KONRAD Frederic [Thu, 11 Apr 2013 14:30:00 +0000 (16:30 +0200)]
virtio-net-ccw: switch to the new API.
Here the virtio-net-ccw is modified for the new API. The device
virtio-net-ccw extends virtio-ccw-device as before. It creates and
connects a virtio-net-device during the init. The properties are not modified.
KONRAD Frederic [Thu, 11 Apr 2013 14:29:59 +0000 (16:29 +0200)]
virtio-net-s390: switch to the new API.
Here the virtio-net-s390 is modified for the new API. The device
virtio-net-s390 extends virtio-s390-device as before. It creates and
connects a virtio-net-device during the init. The properties are not modified.
KONRAD Frederic [Thu, 11 Apr 2013 14:29:58 +0000 (16:29 +0200)]
virtio-net-pci: switch to the new API.
Here the virtio-net-pci is modified for the new API. The device
virtio-net-pci extends virtio-pci. It creates and connects a
virtio-net-device during the init. The properties are not changed.
Anthony Liguori [Tue, 16 Apr 2013 14:45:20 +0000 (09:45 -0500)]
i440fx-test: add test for PAM functionality
This tests PAM settings for the i440fx. This test does a lot of
byte MMIO which is fairly slow with qtest today. But the test
does complete in under 2 seconds.
We don't fully emulate PAM largely because of limitations with
KVM so we #if 0 that part of the test case.
Anthony Liguori [Tue, 16 Apr 2013 14:45:19 +0000 (09:45 -0500)]
i440fx-test: add test to compare default register values
This test compares all of the default register values against the
spec. It turns out we deviate in quite a few places. These
places are really only visible to the BIOS though which is why
this hasn't created any problems.
The deviation actually happens in the core PCI layer so I suspect
it's not a simple fix if we really care to fix it. For now, just
disable the affected checks.
Anthony Liguori [Tue, 16 Apr 2013 14:45:15 +0000 (09:45 -0500)]
qtest: don't use system command to avoid double fork
Currently we waitpid on the child process we spawn off that does
nothing more than system() another process. While this does not
appear to be incorrect, it's wasteful and confusing so get rid of
it.
Anthony Liguori [Wed, 17 Apr 2013 15:14:07 +0000 (10:14 -0500)]
Merge remote-tracking branch 'bonzini/migration-writev' into staging
# By Paolo Bonzini
# Via Paolo Bonzini
* bonzini/migration-writev:
qemu-file: do not use stdio for qemu_fdopen
iov: handle partial writes from sendmsg and recvmsg
iov: reorganize iov_send_recv, part 3
iov: reorganize iov_send_recv, part 2
iov: reorganize iov_send_recv, part 1
qemu-file: drop socket_put_buffer
Paolo Bonzini [Wed, 27 Mar 2013 16:36:32 +0000 (17:36 +0100)]
qemu-file: do not use stdio for qemu_fdopen
This uses system calls directly for Unix file descriptors, so that the
efficient writev_buffer can be used. Pay attention to the possibility
of partial writes in writev.
Paolo Bonzini [Wed, 27 Mar 2013 16:36:31 +0000 (17:36 +0100)]
iov: handle partial writes from sendmsg and recvmsg
Partial writes can still happen in sendmsg and recvmsg, if a
signal is received in the middle of a write. To handle this,
retry the operation with a new offset/bytes pair.
This happens when starting a guest with '-device virtio-rng-pci',
issuing a 'cat /dev/hwrng' in the guest, while also doing 'cat
/dev/random' on the host.
Paolo Bonzini [Tue, 9 Apr 2013 15:43:43 +0000 (17:43 +0200)]
migration: initialize RAM to zero
Using qemu_memalign only leaves the RAM zero by chance, because libc
will usually use mmap to satisfy our huge requests. But memory will
not be zero when using MALLOC_PERTURB_ with a nonzero value. In the
case of incoming migration, this breaks a recently-introduced
invariant (commit f1c7279, migration: do not sent zero pages in
bulk stage, 2013-03-26).
To fix this, use mmap ourselves to get a well-aligned, always zero
block for the RAM. Mmap-ed memory is easy to "trim" at the sides.
This also removes the need to do something special on valgrind
(see commit c2a8238a, Support running QEMU on Valgrind, 2011-10-31),
thus effectively reverts that patch.
