Alexander Graf [Thu, 17 May 2012 14:58:55 +0000 (16:58 +0200)]
dt: add helper for phandle allocation
Phandle references work by having 2 pieces:
- a "phandle" 1-cell property in the device tree node
- a reference to the same value in a property we want to point
to the other node
To generate the 1-cell property, we need an allocation mechanism that
gives us a unique number space. This patch adds an allocator for these
properties.
Alexander Graf [Thu, 17 May 2012 13:33:54 +0000 (15:33 +0200)]
dt: add helper for empty dt creation
We want to get rid of the concept of loading an external device tree and instead
generate our own. However, to do this we need to also create a device tree
template programatically.
This patch adds a helper to create an empty device tree in memory.
Alexander Graf [Thu, 17 May 2012 13:23:39 +0000 (15:23 +0200)]
dt: add helper for phandle enumeration
This patch adds a helper to search for a node's phandle by its path. This
is especially useful when the phandle is part of an array, not just a single
cell in which case qemu_devtree_setprop_phandle would be the easy choice.
Usually we want to know when creating a subtree fails. However, while
introducing this patch set we have to modify the device tree and some
times have the code to create a subtree in both the binary tree and
the dynamically created tree.
So ignore failures about this for now and enable them once we got rid
of the binary device tree.
Alexander Graf [Thu, 17 May 2012 12:11:52 +0000 (14:11 +0200)]
dt: add helper for phandle references
Phandles are the fancy device tree name for "pointer to another node".
To create a phandle property, we most likely want to reference to the
node we're pointing to by its path. So create a helper that allows
us to do so.
Alexander Graf [Thu, 17 May 2012 10:47:57 +0000 (12:47 +0200)]
dt: add helpers for multi-cell adds
We have device tree helpers that allow us to create single cell (u32)
wide properties. However, when creating properties that contain an array of
cells, we need to jump through hoops, manually passing in an array with
converted endianness.
To ease the pain of this, create a generic macro helper that allows us
to pass the cells as arguments.
Alexander Graf [Thu, 17 May 2012 09:40:42 +0000 (11:40 +0200)]
dt: allow add_subnode to create root subnodes
Our subnode creation helper can't handle creation of root subnodes,
like "/memory". Fix this by allowing the parent node to be an empty
string, indicating the root node.
This adds a qemu-specific hypervisor call to the pseries machine
which allows to do what amounts to memmove, memcpy and xor over
regions of physical memory such as the framebuffer.
This is the simplest way to get usable framebuffer speed from
SLOF since the framebuffer isn't mapped in the VRMA and so would
otherwise require an hcall per 8 bytes access.
The performance is still not great but usable, and can be improved
with a more complex implementation of the hcall itself if needed.
This also adds some documentation for the qemu-specific hypercalls
that we add to PAPR along with a new qemu,hypertas-functions property
that mirrors ibm,hypertas-functions and provides some discoverability
for the new calls.
Note: I chose note to advertise H_RTAS to the guest via that mechanism.
This is done on purpose, the guest uses the normal RTAS interfaces
provided by qemu (including SLOF) which internally calls H_RTAS.
We might in the future implement part (or even all) of RTAS inside the
guest like IBM's firmware does and replace H_RTAS with some finer grained
set of private hypercalls.
The core tcg/kvm code for ppc64 now has at least the outline
capability to support pagesizes beyond the standard 4k and 16MB. The
CPUState is initialized with information advertising the available
pagesizes and their correct encodings, and under the right KVM setup
this will be populated with page sizes beyond the standard.
Obviously guests can't use the extra page sizes unless they know
they're present. For the pseries machine, at least, there is a
defined method for conveying exactly this information, the
"ibm-segment-page-sizes" property in the guest device tree.
This patch generates this property using the supported page size
information that's already in the CPUState.
ppc64: Rudimentary Support for extra page sizes on server CPUs
More recent Power server chips (i.e. based on the 64 bit hash MMU)
support more than just the traditional 4k and 16M page sizes. This
can get quite complicated, because which page sizes are supported,
which combinations are supported within an MMU segment and how these
page sizes are encoded both in the SLB entry and the hash PTE can vary
depending on the CPU model (they are not specified by the
architecture). In addition the firmware or hypervisor may not permit
use of certain page sizes, for various reasons. Whether various page
sizes are supported on KVM, for example, depends on whether the PR or
HV variant of KVM is in use, and on the page size of the memory
backing the guest's RAM.
