Two years and a little over a month elapsed between the pkgsrc release
and the latest version 2.2.1 (5 May 2011). There were hundreds of
changes, but the vast majority were bug fixes and improvements on
existing functionality.
The only major addition was implementation of IDE.
See changelog in DOC/ChangeLog for details.
* Update configure script to force all platforms to use same installation
path as NetBSD
* The PLIST was built with fc-cache detected. This implicates fontconfig
as a dependency (without it, font still install but with a diff. name)
* like NetBSD, define HAVE_STDINT_H (other platforms may need it too)
Revbump required because package would have installed before without
PKG_DEVELOPER=yes, albeit incorrectly.
Version 2.1a is not compatible with zlib 1.26+. Luckily this problem
was fixed in version 2.2. Three patches were removed, and a few were
tweaked.
src/linux.c and src/win.c now have dos line endings. I used the
technique seen in textproc/FlightCrew to convert these to unix
endings after extraction. That should allow this package to build
on all platforms where zip handling of line ending differs.
Also defined GPLv3 license.
Several files needed the <cstring> include for functions like memset.
Additional, Tinderbox caught this error:
=== Checking filesystem state after all packages deleted
================================================================
list of extra files and directories in / (not present on clean system
but present after everything was deinstalled)
34217227226 0 lrwxr-xr-x 1 root wheel 65 Aug 12 00:13
usr/local/bin/openmsx ->
/work/emulators/openmsx/work/.destdir/usr/pkg/openmsx/bin/openmsx
The package was installing a symbolic link from /usr/local/bin to the
destdir! Luckily this feature is switchable with a variable. Fixed.
Now works with NetBSD 4, 5 & 6 which would have previously failed.
Comment out some unusual sector subtractions which handled values in the
multigigabyte range.
Now allows all tested NetBSD versions (3, 4, 5 & 6) to mount ISO images
Changelog:
# System emulation
## All targets
* qdev properties of type hex8 and hex32 used to accept hexadecimal values not prefixed with "0x"; the prefix is mandatory starting with this version. These properties are: iobase, membase, io_base for all devices that support the properties; vram_size for sysbus-g364 and SUNW,tcx; version for sb16 and iommu; ctl_iobase and data_iobase for fw_cfg; readback for debugcon; elcr_addr and elcr_mask for i8259.
* PCI addresses can still be addressed with a pair of hexadecimal device and function without a "0x" prefix.
* -kernel, -initrd and -append are now aliases for suboptions of -machine (for example -machine kernel=foo), and as such they are also available with -readconfig.
* PCI-to-PCI bridges are supported [...]
* PCI emulation includes a standard hot-plug controller [...]
* The coroutine backend can now be configured using configure --with-coroutine=.... A new coroutine backend sigaltstack is available for platforms that don't support the default ucontext backend.
* Last but not least: the SDL user interface now uses the new QEMU icon.
## ARM
* The syborg machine type has been removed since the Symbian Virtual Platform is no longer relevant with the disbanding of Symbian.
* A device tree can be passed to the kernel using -dtb option (or alternatively -machine dtb=...).
* New 'nuri' and 'smdkc210' models of Samsung Exynos4210 based devboards.
* New 'highbank' model of the Calxeda Highbank.
* New 'vexpress-a15' model of the Versatile Express Cortex-A15.
## PowerPC
* The Bamboo machine now works with TCG, before only KVM was able to execute 440 code.
* Pseries handles PCI, allowing for virtio devices with -M pseries.
* Pseries works with PR KVM allowing for -M pseries -enable-kvm to work on G5s.
* We can now emulate e500mc cores, but no e500mc based board is emulated yet. You need to use -M mpc8544ds and a special guest kernel.
* Timers on ppc405 now work and don't segfault QEMU.
## S/390
* Added support to -kernel to allow booting of ELF binaries.
* Fixed -kernel to allow booting of newer guest kernels.
* Devices can now be hotplug add'ed (remove is not there yet).
* Rebooting a virtual machine now works.
## SPARC
* Added interrupt controller and support of vector interrupts.
* Fixed handling 32 bit instructions on a sparc64 CPU.
* Fixed block load instructions regression happened in 1.0 release.
* Fixed BCD mask in m48t59 emulation, so that Solaris 2.5.1 doesn't hang on sun4m emulation when day of month >21.