Anthony Liguori [Tue, 16 Apr 2013 15:28:58 +0000 (10:28 -0500)]
Merge remote-tracking branch 'kraxel/usb.80' into staging
# By Gerd Hoffmann (6) and Hans de Goede (1)
# Via Gerd Hoffmann
* kraxel/usb.80:
use libusb for usb-host
xhci: fix address device
xhci: use slotid as device address
xhci: fix portsc writes
xhci: add xhci_cap_write
xhci: remove leftover debug printf
usb-serial: Remove double call to qemu_chr_add_handlers( NULL )
Anthony Liguori [Tue, 16 Apr 2013 15:28:51 +0000 (10:28 -0500)]
Merge remote-tracking branch 'spice/spice.v69' into staging
# By Hans de Goede (5) and others
# Via Gerd Hoffmann
* spice/spice.v69:
spice-qemu-char: vmc_write: Don't write more bytes then we're asked too
spice-qemu-char: Remove intermediate buffer
spice-qemu-char: Add watch support
spice-qemu-char: Remove #ifdef-ed code for old spice-server compat
virtio-console: Remove any pending watches on close
virtio-console: Also throttle when less was written then requested
spice: (32 bit only) fix surface cmd tracking destruction
qxl: add 2000x2000 and 2048x2048 video modes
qxl: add 4k + 8k resolutions
Anthony Liguori [Tue, 16 Apr 2013 15:28:36 +0000 (10:28 -0500)]
Merge remote-tracking branch 'afaerber/qom-cpu' into staging
# By Igor Mammedov (8) and others
# Via Andreas Färber
* afaerber/qom-cpu:
target-cris: Override do_interrupt for pre-v32 CPU cores
qdev: Set device's parent before calling realize() down inheritance chain
cpu: Pass CPUState to *cpu_synchronize_post*()
target-i386: Split out CPU creation and features parsing
target-i386/cpu.c: Coding style fixes
ioapic: Replace FROM_SYSBUS() with QOM type cast
kvmvapic: Replace FROM_SYSBUS() with QOM type cast
target-i386: Split APIC creation from initialization in x86_cpu_realizefn()
target-i386: Consolidate error propagation in x86_cpu_realizefn()
qdev: Add qdev property for bool type
target-i386: Improve -cpu ? features output
target-i386: Fix including "host" in -cpu ? output
Gerd Hoffmann [Fri, 30 Nov 2012 15:02:11 +0000 (16:02 +0100)]
use libusb for usb-host
Reimplement usb-host on top of libusb.
Reasons to do this:
(1) Largely rewritten from scratch, nice opportunity to kill historical
cruft.
(2) Offload usbfs handling to libusb.
(3) Have a single portable code base instead of bsd + linux variants.
(4) Bring usb-host support to any platform supported by libusbx.
For now this goes side-by-side to the existing code. That is only to
simplify regression testing though, at the end of the day I want remove
the old code and support libusb exclusively. Merge early in 1.5 cycle,
remove the old code after 1.5 release or something like this.
Thanks to qdev the old and new code can coexist nicely on linux. Just
use "-device usb-host-linux" to use the old linux driver instead of the
libusb one (which takes over the "usb-host" name).
The bsd driver isn't qdev'ified so it isn't that easy for bsd.
I didn't bother making it runtime switchable, so you have to rebuild
qemu with --disable-libusb to get back the old code.
Gerd Hoffmann [Thu, 21 Mar 2013 09:59:05 +0000 (10:59 +0100)]
xhci: use slotid as device address
Is good enougth for unique device addresses and avoids the need for any
state for device addressing. Makes live migration support easier. Also
makes device->slot lookups trivial.
Hans de Goede [Fri, 5 Apr 2013 09:30:24 +0000 (11:30 +0200)]
spice-qemu-char: vmc_write: Don't write more bytes then we're asked too
This one took me eons to debug, but I've finally found it now, oh well.
The usage of the MIN macro in this line:
last_out = MIN(len, qemu_chr_be_can_write(scd->chr));
Causes qemu_chr_be_can_write to be called *twice*, since the MIN macro
evaluates its arguments twice (bad MIN macro, bad!). And the result of
the call can change between the 2 calls since the guest may have consumed
some data from the virtio ringbuffer between the calls!
When this happens it is possible for qemu_chr_be_can_write to return less
then len in the call made for the comparision, and then to return more then
len in the actual call for the return-value of MIN, after which we will end
up writing len data + some extra garbage, not good.
This patch fixes this by only calling qemu_chr_be_can_write once.
Alon Levy [Fri, 5 Apr 2013 09:30:23 +0000 (11:30 +0200)]
spice-qemu-char: Remove intermediate buffer
virtio-serial's buffer is valid when it calls us, and we don't
access it otherwise: vmc_read is only called in response to wakeup,
or else we set datalen=0 and throttle. Then vmc_read is called back,
we return 0 (not accessing the buffer) and set the timer to unthrottle.
