I "reverse engineered" the following permissions between the -socket
sub-options:
fd listen connect mcast udp | localaddr
fd x . . . . | .
listen . x . . . | .
connect . . x . . | .
mcast . . . x . | x
udp . . . . x | x
-------------------------------------------+
localaddr . . . x x x
I transformed the code accordingly. The real fix would be to embed "fd",
"listen", "connect", "mcast" and "udp" in a separate union. However
OptsVisitor's enum parser only supports the type=XXX QemuOpt instance as
union discriminator.
Based on "is_netdev", the QemuOpts-rooted QemuOpt-list is parsed as a
Netdev or a NetLegacy. The original meat of net_client_init() is moved to
and simplified in net_client_init1():
Fields not common between -net and -netdev are clearly separated. Getting
the name for the init functions is cleaner: Netdev::id is mandatory, and
all init functions handle a NULL NetLegacy::name. NetLegacy::vlan
explicitly depends on -net (see below).
Verifying the "type=" option for -netdev can be turned into a switch.
Format validation with qemu_opts_validate() can be removed because the
visitor covers it. Relatedly, the "net_client_types" array is reduced to
an array of init functions that can be directly indexed by opts->kind.
(Help text is available in the schema JSON.)
The outermost negation in the condition around qemu_find_vlan() was
flattened, because it expresses the dependent code's requirements more
clearly.
VLAN lookup is avoided if there's no init function to pass the VLAN to.
Whenever the value of type=... is needed, we substitute
NetClientOptionsKind_lookup[kind].
The individual init functions are not converted yet, thus the original
QemuOpts instance is passed transparently.
v1->v2:
- NetLegacy::name is optional. Tracked it through all init functions: they
all handle a NULL name. Updated commit message accordingly.
v2->v3:
- NetLegacy::id is allowed and takes precedence over NetLegacy::name.
NetdevTapOptions::sndbuf and NetdevDumpOptions::len use the new "size"
type.
v1->v2:
- NetLegacy::name is optional
- NetLegacyNicOptions::vectors is of type uint32
- NetdevVdeOptions::port and ::mode are of type uint16
- NetLegacy::vlan has type int32
v2->v3:
- NetLegacy::id is allowed and takes precedence over NetLegacy::name
- replace "@traits" with "@opts" in NetLegacy & Netdev descriptions
style QemuOpts objects into "native" C structures. After defining the type
tree in the qapi schema (see below), a root type traversal with this
visitor linked to the underlying QemuOpts object will build the "native" C
representation of the option.
The type tree in the schema, corresponding to an option with a
discriminator, must have the following structure:
struct
scalar member for non-discriminated optarg 1 [*]
list for repeating non-discriminated optarg 2 [*]
wrapper struct
single scalar member
union
struct for discriminator case 1
scalar member for optarg 3 [*]
list for repeating optarg 4 [*]
wrapper struct
single scalar member
scalar member for optarg 5 [*]
struct for discriminator case 2
...
The "type" optarg name is fixed for the discriminator role. Its schema
representation is "union of structures", and each discriminator value must
correspond to a member name in the union.
If the option takes no "type" descriminator, then the type subtree rooted
at the union must be absent from the schema (including the union itself).
Optarg values can be of scalar types str / bool / integers / size.
Members marked with [*] may be defined as optional in the schema,
describing an optional optarg.
Repeating an optarg is supported; its schema representation must be "list
of structure with single mandatory scalar member". If an optarg is not
described as repeating in the schema (ie. it is defined as a scalar field
instead of a list), its last occurrence will take effect. Ordering between
differently named optargs is not preserved.
A mandatory list (or an optional one which is reported to be available),
corresponding to a repeating optarg, has at least one element after
successful parsing.
v1->v2:
- Update opts_type_size() prototype to uint64_t.
- Add opts_type_uint64() for options needing the full uint64_t range.
(Internals could be extracted to "cutils.c".)
- Allow negative values in opts_type_int().
- Rebase to nested Makefiles.
v2->v3:
- Factor opts_visitor_insert() out of opts_start_struct() and call it
separately for opts_root->id if there's any.
- Don't require non-negative values in opts_type_int()'s error message.
- g_malloc0() may return NULL for zero-sized requests. Support empty
structures by requesting 1 byte for them instead.
Paolo Bonzini [Tue, 17 Jul 2012 14:17:04 +0000 (16:17 +0200)]
qapi: fix error propagation
Don't overwrite / leak previously set errors.
Make traversal cope with missing mandatory sub-structs.