* Fixed initrd loading in qemu-system-sparc64.
## x86
* NMIs are correctly injected through the LAPIC (rather than sent directly to the guest CPU) and respect the LINT1 mask, thus fixing kdump.
* The -rtc-td-hack option is now available as a qdev property (-global mc146818rtc.lost_tick_policy=slew) and as such can be specified in a -readconfig configuration file.
* When the system is suspended to RAM, QEMU will now stop executing the guest until a wakeup event occurs. Implemented wakeup events include key presses, mouse button presses, RTC alarm, ACPI timer expiration, serial console input (off by default; enable with -global isa-serial.wakeup=1), and the system_wakeup monitor command.
* KVM can optionally use kernel-based emulation of the local APIC, IOAPIC, i8259 PIC and i8254 PIT. This is supported with the -machine ...,kernel_irqchip=on command-line option. MSI is not supported yet when using the kernel_irqchip option.
* Emulation of a PC System Flash device. The feature is enabled by adding a pflash drive. This feature is not supported when KVM is enabled. See Features/PC System Flash for more information.
* KVM guests support Hyper-V enlightenments. These are enabled by the -cpu features hv_spinlocks, hv_relaxed and hv_vapic.
## Device emulation
* The virtio-scsi device is now supported by QEMU. The guest driver will be supported in Linux starting at version 3.4.
* QEMU now includes experimental support for USB 3.0 (xHCI).
* Various improvement on the floppy emulation, most notably media change has been fixed
## Audio devices
* Audio devices can now use volume control capabilities exposed by the PulseAudio and Spice backends. (Note: all audio devices do not expose their volume control unless --enable-mixemu is turned on).
## Block devices
* QEMU's NBD client implementation and the qemu-nbd server both fully support asynchronous I/O.
* The rbd block driver supports the discard operation now
* The vpc block driver supports the Fixed Disk subformat of VHD images now
* The new QMP command transaction can be used to perform a set of multiple snapshots atomically, rolling back to the original images if there is a problem along the way. See Features/SnapshotsMultipleDevices for more information.
* The new monitor command block_stream lets guest copy data from the backing file to the current image while the guest is running. This lets users enables quick provisioning of new virtual machines using shared remote storage, and lets the guest transition incrementally to fast local storage. An alternative to streaming is the copy-on-read option of the -drive command-line option, which only transfers data when the guest needs it.
* An experimental extension of the qcow2 file format has been introduced. With the implementation as of QEMU 1.1, the most important addition is zero clusters, which allows image streaming and copy-on-read to leave images sparse if the backing file they are copying from is sparse. In order to enable the extension, use qemu-img create -f qcow2 -o compat=1.1 <filename> <size> for creating the image. Note that older versions of QEMU won't be able to read such images.
* I/O throttling is experimentally supported using the new -drive options bps/bps_rd/bps_wr/iops/iops_rd/iops_wr. It is expected to work with virtio-blk and IDE harddisks, but may lead to hangs when used with CD-ROM or floppy emulation or other devices.
* qemu-io supports new options to enable tracing and to choose a cache modes
## Network devices
* QEMU supports a new type of network device, bridge (used with -net bridge or -netdev bridge). The new type is similar to tap, but uses a helper program instead of a script to attach the device to a bridge. The helper program can then be installed as setuid. The helper program supports a simple ACL and configuration mechanism, see the commit message and feature page for documentation.
## Live Migration, Save/Restore
* Live migration (or save/restore) from QEMU releases prior to 0.13 to QEMU 1.1 is not supported.
* Live migration supports IPv6. IPv6 addresses can be expressed as tcp:host:port. Brackets around a numeric host address are required if a port is also specified, otherwise they can be omitted.
# Guest agent
* qemu-ga has been ported to Windows.
* Interfaces added for suspending guests to disk/ram, and retrieving information about network interfaces
* Interfaces for filesystem freeze have been hardened, and no longer depend on guest agent runtime state to determine whether a system is currently frozen.
* An easier to use reset mechanism has been added: guest-sync-delimited.
# Host support
* ARM hosts are now supported again (they were broken in 1.0).
* Sockets and SLIRP on Windows hosts was broken in 0.14 and works again.
* 64-bit Windows hosts are now supported.
# User-mode emulation
* User-mode emulation can provide some information from the /proc filesystem.
