* remotes/rth/tags/pull-tcg-20211016: (24 commits)
Revert "cpu: Move cpu_common_props to hw/core/cpu.c"
target/xtensa: Drop check for singlestep_enabled
target/tricore: Drop check for singlestep_enabled
target/sh4: Drop check for singlestep_enabled
target/s390x: Drop check for singlestep_enabled
target/rx: Drop checks for singlestep_enabled
target/riscv: Remove exit_tb and lookup_and_goto_ptr
target/riscv: Remove dead code after exception
target/ppc: Drop exit checks for singlestep_enabled
target/openrisc: Drop checks for singlestep_enabled
target/mips: Drop exit checks for singlestep_enabled
target/mips: Fix single stepping
target/microblaze: Drop checks for singlestep_enabled
target/microblaze: Check CF_NO_GOTO_TB for DISAS_JUMP
target/m68k: Drop checks for singlestep_enabled
target/i386: Drop check for singlestep_enabled
target/i386: Check CF_NO_GOTO_TB for dc->jmp_opt
target/hppa: Drop checks for singlestep_enabled
target/arm: Drop checks for singlestep_enabled
target/hexagon: Drop checks for singlestep_enabled
...
Despite a comment saying why cpu_common_props cannot be placed in
a file that is compiled once, it was moved anyway. Revert that.
Since then, Property is not defined in hw/core/cpu.h, so it is now
easier to declare a function to install the properties rather than
the Property array itself.
As per an ancient comment in mips_tr_translate_insn about the
expectations of gdb, when restarting the insn in a delay slot
we also re-execute the branch. Which means that we are
expected to execute two insns in this case.
This has been broken since 8b86d6d2580, where we forced max_insns
to 1 while single-stepping. This resulted in an exit from the
translator loop after the branch but before the delay slot is
translated.
Increase the max_insns to 2 for this case. In addition, bypass
the end-of-page check, for when the branch itself ends the page.
block-backend: update blk_co_pwrite() and blk_co_pread() wrappers
Make bytes argument int64_t to be consistent with modern block-layer.
Callers should be OK with it as type becomes wider.
What is inside functions?
- Conversion from int64_t to size_t. Still, we
can't have a buffer larger than SIZE_MAX, therefore bytes should not be
larger than SIZE_MAX as well. Add an assertion.
- Passing to blk_co_pwritev() / blk_co_preadv() which already has
int64_t bytes argument.
block-backend: drop INT_MAX restriction from blk_check_byte_request()
blk_check_bytes_request is called from blk_co_do_preadv,
blk_co_do_pwritev_part, blk_co_do_pdiscard and blk_co_copy_range
before (maybe) calling throttle_group_co_io_limits_intercept() (which
has int64_t argument) and then calling corresponding bdrv_co_ function.
bdrv_co_ functions are OK with int64_t bytes as well.
So dropping the check for INT_MAX we just get same restrictions as in
bdrv_ layer: discard and write-zeroes goes through
bdrv_check_qiov_request() and are allowed to be 64bit. Other requests
go through bdrv_check_request32() and still restricted by INT_MAX
boundary.
block-backend: convert blk_aio_ functions to int64_t bytes paramter
1. Convert bytes in BlkAioEmAIOCB:
aio->bytes is only passed to already int64_t interfaces, and set in
blk_aio_prwv, which is updated here.
2. For all updated functions the parameter type becomes wider so callers
are safe.
3. In blk_aio_prwv we only store bytes to BlkAioEmAIOCB, which is
updated here.
4. Other updated functions are wrappers on blk_aio_prwv.
Note that blk_aio_preadv and blk_aio_pwritev become safer: before this
commit, it's theoretically possible to pass qiov with size exceeding
INT_MAX, which than converted to int argument of blk_aio_prwv. Now it's
converted to int64_t which is a lot better. Still add assertions.
block-backend: convert blk_co_copy_range to int64_t bytes
Function is updated so that parameter type becomes wider, so all
callers should be OK with it.
Look at blk_co_copy_range() itself: bytes is passed only to
blk_check_byte_request() and bdrv_co_copy_range(), which already have
int64_t bytes parameter, so we are OK.
Note that requests exceeding INT_MAX are still restricted by
blk_check_byte_request().
block-backend: convert blk_foo wrappers to use int64_t bytes parameter
Convert blk_pdiscard, blk_pwrite_compressed, blk_pwrite_zeroes.
