A strong limitation of QOM right now is that unconverted ports
(e.g. all...) do not give a canonical path to devices that are
part of the board. This in turn makes it impossible to replace
PROP_PTR with a QOM link for example.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Since QOM'ification, qdev_try_create() uses object_new() internally,
which asserts "type != NULL" when the type is not registered.
This was revealed by the combination of kvmclock's kvm_enabled() check
and early QOM type registration.
Check whether the class exists before calling object_new(), so that
the caller (e.g., qdev_create) can fail gracefully, telling us which
device could not be created.
Signed-off-by: Andreas Färber <afaerber@suse.de>
Cc: Anthony Liguori <aliguori@codemonkey.ws>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Replace device_init() with generalized type_init().
While at it, unify naming convention: type_init([$prefix_]register_types)
Also, type_init() is a function, so add preceding blank line where
necessary and don't put a semicolon after the closing brace.
Signed-off-by: Andreas Färber <afaerber@suse.de>
Cc: Anthony Liguori <anthony@codemonkey.ws>
Cc: malc <av1474@comtv.ru>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Similarly, use the object properties also to set the default
values of the qdev properties. This requires reordering
registration and initialization.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Drop the special free callback. Instead, register a "regular"
release method in the non-legacy property.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In some cases, a legacy property does need a special print method
but not a special parse method. In this case, we can reuse the get/set
from the static (non-legacy) property.
If neither parse nor print is needed, though, do not register the
legacy property at all. The previous patch ensures that the right
fallback will be used.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
There's no need to call into ->parse and ->print manually. The
QOM legacy properties do that for us.
Furthermore, in some cases legacy and static properties have exactly
the same behavior, and we could drop the legacy properties right away.
Add an appropriate fallback to prepare for this.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This is mostly code movement although not entirely. This makes properties part
of the Object base class which means that we can now start using Object in a
meaningful way outside of qdev.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
I'm sure the intentions were good here, but there's no reason this should be in
qdev. Move it to qemu-char where it belongs.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Note that the FIXME gets fixed in series 4/4. We need to convert BusState to
QOM before we can make parent_bus a link.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
This gets us closer to being able to object_new() a qdev type and have a
functioning object verses having to call qdev_create().
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Limit them to the device_add functionality. Device aliases were a hack based
on the fact that virtio was modeled the wrong way. The mechanism for aliasing
is very limited in that only one alias can exist for any device.
We have to support it for the purposes of compatibility but we only need to
support it in device_add so restrict it to that piece of code.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
---
v1 -> v2
- Use a table for aliases (Paolo)
This allows us to drop per-Device registration functions by allowing the
class_init functions to overload qdev methods.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Now DeviceInfo is no longer used after object construction. All of the
relevant members have been moved to DeviceClass.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Since we are still dynamically creating TypeInfo, we need to chain the
class_init function in order to be able to make use of it within subclasses of
TYPE_DEVICE.
This will disappear once we register TypeInfos directly.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
In order to introduce inheritance while still using the qdev registration
interfaces, we need to be able to use a parent other than TYPE_DEVICE. Add a
new interface that allows this.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Right now, DeviceInfo acts as the class for qdev. In order to switch to a
proper ObjectClass derivative, we need to ween all of the callers off of
interacting directly with the info pointer.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
This is a very shallow integration. We register a TYPE_DEVICE but only use
QOM as basically a memory allocator. This will make all devices show up as
QOM objects but they will all carry the TYPE_DEVICE.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
---
v1 -> v2
- update for new location of object.h
Improve VGA selection logic, push check for device availabilty to vl.c.
Create the devices at board level unconditionally.
Remove now unused pci_try_create*() functions.
Make PCI VGA devices optional.
Reviewed-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Commit 8eb0283 broken device_del by having too overzealous reference counting
checks. Move the reference count checks to qdev_free(), make sure to remove
the parent link on free, and decrement the reference count on property removal.