* On 64-bit hosts user-mode emulation now defaults to reserving 0xf7000000 bytes of address space for the guest. This significantly reduces the likelihood of QEMU having to fail a guest mmap() request when there is still memory available. This reservation can be overridden with the -R command line option.
# Testing
* A make check target has been implemented and runs some quick sanity tests
* qemu-iotests, which was previously hosted in an external git repository, has been merged into the QEMU source tree
# Build dependencies
* Building QEMU requires glib 2.12 on POSIX systems, and glib 2.20 on Win32 systems.
This package starting failing this week. It was due to an change in the
system binutils which turned on link time optimization support. This
was intended for external compilers.
The package picked up that the linker was capable of supporting -flto, but
then just assumed the compiler also was! "4.4.7" was added to the list of
compilers to exclude. Earlier versions are not necessary because the
-flto switch would be suppressed anyway due to binutils not supporting it.
The current package won't compile if using libz >= 1.2.6 due to a typedef
change of GzFile. Luckily this packages used macros for gz* functions
which significantly cut down the number of lines to be patched.
As previously predicted, the horrible code in archive.c wouldn't last long
before breaking. There were a few dozen errors erupting when libz 1.2.7
header was used. The previous patch-af was appended, but the previous fix
for linux was changed because that was broken too.
Info: typedef of gzFile changed from void* at version 1.2.6, so its use
is no longer optional with libz versions >= 1.2.6.
The current package won't compile if using libz >= 1.2.6 due to a typedef
change of GzFile. Fix inspired by r1091 at vbam.svn.sourceforge.net
Package broke after DragonFly upgraded system libz from 1.2.5 to 1.2.7
* based on wip/qemu
Changelog:
1.1.0
System emulation
All targets
qdev properties of type hex8 and hex32 used to accept hexadecimal values not prefixed with "0x"; the prefix is mandatory starting with this version. These properties are: iobase, membase, io_base for all devices that support the properties; vram_size for sysbus-g364 and SUNW,tcx; version for sb16 and iommu; ctl_iobase and data_iobase for fw_cfg; readback for debugcon; elcr_addr and elcr_mask for i8259.
PCI addresses can still be addressed with a pair of hexadecimal device and function without a "0x" prefix.
-kernel, -initrd and -append are now aliases for suboptions of -machine (for example -machine kernel=foo), and as such they are also available with -readconfig.
PCI-to-PCI bridges are supported [...]
PCI emulation includes a standard hot-plug controller [...]
The coroutine backend can now be configured using configure --with-coroutine=.... A new coroutine backend sigaltstack is available for platforms that don't support the default ucontext backend.
Last but not least: the SDL user interface now uses the new QEMU icon.
ARM
The syborg machine type has been removed since the Symbian Virtual Platform is no longer relevant with the disbanding of Symbian.
A device tree can be passed to the kernel using -dtb option (or alternatively -machine dtb=...).
New 'nuri' and 'smdkc210' models of Samsung Exynos4210 based devboards.
New 'highbank' model of the Calxeda Highbank.
New 'vexpress-a15' model of the Versatile Express Cortex-A15.
PowerPC
The Bamboo machine now works with TCG, before only KVM was able to execute 440 code.
Pseries handles PCI, allowing for virtio devices with -M pseries.
Pseries works with PR KVM allowing for -M pseries -enable-kvm to work on G5s.
We can now emulate e500mc cores, but no e500mc based board is emulated yet. You need to use -M mpc8544ds and a special guest kernel.
Timers on ppc405 now work and don't segfault QEMU.
S/390
Added support to -kernel to allow booting of ELF binaries.
Fixed -kernel to allow booting of newer guest kernels.
Devices can now be hotplug add'ed (remove is not there yet).
Rebooting a virtual machine now works.
SPARC
Added interrupt controller and support of vector interrupts.
Fixed handling 32 bit instructions on a sparc64 CPU.
Fixed block load instructions regression happened in 1.0 release.
Fixed BCD mask in m48t59 emulation, so that Solaris 2.5.1 doesn't hang on sun4m emulation when day of month >21.
Fixed initrd loading in qemu-system-sparc64.
x86
NMIs are correctly injected through the LAPIC (rather than sent directly to the guest CPU) and respect the LINT1 mask, thus fixing kdump.
The -rtc-td-hack option is now available as a qdev property (-global mc146818rtc.lost_tick_policy=slew) and as such can be specified in a -readconfig configuration file.
