215121609.38570-1-vsementsov@virtuozzo.com>
[hreitz: Rebased on block GS/IO split]
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/curl.c: Check error return from curl_easy_setopt()

Coverity points out that we aren't checking the return value
from curl_easy_setopt() for any of the calls to it we make
in block/curl.c.

Some of these options are documented as always succeeding (e.g.
CURLOPT_VERBOSE) but others have documented failure cases (e.g.
CURLOPT_URL).  For consistency we check every call, even the ones
that theoretically cannot fail.

Fixes: Coverity CID 1459336, 1459482, 1460331
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20220222152341.850419-3-peter.maydell@linaro.org>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block/curl.c: Set error message string if curl_init_state() fails

In curl_open(), the 'out' label assumes that the state->errmsg string
has been set (either by curl_easy_perform() or by manually copying a
string into it); however if curl_init_state() fails we will jump to
that label without setting the string.  Add the missing error string
setup.

(We can't be specific about the cause of failure: the documentation
of curl_easy_init() just says "If this function returns NULL,
something went wrong".)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20220222152341.850419-2-peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

ide: Increment BB in-flight counter for TRIM BH

When we still have an AIOCB registered for DMA operations, we try to
settle the respective operation by draining the BlockBackend associated
with the IDE device.

However, this assumes that every DMA operation is associated with an
increment of the BlockBackend’s in-flight counter (e.g. through some
ongoing I/O operation), so that draining the BB until its in-flight
counter reaches 0 will settle all DMA operations.  That is not the case:
For TRIM, the guest can issue a zero-length operation that will not
result in any I/O operation forwarded to the BlockBackend, and also not
increment the in-flight counter in any other way.  In such a case,
blk_drain() will be a no-op if no other operations are in flight.

It is clear that if blk_drain() is a no-op, the value of
s->bus->dma->aiocb will not change between checking it in the `if`
condition and asserting that it is NULL after blk_drain().

The particular problem is that ide_issue_trim() creates a BH
(ide_trim_bh_cb()) to settle the TRIM request: iocb->common.cb() is
ide_dma_cb(), which will either create a new request, or find the
transfer to be done and call ide_set_inactive(), which clears
s->bus->dma->aiocb.  Therefore, the blk_drain() must wait for
ide_trim_bh_cb() to run, which currently it will not always do.

To fix this issue, we increment the BlockBackend's in-flight counter
when the TRIM operation begins (in ide_issue_trim(), when the
ide_trim_bh_cb() BH is created) and decrement it when ide_trim_bh_cb()
is done.

Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2029980
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220120142259.120189-1-hreitz@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Tested-by: John Snow <jsnow@redhat.com>

tests/tcg/s390x: Fix the exrl-trt* tests with Clang

The exrl-trt* tests use two pre-initialized variables for the
results of the assembly code:

    uint64_t r1 = 0xffffffffffffffffull;
    uint64_t r2 = 0xffffffffffffffffull;

But then the assembly code copies over the full contents
of the register into the output variable, without taking
care of this pre-initialized values:

        "    lgr %[r1],%%r1\n"
        "    lgr %[r2],%%r2\n"

The code then finally compares the register contents to
a value that apparently depends on the pre-initialized values:

    if (r2 != 0xffffffffffffffaaull) {
        write(1, "bad r2\n", 7);
        return 1;
    }

This all works with GCC, since the 0xffffffffffffffff got into
the r2 register there by accident, but it fails completely with
Clang.

Let's fix this by declaring the r1 and r2 variables as proper
register variables instead, so the pre-initialized values get
correctly passed into the inline assembly code.

Message-Id: <20220301092431.1448419-1-thuth@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/s390x: Fix mvc, mvo and pack tests with Clang

These instructions use addressing with a "base address", meaning
that if register r0 is used, it is always treated as zero, no matter
what value is stored in the register. So we have to make sure not
to use register r0 for these instructions in our tests. There was
no problem with GCC so far since it seems to always pick other
registers by default, but Clang likes to chose register r0, too,
so we have to use the "a" constraint to make sure that it does
not pick r0 here.

Message-Id: <20220301093911.1450719-1-thuth@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

accel/tcg: Remove pointless CPUArchState casts

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220305233415.64627-2-philippe.mathieu.daude@gmail.com>

target/i386: Remove pointless CPUArchState casts

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220305233415.64627-3-philippe.mathieu.daude@gmail.com>

target: Use ArchCPU as interface to target CPU

ArchCPU is our interface with target-specific code. Use it as
a forward-declared opaque pointer (abstract type), having its
structure defined by each target.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-15-f4bug@amsat.org>

target: Introduce and use OBJECT_DECLARE_CPU_TYPE() macro

Replace the boilerplate code to declare CPU QOM types
and macros, and forward-declare the CPU instance type.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-14-f4bug@amsat.org>

target: Use CPUArchState as interface to target-specific CPU state

While CPUState is our interface with generic code, CPUArchState is
our interface with target-specific code. Use CPUArchState as an
abstract type, defined by each target.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-13-f4bug@amsat.org>

target: Use forward declared type instead of structure type

The CPU / CPU state are forward declared.

