* remotes/ehabkost/tags/x86-and-machine-pull-request:
hw/i386: Rename apicid_from_topo_ids to x86_apicid_from_topo_ids
hw/i386: Update structures to save the number of nodes per package
hw/i386: Remove unnecessary initialization in x86_cpu_new
machine: Add SMP Sockets in CpuTopology
hw/i386: Consolidate topology functions
hw/i386: Introduce X86CPUTopoInfo to contain topology info
cpu: Use DeviceClass reset instead of a special CPUClass reset
machine/memory encryption: Disable mem merge
hw/i386: Rename X86CPUTopoInfo structure to X86CPUTopoIDs
i386: Add 2nd Generation AMD EPYC processors
i386: Add missing cpu feature bits in EPYC model
target/i386: Add new property note to versioned CPU models
target/i386: Add Denverton-v2 (no MPX) CPU model
Peter Maydell [Thu, 19 Mar 2020 12:33:50 +0000 (12:33 +0000)]
Merge remote-tracking branch 'remotes/cleber/tags/python-next-pull-request' into staging
Python and tests (mostly acceptance) patches 2020-03-17
# gpg: Signature made Wed 18 Mar 2020 00:16:03 GMT
# gpg: using RSA key 7ABB96EB8B46B94D5E0FE9BB657E8D33A5F209F3
# gpg: Good signature from "Cleber Rosa <[email protected]>" [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: 7ABB 96EB 8B46 B94D 5E0F E9BB 657E 8D33 A5F2 09F3
* remotes/cleber/tags/python-next-pull-request:
tests/docker: make "buildah bud" output similar to "docker build"
tests/docker: add CentOS 8 Dockerfile
Acceptance tests: add make targets to download images
Acceptance test: add "boot_linux" tests
Acceptance tests: introduce BUILD_DIR and SOURCE_DIR
python/qemu/qmp.py: QMP debug with VM label
* remotes/jnsnow/tags/ide-pull-request:
hw/ide: Remove unneeded inclusion of hw/ide.h
hw/ide: Move MAX_IDE_DEVS define to hw/ide/internal.h
hw/ide: Do ide_drive_get() within pci_ide_create_devs()
hw/ide/pci.c: Coding style update to fix checkpatch errors
hw/ide: Remove now unneded #include "hw/pci/pci.h" from hw/ide.h
hw/ide: Get rid of piix4_init function
hw/isa/piix4.c: Introduce variable to store devfn
hw/ide: Get rid of piix3_init functions
hd-geo-test: Clean up use of buf[] in create_qcow2_with_mbr()
via-ide: always use legacy IRQ 14/15 routing
via-ide: allow guests to write to PCI_CLASS_PROG
via-ide: initialise IDE controller in legacy mode
via-ide: ensure that PCI_INTERRUPT_LINE is hard-wired to its default value
pci: Honour wmask when resetting PCI_INTERRUPT_LINE
ide/via: Get rid of via_ide_init()
via-ide: move registration of VMStateDescription to DeviceClass
cmd646: remove unused pci_cmd646_ide_init() function
dp264: use pci_create_simple() to initialise the cmd646 device
cmd646: register vmstate_ide_pci VMStateDescription in DeviceClass
cmd646: register cmd646_reset() function in DeviceClass
* remotes/armbru/tags/pull-qapi-2020-03-17: (30 commits)
net: Track netdevs in NetClientState rather than QemuOpt
net: Complete qapi-fication of netdev_add
qmp: constify QmpCommand and list
qapi: Mark deprecated QMP parts with feature 'deprecated'
qapi: New special feature flag "deprecated"
qapi: Replace qmp_dispatch()'s TODO comment by an explanation
qapi: Simplify how qmp_dispatch() gets the request ID
qapi: Simplify how qmp_dispatch() deals with QCO_NO_SUCCESS_RESP
qapi: Inline do_qmp_dispatch() into qmp_dispatch()
qapi: Add feature flags to struct members
qapi/schema: Call QAPIDoc.connect_member() in just one place
qapi/schema: Rename QAPISchemaObjectType{Variant,Variants}
qapi/schema: Reorder classes so related ones are together
qapi/schema: Change _make_features() to a take feature list
qapi/introspect: Factor out _make_tree()
qapi/introspect: Rename *qlit* to reduce confusion
qapi: Consistently put @features parameter right after @ifcond
qapi: Add feature flags to remaining definitions
qapi/schema: Clean up around QAPISchemaEntity.connect_doc()
tests/test-qmp-event: Check event is actually emitted
...