Don't try to end a container that could not be started.
v1->v2:
- unchanged
v2->v3:
- instead of examining, assert that we never overwrite errors with
error_set()
- allow visitors to set a NULL struct pointer successfully, so traversal
of incomplete objects can continue
- check for a NULL "obj" before accessing "(*obj)->has_XXX" (this is not a
typo, "obj != NULL" implies "*obj != NULL" here)
- fix start_struct / end_struct balance for unions as well
Peter Maydell [Thu, 19 Jul 2012 15:52:27 +0000 (16:52 +0100)]
cpus.c: Make all_cpu_threads_idle() static
Commit 946fb27c1 moved all the uses of all_cpu_threads_idle()
into cpus.c. This means we can mark the function 'static'
(again), if we shuffle it a bit earlier in the source file.
By default mingw-gcc is trying to pack structures the way to
preserve binary compatibility with MS Visual C what leads to
incorrect and unexpected padding in the PCI bus ranges property of
the sPAPR PHB.
The patch replaces __attribute__((packed)) with more strict QEMU_PACKED
which actually is __attribute__((gcc_struct, packed)) on Windows.
Peter Maydell [Fri, 20 Jul 2012 12:34:50 +0000 (13:34 +0100)]
hw/vexpress.c: Allow >4GB of RAM for Cortex-A15 daughterboard
Now that we have LPAE support and can handle passing 64 bit
RAM sizes to Linux via the device tree, we can lift the
restriction in the Versatile Express A15 daughterboard model
on not having more than 2GB of RAM. Allow up to 30GB, which
is the maximum that can fit in the address map before running
into the (unmodelled) aliases of the first 2GB.
Peter Maydell [Fri, 20 Jul 2012 12:34:50 +0000 (13:34 +0100)]
hw/arm_boot.c: Support DTBs which use 64 bit addresses
Support the case where the device tree blob specifies that
#address-cells and #size-cells are greater than 1. (This
is needed for device trees which can handle 64 bit physical
addresses and thus total RAM sizes over 4GB.)
Peter Maydell [Fri, 20 Jul 2012 12:34:50 +0000 (13:34 +0100)]
hw/arm_boot.c: Check for RAM sizes exceeding ATAGS capacity
The legacy ATAGS format for passing information to the kernel only
allows RAM sizes which fit in 32 bits; enforce this restriction
rather than silently doing something weird.
Peter Maydell [Fri, 20 Jul 2012 12:34:50 +0000 (13:34 +0100)]
hw/arm_boot.c: Consistently use ram_size from arm_boot_info struct
Clean up the mix of getting the RAM size from the global ram_size
and from the ram_size field in the arm_boot_info structure, so
that we always use the structure field.
Peter Maydell [Fri, 20 Jul 2012 12:34:49 +0000 (13:34 +0100)]
hw/pl011.c: Avoid crash on read when no chr backend present
Add a missing guard that meant we would segfault if the guest read
UARTDR on a PL011 serial device which had no chr backend connected.
(This didn't happen for Linux guests because Linux reads the flags
register and doesn't try to read the UART if it's empty.)
Juan Quintela [Thu, 28 Jun 2012 17:41:52 +0000 (19:41 +0200)]
ram: save_live_setup() don't need to sent pages
We should send pages on interate phase, not in setup one. This was a
"bug". Just removing the loop does what we want. Tested that it
works with current ram_load().
Juan Quintela [Thu, 28 Jun 2012 13:31:37 +0000 (15:31 +0200)]
savevm: split save_live into stage2 and stage3
We split it into 2 functions, foo_live_iterate, and foo_live_complete.
At this point, we only remove the bits that are for the other stage,
functionally this is equivalent to previous code.
Juan Quintela [Thu, 28 Jun 2012 13:11:57 +0000 (15:11 +0200)]
savevm: split save_live_setup from save_live_state
This patch splits stage 1 to its own function for both save_live
users, ram and block. It is just a copy of the function, removing the
parts of the other stages. Optimizations would came later.
Juan Quintela [Wed, 27 Jun 2012 08:59:15 +0000 (10:59 +0200)]
savevm: introduce is_active method
Enable the creation of a method to tell migration if that section is
active and should be migrate. We use it for blk-migration, that is
normally not active. We don't create the method for RAM, as setups
without RAM are very strange O:-)
Jan Kiszka [Thu, 19 Jul 2012 15:57:51 +0000 (17:57 +0200)]
audio: Make PC speaker audio card available by default
[ The following text is in the "ISO-8859-1" character set. ]
[ Your display is set for the "KOI8-R" character set. ]
[ Some special characters may be displayed incorrectly. ]
msi/msix: added API to set MSI message address and data
Added (msi|msix)_set_message() function for whoever might
want to use them.