Also make datalen int and not ssize_t (to fit spice_chr_write signature).
HdG: Update to apply to spice-qemu-char with new gio-channel based
flowcontrol support.
The uint32_t *app argument doesn't exist in real hardware. It was a hack in
xilinx_axidma/enet to fake the (secondary) control stream connection. Removed
the argument and added the second stream to axienet/dma.
xilinx_axienet/dma: Implement rx path flow control
Implement flow control for the RX data path from xilinx_axienet->xilinx_axidma.
On short return from axidma, then ethernet sets up the notify callback to resume
transfer from where it left off.
This also allows the ethernet to track whether there is an in progress transaction
and return false from ethernet can_receive() as appropriate.
If the DMA backs up or is disabled it waits for enablement. When the rx stream IO
region is touched, the can_push() notify function is called if set.
Add basic flow control to stream. A stream slave may return short, indicating
that it is not capable of accepting any more data at the present time. Polling
or a callback can be used via the can_push() function to determine when the
slave can receive again.
Create a separate child object to proxy the stream slave connection. This is
setup for future work where a second stream slave connection is needed. The
new child object is created at qdev init time and is linked back to the parent
(the ethernet device itself) automatically.
Stream slave masters differentiate which slave connection they are connected to
by linking to the proxy object rather than the parent.
Create a separate child object to proxy the stream slave connection. This is
setup for future work where a second stream slave connection is needed. The
new child object is created at qdev init time and is linked back to the parent
(the ethernet device itself) automatically.
Stream slave masters differentiate which slave connection they are connected to
by linking to the proxy object rather than the parent.
Explicitly make the ethernet a child of the machine. This is needed to set
and use links pre-realize. Also makes the ethernet initialization consistent
with its peer DMA.
The prescribed transition from SysBusDevice::init to Device::realize. I'm going
with Andreas suggestion to move the sysbus foo to Object::init for early IRQ
visibility.
The prescribed transition from SysBusDevice::init to Device::realize. Im going
with Andreas suggestion to move the sysbus foo to Object::init for early IRQ
visibility.
Gerd Hoffmann [Fri, 15 Mar 2013 14:45:54 +0000 (15:45 +0100)]
console: allow pinning displaychangelisteners to consoles
DisplayChangeListener gets a new QemuConsole field, which can be set to
non-NULL before registering. This will pin the QemuConsole, so that
particular DisplayChangeListener will not follow console switches.
spice+gtk (which don't support text console input anyway) are switched
over to be pinned to console 0, which usually is the graphical display.
Gerd Hoffmann [Tue, 19 Mar 2013 14:01:02 +0000 (15:01 +0100)]
xen: re-enable refresh interval reporting for xenfb
xenfb informs the guest about the gui refresh interval so it can avoid
pointless work. That logic was temporarely disabled for the
DisplayState reorganization. Restore it now, with a proper interface
for it.
Gerd Hoffmann [Thu, 14 Mar 2013 10:56:16 +0000 (11:56 +0100)]
console: gui timer fixes
Make gui update rate adaption code in gui_update() actually work.
Sprinkle in a tracepoint so you can see the code at work. Remove
the update rate adaption code in vnc and make vnc simply use the
generic bits instead.
Gerd Hoffmann [Wed, 13 Mar 2013 09:14:52 +0000 (10:14 +0100)]
console: zap g_width + g_height
We have a surface per QemuConsole now, so there is no need to keep
track of the QemuConsole size any more as we can query the surface
size directly at any time.
Gerd Hoffmann [Tue, 12 Mar 2013 13:48:31 +0000 (14:48 +0100)]
console: simplify screendump
Screendumps are alot simpler as we can update non-active
QemuConsoles now. So we only need to update the QemuConsole
we want write out, then dump the DisplaySurface content into
a ppm file. Done.
No console switching needed. No special support code in the
gfx card emulation needed. Zap it all. Also move ppm_save
out of the vga code and next to the qmp_screendump function.
For now screen dumping is limited to console #0 (like it used
to be), even though it is dead simple to extend it to other
consoles. I wanna finish the console cleanup before setting
new qapi interfaces into stone.
Gerd Hoffmann [Tue, 12 Mar 2013 13:39:22 +0000 (14:39 +0100)]
console: give each QemuConsole its own DisplaySurface
Go away from the global DisplaySurface, give one to each QemuConsole
instead. With this patch applied it is possible to call
graphics_hw_* functions with qemu consoles which are not the current
foreground console.