Peter Maydell [Wed, 18 Mar 2020 20:25:23 +0000 (20:25 +0000)]
Merge remote-tracking branch 'remotes/stsquad/tags/pull-testing-and-gdbstub-170320-1' into staging
Testing and gdbstub updates:
- docker updates for VirGL
- re-factor gdbstub for static GDBState
- re-factor gdbstub for dynamic arrays
- add SVE support to arm gdbstub
- add some guest debug tests to check-tcg
- add aarch64 userspace register tests
- remove packet size limit to gdbstub
- simplify gdbstub monitor code
- report vContSupported in gdbstub to use proper single-step
# gpg: Signature made Tue 17 Mar 2020 17:47:46 GMT
# gpg: using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <[email protected]>" [full]
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8 DF35 FBD0 DB09 5A9E 2A44
* remotes/stsquad/tags/pull-testing-and-gdbstub-170320-1: (28 commits)
gdbstub: Fix single-step issue by confirming 'vContSupported+' feature to gdb
gdbstub: do not split gdb_monitor_write payload
gdbstub: change GDBState.last_packet to GByteArray
tests/tcg/aarch64: add test-sve-ioctl guest-debug test
tests/tcg/aarch64: add SVE iotcl test
tests/tcg/aarch64: add a gdbstub testcase for SVE registers
tests/guest-debug: add a simple test runner
configure: allow user to specify what gdb to use
tests/tcg/aarch64: userspace system register test
target/arm: don't bother with id_aa64pfr0_read for USER_ONLY
target/arm: generate xml description of our SVE registers
target/arm: default SVE length to 64 bytes for linux-user
target/arm: explicitly encode regnum in our XML
target/arm: prepare for multiple dynamic XMLs
gdbstub: extend GByteArray to read register helpers
target/i386: use gdb_get_reg helpers
target/m68k: use gdb_get_reg helpers
target/arm: use gdb_get_reg helpers
gdbstub: add helper for 128 bit registers
gdbstub: move mem_buf to GDBState and use GByteArray
...
* remotes/armbru/tags/pull-error-2020-03-17:
hw/sd/ssi-sd: fix error handling in ssi_sd_realize
xen-block: Use one Error * variable instead of two
hw/misc/ivshmem: Use one Error * variable instead of two
Use &error_abort instead of separate assert()
Peter Maydell [Wed, 18 Mar 2020 15:07:57 +0000 (15:07 +0000)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-5.0-20200317' into staging
ppc patch queue 2020-03-17
Here's my final pull request for the qemu-5.0 soft freeze. Sorry this
is just under the wire - I hit some last minute problems that took a
while to fix up and retest.
Highlights are:
* Numerous fixes for the FWNMI feature
* A handful of cleanups to the device tree construction code
* Numerous fixes for the spapr-vscsi device
* A number of fixes and cleanups for real mode (MMU off) softmmu
handling
* Fixes for handling of the PAPR RMA
* Better handling of hotplug/unplug events during boot
* Assorted other fixes
* remotes/dgibson/tags/ppc-for-5.0-20200317: (45 commits)
pseries: Update SLOF firmware image
ppc/spapr: Ignore common "ibm,nmi-interlock" Linux bug
ppc/spapr: Implement FWNMI System Reset delivery
target/ppc: allow ppc_cpu_do_system_reset to take an alternate vector
ppc/spapr: Allow FWNMI on TCG
ppc/spapr: Fix FWNMI machine check interrupt delivery
ppc/spapr: Add FWNMI System Reset state
ppc/spapr: Change FWNMI names
ppc/spapr: Fix FWNMI machine check failure handling
spapr: Rename DT functions to newer naming convention
spapr: Move creation of ibm,architecture-vec-5 property
spapr: Move creation of ibm,dynamic-reconfiguration-memory dt node
spapr/rtas: Reserve space for RTAS blob and log
pseries: Update SLOF firmware image
ppc/spapr: Move GPRs setup to one place
target/ppc: Fix rlwinm on ppc64
spapr/xive: use SPAPR_IRQ_IPI to define IPI ranges exposed to the guest
hw/scsi/spapr_vscsi: Convert debug fprintf() to trace event
hw/scsi/spapr_vscsi: Prevent buffer overflow
hw/scsi/spapr_vscsi: Do not mix SRP IU size with DMA buffer size
...
Babu Moger [Wed, 11 Mar 2020 22:53:06 +0000 (17:53 -0500)]
machine: Add SMP Sockets in CpuTopology
Store the smp sockets in CpuTopology. The socket information required to
build the apic id in EPYC mode. Right now socket information is not passed
to down when decoding the apic id. Add the socket information here.
Peter Maydell [Tue, 3 Mar 2020 10:05:11 +0000 (10:05 +0000)]
cpu: Use DeviceClass reset instead of a special CPUClass reset
The CPUClass has a 'reset' method. This is a legacy from when
TYPE_CPU used not to inherit from TYPE_DEVICE. We don't need it any
more, as we can simply use the TYPE_DEVICE reset. The 'cpu_reset()'
function is kept as the API which most places use to reset a CPU; it
is now a wrapper which calls device_cold_reset() and then the
tracepoint function.
