allow changing the speed of a running migration

This patch allow us to call migrate_set_speed on running
migrations. This should allow mgmt tools to increase the allocated
bandwidth of a running migration if there is no progress, and they
really want the migration to succeed.

Signed-off-by: Glauber Costa <glommer@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

introduce set_rate_limit function for QEMUFile

This patch converts the current callers of qemu_fopen_ops().

Signed-off-by: Glauber Costa <glommer@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

augment info migrate with page status

This patch augments info migrate output with status about:
* ram bytes remaining
* ram bytes transferred
* ram bytes total

This should be enough for management tools to realize
whether or not there is progress in migration. We can
add more information later on, if the need arrives

[v2: fixes bytes_transferred type]

Signed-off-by: Glauber Costa <glommer@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Don't send all gratuitous packets at once.

Use timer to separate them in time.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Minimal ethernet frame length is 64 bytes.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Fix segv when passing an unknown protocol

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Partially revert e20a8dff4c0da3827764924139d3bb73962f5d5a

From Paul Brook:

 "the fdc is tied to the ISA DMA engine. We don't currently have a target
  independent method of handling inter-device data transfer."

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Revert "Fix arm-softmmu breakage"

This reverts commit ec6bd8dea77478f32981a5df49f66ca2430ad19d.

This broke any target that uses virtio.  Virtio devices live in libhw and
without whole-archive, the constructors will never be called for virtio.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Merge commit 'garage/master' into garage-push

changelog update

Merge commit 'juri/riku-devel'

make it impossible to build a empty sb1 devkit

Introduce is_default field for QEMUMachine

f80f9ec changed the order that machines are registered which had the effect of
changing the default machine.  This changeset introduces a new is_default field
so that machine types can declare that they are the default for an architecture.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Refactor how display drivers are selected

My previous commit, f92f8afebe,  broke -vnc (spotted by Glauber Costa).  This
is because it's necessary to tell when the no special display parameters have
been passed and default to SDL or VNC appropriately.

This refactors the display selection logic to be less complicated which has
the effect of fixing the regression mentioned above.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Fix arm-softmmu breakage

Don't use whole-archive for hwlib or libqemu objects

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Compile most Xen files only once

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Compile fdc, escc and SCSI controllers only once

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Let the venomous poison spread to more identifiers

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Convert machine registration to use module init functions

This cleans up quite a lot of #ifdefs, extern variables, and other ugliness.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Eliminate --disable-gfx-check and make VNC default when SDL not available

 --disable-gfx-check predates VNC server support.  It made sense back then
because the only thing you could do without SDL was use -nographic mode or
similar tricks.  Since this is a very advanced mode of operation, gfx-check
provided a good safety net for casual users.

A casual user is very likely to use VNC to interact with a guest.  In fact, it's
often frustrating to install QEMU on a server and have to specify
disable-gfx-check when you only want to use VNC.

This patch eliminates disable-gfx-check and makes SDL behave like every other
optional dependency.  If SDL is not available, instead of failing ungracefully
if no special options are specified, we default to -vnc localhost:0,to=99.
When we do default to VNC, we also print a message to tell the user that we've
done this include which port we're currently listening on.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Do not attempt to allocate sn_tab when there are no snapshots

This was caught by a7d27b536ffc0773028a90f14580552c0c3ddb2a which
aborted on this attempt, thanks to Alex Ivanov for report.

Signed-off-by: malc <av1474@comtv.ru>

Cast pointer arguments of get/setsockopt, send to void * to keep GCC
from producing a warning about pointer type mismatches with Winsock

Signed-off-by: malc <av1474@comtv.ru>

kvm: add error message for when SMP is requested

Right now, if you try e.g. '-smp 2' you just get 'failed to
initialize KVM'.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>

Remove initrd warning message

Signed-off-by: Richard W.M. Jones <rjones@redhat.com>

initialize struct sigevent before timer_create

When qemu is run under valgrind, valgrind shows the following output
on exit:

==3648== 1 errors in context 2 of 2:
==3648== Syscall param timer_create(evp) points to uninitialised byte(s)
==3648==    at 0x54E936A: timer_create (in /lib/librt-2.9.so)
==3648==    by 0x405DCF: dynticks_start_timer (vl.c:1549)
==3648==    by 0x40A966: main (vl.c:1726)
==3648==  Address 0x7fefffb34 is on thread 1's stack
==3648==  Uninitialised value was created by a stack allocation
==3648==    at 0x405D60: dynticks_start_timer (vl.c:1534)

This patch is a simple fix to remove this potential problem.