This patch adds information to the CPUState and cpu defs to describe
the supported page sizes and encodings. Since TCG does not yet
support any extended page sizes, we just set this to NULL in the
static CPU definitions, expanding this to the default 4k and 16M page
sizes when we initialize the cpu state. When using KVM, however, we
instead determine available page sizes using the new
KVM_PPC_GET_SMMU_INFO call. For old kernels without that call, we use
some defaults, with some guesswork which should do the right thing for
existing HV and PR implementations. The fallback might not be correct
for future versions, but that's ok, because they'll have
KVM_PPC_GET_SMMU_INFO.
Fabien Chouteau [Mon, 21 May 2012 06:11:06 +0000 (06:11 +0000)]
booke_206_tlbwe: Discard invalid bits in MAS2
The size of EPN field in MAS2 depends on page size. This patch adds a
mask to discard invalid bits in EPN field.
Definition of EPN field from e500v2 RM:
EPN Effective page number: Depending on page size, only the bits
associated with a page boundary are valid. Bits that represent offsets
within a page are ignored and should be cleared.
There is a similar (but more complicated) definition in PowerISA V2.06.
Alexander Graf [Fri, 18 May 2012 22:48:50 +0000 (00:48 +0200)]
PPC: mpc8544ds: Span initial TLB entry over as much RAM as we need
The initial TLB entry is supposed to help us run the guest -kernel payload.
This means the guest needs to be able to access its own memory, the initrd
memory and the device tree.
So far we only statically reserved a TLB entry from [0;256M[. This patch
fixes it to span from [0;dt_end[, allowing the guest payload to access
everything initially.
Blue Swirl [Wed, 30 May 2012 04:23:40 +0000 (04:23 +0000)]
ppc: Move load and store helpers, switch to AREG0 free mode
Add an explicit CPUPPCState parameter instead of relying on AREG0
and rename op_helper.c (which only contains load and store helpers)
to mem_helper.c. Remove AREG0 swapping in
tlb_fill().
Switch to AREG0 free mode. Use cpu_ld{l,uw}_code in translation
and interrupt handling, cpu_{ld,st}{l,uw}_data in loads and stores.
Blue Swirl [Wed, 30 May 2012 04:23:32 +0000 (04:23 +0000)]
ppc: Avoid a warning with the next patch
When the code is moved together by the next patch, compiler
detects a possible uninitialized variable use. Avoid the warning
by initializing the variables.
Blue Swirl [Sun, 10 Jun 2012 10:18:54 +0000 (10:18 +0000)]
libcacard: build fixes
Link trace objects to fix these errors:
LINK vscclient
oslib-posix.o: In function `trace_qemu_vfree':
/src/qemu/obj-amd64/./trace.h:39: undefined reference to `trace1'
oslib-posix.o: In function `trace_qemu_memalign':
/src/qemu/obj-amd64/./trace.h:31: undefined reference to `trace3'
oslib-posix.o: In function `trace_qemu_vmalloc':
/src/qemu/obj-amd64/./trace.h:35: undefined reference to `trace2'
The timer was deadlocking when the interval was set too low. It would cause a
flood of timer events and the CPU would halt indefinately. This is a known issue
and theres a generic workaround in place in ptimer on ptimer_set_limit(),
however the Xilinx timer uses ptimer_set_count() instead of set_limit. Changed
the call to set_count() to an equivalent call of set_limit() instead, which
brings the workaround into play.
Crístian Viana [Wed, 30 May 2012 03:35:51 +0000 (00:35 -0300)]
Allow machines to configure the QEMU_VERSION that's exposed via hardware
QEMU exposes its version to the guest's hardware and in some cases that is wrong
(e.g. Windows prints messages about driver updates when you switch
the QEMU version).
There is a new field now on the struct QEmuMachine, hw_version, which may
contain the version that the specific machine should report. If that field is
set, then that machine will report that version to the guest.
Gleb Natapov [Mon, 4 Jun 2012 11:31:55 +0000 (14:31 +0300)]
Add PIIX4 properties to control PM system states.