This change should not cause CPU objects to be reset more often
than they are at the moment, because:
* nobody is directly calling device_cold_reset() or
qdev_reset_all() on CPU objects
* no CPU object is on a qbus, so they will not be reset either
by somebody calling qbus_reset_all()/bus_cold_reset(), or
by the main "reset sysbus and everything in the qbus tree"
reset that most devices are reset by
Note that this does not change the need for each machine or whatever
to use qemu_register_reset() to arrange to call cpu_reset() -- that
is necessary because CPU objects are not on any qbus, so they don't
get reset when the qbus tree rooted at the sysbus bus is reset, and
this isn't being changed here.
All the changes to the files under target/ were made using the
included Coccinelle script, except:
(1) the deletion of the now-inaccurate and not terribly useful
"CPUClass::reset" comments was done with a perl one-liner afterwards:
perl -n -i -e '/ CPUClass::reset/ or print' target/*/*.c
(2) this bit of the s390 change was done by hand, because the
Coccinelle script is not sophisticated enough to handle the
parent_reset call being inside another function:
When a host is running with memory encryption, the memory isn't visible
to the host kernel; attempts to merge that memory are futile because
what it's really comparing is encrypted memory, usually encrypted
with different keys.
Automatically turn mem-merge off when memory encryption is specified.
Babu Moger [Tue, 3 Mar 2020 19:56:58 +0000 (13:56 -0600)]
hw/i386: Rename X86CPUTopoInfo structure to X86CPUTopoIDs
Rename few data structures related to X86 topology. X86CPUTopoIDs will
have individual arch ids. Next patch introduces X86CPUTopoInfo which will
have all topology information(like cores, threads etc..).
Moger, Babu [Thu, 7 Nov 2019 18:01:04 +0000 (18:01 +0000)]
i386: Add 2nd Generation AMD EPYC processors
Adds the support for 2nd Gen AMD EPYC Processors. The model display
name will be EPYC-Rome.
Adds the following new feature bits on top of the feature bits from the
first generation EPYC models.
perfctr-core : core performance counter extensions support. Enables the VM to
use extended performance counter support. It enables six
programmable counters instead of four counters.
clzero : instruction zeroes out the 64 byte cache line specified in RAX.
xsaveerptr : XSAVE, XSAVE, FXSAVEOPT, XSAVEC, XSAVES always save error
pointers and FXRSTOR, XRSTOR, XRSTORS always restore error
pointers.
wbnoinvd : Write back and do not invalidate cache
ibpb : Indirect Branch Prediction Barrier
amd-stibp : Single Thread Indirect Branch Predictor
clwb : Cache Line Write Back and Retain
xsaves : XSAVES, XRSTORS and IA32_XSS support
rdpid : Read Processor ID instruction support
umip : User-Mode Instruction Prevention support
The Reference documents are available at
https://developer.amd.com/wp-content/resources/55803_0.54-PUB.pdf
https://www.amd.com/system/files/TechDocs/24594.pdf
Depends on following kernel commits: 40bc47b08b6e ("kvm: x86: Enumerate support for CLZERO instruction") 504ce1954fba ("KVM: x86: Expose XSAVEERPTR to the guest") 6d61e3c32248 ("kvm: x86: Expose RDPID in KVM_GET_SUPPORTED_CPUID") 52297436199d ("kvm: svm: Update svm_xsaves_supported")
Moger, Babu [Thu, 7 Nov 2019 18:00:57 +0000 (18:00 +0000)]
i386: Add missing cpu feature bits in EPYC model
Adds the following missing CPUID bits:
perfctr-core : core performance counter extensions support. Enables the VM
to use extended performance counter support. It enables six
programmable counters instead of 4 counters.
clzero : instruction zeroes out the 64 byte cache line specified in RAX.
xsaveerptr : XSAVE, XSAVE, FXSAVEOPT, XSAVEC, XSAVES always save error
pointers and FXRSTOR, XRSTOR, XRSTORS always restore error
pointers.
ibpb : Indirect Branch Prediction Barrie.
xsaves : XSAVES, XRSTORS and IA32_XSS supported.
Depends on following kernel commits: 40bc47b08b6e ("kvm: x86: Enumerate support for CLZERO instruction") 504ce1954fba ("KVM: x86: Expose XSAVEERPTR to the guest") 52297436199d ("kvm: svm: Update svm_xsaves_supported")
These new features will be added in EPYC-v3. The -cpu help output after the change.
x86 EPYC-v1 AMD EPYC Processor
x86 EPYC-v2 AMD EPYC Processor (with IBPB)
x86 EPYC-v3 AMD EPYC Processor
Cleber Rosa [Thu, 12 Mar 2020 19:36:13 +0000 (15:36 -0400)]
tests/docker: make "buildah bud" output similar to "docker build"
Podman users will most often be using buildah to build containers.
Among the differences between "buildah bud|build-using-dockerfile" and
a traditional "docker build" is that buildah does not run a container
during build.
To the best of my knowledge and experiments, this means that runtime
variables, such as ENV from one base image will not propagate into
another. The end result is that the location for the cross compiler
binaries, defined in the base "qemu/debian9-mxe" image, are not passed
through this image. Consequently, the cross compilers are not on PATH
and the build fails.