When the system is suspended to RAM, QEMU will now stop executing the guest until a wakeup event occurs. Implemented wakeup events include key presses, mouse button presses, RTC alarm, ACPI timer expiration, serial console input (off by default; enable with -global isa-serial.wakeup=1), and the system_wakeup monitor command.
KVM can optionally use kernel-based emulation of the local APIC, IOAPIC, i8259 PIC and i8254 PIT. This is supported with the -machine ...,kernel_irqchip=on command-line option. MSI is not supported yet when using the kernel_irqchip option.
Emulation of a PC System Flash device. The feature is enabled by adding a pflash drive. This feature is not supported when KVM is enabled. See Features/PC System Flash for more information.
KVM guests support Hyper-V enlightenments. These are enabled by the -cpu features hv_spinlocks, hv_relaxed and hv_vapic.
Device emulation
The virtio-scsi device is now supported by QEMU. The guest driver will be supported in Linux starting at version 3.4.
QEMU now includes experimental support for USB 3.0 (xHCI).
Various improvement on the floppy emulation, most notably media change has been fixed
Audio devices
Audio devices can now use volume control capabilities exposed by the PulseAudio and Spice backends. (Note: all audio devices do not expose their volume control unless --enable-mixemu is turned on).
Block devices
QEMU's NBD client implementation and the qemu-nbd server both fully support asynchronous I/O.
The rbd block driver supports the discard operation now
The vpc block driver supports the Fixed Disk subformat of VHD images now
The new QMP command transaction can be used to perform a set of multiple snapshots atomically, rolling back to the original images if there is a problem along the way. See Features/SnapshotsMultipleDevices for more information.
The new monitor command block_stream lets guest copy data from the backing file to the current image while the guest is running. This lets users enables quick provisioning of new virtual machines using shared remote storage, and lets the guest transition incrementally to fast local storage. An alternative to streaming is the copy-on-read option of the -drive command-line option, which only transfers data when the guest needs it.
An experimental extension of the qcow2 file format has been introduced. With the implementation as of QEMU 1.1, the most important addition is zero clusters, which allows image streaming and copy-on-read to leave images sparse if the backing file they are copying from is sparse. In order to enable the extension, use qemu-img create -f qcow2 -o compat=1.1 <filename> <size> for creating the image. Note that older versions of QEMU won't be able to read such images.
I/O throttling is experimentally supported using the new -drive options bps/bps_rd/bps_wr/iops/iops_rd/iops_wr. It is expected to work with virtio-blk and IDE harddisks, but may lead to hangs when used with CD-ROM or floppy emulation or other devices.
qemu-io supports new options to enable tracing and to choose a cache modes
Network devices
QEMU supports a new type of network device, bridge (used with -net bridge or -netdev bridge). The new type is similar to tap, but uses a helper program instead of a script to attach the device to a bridge. The helper program can then be installed as setuid. The helper program supports a simple ACL and configuration mechanism, see the commit message and feature page for documentation.
Live Migration, Save/Restore
Live migration (or save/restore) from QEMU releases prior to 0.13 to QEMU 1.1 is not supported.
Live migration supports IPv6. IPv6 addresses can be expressed as tcp:host:port. Brackets around a numeric host address are required if a port is also specified, otherwise they can be omitted.
Guest agent
qemu-ga has been ported to Windows.
Interfaces added for suspending guests to disk/ram, and retrieving information about network interfaces
Interfaces for filesystem freeze have been hardened, and no longer depend on guest agent runtime state to determine whether a system is currently frozen.
An easier to use reset mechanism has been added: guest-sync-delimited.
Host support
ARM hosts are now supported again (they were broken in 1.0).
Sockets and SLIRP on Windows hosts was broken in 0.14 and works again.
64-bit Windows hosts are now supported.
User-mode emulation
User-mode emulation can provide some information from the /proc filesystem.
On 64-bit hosts user-mode emulation now defaults to reserving 0xf7000000 bytes of address space for the guest. This significantly reduces the likelihood of QEMU having to fail a guest mmap() request when there is still memory available. This reservation can be overridden with the -R command line option.
Testing
A make check target has been implemented and runs some quick sanity tests
qemu-iotests, which was previously hosted in an external git repository, has been merged into the QEMU source tree
Build dependencies
Building QEMU requires glib 2.12 on POSIX systems, and glib 2.20 on Win32 systems.
