When a VM state change handler changes VM state, other VM state change
handlers can see the state transitions out of order.
bmdma_map(), scsi_disk_init() and virtio_blk_init() install VM state
change handlers to restart DMA. These handlers can vm_stop() by
running into a write error on a drive with werror=stop. This throws
the VM state change handler callback into disarray. Here's an example
case I observed:
0. The virtual IDE drive goes south. All future writes return errors.
1. Something encounters a write error, and duly stops the VM with
vm_stop().
2. vm_stop() calls vm_state_notify(0).
3. vm_state_notify() runs the callbacks in list vm_change_state_head.
It contains ide_dma_restart_cb() installed by bmdma_map(). It also
contains audio_vm_change_state_handler() installed by audio_init().
When sending a copyrect command to the vnc client, we must also update
the local server surface. Otherwise the server's and the client's idea
of the screen content run out of sync and screen updates don't work
correctly.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Paolo Bonzini [Mon, 27 Jul 2009 21:17:51 +0000 (23:17 +0200)]
fix migration to obey -S
Since migration returns right away, starting the VM right
after calling qemu_start_incoming_migration is wrong even
if -S is not passed. We have to do this after migration
has completed.
Cc: Glauber Costa <glommer@redhat.com> Cc: Anthony Liguori <aliguori@us.ibm.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Both input and output streams may be in SND_PCM_STATE_SUSPENDED
after the host is suspended and resumed, meaning "Hardware is
suspended". snd_pcm_readi() and snd_pcm_writei() will return
-ESTRPIPE if called while the stream is in this state.
Call snd_pcm_resume() to enable audio output and capture after
host resume.
Juan Quintela [Mon, 27 Jul 2009 14:12:41 +0000 (16:12 +0200)]
Remove SP_CFLAGS and SP_LDFLAGS
SP_CFLAGS and SP_LDFLAGS are only used as initial values for ARCH_CFLAGS/ARCH_LDFLAGS. Call it directly ARCH_*. Once there, use the same indentantion that the rest of the file
Signed-off-by: Juan Quintela <quintela@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Filip Navara [Mon, 27 Jul 2009 15:02:04 +0000 (10:02 -0500)]
Remove setvbuf(<handle>, NULL, _IOLBF, 0) calls for Win32
On Win32 the setvbuf function requires the last parameter to be size between 2 and INT_MAX bytes, so the calls always failed. Since the whole point of the calls is to set line-buffered mode for the file handle and that's not supported on Win32 anyway, conditionally remove them.
Signed-off-by: Filip Navara <filip.navara@gmail.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Glauber Costa [Fri, 24 Jul 2009 20:20:23 +0000 (16:20 -0400)]
fix broken migration
While fixing migration with -S, commit 89befdd1a6b18215153b8976682d57b7d03d5782 broke the rest of us. Poor
glommer, with a poor family, spare him his life from this monstruosity.
Since the unconditional vm_start, not autostart was the villain, I'm putting
back autostart. Let me know if you prefer other solutions, it doesn't really matter,
doesn't really matter to me.
Any way the wind blows...
Signed-off-by: Glauber Costa <glommer@redhat.com> CC: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Ed Swierk [Thu, 23 Jul 2009 21:13:34 +0000 (14:13 -0700)]
slirp: Use monotonic clock if available (v2)
Calling gettimeofday() to compute a time interval can cause problems if
the system clock jumps forwards or backwards; replace updtime() with
qemu_get_clock(rt_clock), which calls clock_gettime(CLOCK_MONOTONIC) if
it is available.
Also remove some useless macros.
Signed-off-by: Ed Swierk <eswierk@aristanetworks.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
The UDP emulation code for talk has been commented out since the
beginning of time, and unless someone who runs CU-SeeMe on qemu with
user-mode networking can vouch that the special magic (a) is necessary
and (b) works, let's get rid of the code.
Signed-off-by: Ed Swierk <eswierk@aristanetworks.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Alexander Graf [Mon, 27 Jul 2009 10:49:56 +0000 (12:49 +0200)]
Use Little Endian for Dirty Log
We currently use host endian long types to store information
in the dirty bitmap.
This works reasonably well on Little Endian targets, because the
u32 after the first contains the next 32 bits. On Big Endian this
breaks completely though, forcing us to be inventive here.
So Ben suggested to always use Little Endian, which looks reasonable.
We only have dirty bitmap implemented in Little Endian targets so far
and since PowerPC would be the first Big Endian platform, we can just
as well switch to Little Endian always with little effort without
breaking existing targets.
This is the userspace part of the patch. It shouldn't change anything
for existing targets, but help PowerPC.
It replaces my older patch called "Use 64bit pointer for dirty log".
Signed-off-by: Alexander Graf <agraf@suse.de> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Alexander Graf [Fri, 24 Jul 2009 23:16:44 +0000 (01:16 +0200)]
Use 64bit pointer for dirty log
Dirty logs currently get written with native "long" size. On little endian
it doesn't matter if we use uint64_t instead though, because we'd still end
up using the right bytes.
On big endian, this does become a bigger problem, so we need to ensure that
kernel and userspace talk the same language, which means getting rid of "long"
and using a defined size instead.
So I decided to use 64 bit types at all times. This doesn't break existing
targets but will in conjunction with a patch I'll send to the KVM ML make
dirty logs work with 32 bit userspace on 64 kernel with big endian.