Currently msi_notify()/msix_notify() write to these vectors to
signal the guest about an interrupt so the correct values have to
written there by the guest or QEMU.
For example, POWER guest never initializes MSI/MSIX vectors, instead
it uses RTAS hypercalls. So in order to support MSIX for virtio-pci on
POWER we have to initialize MSI/MSIX message from QEMU.
Jan Kiszka [Mon, 2 Jul 2012 12:38:47 +0000 (14:38 +0200)]
pci: Add INTx routing notifier
This per-device notifier shall be triggered by any interrupt router
along the path of a device's legacy interrupt signal on routing changes.
For simplicity reasons and as this is a slow path anyway, no further
details on the routing changes are provided. Instead, the callback is
expected to use pci_device_route_intx_to_irq to check the effect of the
change.
Will be used by KVM PCI device assignment and VFIO.
Device assigned on KVM needs to know the mode
(enabled/inverted/disabled) and the IRQ number that a given device
triggers in the attached interrupt controller.
Add a PCI IRQ path discovery function that walks from a given device to
the host bridge, and gets this information. For
this purpose, a host bridge callback function is introduced:
route_intx_to_irq. It is so far only implemented by the PIIX3, other
host bridges can be added later on as required.
Will be used for KVM PCI device assignment and VFIO.
Simpletrace v2: Support multiple arguments, strings.
Existing simpletrace backend allows to trace at max 6 args and does not
support strings. This newer tracelog format gets rid of fixed size records
and therefore allows to trace variable number of args including strings.
Going forward with simpletrace v2 variable size trace records, we cannot
have a generic function to print trace event info and therefore this
interface becomes invalid.
As per Stefan Hajnoczi:
"This command is only available from the human monitor. It's not very
useful because it historically hasn't been able to pretty-print events
or show them in the right order (we use a ringbuffer but it prints
them out from index 0).
Therefore, I don't think we're under any obligation to keep this
command around. No one has complained about it's limitations - I
think this is a sign that no one has used it. I'd be okay with a
patch that removes it."
trace: added ability to comment out events in the list
It is convenient for debug to be able to switch on/off some events easily.
The only possibility now is to remove event name from the file completely
and type it again when we want it back.
The patch adds '#' symbol handling as a comment specifier.
Anthony Liguori [Wed, 18 Jul 2012 19:44:50 +0000 (14:44 -0500)]
Merge remote-tracking branch 'kwolf/for-anthony' into staging
* kwolf/for-anthony: (41 commits)
fdc-test: Clean up a bit
fdc-test: introduce test_relative_seek
fdc: fix relative seek
qemu-iotests: Valgrind support
coroutine-ucontext: Help valgrind understand coroutines
qemu-io: Fix memory leaks
hw/block-common: Factor out fall back to legacy -drive cyls=...
blockdev: Don't limit DriveInfo serial to 20 characters
hw/block-common: Factor out fall back to legacy -drive serial=...
hw/block-common: Move BlockConf & friends from block.h
Relax IDE CHS limits from 16383,16,63 to 65535,16,255
blockdev: Drop redundant CHS validation for if=ide
hd-geometry: Compute BIOS CHS translation in one place
qtest: Test we don't put hard disk info into CMOS for a CD-ROM
ide pc: Put hard disk info into CMOS only for hard disks
block: Geometry and translation hints are now useless, purge them
qtest: Cover qdev property for BIOS CHS translation
ide: qdev property for BIOS CHS translation
qdev: New property type chs-translation
qdev: Collect private helpers in one place
...
Anthony Liguori [Wed, 18 Jul 2012 19:44:43 +0000 (14:44 -0500)]
Merge remote-tracking branch 'qemu-kvm/uq/master' into staging
* qemu-kvm/uq/master:
virtio: move common irqfd handling out of virtio-pci
virtio: move common ioeventfd handling out of virtio-pci
event_notifier: add event_notifier_set_handler
memory: pass EventNotifier, not eventfd
ivshmem: wrap ivshmem_del_eventfd loops with transaction
ivshmem: use EventNotifier and memory API
event_notifier: add event_notifier_init_fd
event_notifier: remove event_notifier_test
event_notifier: add event_notifier_set
apic: Defer interrupt updates to VCPU thread
apic: Reevaluate pending interrupts on LVT_LINT0 changes
apic: Resolve potential endless loop around apic_update_irq
kvm: expose tsc deadline timer feature to guest
kvm_pv_eoi: add flag support
kvm: Don't abort on kvm_irqchip_add_msi_route()
Anthony Liguori [Tue, 17 Jul 2012 23:58:20 +0000 (18:58 -0500)]
build: fix build breakage due to bad merge
34bb443ee74df3780d86044dbf6492eb798c5807 broke the build but in a subtle way.