This patch adds two things. First it allows QEMU to distinguish between
regular powerdown and S4 powerdown. Later separate QMP notification will
be added for S4 powerdown. Second it allows S3/S4 states to be disabled
from QEMU command line. Some guests known to be broken with regards to
power management, but allow to use it anyway. Using new properties
management will be able to disable S3/S4 for such guests.
Supported system state are passed to a firmware using new fw_cfg file.
The file contains 6 byte array. Each byte represents one system
state. If byte at offset X has its MSB set it means that system state
X is supported and to enter it guest should use the value from lowest 3
bits.
Anthony Liguori [Mon, 18 Jun 2012 15:35:16 +0000 (10:35 -0500)]
Merge remote-tracking branch 'afaerber-or/qom-next-2' into staging
* afaerber-or/qom-next-2: (22 commits)
qom: Push error reporting to object_property_find()
qdev: Remove qdev_prop_exists()
qbus: Initialize in standard way
qbus: Make child devices links
qdev: Connect busses with their parent devices
qdev: Convert busses to QEMU Object Model
qdev: Move SysBus initialization to sysbus.c
qdev: Use wrapper for qdev_get_path
qdev: Remove qdev_prop_set_defaults
qdev: Clean up global properties
qdev: Move bus properties to abstract superclasses
qdev: Move bus properties to a separate global
qdev: Push "type" property up to Object
arm_l2x0: Rename "type" property to "cache-type"
m48t59: Rename "type" property to "model"
qom: Assert that public types have a non-NULL parent field
qom: Drop type_register_static_alias() macro
qom: Make Object a type
qom: Add class_base_init
qom: Add object_child_foreach()
...
Anthony Liguori [Mon, 18 Jun 2012 15:35:13 +0000 (10:35 -0500)]
Merge remote-tracking branch 'qmp/queue/qmp' into staging
* qmp/queue/qmp:
build: install qmp-commands.txt
Add rate limiting of RTC_CHANGE, BALLOON_CHANGE & WATCHDOG events
Add event notification for guest balloon changes
Fix some more license versions (GPL2+ instead of GPL2)
monitor: Fix memory leak with readline completion
qmp: do not include monitor.h from qapi-types-core.h
qmp: include monitor.h when needed
kvm: add missing include files
Anthony Liguori [Mon, 18 Jun 2012 15:34:59 +0000 (10:34 -0500)]
Merge remote-tracking branch 'kwolf/for-anthony' into staging
* kwolf/for-anthony: (39 commits)
qemu-iotests: add 036 autoclear feature bit test
qemu-iotests: add qcow2.py set-feature-bit command
fdc-test: introduced qtest read_without_media
fdc: fix implied seek while there is no media in drive
qcow2: fix autoclear image header update
xen: Don't peek behind the BlockDriverState abstraction
xen: Don't change -drive if=xen device name during machine init
block: Replace bdrv_get_format() by bdrv_get_format_name()
qemu-img: document qed format on qemu-img man page
qemu-iotests: COW with many AIO requests on the same cluster
qemu-iotests: Some backing file COW tests
qcow2: Fix avail_sectors in cluster allocation code
qcow2: Simplify calculation for COW area at the end
qcow2: always operate caches in writeback mode
ide: support enable/disable write cache
block: always open drivers in writeback mode
block: add bdrv_set_enable_write_cache
block: copy enable_write_cache in bdrv_append
savevm: flush after saving vm state
block: flush in writethrough mode after writes
...
Anthony Liguori [Mon, 18 Jun 2012 15:18:44 +0000 (10:18 -0500)]
Merge remote-tracking branch 'mst/tags/for_anthony' into staging
* mst/tags/for_anthony:
pci_bridge_dev: fix error path in pci_bridge_dev_initfn()
qdev: release parent properties on dc->init failure
msi: Use msi/msix_present more consistently
msi: Invoke msi/msix_write_config from PCI core
msi: Guard msi/msix_write_config with msi_present
msi: Invoke msi/msix_reset from PCI core
msi: Guard msi_reset with msi_present
ahci: Clean up reset functions
intel-hda: Fix reset of MSI function
ahci: Fix reset of MSI function
rtl8139: honor RxOverflow flag in can_receive method
shpc: unparent device before free
Anthony Liguori [Mon, 18 Jun 2012 15:18:26 +0000 (10:18 -0500)]
Merge remote-tracking branch 'sweil/for-anthony' into staging
* sweil/for-anthony:
Fix some more license versions (GPL2+ instead of GPL2)
dump: Fix license version (GPL2+ instead of GPL2)
configure: Fix build for some versions of glibc (9pfs)
monitor: Fix memory leak with readline completion
Paolo Bonzini [Thu, 12 Apr 2012 16:00:18 +0000 (18:00 +0200)]
qom: Push error reporting to object_property_find()
Avoids duplicated error_set().