Cleber Rosa [Tue, 17 Mar 2020 14:16:54 +0000 (10:16 -0400)]
Acceptance tests: add make targets to download images
The newly introduced "boot linux" tests make use of Linux images that
are larger than usual, and fall into what Avocado calls "vmimages",
and can be referred to by name, version and architecture.
The images can be downloaded automatically during the test. But, to
make for more reliable test results, this introduces a target that
will download the vmimages for the architectures that have been
configured and are available for the currently used distro (Fedora
31).
Cleber Rosa [Tue, 17 Mar 2020 14:16:53 +0000 (10:16 -0400)]
Acceptance test: add "boot_linux" tests
This acceptance test, validates that a full blown Linux guest can
successfully boot in QEMU. In this specific case, the guest chosen is
Fedora version 31.
* x86_64, pc-i440fx and pc-q35 machine types, with TCG and KVM as
accelerators
* aarch64 and virt machine type, with TCG and KVM as accelerators
* ppc64 and pseries machine type with TCG as accelerator
* s390x and s390-ccw-virtio machine type with TCG as accelerator
The Avocado vmimage utils library is used to download and cache the
Linux guest images, and from those images a snapshot image is created
and given to QEMU. If a qemu-img binary is available in the build
directory, it's used to create the snapshot image, so that matching
qemu-system-* and qemu-img are used in the same test run. If qemu-img
is not available in the build tree, one is attempted to be found
installed system-wide (in the $PATH). If qemu-img is not found in the
build dir or in the $PATH, the test is canceled.
The method for checking the successful boot is based on "cloudinit"
and its "phone home" feature. The guest is given an ISO image with
the location of the phone home server, and the information to post
(the instance ID). Upon receiving the correct information, from the
guest, the test is considered to have PASSed.
This test is currently limited to user mode networking only, and
instructs the guest to connect to the "router" address that is hard
coded in QEMU.
To create the cloudinit ISO image that will be used to configure the
guest, the pycdlib library is also required and has been added as
requirement to the virtual environment created by "check-venv".
The console output is read by a separate thread, by means of the
Avocado datadrainer utility module.
Cleber Rosa [Tue, 17 Mar 2020 14:16:52 +0000 (10:16 -0400)]
Acceptance tests: introduce BUILD_DIR and SOURCE_DIR
Some tests may benefit from using resources from a build directory.
This introduces three variables that can help tests find resources in
those directories.
First, a BUILD_DIR is assumed to exist, given that the primary form of
running the acceptance tests is from a build directory (which may or
may not be the same as the source tree, that is, the SOURCE_DIR).
If the directory containing the acceptance tests happens to be a link
to a directory, it's assumed to it points to the source tree
(SOURCE_DIR), which is the behavior defined on the QEMU Makefiles. If
the directory containing the acceptance tests is not a link, then a
in-tree build is assumed, and the BUILD_DIR and SOURCE_DIR have the
same value.
Oksana Vohchana [Mon, 16 Mar 2020 10:32:03 +0000 (12:32 +0200)]
python/qemu/qmp.py: QMP debug with VM label
QEMUMachine writes some messages to the default logger.
But it sometimes hard to read the output if we have requests to
more than one VM.
This patch adds a label to the logger in the debug mode.
Eric Blake [Tue, 17 Mar 2020 20:17:11 +0000 (15:17 -0500)]
net: Track netdevs in NetClientState rather than QemuOpt
As mentioned in the previous patch, our use of QemuOpt group "netdev"
has two purposes: collect the CLI arguments, and serve as a witness
for monitor hotplug actions. As the latter didn't use anything but an
id, it felt rather unclean to have to touch QemuOpts at all when going
through QMP, so let's instead track things with a bool field in
NetClientState.
Eric Blake [Tue, 17 Mar 2020 20:17:10 +0000 (15:17 -0500)]
net: Complete qapi-fication of netdev_add
We've had all the required pieces for doing a type-safe representation
of netdev_add as a flat union for quite some time now (since 0e55c381f6 in v2.7.0, released in 2016), but did not make the final
switch to using it because of concern about whether a command-line
regression in accepting "1" in place of 1 for integer arguments would
be problematic. Back then, we did not have the deprecation cycle to
allow us to make progress. But now that we have waited so long, other
problems have crept in: for example, our desire to add
qemu-storage-daemon is hampered by the inability to express net
objects, and we are unable to introspect what we actually accept.
Additionally, our round-trip through QemuOpts silently eats any
argument that expands to an array, rendering dnssearch, hostfwd, and
guestfwd useless through QMP:
So without further ado, let's turn on proper QAPI. netdev_add() was a
trivial wrapper around net_client_init(), which did a few steps prior
to calling net_client_init1(); with this patch, we now skip directly
to net_client_init1(). In addition to fixing array parameters, the
following additional differences occur:
- {"execute": "netdev_add", "arguments": {"type": "help"}}
no longer attempts to print help to stdout and exit. Bug fix, broken
in 547203ead4 'net: List available netdevs with "-netdev help"',
v2.12.0.