Reported-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
The peripheral[-anon] containers are initialized lazily but since they sit on
sysbus, they can not be created after realize. This was causing an abort() to
occur during hotplug if no -device option was used.
This was spotted by qemu-test::device-add.sh
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Push legacy properties into a "legacy-..." namespace, and make them
available with correct types too.
For now, all properties come in both variants. This need not be the
case for string properties. We will revisit this after -device is
changed to actually use the legacy properties.
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
qdev_property_get and qdev_property_set can generate permission
denied errors themselves. Do not duplicate this functionality in
qdev_get/set_legacy_property, and clean up excessive indentation.
Reviewed-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Reviewed-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
We first add a 'peripheral' container to the root device that we add user
created devices to. This provides all user created devices with a unique and
isolated namespace.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Links represent an ephemeral relationship between devices. They are meant to
replace the qdev concept of busses by allowing more informal relationships
between devices.
Links are fairly limited in their usefulness without implementing QOM-style
subclassing and interfaces.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
There are two types of supported paths--absolute paths and partial paths.
Absolute paths are derived from the root device and can follow child<> or
link<> properties. Since they can follow link<> properties, they can be
arbitrarily long. Absolute paths look like absolute filenames and are prefixed
with a leading slash.
Partial paths are look like relative filenames. They do not begin with a
prefix. The matching rules for partial paths are subtle but designed to make
specifying devices easy. At each level of the composition tree, the partial
path is matched as an absolute path. The first match is not returned. At
least two matches are searched for. A successful result is only returned if
only one match is founded. If more than one match is found, a flag is returned
to indicate that the match was ambiguous.
At the end of the day, partial path support means that if you create a device
called 'ide0', you can just say 'ide0' as the path name and it will Just Work.
If we internally create a device called 'i440fx', you can just say 'i440fx' and
it will Just Work and long as you don't do anything silly.
A management tool should probably always use absolute paths since then they
don't have to deal with the possibility of ambiguity.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
The canonical path is the path in the composition tree from the root to the
device. This is effectively the name of the device.
This is an incredibly unefficient implementation that will be optimized in
a future patch.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
This is based on Jan's suggestion for how to do unique naming. The root device
is the root of composition. All devices are reachable via child<> links from
this device.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Expose all legacy properties through the new QOM property mechanism. The qdev
property types are exposed through the 'legacy<>' namespace. They are always
visited as strings since they do their own string parsing.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
qdev properties are settable only during construction and static to classes.
This isn't flexible enough for QOM.
This patch introduces a property interface for qdev that provides dynamic
properties that are tied to objects, instead of classes. These properties are
Visitor based instead of string based too.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
SCSI buses will need to read the children list first-to-last. This
requires using a QTAILQ, because hell breaks loose if you just try
inserting at the tail (thus reversing the order of all existing
visits from last-to-first to first-to-tail).
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Make qdev_device_help print both device and bus properties.
Helps libvirt to figure whenever bus properties such as
PCI.multifunction are supported present or not.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
If an attempt to create a qdev device on the default sysbus (by passing
NULL as the bus to qdev_create) fails, print a useful error message
rather than crashing trying to dereference a NULL pointer.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
qbus_reset_all_fn was registered twice, so a lot of device reset
functions were also called twice when QEMU started.
Which was introduced by 80376c3fc2
This patch fixes it by making the main_system_bus creation not register
reset handler.
Cc: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Tested-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Device models rely on the core invoking their reset handlers after init.
We do this in the cold-plug case, but so far we miss this step after
hot-plug.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Drop the open-coded MAC assignment from net_init_nic and replace it with
standard qemu_macaddr_default_if_unset which is also used by qdev. That
avoid creating colliding MACs when instantiating NICs via different
mechanisms.
This change requires to store the MAC as MACAddr in NICInfo, and the
remaining nd_table users need to be updated.
Based on suggestion by Peter Maydell.