Signed-off-by: Jean-Christophe DUBOIS <jcd@tribudubois.net>

Fix NULL alarm_timer pointer at exit

This fixes a SIGSEGV error on qemu exit.

Here is the valgrind output related to this error

==3648== Process terminating with default action of signal 11 (SIGSEGV)
==3648==  Access not within mapped region at address 0x8
==3648==    at 0x40636B: host_alarm_handler (vl.c:1345)
==3648==    by 0x52D807F: (within /lib/libpthread-2.9.so)
==3648==    by 0x5C0A12E: tcsetattr (in /lib/libc-2.9.so)
==3648==    by 0x4DD601: term_exit (qemu-char.c:700)
==3648==    by 0x5B636EC: exit (in /lib/libc-2.9.so)
==3648==    by 0x5B4B5AC: (below main) (in /lib/libc-2.9.so)

This simple fix check for a valid pointer as host_alarm_handler is
also called after alarm_timer is released in the exit path.

Signed-off-by: Jean-Christophe DUBOIS <jcd@tribudubois.net>

keep initrd in below 4g area.

initrd must be kept on the memory area below 4g. By not doing this,
we're seeing guests break while using -initrd and values of -mem
superior to 4096.

Signed-off-by: Glauber Costa <glommer@redhat.com>

migrate.c: migrate_fd_put_buffer: Do not busyloop: stop writing if EWOULDBLOCK

The migration code is non-blocking, designed for live migration.

Practically migrate_fd_put_buffer busy-loops trying to write, as
on many machines EWOULDBLOCK==EAGAIN (look in include/asm-generic/errno.h).

Signed-off-by: Uri Lublin <uril@redhat.com>

kvm: work around supported cpuid ioctl() brokenness

KVM_GET_SUPPORTED_CPUID has been known to fail to return -E2BIG
when it runs out of entries. Detect this by always trying again
with a bigger table if the ioctl() fills the table.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>

remove gcc 3.x requirement from documentation

This text is no longer accurate. After the patch is applied, the
generated version at http://www.nongnu.org/qemu/qemu-doc.html should be
regenerated.

This patch is also a candidate for the stable branch. (The URL above is
probably generated from the stable branch anyways, so maybe it goes
without saying.)

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

Remove dead code

Signed-off-by: malc <av1474@comtv.ru>

Format per CODING_STYLE

Signed-off-by: malc <av1474@comtv.ru>

Abort on attempts to allocate zero bytes

http://marc.info/?t=124267873300015&r=1&w=2

Signed-off-by: malc <av1474@comtv.ru>

Unbreak out-of-tree builds

Signed-off-by: malc <av1474@comtv.ru>

Hardware convenience library

The only target dependency for most hardware is sizeof(target_phys_addr_t).
Build these files into a convenience library, and use that instead of
building for every target.

Remove and poison various target specific macros to avoid bogus target
dependencies creeping back in.

Big/Little endian is not handled because devices should not know or care
about this to start with.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Disable >4G ram support on 32-bit targets

If the target only has a 32-bit physical address space then
the code to map >4G ram breaks horribly, and causes compiler warnings.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Only define __llseek if it is going to be used

Signed-off-by: Paul Brook <paul@codesourcery.com>

Avoid implicit truncation compiler warnings

Signed-off-by: Paul Brook <paul@codesourcery.com>

Remove obsolete BIOS_SIZE from sysemu.h

BIOS_SIZE is no longer needed by vl.c, so there's no point having it in
sysemu.h.

Signed-off-by: Paul Brook <paul@codesourcery.com>

cris: First shot at qdev for CRIS interrupts.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

etrax: Don't keep the passed irq pointer.

Copy passed irq object at channel connect.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

etrax: Remove unused eth irq line.

The ethernet blocks irq line to report errors is unimplemented in QEMU.
Remove it for now.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

Syborg virtio bindings.