- {"execute": "netdev_add", "arguments': {... "ipv6-net": "..." }}
no longer attempts to desugar the undocumented ipv6-net magic string
into the proper "ipv6-prefix" and "ipv6-prefixlen". Undocumented
misfeature, introduced in commit 7aac531ef2 "qapi-schema, qemu-options
& slirp: Adding Qemu options for IPv6 addresses", v2.6.0.
- {'execute':'netdev_add',
'arguments':{'id':'net2', 'type':'hubport', 'hubid':"2"}}
{"error": {"class": "GenericError", "desc": "Invalid parameter type for 'hubid', expected: integer"}}
Used to succeed: since our command line treats everything as strings,
our not-so-round-trip conversion from QAPI -> QemuOpts -> QAPI lost
the original typing and turned everything into a string; now that we
skip the QemuOpts, the JSON input has to match the exact QAPI type.
But this stricter QMP is desirable, and introspection is sufficient
for any affected applications to make sure they use it correctly.
In qmp_netdev_add(), we still have to create a QemuOpts object so that
qmp_netdev_del() will be able to remove a hotplugged network device;
but the opts->head remains empty since we now manage all parsing
through the QAPI object rather than QemuOpts; a separate patch will
address the abuse of QemuOpts as a witness for whether a
NetClientState is a netdev. In the meantime, our argument that we are
okay requires auditing all uses of option group "netdev":
- qemu_netdev_opts: option group definition, empty .desc[]
- CLI (CLI netdev parsing ends before monitors start, so while
monitors can mess with CLI netdevs, CLI cannot mess with
monitor netdevs):
- main() case QEMU_OPTION_netdev: store CLI definition
- main() case QEMU_OPTION_readconfig, case QEMU_OPTION_writeconfig:
similar, dealing only with CLI
- net_init_clients(): Pass CLI to net_client_init()
- Monitor:
- hmp_netdev_add(): straightforward parse into net_client_init()
- qmp_netdev_add(): subject of this patch, used to add full
object to option group, now just adds bare-bones id
- qmp_netdev_del(), netdev_del_completion(): check the option group
solely for id, as a 'is this a netdev' predicate
Management applications could conceivably use this for static
checking.
The deprecated commands are change, cpu-add, migrate-set-cache-size,
migrate_set_downtime, migrate_set_speed, query-cpus, query-events,
query-migrate-cache-size.
The deprecated command arguments are block-commit arguments @base and
@top, and block_set_io_throttle, blockdev-change-medium,
blockdev-close-tray, blockdev-open-tray, eject argument @device.
The deprecated command results are query-cpus-fast result @arch,
query-block result @dirty-bitmaps, query-named-block-nodes result
@encryption_key_missing and result @dirty-bitmaps's member @status.
Same for query-block result @inserted, which mirrors
query-named-block-nodes.
Unlike regular feature flags, the new special feature flag
"deprecated" is recognized by the QAPI generator. For now, it's only
permitted with commands, events, and struct members. It will be put
to use shortly.
qapi: Simplify how qmp_dispatch() gets the request ID
We convert the request object to a QDict twice: first in
qmp_dispatch() to get the request ID, and then again in
qmp_dispatch_check_obj(), which converts to QDict, then checks and
returns it. We can't get the request ID from the latter, because it's
null when the qdict flunks the checks.
Move the checked conversion to QDict from qmp_dispatch_check_obj() to
qmp_dispatch(), and drop the duplicate there.
qapi: Inline do_qmp_dispatch() into qmp_dispatch()
Both functions check @request is a QDict, and both have code for
QCO_NO_SUCCESS_RESP. This wasn't the case back when they were
created. It's a sign of muddled responsibilities. Inline. The next
commits will clean up some more.
qapi/schema: Call QAPIDoc.connect_member() in just one place
The .connect_doc() of classes that have QAPISchemaMember connect them
to their documentation. Change them to delegate the actual work to
new QAPISchemaMember.connect_doc(). Matches the .connect_doc() that
already exist.
qapi/schema: Change _make_features() to a take feature list
QAPISchema._make_features() takes a definition expression, and
extracts its 'features' member. The other ._make_FOO() leave
destructuring expressions to their callers. Change ._make_features()
to match them.
The value of @qmp_schema_qlit is generated from an expression tree.
Tree nodes are created in several places. Factor out the common code
into _make_tree(). This isn't much of a win now. It will pay off
when we add feature flags in the next few commits.
qapi/introspect: Rename *qlit* to reduce confusion
We generate the value of qmp_schema_qlit from an expression tree. The
function doing that is named to_qlit(), and its inputs are accumulated
in QAPISchemaGenIntrospectVisitor._qlits. We call both its input and
its output "qlit". This is confusing.