  $ git grep -E 'struct [A-Za-z]+CPU\ \*'
  target/arm/hvf_arm.h:16:void hvf_arm_set_cpu_features_from_host(struct ARMCPU *cpu);
  target/openrisc/cpu.h:234:    int (*cpu_openrisc_map_address_code)(struct OpenRISCCPU *cpu,
  target/openrisc/cpu.h:238:    int (*cpu_openrisc_map_address_data)(struct OpenRISCCPU *cpu,

  $ git grep -E 'struct CPU[A-Za-z0-9]+State\ \*'
  target/mips/internal.h:137:    int (*map_address)(struct CPUMIPSState *env, hwaddr *physical, int *prot,
  target/mips/internal.h:139:    void (*helper_tlbwi)(struct CPUMIPSState *env);
  target/mips/internal.h:140:    void (*helper_tlbwr)(struct CPUMIPSState *env);
  target/mips/internal.h:141:    void (*helper_tlbp)(struct CPUMIPSState *env);
  target/mips/internal.h:142:    void (*helper_tlbr)(struct CPUMIPSState *env);
  target/mips/internal.h:143:    void (*helper_tlbinv)(struct CPUMIPSState *env);
  target/mips/internal.h:144:    void (*helper_tlbinvf)(struct CPUMIPSState *env);
  target/xtensa/cpu.h:347:    struct CPUXtensaState *env;
  ...

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-12-f4bug@amsat.org>

target/hexagon: Add missing 'hw/core/cpu.h' include

HexagonCPU field parent_class is of type CPUClass, which
is declared in "hw/core/cpu.h".

Reviewed-by: Damien Hedde <damien.hedde@greensocs.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-11-f4bug@amsat.org>

target: Include missing 'cpu.h'

These target-specific files use the target-specific CPU state
but lack to include "cpu.h"; i.e.:

    ../target/riscv/pmp.h:61:23: error: unknown type name 'CPURISCVState'
    void pmpcfg_csr_write(CPURISCVState *env, uint32_t reg_index,
                          ^
    ../target/nios2/mmu.h:43:18: error: unknown type name 'CPUNios2State'
    void mmu_flip_um(CPUNios2State *env, unsigned int um);
                     ^
    ../target/microblaze/mmu.h:88:19: error: unknown type name 'CPUMBState'; did you mean 'CPUState'?
    uint32_t mmu_read(CPUMBState *env, bool ea, uint32_t rn);
                      ^~~~~~~~~~
                      CPUState

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-10-f4bug@amsat.org>

Hexagon (target/hexagon) convert to OBJECT_DECLARE_TYPE

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220211033034.21107-1-tsimpson@quicinc.com>
[PMD: Add missing "qom/object.h" include]
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

target/i386/tcg/sysemu: Include missing 'exec/exec-all.h' header

excp_helper.c requires "exec/exec-all.h" for tlb_set_page_with_attrs()
and misc_helper.c for tlb_flush().

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-8-f4bug@amsat.org>

cpu: Add missing 'exec/exec-all.h' and 'qemu/accel.h' headers

cpu.c requires "exec/exec-all.h" to call tlb_flush() and
"qemu/accel.h" to call accel_cpu_realizefn().

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-6-f4bug@amsat.org>

exec/cpu_ldst: Include 'cpu.h' to get target_ulong definition

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-5-f4bug@amsat.org>

meson: Display libfdt as disabled when system emulation is disabled

When configuring QEMU with --disable-system, meson keeps showing
libfdt as "auto". Mark it as disabled instead.

Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220214183144.27402-2-f4bug@amsat.org>

softmmu: Build target-agnostic objects once

Various softmmu objects aren't target specific. Move them
to the generic softmmu source set.

For our 31 softmmu targets, this is in total 330 objects
less to build!

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-22-f4bug@amsat.org>

softmmu: Add qemu_init_arch_modules()

module_allow_arch() is the single target-specific call in the
whole vl.c file. Move the module initialization out to arch_init.c,
that way we'll be able to build vl.o once for all targets (the
next commit).

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-21-f4bug@amsat.org>

exec/cpu: Make address_space_init/reloading_memory_map target agnostic

cpu_address_space_init() and cpu_reloading_memory_map() are
target-agnostic, but are declared in "exec/exec-all.h" which
contains target-specific declarations. Any target-agnostic
source including "exec/exec-all.h" becomes target-specific and
we have to compile it N times for the N targets built.  In order
to avoid that, move the declarations to "exec/cpu-common.h" which
only contains target-agnostic declarations.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-20-f4bug@amsat.org>

exec/gdbstub: Make gdb_exit() / gdb_set_stop_cpu() target agnostic

gdb_exit() and gdb_set_stop_cpu() prototypes don't have to be
target specific. Remove this limitation to be able to build
softmmu/cpus.c and softmmu/runstate.c once for all targets.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-19-f4bug@amsat.org>

misc: Add missing "sysemu/cpu-timers.h" include

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-18-f4bug@amsat.org>

misc: Remove unnecessary "sysemu/cpu-timers.h" include

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-17-f4bug@amsat.org>

softmmu/cpu-timers: Remove unused 'exec/exec-all.h' header

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-16-f4bug@amsat.org>

softmmu/physmem: Remove unnecessary include

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-14-f4bug@amsat.org>

softmmu/globals: Remove unused 'hw/i386/*' headers

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-13-f4bug@amsat.org>

accel: Introduce AccelOpsClass::cpus_are_resettable()

Add cpus_are_resettable() to AccelOps, and implement it for the
KVM accelerator.

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-12-f4bug@amsat.org>

accel: Introduce AccelOpsClass::cpu_thread_is_idle()

Add cpu_thread_is_idle() to AccelOps, and implement it for the
KVM / WHPX accelerators.

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-11-f4bug@amsat.org>

softmmu/cpus: Code movement

We want cpu_thread_is_idle() to use cpus_accel, so declare this
variable earlier.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220207075426.81934-10-f4bug@amsat.org>