Gerd Hoffmann [Tue, 12 Mar 2013 12:44:38 +0000 (13:44 +0100)]
console: rename vga_hw_*, add QemuConsole param
Add QemuConsole parameter to vga_hw_*, so the interface allows to update
non-active consoles (the actual code can't handle this yet, see next
patch). Passing NULL is allowed and updates the active console, like
the functions do today.
While touching all vga_hw_* calls anyway rename that to the functions to
hardware-neutral graphics_hw_*
Gerd Hoffmann [Thu, 7 Mar 2013 16:08:29 +0000 (17:08 +0100)]
console: displaystate init revamp
We have only one DisplayState, so there is no need for the "next"
linking, rip it. Also consolidate all displaystate initialization
into init_displaystate(). This function is called by vl.c after
creating the devices (and thus all QemuConsoles) and before
initializing DisplayChangeListensers (aka gtk/sdl/vnc/spice ui).
Gerd Hoffmann [Mon, 25 Mar 2013 10:44:21 +0000 (11:44 +0100)]
hw/vmware_vga.c: various vmware vga fixes.
Hardcode depth to 32 bpp. It effectively was that way before because
that is the default surface depth, this just makes it explicit in the
code.
Rename depth to new_depth to make it consistent with the new_width +
new_height names. In theory we can make new_depth changeable (i.e.
allow the guest to fill in -- say -- 16 there). In practice the guests
don't try, the X-Server refuses to start if you ask it to use 16bpp
depth (via DefaultDepth in the Screen section).
Always return the correct rmask+gmask+bmask values for the given
new_depth.
Fix mode setting to also verify at new_depth to make sure we have a
correct DisplaySurface, even if the current video mode happes to be
16bpp (set by vgabios via bochs vbe interface). While being at it
switch over to use qemu_create_displaysurface_from, so the surface is
backed by guest-visible video memory and we save a memcpy.
Igor Mitsyanko [Tue, 19 Mar 2013 19:44:56 +0000 (23:44 +0400)]
hw/vmware_vga.c: fix screen resize bug introduced after console revamp
In vmsvga display update function, a pointer to DisplaySurface must be acquired
after a call to vmsvga_check_size since this function might replace current
DisplaySurface with a new one.
Igor Mitsyanko [Tue, 19 Mar 2013 19:44:55 +0000 (23:44 +0400)]
exynos4210_fimd.c: fix display resize bug introduced after console revamp
In exynos4210 display update function, we were acquiring DisplaySurface
pointer before calling screen resize function, not paying attention that resize
procedure can replace current DisplaySurface with newly allocated one.
Right thing to do is to initialize DisplaySurface AFTER a call to resize function.
Andreas Färber [Mon, 18 Feb 2013 18:59:39 +0000 (19:59 +0100)]
target-cris: Override do_interrupt for pre-v32 CPU cores
Instead of forwarding from cris_cpu_do_interrupt() to do_interruptv10(),
override CPUClass::do_interrupt with crisv10_cpu_do_interrupt() in the
newly introduced class_init functions.
Igor Mammedov [Thu, 11 Apr 2013 14:51:56 +0000 (16:51 +0200)]
qdev: Set device's parent before calling realize() down inheritance chain
Currently device_set_realized() sets parent only after device was realized,
but qdev_device_add() sets it before device is realized.
Make behavior consistent and alter device_set_realized() to behave like
qdev_device_add().
It will allow to set link<> properties in realize() method in classes
inherited from DEVICE.
Igor Mammedov [Thu, 11 Apr 2013 14:51:40 +0000 (16:51 +0200)]
target-i386: Split out CPU creation and features parsing
Move CPU creation and features parsing into a separate cpu_x86_create()
function, so that board would be able to set board-specific CPU
properties before CPU is realized.
Keep cpu_x86_init() for compatibility with the code that uses cpu_init()
and doesn't need to modify CPU properties.
Igor Mammedov [Fri, 5 Apr 2013 14:36:54 +0000 (16:36 +0200)]
target-i386: Split APIC creation from initialization in x86_cpu_realizefn()
When APIC is hotplugged during CPU hotplug, device_set_realized()
calls device_reset() on it. And if QEMU runs in KVM mode, following
call chain will fail:
apic_reset_common()
-> kvm_apic_vapic_base_update()
-> kvm_vcpu_ioctl(cpu->kvm_fd,...)
due to cpu->kvm_fd not being initialized yet.
cpu->kvm_fd is initialized during qemu_init_vcpu() but x86_cpu_apic_init()
can't be moved after it because kvm_init_vcpu() -> kvm_arch_reset_vcpu()
relies on APIC to determine if CPU is BSP for setting initial env->mp_state.
So split APIC device creation from its initialization and realize APIC
after CPU is created, when it's safe to call APIC's reset method.
projects / qemu.git / log