Signed-off-by: Alexander Graf <agraf@suse.de> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Beth Kon [Fri, 24 Jul 2009 16:26:59 +0000 (12:26 -0400)]
HPET fixes for reg writes
This patch addresses the problems found by Andriy Gapon:
- The code was incorrectly overwriting the high order 32
bits of the timer and hpet config registers. This didn't show up
in testing because linux and windows use hpet in legacy mode,
where the high order 32 bits (advertising available interrupts)
of the timer config register are ignored, and the high order 32
bits of the hpet config register are reserved and unused.
- The mask for level-triggered interrupts was off by a bit. (hpet
doesn't currently support level-triggered interrupts).
In addition, I removed some unused #defines, and corrected the ioapic
interrupt values advertised. I'd set this up early in hpet development
and never went back to correct it, and no bugs resulted since linux and
windows use hpet in legacy mode where available interrupts are ignored.
Signed-off-by: Beth Kon <eak@us.ibm.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Jan Kiszka [Wed, 22 Jul 2009 15:03:52 +0000 (17:03 +0200)]
slirp: Fix guestfwd for incoming data
Unless a virtual server address was explicitly defined (which is
impossible with the legacy -net channel format), guestfwd did not
properly forwarded host->guest packets. This patch fixes it.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Implementing a alternative way to specify the filename should be
just a few lines of code now once we decided how the cmd line syntax
should look like.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
QemuOpts: framework for storing and parsing options.
This stores device parameters in a better way than unparsed strings.
New types:
QemuOpt - one key-value pair.
QemuOpts - group of key-value pairs, belonging to one
device, i.e. one drive.
QemuOptsList - list of some kind of devices, i.e. all drives.
Functions are provided to work with these types. The plan is that some
day we will pass around QemuOpts pointers instead of strings filled with
"key1=value1,key2=value2".
Check out the next patch to see all this in action ;)
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
cleanup pretty simliar to the drives_table removal patch:
- drop the table and make a linked list out of it.
- pass around struct pointers instead of table indices.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
-drive accepts the new id= now, allowing to explicitely name your
drives. They will show up with that name in "info block" if specified,
otherwise the existing namimg scheme is used to autogenerate one.
There is also a new function to lookup drives by name. Not used yet.
The plan is to link disk drivers and drives using the drive id instead
of passing around pointers to BlockDriveState.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
First step cleaning up the drives handling. This one does nothing but
removing drives_table[], still it became seriously big.
drive_get_index() is gone and is replaced by drives_get() which hands
out DriveInfo pointers instead of a table index. This needs adaption in
*tons* of places all over.
The drives are now maintained as linked list.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
This patch implements a parser and qdev tree walker for bus paths and
adds qdev_device_add on top of this.
A bus path can be:
(1) full path, i.e. /i440FX-pcihost/pci.0/lsi/scsi.0
(2) bus name, i.e. "scsi.0". Best used together with id= to make
sure this is unique.
(3) relative path starting with a bus name, i.e. "pci.0/lsi/scsi.0"
For the (common) case of a single child bus being attached to a device
it is enougth to specify the device only, i.e. "pci.0/lsi" will be
accepted too.
qdev_device_add() adds devices and accepts bus= parameters to find the
bus the device should be attached to. Without bus= being specified it
takes the first bus it finds where the device can be attached to (i.e.
first pci bus for pci devices, ...).
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Create a default bus name if none is passed to qbus_create().
If the parent device has DeviceState->id set it will be used to create
the bus name,. i.e. -device lsi,id=foo will give you a scsi bus named
"foo.0".
If there is no id BusInfo->name (lowercased) will be used instead, i.e.
-device lsi will give you a scsi bus named "scsi.0".
A scsi adapter with two scsi busses would have "scsi.0" and "scsi.1" or
"$id.0" and "$id.1" busses. The numbers of the child busses are per
device, i.e. when adding two lsi adapters both will have a "*.0" child
bus.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Mark McLoughlin [Wed, 22 Jul 2009 09:02:51 +0000 (10:02 +0100)]
Add a pc-0.11 machine type and make the pc type an alias
The pc-0.11 type allows users of qemu-0.11 to use a machine type which
they know will remain compatible when the upgrade to qemu-0.12.
Management tools may choose to canonicalize the 'pc' machine type to
'pc-0.11' so that if the 'pc' alias changes target in future versions
of qemu, the machine type used will remain compatible.
Signed-off-by: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Mark McLoughlin [Wed, 22 Jul 2009 09:02:50 +0000 (10:02 +0100)]
Add machine type aliases
Add an 'alias' field to QEMUMachine and display it in the output of
'qemu -M ?' with an '(aliased to foo)' suffix.
Aliases can change targets in newer versions of qemu, so management tools
may choose canonicalize machine types to ensure that if a user chooses an
alias, that the actual machine type used will remain compatible in
future.
This is intended to mimic a symlink to a machine description file.
Signed-off-by: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Mark McLoughlin [Wed, 22 Jul 2009 08:11:42 +0000 (09:11 +0100)]
Add support for fd=name to tap and socket networking
This allows a program to initialize a host networking device using a
file descriptor passed over a unix monitor socket.
The program must first pass the file descriptor using SCM_RIGHTS
ancillary data with the getfd monitor command. It then may do
"host_net_add tap fd=name" to use the named file descriptor.
Signed-off-by: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>