Changelog:
1.0.1
* Version 1.0.1
* Merge branch 's390-1.0' of git://repo.or.cz/qemu/agraf
* Merge branch 'ppc-1.0' of git://repo.or.cz/qemu/agraf
* e1000: bounds packet size against buffer size
* s390: fix cpu hotplug / cpu activity on interrupts
* s390x: add TR function for EXECUTE
* pseries: Don't try to munmap() a malloc()ed TCE table
* pseries: Populate "/chosen/linux,stdout-path" in the FDT
* pseries: Add a routine to find a stable "default" vty and use it
* pseries: Emit device tree nodes in reg order
* kvm-ppc: halt secondary cpus when guest reset
* pseries: Fix array overrun bug in PCI code
* console: Fix segfault on screendump without VGA adapter
* Merge branch 'master' of ssh://git.qemu.org/pub/git/qemu-stable-1.0
* rbd: always set out parameter in qemu_rbd_snap_list
* Documentation: Add qemu-img -t parameter in man page
* qemu-img rebase: Fix for undersized backing files
* coroutine: switch per-thread free pool to a global pool
* qiov: prevent double free or use-after-free
* PPC: Fix linker scripts on ppc hosts
* target-sh4: ignore ocbp and ocbwb instructions
* usb-ohci: td.cbp incorrectly updated near page end
* usb-host: properly release port on unplug & exit
* usb-storage: cancel I/O on reset
* Fix parse of usb device description with multiple configurations
* pc: fix event_idx compatibility for virtio devices
* pc: add pc-0.15
* cris: Handle conditional stores on CRISv10
* configure: Enable build by default PIE / read-only relocation sections on OpenBSD amd64/i386.
* target-i386: fix cmpxchg instruction emulation
* hw/9pfs: Use the correct signed type for different variables
* hw/9pfs: replace iovec manipulation with QEMUIOVector
* hw/9pfs: Use the correct file descriptor in Fsdriver Callback
* hw/9pfs: Add qdev.reset callback for virtio-9p-pci device
* hw/9pfs: Reset server state during TVERSION
* hw/9pfs: use migration blockers to prevent live migration when virtfs export path is mounted
* hw/9pfs: Improve portability to older systems
* exec.c: Fix subpage memory access to RAM MemoryRegion
* malta: Fix regression (i8259 interrupts did not work)
General
i386-softmmu is no longer named qemu but instead referred to as qemu-system-i386 for better consistency with other targets. A new tool is likely to be introduced that uses the qemu name so distributions are advised to not undo this change.
QEMU now uses a separate thread for VCPU execution. This merges the biggest difference between the qemu-kvm tree and upstream QEMU.
A new memory dispatch API has been added internally. A new monitor command "info mtree" can show the hierarchy of memory regions in the guest.
QEMU now has a build dependency on glib and makes extensive use of glib.
QEMU now can run on more hosts. Hosts without a native code generator can use the TCG interpreter (TCI). See Features/TCI for more information.
Block devices (disks)
QEMU now supports I/O latency accounting in the monitor command "info blockstats".
Errors are now tracked per device and are shown by the monitor command "info block".
All image formats now support asynchronous operation. IDE and SCSI emulation will use this feature, while other devices (notably floppy and SD) will not.
IDE/ATAPI
A large number of bugs were fixed regarding CD media change and tray locking.
SCSI
Memory management errors could crash QEMU when scsi-disk encountered I/O errors. Many instances of this problem were fixed.
The accuracy of error handling for SCSI emulation has been greatly improved.
SCSI devices can now be addressed by channel, target (id) and LUN. Not all emulated HBAs will support this feature (in particular, the LSI controller will not).
Block device pass through is now supported through a new scsi-block device. The scsi-block device works with block devices (like /dev/sda or /dev/sr0) rather than /dev/sgN devices, and is more efficient because it does not consume arbitrary amounts of memory when the guest does large data transfers.
SCSI CD-ROMs now report media changed events.
SCSI CD-ROMs now support DVD images.
Bugfixes for IDE media change also apply to SCSI.
SCSI devices now report a unit attention condition when the system is started or reset. This may cause problems with old firmware versions.
VDI
Now supports discarded blocks in dynamically-sized images.