The patch on the ML was actually based on the 1.1.1 stable branch
(accidentally). When I merged it from the ML, the fuzzing got resolved in a
such a way that the newly introduced Makefile dependency on GENERATED_HEADERS
got lost (that was not in 1.1.1).
The dist bits also got duplicated (but this is just cosmetic).
Michal Novotny [Mon, 16 Jul 2012 13:54:38 +0000 (15:54 +0200)]
qemu-options.hx: Fix set_password and expire_password description
The description for set_password and expire_password commands is
incomplete. This patch fixes the man page that is being generated
to match the real behaviour of these functions.
blockdev: Don't limit DriveInfo serial to 20 characters
All current users (IDE, SCSI and virtio-blk) happen to share this 20
characters limit. Still, it should be left to device models. They
already enforce their limits. They have to, as the DriveInfo limit
only affects legacy -drive serial=..., not the qdev properties.
usb-storage, which doesn't limit serial number length, also uses
DriveInfo for -usbdevice. But that doesn't provide access to
DriveInfo serial.
Relax IDE CHS limits from 16383,16,63 to 65535,16,255
New limits straight from ATA4 6.2 Register delivered data transfer
command sector addressing.
I figure the old sector limit 63 was blindly copied from the BIOS
int 13 limit. Doesn't apply to the hardware. No idea where the old
cylinder limit comes from.
block: Geometry and translation hints are now useless, purge them
There are two producers of these hints: drive_init() on behalf of
-drive, and hd_geometry_guess().
The only consumer of the hint is hd_geometry_guess().
The callers of hd_geometry_guess() call it only when drive_init()
didn't set the hints. Therefore, drive_init()'s hints are never used.
Thus, hd_geometry_guess() only ever sees hints it produced itself in a
prior call. Only the first call computes something, subsequent calls
just repeat the first call's results. However, hd_geometry_guess() is
never called more than once: the device models don't, and the block
device is destroyed on unplug. Thus, dropping the repeat feature
doesn't break anything now.
If a block device wasn't destroyed on unplug and could be reused with
a new device, then repeating old results would be wrong. Thus,
dropping the repeat feature prevents future breakage.
This renders the hints unused. Purge them from the block layer.
This isn't quite orthodox. CHS translation is firmware configuration,
communicated via the RTC's CMOS RAM, not a property of the disk. But
it's best to treat it just like geometry anyway.
Maintain backward compatibility exactly like for geometry: fall back
to DriveInfo's translation, set with -drive trans=...
Bonus: info qtree now shows the translation. Except when it shows
"auto": that's resolved by pc_cmos_init_late(). To be addressed
shortly.
ide pc: Cut out the block layer geometry middleman
PC BIOS setup needs IDE geometry information. Get it directly from
the device model rather than through the block layer. In preparation
of purging geometry from the block layer, which will happen later in
this series.
hd-geometry: Cut out block layer translation middleman
hd_geometry_guess() picks geometry and translation. Callers can get
the geometry directly, via parameters, but for translation they need
to go through the block layer.
Add a parameter for translation, so it can optionally be gotten just
like geometry. In preparation of purging translation from the block
layer, which will happen later in this series.
hd-geometry: Clean up confusing use of prior translation hint
When hd_geometry_guess() picks a geometry, it also picks the
appropriate translation, but only when the prior translation hint is
BIOS_ATA_TRANSLATION_AUTO. Looks wrong, because such a prior
translation would be passed to the BIOS whether it's suitable for the
geometry or not.
Fortunately, that can't happen. There are just two ways for the
translation hint to get set to something other than
BIOS_ATA_TRANSLATION_AUTO: drive_init() on behalf of -drive trans=...,
and hd_geometry_guess(). Both set it only when they also set a valid
geometry hint, i.e. one with a non-zero number of cylinders.
Since hd_geometry_guess() returns right away when it finds a valid
geometry hint, translation can only be BIOS_ATA_TRANSLATION_AUTO in
the remainder of the function.
projects / qemu.git / log