These are just wrappers for functions with int64_t argument, so allow
passing int64_t as well. Parameter type becomes wider so all callers
should be OK with it.
Note that requests exceeding INT_MAX are still restricted by
blk_check_byte_request().
Note also that we don't (and are not going to) convert blk_pwrite and
blk_pread: these functions return number of bytes on success, so to
update them, we should change return type to int64_t as well, which
will lead to investigating and updating all callers which is too much.
block-backend: convert blk_co_pdiscard to int64_t bytes
We updated blk_do_pdiscard() and its wrapper blk_co_pdiscard(). Both
functions are updated so that the parameter type becomes wider, so all
callers should be OK with it.
Look at blk_do_pdiscard(): bytes is passed only to
blk_check_byte_request() and bdrv_co_pdiscard(), which already have
int64_t bytes parameter, so we are OK.
Note that requests exceeding INT_MAX are still restricted by
blk_check_byte_request().
For both updated functions, the type of bytes becomes wider, so all callers
should be OK with it.
blk_co_preadv() only passes its arguments to blk_do_preadv().
blk_do_preadv() passes bytes to:
- trace_blk_co_preadv, which is updated too
- blk_check_byte_request, throttle_group_co_io_limits_intercept,
bdrv_co_preadv, which are already int64_t.
Note that requests exceeding INT_MAX are still restricted by
blk_check_byte_request().
Rename size and make it int64_t to correspond to modern block layer,
which always uses int64_t for offset and bytes (not in blk layer yet,
which is a task for following commits).
All callers pass int or unsigned int.
So, for bytes in [0, INT_MAX] nothing is changed, for negative bytes we
now fail on "bytes < 0" check instead of "bytes > INT_MAX" check.
Note, that blk_check_byte_request() still doesn't allow requests
exceeding INT_MAX.
* remotes/kwolf/tags/for-upstream:
vl: Enable JSON syntax for -device
qdev: Base object creation on QDict rather than QemuOpts
virtio-net: Avoid QemuOpts in failover_find_primary_device()
virtio-net: Store failover primary opts pointer locally
qdev: Add Error parameter to hide_device() callbacks
qemu-option: Allow deleting opts during qemu_opts_foreach()
softmmu/qdev-monitor: add error handling in qdev_set_id
qdev: Make DeviceState.id independent of QemuOpts
qdev: Avoid using string visitor for properties
iotests/051: Fix typo
iotests/245: Fix type for iothread property
qom: Reduce use of error_propagate()
net/vhost-vdpa: Fix device compatibility check
net/vhost-user: Fix device compatibility check
net: Introduce NetClientInfo.check_peer_type()
Merge remote-tracking branch 'remotes/thuth/tags/pull-request-2021-10-15' into staging
* Check kernel command line size on s390x
* Simplification of one of the SIGP instructions on s390x
* Cornelia stepping down as maintainer in some subsystems
* Update the dtc submodule to a proper release version
* remotes/thuth/tags/pull-request-2021-10-15:
dtc: Update to version 1.6.1
s390x virtio-ccw machine: step down as maintainer
s390x/kvm: step down as maintainer
vfio-ccw: step down as maintainer
s390x: sigp: Force Set Architecture to return Invalid Parameter
s390x/ipl: check kernel command line size
Kevin Wolf [Fri, 8 Oct 2021 13:34:42 +0000 (15:34 +0200)]
vl: Enable JSON syntax for -device
Like we already do for -object, introduce support for JSON syntax in
-device, which can be kept stable in the long term and guarantees that a
single code path with identical behaviour is used for both QMP and the
command line. Compared to the QemuOpts based code, the parser contains
less surprises and has support for non-scalar options (lists and
structs). Switching management tools to JSON means that we can more
easily change the "human" CLI syntax from QemuOpts to the keyval parser
later.
In the QAPI schema, a feature flag is added to the device-add command to
allow management tools to detect support for this.
Kevin Wolf [Fri, 8 Oct 2021 13:34:41 +0000 (15:34 +0200)]
qdev: Base object creation on QDict rather than QemuOpts
QDicts are both what QMP natively uses and what the keyval parser
produces. Going through QemuOpts isn't useful for either one, so switch
the main device creation function to QDicts. By sharing more code with
the -object/object-add code path, we can even reduce the code size a
bit.