CC: Markus Armbruster <armbru@redhat.com>
CC: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
error_report() prepends location, and appends a newline. The message
constructed from the arguments should not contain a newline. Fix the
obvious offenders.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Diagnose the case where the user asked for a NIC via "-net nic"
but the board didn't instantiate that NIC (for example where the
user asked for two NICs but the board only supports one). Note
that this diagnostic doesn't apply to NICs created through -device,
because those are always instantiated.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
In previous life qdev_init_nofail() used to call hw_error() which
did register dump and other scary things. Now it calls
error_report() and does a regular exit(1). Fix the comment
to match reality.
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Before commit 622b520f, index=12 meant bus=1,unit=5.
Since the commit, it means bus=0,unit=12. The drive is created, but
not the guest device. That's because the controllers we use with
if=scsi drives (lsi53c895a and esp) support only 7 units, and
scsi_bus_legacy_handle_cmdline() ignores drives with unit numbers
exceeding that limit.
Changing the mapping of index to bus, unit is a regression. Breaking
-drive invocations that used to work just makes it worse.
Revert the part of commit 622b520f that causes this, and clean up
some.
Note that the fix only affects if=scsi. You can still put more than 7
units on a SCSI bus with -device & friends.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
qdev_init_bdrv() doesn't belong into qdev.c; it's about drives, not
qdevs. Rename to drive_get_next, move to blockdev.c, drop the bogus
DeviceState argument, and return DriveInfo instead of
BlockDriverState.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Create a trivial interface to track whether the machine has been
modified since boot. Adding or removing devices will trigger this
to return true. An example usage scenario for such an interface is
the rtl8139 driver which includes a cpu_register_io_memory() value
in it's migration stream. For the majority of migrations, where
no hotplug has occured in the machine, this works correctly. Once
the machine is modified, we can use this interface to detect that
and include a subsection for the device to prevent migrations to
rtl8139 versions with this bug.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch exports qdev_find_recursive() for later use.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Stefan Weil reported the regression caused by
ec990eb622 as follows
> The second regression also occurs with MIPS malta.
> Networking no longer works with the default pcnet nic.
>
> This is caused because the reset function for pcnet is no
> longer called during system boot. The result in an invalid
> mac address (all zero) and a non-working nic.
>
> For this second regression I still have no simple solution.
> Of course mips_malta.c should be converted to qdev which
> would fix both problems (but only for malta system emulation).
The issue is, it is assumed that all qbuses, qdeves are under
main_system_bus. But there are qbuses whose parent is NULL. So it
is necessary to trigger reset for those qbuses.
(On the other hand, if NULL is passed to qdev_create(), its parent bus
is main_system_bus.)
Ideally those buses should be moved under bus controller
device which is qdev. But it's not done yet.
So register qbus reset handler for qbus whose parent is NULL.
Reported-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: "Michael S. Tsirkin" <mst@redhat.com>
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Every system should have some sort of main system bus,
so sysbus_get_default should always return a valid bus.
Without this patch, at least mipssim and malta no longer
start but raise a null pointer access exception (caused by
commit ec990eb622).
Cc: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
If bootindex is specified on command line a string that describes device
in firmware readable way is added into sorted list. Later this list will
be passed into firmware to control boot order.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Introduce a helper function which triggers reset from a given device.
Will be used by pci bus emulation.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
and make it called via qbus_reset_all().
The qbus reset callback will be used by pci bus reset.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch changes the reset handling so that qdev has no knowledge of the
global system reset. Instead, a new bus/device level function is introduced
that allows all devices/buses on the bus/device to be reset using a depth
first transversal.
N.B. we have to expose the implicit system bus because we have various hacks
that result in an implicit system bus existing. Instead, we ought to have an
explicitly created system bus that we can trigger reset from. That's a topic
for a future patch though.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
There are some cases where you want to walk the busses, in particular, when
searching for a bus either by name or DeviceInfo.