Implement Syborg device bindings for virtio-net.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Separate virtio PCI code

Split the PCI host bindings from the VRing transport implementation.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Only shutdown video subsytem in sdl_cleanup

Depending on the order in which atexit handlers are called SDL might
try to join on an audio thread without said thread ever being notified
that it must stop, hence QEMU will forever block in pthread_join call.

Signed-off-by: malc <av1474@comtv.ru>

Consistently use uint64_t for int properties

I apparently failed to do this properly on the first attempt.

Signed-off-by: Paul Brook <paul@codesourcery.com>

ETRAX-PIC: Untabify.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX-TIMER: qdevify.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX-TIMER: Untabify.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX-SER: Untabify.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX-SER: qdevify.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX: Simplify PIC interface.

Instead of exporting a custom structure to represent different
interrupt types, just export the irq array and have the top
elements point to the NMI lines.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

support ELF_HWCAP for PPPC

Signed-off-by: Nathan Froyd <froydnj@codesourcery.com>
Signed-off-by: malc <av1474@comtv.ru>

linux-user: ppc signal handling

Implement setup_{,rt_}frame and do_{,rt_}sigreturn for PPC 32-bit.  Use
the same TARGET_QEMU_ESIGRETURN hack as for MIPS to avoid clobbering
register state on a sigreturn.

Signed-off-by: Nathan Froyd <froydnj@codesourcery.com>
Signed-off-by: malc <av1474@comtv.ru>

target-ppc: expose cpu capability flags

Do this so other pieces of code can make decisions based on the
capabilities of the CPU we're emulating.

Signed-off-by: Nathan Froyd <froydnj@codesourcery.com>
Signed-off-by: malc <av1474@comtv.ru>

ETRAX: Correct passing of kernel command line.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

fix ARMv7 data processing instructions

ARMv7 defines a new behavior for ARM data processing instructions
compared to earlier architecture revisions; when the destination
register is R15, a Branch and Exchange operation is executed rather
than a simple Branch to the target address. This patch corrects the
behavior of the emulation for the aforementioned operations. To be
applied after applying the previous patch in this patch set.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Paul Brook <paul@codesourcery.com>

fix ARMv7 data processing instructions

Modernize parts of target-arm/translate.c in preparation for the
modifications in the subsequent patch in this patch set. This is done
in order to avoid writing new code to target-arm/translate.c that
would use deprecated methods and/or variables.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Paul Brook <paul@codesourcery.com>

Fix module initialization when more than 1 class is in use

Now that we're using enums for module types, it makes no sense to bother keeping
a list of module types when we know exactly how many they are.

Switching to an array simplifies the code and eliminates the aforementioned bug.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Syborg (Symbian Virtual Platform) board

A virtual reference platform for SymbianOS development/debugging.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Fix warning on WIN32

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Fix build on Solaris and WIN32

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

PXA SSI qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

Stellaris SSI qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

SSP bus framework

Signed-off-by: Paul Brook <paul@codesourcery.com>

Stellaris I2C qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

I2C cleanup

Signed-off-by: Paul Brook <paul@codesourcery.com>

Remove bogus omap i2c slave code

Signed-off-by: Paul Brook <paul@codesourcery.com>

SMBus qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

TOSC DAC i2c qdev voncersion

Signed-off-by: Paul Brook <paul@codesourcery.com>

TWL92230 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

WM8750 qdev coversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

TMP105 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

PXA2xx I2C slave qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

LM832x qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

MAX7310 I2C qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

SSD0303 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

I2C qdev support

Signed-off-by: Paul Brook <paul@codesourcery.com>

MusicPal qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

ARM GIC qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

ARM PCI host qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

Virtio-console conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

Virtio-balloon qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

Virtio-blk qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

PL080 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

ARM RealView sytem controller qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

Versatile/PB core qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

PL190 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

ESP SCSI qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

smc91c111 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

ARM timers qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

Integrator/CP core qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

PL181 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

PL031 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

Stellaris ethernet qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

Virtio-net qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

PCI network qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

LSI SCSI qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

PL050 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>

PL011 qdev conversion

Signed-off-by: Paul Brook <paul@codesourcery.com>