This commit doesn't remove the detour through QemuOpts from any code
path yet, but it allows the following commits to do so.
Kevin Wolf [Fri, 8 Oct 2021 13:34:40 +0000 (15:34 +0200)]
virtio-net: Avoid QemuOpts in failover_find_primary_device()
Don't go through the global QemuOptsList, it is state of the legacy
command line parser and we will create devices that are not contained
in it. It is also just the command line configuration and not
necessarily the current runtime state.
Instead, look at the qdev device tree which has the current state of all
existing devices.
Kevin Wolf [Fri, 8 Oct 2021 13:34:39 +0000 (15:34 +0200)]
virtio-net: Store failover primary opts pointer locally
Instead of accessing the global QemuOptsList, which really belong to the
command line parser and shouldn't be accessed from devices, store a
pointer to the QemuOpts in a new VirtIONet field.
This is not the final state, but just an intermediate step to get rid of
QemuOpts in devices. It will later be replaced with an options QDict.
Before this patch, two "primary" devices could be hidden for the same
standby device, but only one of them would actually be enabled and the
other one would be kept hidden forever, so this doesn't make sense.
After this patch, configuring a second primary device is an error.
Kevin Wolf [Fri, 8 Oct 2021 13:34:38 +0000 (15:34 +0200)]
qdev: Add Error parameter to hide_device() callbacks
hide_device() is used for virtio-net failover, where the standby virtio
device delays creation of the primary device. It only makes sense to
have a single primary device for each standby device. Adding a second
one should result in an error instead of hiding it and never using it
afterwards.
Prepare for this by adding an Error parameter to the hide_device()
callback where virtio-net is informed about adding a primary device.
Damien Hedde [Fri, 8 Oct 2021 13:34:36 +0000 (15:34 +0200)]
softmmu/qdev-monitor: add error handling in qdev_set_id
qdev_set_id() is mostly used when the user adds a device (using
-device cli option or device_add qmp command). This commit adds
an error parameter to handle the case where the given id is
already taken.
Also document the function and add a return value in order to
be able to capture success/failure: the function now returns the
id in case of success, or NULL in case of failure.
The commit modifies the 2 calling places (qdev-monitor and
xen-legacy-backend) to add the error object parameter.
Note that the id is, right now, guaranteed to be unique because
all ids came from the "device" QemuOptsList where the id is used
as key. This addition is a preparation for a future commit which
will relax the uniqueness.
Kevin Wolf [Fri, 8 Oct 2021 13:34:35 +0000 (15:34 +0200)]
qdev: Make DeviceState.id independent of QemuOpts
DeviceState.id is a pointer to a string that is stored in the QemuOpts
object DeviceState.opts and freed together with it. We want to create
devices without going through QemuOpts in the future, so make this a
separately allocated string.
Kevin Wolf [Fri, 8 Oct 2021 13:34:34 +0000 (15:34 +0200)]
qdev: Avoid using string visitor for properties
The only thing the string visitor adds compared to a keyval visitor is
list support. git grep for 'visit_start_list' and 'visit.*List' shows
that devices don't make use of this.
In a world with a QAPIfied command line interface, the keyval visitor is
used to parse the command line. In order to make sure that no devices
start using this feature that would make backwards compatibility harder,
just switch away from object_property_parse(), which internally uses the
string visitor, to a keyval visitor and object_property_set().
Kevin Wolf [Fri, 8 Oct 2021 13:34:33 +0000 (15:34 +0200)]
iotests/051: Fix typo
The iothread isn't called 'iothread0', but 'thread0'. Depending on the
order that properties are parsed, the error message may change from the
expected one to another one saying that the iothread doesn't exist.
Kevin Wolf [Fri, 8 Oct 2021 13:34:30 +0000 (15:34 +0200)]
net/vhost-vdpa: Fix device compatibility check
vhost-vdpa works only with specific devices. At startup, it second
guesses what the command line option handling will do and error out if
it thinks a non-virtio device will attach to them.
This second guessing is not only ugly, it can lead to wrong error
messages ('-device floppy,netdev=foo' should complain about an unknown
property, not about the wrong kind of network device being attached) and
completely ignores hotplugging.