User-mode networking (SLIRP)
SLIRP can process ARP replies and gratuitous ARP requests from the guest.
ARM
QEMU now supports the new Cortex-A15 instructions in linux-user mode (via "-cpu any"): VFPv4 fused multiply-accumulate (VFMA, VFMS, VFNMA, VFNMS) and also integer division (UDIV, SDIV).
The vexpress-a9, versatileab, versatilepb and realview-* boards now have audio support.
QEMU is known not to work on ARM hosts in this release. (ARM target emulation is fine.)
pSeries
sPAPR VIO devices can now be created with -device.
Xtensa
QEMU now supports DC232b and FSF xtensa CPU cores.
QEMU now supports sim (similar to Tensilica ISS) and LX60/LX110/LX200 machines.
Migration
QEMU now supports live migration using image files like QCOW2 on shared storage
The post-extract target on this package was supposed to allow normal users
to build this package, but it didn't work for me. It may have been the
attempt at changing $WRKSRC instead of $WRKSRC/src (recursively).
The updated command touches only directories rather than all files and
now handy_sdl builds for me with a non-root user.
* Update glibc rpm to 2.14.1-14.17.1, previous one is not in repo.
* Change /etc/mtab symlink to relative so that linux binary can read it.
Bump PKGREVISION.
disable the Nagle algorithm in the host-side connection. Either
the VM is already doing Nagle, in which case there is no point
in doing it twice, or it has chosen to disable it, in which case
we should respect that choice. This change speeds up GDB remote
debugging over TCP over user mode networking by multiple orders
of magnitude.
* libXp was used by Xaw8, but it had been obsolated, and in pkgsrc,
x11/libXaw/buildlink3.mk had been switched to pick up Xaw7 by default.
* With x11/xorg-cf-files, libXp was offered with XawClientLibs,
but updated to 1.0.4, it was removed.
* And pkgsrc had been switched to use always xorg-cf-files and imake from pkgsrc,
so all platforms should not require libXp from libXaw with Imake.
Bump PKGREVISION.
v1.03:
pecompat: New app
uips: Fixed a bug that caused bytes near the end of the file to be unnecessarily included in the .ips patch
cdpatch: Combining -f and -v now lists all errors, while still ignoring them
cdpatch: -x option now preserves modification time when possible
* Add patch for configure.ac, for making potential ethernet support easier.
Changelog:
2012/03/23 - version 0.9.13 released
A couple of important bugfixes accumulated in last few days:
o USB NatFeat doesn't fail to claim devices in certain cases
o MPFR FPU got fix for restarting of instructions after access error
o JIT CPU version should work even when compiled with newer compilers
o update your mmusetup.cnf in MiNT setup for larger FastRAM (see
provided atari/mmusetup.cnf that is set for 512 MB FastRAM now)
2012/03/19 - version 0.9.12 released
New features:
o MPFR based new precise FPU emulation core. Somewhat slower than the
normally used IEEE or UAE FPU emulation cores (about 5 times)
but should be totally complete and precise. Enabled in Linux binary
MMU builds (aranym-mmu). Hopefully Debian-m68k developers will enjoy it
:-)
o USB NatFeat fully usable: http://wiki.aranym.org/manual#usb
With a FreeMiNT driver you can access your USB flash discs from
the Atari side easily. First enter the SETUP GUI and plug in
the desired USB device. The USB support is compiled in by default
in all linux binary packages.
o NF Clipboard implemented newly for BeOS
o "aranym --version" will now list the built-in FPU core and also all
NatFeats that are compiled in.
Bugfixes:
o memleaks in various files
o blitter memmove fix
o MacOSX project fixes
o small 64bit fixes in nfcdrom
o aliasing issue fixes (in nfosmesa that is not compiled in)
o CPU CAS+CAS2 instructions - several very important fixes!
o some little fixes for newer GCCs in various files
o RPM spec file updated for Fedora, openSUSE and Mandriva. Give
rpmbuild -tb aranym_0.9.12.orig.tar.gz a try!
2011/02/06 - version 0.9.11 released
New features: SCC serial port emulation and USB NatFeat support.
BeOS/Haiku port: fixes and improvements.
MMU emulation: fix for initialization and FPU registers.
Video: fix (refresh palette when switching from true color to bitplane
mode) and an improvement (use current fullscreen size if fixed size is 0x0).