Paolo suggested that we model the return values on how GCC's walkers work which
allows an actor to skip child transversal, or terminate walking with a positive
value that's returned as the qbus_walk_children's result.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Changing block.h or blockdev.h resulted in recompiling most objects.
Move DriveInfo typedef and BlockInterfaceType enum definitions
to qemu-common.h and rearrange blockdev.h use to decrease churn.
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Switch tree to lookup-by-name using qemu_find_opts().
Also hook up virtfs options so qemu_find_opts works for them too.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
When available, we'd like to be able to access the DeviceState
when registering a savevm. For buses with a get_dev_path()
function, this will allow us to create more unique savevm
id strings.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Some of the failures are internal errors, and hw_error() is okay then.
But the common way to fail is bad user input, e.g. -global
isa-fdc.driveA=foo where drive foo has an unsupported rerror value.
exit(1) instead.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
BusState::name is allocated in qbus_create_inplace().
So it should be freed by qbus_free().
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
parse_string() qemu_strdup()s the property value. It is never freed.
It needs to be freed along with the device. Otherwise, the value of
scsi-disk property "ver" gets leaked when hot-unplugging the disk, for
instance.
Call new PropertyInfo method free() from qdev_free(). Implement it
for qdev_prop_string.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Previous commit added QMP documentation to the qemu-monitor.hx
file, it's is a copy of this information.
While it's good to keep it near code, maintaining two copies of
the same information is too hard and has little benefit as we
don't expect client writers to consult the code to find how to
use a QMP command.
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Some legacy users (mostly PC devices) of vmstate_register manage
instance IDs on their own, and that unfortunately in a way that is
incompatible with automatically generated ones. This so far prevents
switching those users to vmstates that are registered by qdev.
To establish a migration path, this patch introduces the concept of
alias IDs. They can be passed to an extended vmstate registration
service, and qdev provides a set service to be used during device init.
find_se will consider the alias in addition to the default ID. We can
then start generating the default ID automatically and writing it on
vmsave, thus converting that format without breaking support for upward
migration.
The user is required specify the highest vmstate version for which the
alias is required. Once this version falls behind the minimum required
for a specific vmstate, an assertion triggers to motivate cleaning up
the obsolete alias.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Note: our device unplug methods don't need conversion work, because
they can't currently fail.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
We sometimes permit omitting the first option name, for example
-device foo is short for -device driver=foo. The name to use
("driver" in the example) is passed as argument to qemu_opts_parse().
For each QemuOptsList, we use at most one such name.
Move the name into QemuOptsList, and pass whether to permit the
abbreviation. This ensures continued consistency, and simplifies the
commit after next in this series.
This reverts commit 3ced9f7a36.
The next commit will convert all of qdev_device_add() to QError, and
it'll be clearer with this partial conversion reverted.
Users can't set them, so qdev_device_help() shouldn't list them. Fix
that. Also make qdev_prop_parse() hide them instead of printing a
meaningless "has no parser" error message.
Their value means nothing to users, so qdev_print_props() shouldn't
print it. Fix by removing their print method.
Their only use is dirty hacks. Document that.
Users can't create them, so qdev_device_help() shouldn't list them.
Fix that.
Also make qdev_device_add() pretend they don't exist. Before, it
rejected them with a "can't be added via command line" message, which
wasn't quite right for monitor command device_add.
error_report() terminates the message with a newline. Strip it it
from its arguments.
This fixes a few error messages lacking a newline:
net_handle_fd_param()'s "No file descriptor named %s found", and
tap_open()'s "vnet_hdr=1 requested, but no kernel support for
IFF_VNET_HDR available" (all three versions).
There's one place that passes arguments without newlines
intentionally: load_vmstate(). Fix it up.
qbus_find() adds an informational line to error messages, and prints
both lines with one qemu_error(). Use error_printf() for the
informational line instead.
While there, simplify: instead of printing buffers filled by
qbus_list_bus() and qbus_list_dev() in one go, make them print it.