Drop the old checks and implement .check_peer_type() instead to fix
this. As a nice side effect, it also removes one more dependency on the
legacy QemuOpts infrastructure and even reduces the code size.
Kevin Wolf [Fri, 8 Oct 2021 13:34:29 +0000 (15:34 +0200)]
net/vhost-user: Fix device compatibility check
vhost-user works only with specific devices. At startup, it second
guesses what the command line option handling will do and error out if
it thinks a non-virtio device will attach to them.
This second guessing is not only ugly, it can lead to wrong error
messages ('-device floppy,netdev=foo' should complain about an unknown
property, not about the wrong kind of network device being attached) and
completely ignores hotplugging.
Drop the old checks and implement .check_peer_type() instead to fix
this. As a nice side effect, it also removes one more dependency on the
legacy QemuOpts infrastructure and even reduces the code size.
Kevin Wolf [Fri, 8 Oct 2021 13:34:28 +0000 (15:34 +0200)]
net: Introduce NetClientInfo.check_peer_type()
Some network backends (vhost-user and vhost-vdpa) work only with
specific devices. At startup, they second guess what the command line
option handling will do and error out if they think a non-virtio device
will attach to them.
This second guessing is not only ugly, it can lead to wrong error
messages ('-device floppy,netdev=foo' should complain about an unknown
property, not about the wrong kind of network device being attached) and
completely ignores hotplugging.
Add a callback where backends can check compatibility with a device when
it actually tries to attach, even on hotplug.
* remotes/bonzini/tags/for-upstream: (26 commits)
configure: automatically parse command line for meson -D options
meson-buildoptions: include list of tracing backends
configure: prepare for auto-generated option parsing
configure: accept "internal" for --enable-capstone/slirp/fdt
configure: remove deprecated --{enable, disable}-git-update
configure, meson: move more compiler checks to Meson
configure: remove obsolete Solaris ar check
configure, meson: move Spice configure handling to meson
configure, meson: move netmap detection to meson
configure, meson: move vde detection to meson
configure, meson: move libaio check to meson.build
configure, meson: move pthread_setname_np checks to Meson
configure, meson: move remaining HAVE_* compiler tests to Meson
meson: HAVE_GDB_BIN is not used by C code
configure, meson: remove CONFIG_GCOV from config-host.mak
configure, meson: get HOST_WORDS_BIGENDIAN via the machine object
configure, meson: move CONFIG_HOST_DSOSUF to Meson
trace: move configuration from configure to Meson
trace: simple: pass trace_file unmodified to config-host.h
configure, meson: move fuzzing configuration to Meson
...
Paolo Bonzini [Thu, 7 Oct 2021 13:08:29 +0000 (15:08 +0200)]
configure: automatically parse command line for meson -D options
Right now meson_options.txt lists about 90 options. Each option
needs code in configure to parse it and pass the option down to Meson as
a -D command-line argument; in addition the default must be duplicated
between configure and meson_options.txt. This series tries to remove
the code duplication by generating the case statement for those --enable
and --disable options, as well as the corresponding help text.
About 80% of the options can be handled completely by the new mechanism.
Eight meson options are not of the --enable/--disable kind. Six more need
to be parsed in configure for various reasons documented in the patch,
but they still have their help automatically generated.
The advantages are:
- less code in configure
- parsing and help is more consistent (for example --enable-blobs was
not supported)
- options are described entirely in one place, meson_options.txt.
This make it more attractive to use Meson options instead of
hand-crafted configure options and config-host.mak
A few options change name: --enable-tcmalloc and --enable-jemalloc
become --enable-malloc={tcmalloc,jemalloc}; --disable-blobs becomes
--disable-install-blobs; --enable-trace-backend becomes
--enable-trace-backends. However, the old names are allowed
for backwards compatibility.
Paolo Bonzini [Wed, 13 Oct 2021 07:51:47 +0000 (09:51 +0200)]
meson-buildoptions: include list of tracing backends
Manually patch the introspection data to include the tracing backends.
This works around a deficiency in Meson that will be fixed by
https://github.com/mesonbuild/meson/pull/9395.
Paolo Bonzini [Thu, 7 Oct 2021 13:08:28 +0000 (15:08 +0200)]
configure: prepare for auto-generated option parsing
Prepare the configure script and Makefile for automatically generated
help and parsing.