Signed-off-by: Paolo Bonzini <[email protected]>
[AF: Also drop error_set() in object_property_del().] Signed-off-by: Andreas Färber <[email protected]>
Anthony Liguori [Wed, 2 May 2012 07:00:20 +0000 (09:00 +0200)]
qdev: Convert busses to QEMU Object Model
This is far less interesting than it sounds. We simply add an Object to each
BusState and then register the types appropriately. Most of the interesting
refactoring will follow in the next patches.
Since we're changing fundamental type names (BusInfo -> BusClass), it all needs
to convert at once. Fortunately, not a lot of code is affected.
Signed-off-by: Anthony Liguori <[email protected]> Signed-off-by: Paolo Bonzini <[email protected]>
[AF: Made all new bus TypeInfos static const.]
[AF: Made qbus_free() call object_delete(), required {qom,glib}_allocated] Signed-off-by: Andreas Färber <[email protected]>
Anthony Liguori [Fri, 3 Feb 2012 18:28:43 +0000 (12:28 -0600)]
qdev: Use wrapper for qdev_get_path
This makes it easier to remove it from BusInfo.
Signed-off-by: Anthony Liguori <[email protected]> Signed-off-by: Paolo Bonzini <[email protected]>
[AF: Drop now unnecessary NULL initialization in scsibus_get_dev_path()] Signed-off-by: Andreas Färber <[email protected]>
Paolo Bonzini [Wed, 28 Mar 2012 16:12:47 +0000 (18:12 +0200)]
qdev: Move bus properties to abstract superclasses
In qdev, each bus in practice identified an abstract superclass, but
this was mostly hidden. In QOM, instead, these abstract classes are
explicit so we can move bus properties there.
All bus property walks are removed, and all device property walks
are changed to look along the class hierarchy instead.
We would have duplicates if class A defines some properties and its
subclass B does not define any, because class_b->props will be
left equal to class_a->props.
The solution here is to reintroduce the class_base_init TypeInfo
callback, that was present in one of the early QOM versions but
removed (on my request...) before committing.
This breaks global bus properties, an obscure feature when used
with the command-line which is actually useful and used when used by
backwards-compatible machine types. So this patch also adjusts the
global bus properties in hw/pc_piix.c to refer to the abstract class.
Globals and other properties must be modified in the same patch to
avoid complications related to initialization ordering.
Paolo Bonzini [Sat, 31 Mar 2012 14:45:54 +0000 (16:45 +0200)]
qom: Make Object a type
Right now the base Object class has a special NULL type. Change this so
that we will be able to add class_init and class_base_init callbacks.
To do this, remove some special casing of ObjectClass that is not really
necessary.
Paolo Bonzini [Wed, 2 May 2012 11:30:55 +0000 (13:30 +0200)]
qom: Add class_base_init
The class_base_init TypeInfo callback was present in one of the early
QOM versions but removed (on my request...) before committing. We
will need it soon, add it.
Paolo Bonzini [Wed, 11 Apr 2012 21:30:20 +0000 (23:30 +0200)]
qom: Add object_child_foreach()
A utility function that will be used to implement hierarchical realization.
Signed-off-by: Paolo Bonzini <[email protected]> Reviewed-by: Anthony Liguori <[email protected]>
[AF: Drop unrelated whitespace change, add Returns: in documentation]
[AF: Use new object_property_is_child() helper.] Signed-off-by: Andreas Färber <[email protected]>
Add rate limiting of RTC_CHANGE, BALLOON_CHANGE & WATCHDOG events
Allow certain event types to be rate limited to avoid flooding
monitor clients. The monitor_protocol_event() method is changed
such that instead of immediately emitting the event to Monitor
instances, it will call a new monitor_protocol_event_queue()
method.