Use "tree" for input, and "qlit" only for output: rename to_qlit() to
_tree_to_qlit(), ._qlits to ._trees, ._gen_qlit() to ._gen_tree().
In v4.1.0, we added feature flags just to struct types (commit 6a8c0b5102^..f3ed93d545), to satisfy an immediate need (commit c9d4070991 "file-posix: Add dynamic-auto-read-only QAPI feature"). In
v4.2.0, we added them to commands (commit 23394b4c39 "qapi: Add
feature flags to commands") to satisfy another immediate need (commit d76744e65e "qapi: Allow introspecting fix for savevm's cooperation
with blockdev").
Add them to the remaining definitions: enumeration types, union types,
alternate types, and events.
qapi/schema: Clean up around QAPISchemaEntity.connect_doc()
QAPISchemaEntity calls doc.connect_feature() in .check(). Improper
since commit ee1e6a1f6c8 split .connect_doc() off .check(). Move the
call. Requires making the children call super().connect_doc() as they
should.
Locally defined helper qdict_cmp_simple() implements just enough of a
comparison to serve here. Replace it by qobject_is_equal(), which
implements all of it.
Building expected data with qdict_put() & friends is tedious to write
and hard to read. Parse them from string literals with
qdict_from_jsonf_nofail() instead.
While there, use initializers instead of assignments for initializing
aggregate event arguments.
Building requests with qdict_put() & friends is tedious to write and
hard to read. Parse them from string literals with
qdict_from_vjsonf_nofail() instead.
tests/test-qmp-cmds: Factor out qmp_dispatch() test helpers
Checking the value of qmp_dispatch() is repetitive. Factor out
helpers do_qmp_dispatch() and do_qmp_dispatch_error(). Without this,
the next commit would make things even more repetitive.
Commit 6a8c0b5102 "qapi: Add feature flags to struct types" neglected
to update section "Client JSON Protocol introspection", and commit 23394b4c39 "qapi: Add feature flags to commands" didn't either. Make
up for that.
qapi: Belatedly update doc comment for @wait deprecation
Commit a9b305ba29 "socket: allow wait=false for client socket"
deprecated use of @wait for client socket chardevs, but neglected to
update char.json's doc comment. Make up for that.
qemu-doc: Belatedly document QMP command arg & result deprecation
A number of deprecated QMP arguments and results were missed in commit eb22aeca65 "docs: document deprecation policy & deprecated features in
appendix" (v2.10.0):
* Commit b33945cfff "block: Accept device model name for
blockdev-open/close-tray" (v2.8.0) deprecated blockdev-open-tray,
blockdev-close-tray argument @device.
* Commit fbe2d8163e "block: Accept device model name for eject"
(v2.8.0) deprecated eject argument @device.
* Commit 70e2cb3bd7 "block: Accept device model name for
blockdev-change-medium" (v2.8.0) deprecated blockdev-change-medium
argument @device.
* Commit 7a9877a026 "block: Accept device model name for
block_set_io_throttle" (v2.8.0) deprecated block_set_io_throttle
argument @device.
* Commit c01c214b69 "block: remove all encryption handling APIs"
(v2.10.0) deprecated query-named-block-nodes result
@encryption_key_missing and query-block result @inserted member
@encryption_key_missing.
* Commit c42e8742f5 "block: Use JSON null instead of "" to disable
backing file" (v2.10.0) deprecated blockdev-add empty string
argument @backing.
* Commit 4db6ceb0b5 "block/dirty-bitmap: add recording and busy
properties" (v4.0.0) deprecated query-named-block-nodes result
@dirty-bitmaps member @status, not just query-block.
Peter Maydell [Tue, 17 Mar 2020 18:33:05 +0000 (18:33 +0000)]
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
* Bugfixes all over the place
* get/set_uint cleanups (Felipe)
* Lock guard support (Stefan)
* MemoryRegion ownership cleanup (Philippe)
* AVX512 optimization for buffer_is_zero (Robert)
* remotes/bonzini/tags/for-upstream: (62 commits)
hw/arm: Let devices own the MemoryRegion they create
hw/arm: Remove unnecessary memory_region_set_readonly() on ROM alias
hw/ppc/ppc405: Use memory_region_init_rom() with read-only regions
hw/arm/stm32: Use memory_region_init_rom() with read-only regions
hw/char: Let devices own the MemoryRegion they create
hw/riscv: Let devices own the MemoryRegion they create
hw/dma: Let devices own the MemoryRegion they create
hw/display: Let devices own the MemoryRegion they create
hw/core: Let devices own the MemoryRegion they create
scripts/cocci: Patch to let devices own their MemoryRegions
scripts/cocci: Patch to remove unnecessary memory_region_set_readonly()
scripts/cocci: Patch to detect potential use of memory_region_init_rom
hw/sparc: Use memory_region_init_rom() with read-only regions
hw/sh4: Use memory_region_init_rom() with read-only regions
hw/riscv: Use memory_region_init_rom() with read-only regions
hw/ppc: Use memory_region_init_rom() with read-only regions
hw/pci-host: Use memory_region_init_rom() with read-only regions
hw/net: Use memory_region_init_rom() with read-only regions
hw/m68k: Use memory_region_init_rom() with read-only regions
hw/display: Use memory_region_init_rom() with read-only regions
...