Because we need to run the script to generate the full help, we
cannot rely on the user supplying the path to a Python interpreter
with --python; therefore, the introspection output is parsed into
shell functions and stored in scripts/. The converter is written
in Python as standard for QEMU, and this commit contains a stub.
The options were deprecated in 6.0. That said, we do not really have a
formal deprecation cycle for build-time changes, since they do not affect
users.
Paolo Bonzini [Thu, 7 Oct 2021 13:06:10 +0000 (15:06 +0200)]
meson: define symbols for all available audio drivers
All drivers are now built by default if the corresponding libraries
are available, similar to how all other modules behave;
--audio-drv-list only governs the default choice of the audio driver.
Adjust the CONFIG_AUDIO_* preprocessor symbols so that they are
based on library availability rather than --audio-drv-list, so that
the tests and -audiodev help follow the new logic.
Paolo Bonzini [Thu, 7 Oct 2021 13:06:09 +0000 (15:06 +0200)]
configure, meson: move audio driver detection to Meson
This brings a change that makes audio drivers more similar to all
other modules. All drivers are built by default, while
--audio-drv-list only governs the default choice of the audio driver.
Meson options are added to disable the drivers, and the next patches
will fix the help messages and command line options, and especially
make the non-default drivers available via -audiodev.
Thomas Huth [Fri, 27 Aug 2021 12:09:01 +0000 (14:09 +0200)]
dtc: Update to version 1.6.1
The dtc submodule is currently pointing to non-release commit. It's nicer
if submodules point to release versions instead and since dtc 1.6.1 is
available now, let's update to that version.
Merge remote-tracking branch 'remotes/rth/tags/pull-tcg-20211013' into staging
Use MO_128 for 16-byte atomic memory operations.
Add cpu_ld/st_mmu memory primitives.
Move helper_ld/st memory helpers out of tcg.h.
Canonicalize alignment flags in MemOp.
* remotes/rth/tags/pull-tcg-20211013:
tcg: Canonicalize alignment flags in MemOp
tcg: Move helper_*_mmu decls to tcg/tcg-ldst.h
target/arm: Use cpu_*_mmu instead of helper_*_mmu
target/sparc: Use cpu_*_mmu instead of helper_*_mmu
target/s390x: Use cpu_*_mmu instead of helper_*_mmu
target/mips: Use 8-byte memory ops for msa load/store
target/mips: Use cpu_*_data_ra for msa load/store
accel/tcg: Move cpu_atomic decls to exec/cpu_ldst.h
accel/tcg: Add cpu_{ld,st}*_mmu interfaces
target/hexagon: Implement cpu_mmu_index
target/s390x: Use MO_128 for 16 byte atomics
target/ppc: Use MO_128 for 16 byte atomics
target/i386: Use MO_128 for 16 byte atomics
target/arm: Use MO_128 for 16 byte atomics
memory: Log access direction for invalid accesses
Pass in the context to each mini-helper, instead of an
incorrectly named "flags". Separate gen_load_fp and
gen_store_fp, away from the integer helpers.
The helper_*_mmu functions were the only thing available
when this code was written. This could have been adjusted
when we added cpu_*_mmuidx_ra, but now we can most easily
use the newest set of interfaces.
target/sparc: Use cpu_*_mmu instead of helper_*_mmu
The helper_*_mmu functions were the only thing available
when this code was written. This could have been adjusted
when we added cpu_*_mmuidx_ra, but now we can most easily
use the newest set of interfaces.
target/s390x: Use cpu_*_mmu instead of helper_*_mmu
The helper_*_mmu functions were the only thing available
when this code was written. This could have been adjusted
when we added cpu_*_mmuidx_ra, but now we can most easily
use the newest set of interfaces.
We should not have been using the helper_ret_* set of
functions, as they are supposed to be private to tcg.
Nor should we have been using the plain cpu_*_data set
of functions, as they do not handle unwinding properly.
These functions are much closer to the softmmu helper
functions, in that they take the complete MemOpIdx,
and from that they may enforce required alignment.
The previous cpu_ldst.h functions did not have alignment info,
and so did not enforce it. Retain this by adding MO_UNALN to
the MemOp that we create in calling the new functions.
Note that we are not yet enforcing alignment for user-only,
but we now have the information with which to do so.
projects / qemu.git / log