This will check to see if the rate limit for the event has been
exceeded, and if so schedule a timer to wakeup at the end of the
rate limit period. If further events arrive before the timer fires,
the previously queued event will be discarded in favour of the new
event. The event will eventually be emitted when the timer fires.
This logic is applied to RTC_CHANGE, BALLOON_CHANGE & WATCHDOG
events, since the data associated with these events is stateless
* monitor.c: Add support for rate limiting
* monitor.h: Define monitor_global_init for one-time setup tasks
* vl.c: Invoke monitor_global_init
* trace-events: Add hooks for monitor event tracing
After setting a balloon target value, applications have to
continually poll 'query-balloon' to determine whether the
guest has reacted to this request. The virtio-balloon backend
knows exactly when the guest has reacted though, and thus it
is possible to emit a JSON event to tell the mgmt application
whenever the guest balloon changes.
This introduces a new 'qemu_balloon_changed()' API which is
to be called by balloon driver backends, whenever they have
a change in balloon value. This takes the 'actual' balloon
value, as would be found in the BalloonInfo struct.
The qemu_balloon_change API emits a JSON monitor event which
looks like:
* balloon.c, balloon.h: Introduce qemu_balloon_changed() for
emitting balloon change events on the monitor
* hw/virtio-balloon.c: Invoke qemu_balloon_changed() whenever
the guest changes the balloon actual value
* monitor.c, monitor.h: Define QEVENT_BALLOON_CHANGE
Stefan Hajnoczi [Thu, 14 Jun 2012 12:58:18 +0000 (13:58 +0100)]
qemu-iotests: add 036 autoclear feature bit test
This new test validates the autoclear feature bit behavior. When QEMU
opens a qcow2v3 image file with an unknown autoclear feature bit the bit
should be cleared in the image file header.
Pavel Hrdina [Wed, 13 Jun 2012 13:43:11 +0000 (15:43 +0200)]
fdc: fix implied seek while there is no media in drive
The Windows uses 'READ' command at the start of an instalation
without checking the 'dir' register. We have to abort the transfer
with an abnormal termination if there is no media in the drive.
Stefan Hajnoczi [Thu, 14 Jun 2012 10:42:23 +0000 (11:42 +0100)]
qcow2: fix autoclear image header update
The autoclear feature bits can be used for qcow2 file format features
that are safe to "drop" by old programs that do not understand the
feature. Upon opening the image file unknown autoclear feature bits are
cleared and the image file header is rewritten, but this was happening
too early in the code when critical header fields were not yet loaded.
Process autoclear feature bits after all necessary header information
has been loaded.
xen: Don't peek behind the BlockDriverState abstraction
First offender is xen_config_dev_blk()'s use of disk->bdrv->filename.
Get the filename from disk->opts instead. Same result, except for
snapshots: there, we now get the filename specified by the user
instead of the name of the temporary image created by bdrv_open().
Should be an improvement.
Second offender is blk_init()'s use of blkdev->bs->drv->format_name.
Simply use the appropriate interface to get the format name.
block: Replace bdrv_get_format() by bdrv_get_format_name()
So callers don't need to know anything about maximum name length.
Returning a pointer is safe, because the name string lives as long as
the block driver it names, and block drivers don't die.
Kevin Wolf [Wed, 13 Jun 2012 15:15:14 +0000 (17:15 +0200)]
qemu-iotests: COW with many AIO requests on the same cluster
This one is a bit more interesting. The COW operation isn't performed
completely synchronously, and therefore dependencies must be handled
correctly when multiple requests write to the same unallocated cluster.
m->nb_available in turn is only used for COW at the end of the
allocation. A COW occurs only if the request wasn't cluster aligned,
which in turn would imply that requested_sectors was less than
avail_sectors (both in the original and in the fixed version). In this
case avail_sectors is ignored and therefore the mistake doesn't cause
any misbehaviour.
Kevin Wolf [Tue, 22 May 2012 14:58:48 +0000 (16:58 +0200)]
qcow2: Simplify calculation for COW area at the end
copy_sectors() always uses the sum (cluster_offset + n_start) or
(start_sect + n_start), so if some value is added to both cluster_offset
and start_sect, and subtracted from n_start, it's cancelled out anyway.
projects / qemu.git / log