Changbin Du [Mon, 16 Mar 2020 17:21:55 +0000 (17:21 +0000)]
gdbstub: Fix single-step issue by confirming 'vContSupported+' feature to gdb
Recently when debugging an arm32 system on qemu, I found sometimes the
single-step command (stepi) is not working. This can be reproduced by
below steps:
1) start qemu-system-arm -s -S .. and wait for gdb connection.
2) start gdb and connect to qemu. In my case, gdb gets a wrong value
(0x60) for PC, which is an another bug.
3) After connected, type 'stepi' and expect it will stop at next ins.
But, it has never stopped. This because:
1) We doesn't report ‘vContSupported’ feature to gdb explicitly and gdb
think we do not support it. In this case, gdb use a software breakpoint
to emulate single-step.
2) Since gdb gets a wrong initial value of PC, then gdb inserts a
breakpoint to wrong place (PC+4).
Not only for the arm target, Philippe has also encountered this on MIPS.
Probably gdb has different assumption for different architectures.
Since we do support ‘vContSupported’ query command, so let's tell gdb that
we support it.
Before this change, gdb send below 'Z0' packet to implement single-step:
gdb_handle_packet: Z0,4,4
After this change, gdb send "vCont;s.." which is expected:
gdb_handle_packet: vCont?
put_packet: vCont;c;C;s;S
gdb_handle_packet: vCont;s:p1.1;c:p1.-1
Alex Bennée [Mon, 16 Mar 2020 17:21:52 +0000 (17:21 +0000)]
tests/tcg/aarch64: add test-sve-ioctl guest-debug test
This test exercises the gdbstub while runing the sve-iotcl test. I
haven't plubmed it into make system as we need a way of verifying if
gdb has the right support for SVE.
Alex Bennée [Mon, 16 Mar 2020 17:21:51 +0000 (17:21 +0000)]
tests/tcg/aarch64: add SVE iotcl test
This is a fairly bare-bones test of setting the various vector sizes
for SVE which will only fail if the PR_SVE_SET_VL can't reduce the
user-space vector length by powers of 2.
However we will also be able to use it in a future test which
exercises the GDB stub.
Alex Bennée [Mon, 16 Mar 2020 17:21:50 +0000 (17:21 +0000)]
tests/tcg/aarch64: add a gdbstub testcase for SVE registers
A very simple test case which sets and reads SVE registers while
running a test case. We don't really need to compile a SVE binary for
this case but we will later so keep it simple for now.
Alex Bennée [Mon, 16 Mar 2020 17:21:49 +0000 (17:21 +0000)]
tests/guest-debug: add a simple test runner
The test runners job is to start QEMU with guest debug enabled and
then spawn a gdb process running a test script that exercises the
functionality it wants to test.
Alex Bennée [Mon, 16 Mar 2020 17:21:47 +0000 (17:21 +0000)]
tests/tcg/aarch64: userspace system register test
This tests a bunch of registers that the kernel allows userspace to
read including the CPUID registers. We need a SVE aware compiler as we
are testing the id_aa64zfr0_el1 register in the set.
Alex Bennée [Mon, 16 Mar 2020 17:21:46 +0000 (17:21 +0000)]
target/arm: don't bother with id_aa64pfr0_read for USER_ONLY
For system emulation we need to check the state of the GIC before we
report the value. However this isn't relevant to exporting of the
value to linux-user and indeed breaks the exported value as set by
modify_arm_cp_regs.
Alex Bennée [Mon, 16 Mar 2020 17:21:44 +0000 (17:21 +0000)]
target/arm: default SVE length to 64 bytes for linux-user
The Linux kernel chooses the default of 64 bytes for SVE registers on
the basis that it is the largest size on known hardware that won't
grow the signal frame. We still honour the sve-max-vq property and
userspace can expand the number of lanes by calling PR_SVE_SET_VL.
This should not make any difference to SVE enabled software as the SVE
is of course vector length agnostic.
Alex Bennée [Mon, 16 Mar 2020 17:21:42 +0000 (17:21 +0000)]
target/arm: prepare for multiple dynamic XMLs
We will want to generate similar dynamic XML for gdbstub support of
SVE registers (the upstream doesn't use XML). To that end lightly
rename a few things to make the distinction.
Alex Bennée [Mon, 16 Mar 2020 17:21:41 +0000 (17:21 +0000)]
gdbstub: extend GByteArray to read register helpers
Instead of passing a pointer to memory now just extend the GByteArray
to all the read register helpers. They can then safely append their
data through the normal way. We don't bother with this abstraction for
write registers as we have already ensured the buffer being copied
from is the correct size.
Alex Bennée [Mon, 16 Mar 2020 17:21:36 +0000 (17:21 +0000)]
gdbstub: move mem_buf to GDBState and use GByteArray
This is in preparation for further re-factoring of the register API
with the rest of the code. Theoretically the read register function
could overwrite the MAX_PACKET_LENGTH buffer although currently all
registers are well within the size range.
Alex Bennée [Mon, 16 Mar 2020 17:21:35 +0000 (17:21 +0000)]
gdbstub: move str_buf to GDBState and use GString
Rather than having a static buffer replace str_buf with a GString
which we know can grow on demand. Convert the internal functions to
take a GString instead of a char * and length.
Alex Bennée [Mon, 16 Mar 2020 17:21:33 +0000 (17:21 +0000)]
gdbstub: make GDBState static and have common init function
Instead of allocating make this entirely static. We shall reduce the
size of the structure in later commits and dynamically allocate parts
of it. We introduce an init and reset helper function to keep all the
manipulation in one place.
travis.yml: Set G_MESSAGES_DEBUG do report GLib errors
Since commit f5852efa293 we can display GLib errors with the QEMU
error reporting API. Set it to the 'error' level, as this helps
understanding failures from QEMU calls to GLib on Travis-CI.
Building the qemu:debian-amd64 fails when building VirGL:
make[2]: Entering directory '/usr/src/virglrenderer/src/gallium/auxiliary'
CC cso_cache/cso_cache.lo
CC cso_cache/cso_hash.lo
CC os/os_misc.lo
CC util/u_debug.lo
CC util/u_debug_describe.lo
CC util/u_format.lo
GEN util/u_format_table.c
Traceback (most recent call last):
File "./util/u_format_table.py", line 168, in <module>
main()
File "./util/u_format_table.py", line 164, in main
write_format_table(formats)
File "./util/u_format_table.py", line 132, in write_format_table
print(" %s,\t/* is_array */" % (bool_map(format.is_array()),))
File "/usr/src/virglrenderer/src/gallium/auxiliary/util/u_format_parse.py", line 164, in is_array
return self.array_element() != None
File "/usr/src/virglrenderer/src/gallium/auxiliary/util/u_format_parse.py", line 73, in __eq__
return self.type == other.type and self.norm == other.norm and self.pure == other.pure and self.size == other.size
AttributeError: 'NoneType' object has no attribute 'type'
make[2]: Leaving directory '/usr/src/virglrenderer/src/gallium/auxiliary'
make[2]: *** [Makefile:906: util/u_format_table.c] Error 1
make[1]: *** [Makefile:631: install-recursive] Error 1
VirGL commits a8962eda1..a613dcc82 fix this problem.
Update to VirGL 0.8.0 which contains them.
An ALTERNATIVE's value can only be a type name. Arrays are not
supported, yet. The text gets it right: "The form STRING is shorthand
for { 'type': STRING }." The grammar doesn't. Fix it.
hw/sd/ssi-sd: fix error handling in ssi_sd_realize
It's wrong to use same err object as errp parameter for several
function calls without intermediate checking for error: we'll crash if
try to set err object twice. Fix that.
Peter Maydell [Tue, 17 Mar 2020 16:23:03 +0000 (16:23 +0000)]
Merge remote-tracking branch 'remotes/amarkovic/tags/mips-queue-mar-17-2020' into staging
MIPS queue for March 17th, 2020
# gpg: Signature made Tue 17 Mar 2020 12:53:50 GMT
# gpg: using RSA key D4972A8967F75A65
# gpg: Good signature from "Aleksandar Markovic <[email protected]>" [full]
# Primary key fingerprint: 8526 FBF1 5DA3 811F 4A01 DD75 D497 2A89 67F7 5A65
* remotes/amarkovic/tags/mips-queue-mar-17-2020:
MAINTAINERS: Add a file to the main MIPS section
MAINTAINERS: Adjust maintainer's email
MAINTAINERS: Adjust maintainer's status for some MIPS items
BALATON Zoltan [Tue, 17 Mar 2020 15:05:37 +0000 (16:05 +0100)]
hw/ide: Do ide_drive_get() within pci_ide_create_devs()
The pci_ide_create_devs() function takes a hd_table parameter but all
callers just pass what ide_drive_get() returns so we can do it locally
simplifying callers and removing hd_table parameter.
BALATON Zoltan [Tue, 17 Mar 2020 15:05:37 +0000 (16:05 +0100)]
hw/ide/pci.c: Coding style update to fix checkpatch errors
Spaces are required around a + operator and if statements should have
braces even for single line. Also make it simpler by reversing the
condition instead of breaking the loop.
BALATON Zoltan [Tue, 17 Mar 2020 15:05:37 +0000 (16:05 +0100)]
hw/ide: Remove now unneded #include "hw/pci/pci.h" from hw/ide.h
After previous patches we don't need hw/pci/pci.h any more in
hw/ide.h. Some files depended on implicit inclusion by this header
which are also fixed up here.