ide: add VMSTATE_IDE_BUS and VMSTATE_IDE_DRIVES

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

ide: microdrive cycle field is set as uint8_t

Really, it is used as a boolean
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

ide: port ide_drive to vmstate

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

vmstate: add VMSTATE_BUFFER_TEST

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

ide: change identify_data type to uint8_t

It is used everywhere as uint8_t except in one place.  Cast to uint16_t
in that place
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

ide: port idebus to vmstate

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

mc146818rtc: remove rtc_mm_init()

It was used for Acer Pica 61 emulation, removed in 2008
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

mc146818rtc: fix indentation

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

mc146818rtc: port rtc to vmstate

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

mc145818rtc: fix saving of rtc-td hack properly upgrading the version number

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

usb-uhci: port to vmstate

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

usb-uhci: Add num_ports_vmstate to UHCIState

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Add chardev option to disable signal.

If I am using vga and serial which is stdio and hit C-c on
serial console, qemu terminates. That is annoying for me.
So make it configurable whether signal is generated when C-c is hit.

Signed-off-by: Kusanagi Kouichi <slash@ac.auone-net.jp>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

hotplug: more fixes for scsi disk hotplug.

Changes:
 - create common scsi_hot_add function for adding a scsi disk to
   a adapter.
 - Add sanity checks.  You can't drive_add disks to the VGA any more.
 - Ignore the unit value calculated by drive_init, add a comment
   explaining why.  Review the patch to find the details ;)

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

hotplug: fix "pci_add storage if=scsi"

Explicitly add the drive to the bus of the newly created scsi adapter
instead of hoping that scsi_bus_legacy_handle_cmdline() picks it up
correctly.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

isa: configure serial+parallel by index.

This patch adds a 'index' property to the isa-parallel and isa-serial
devices.  This can be used to create devices with the default isa irqs
and ioports by simply specifying the index, i.e.

   -device isa-serial,index=1

instead of

   -device isa-serial,iobase=0x2f8,irq=3

for ttyS1 aka com2.  Likewise for parallel ports.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

pc.c: only load e1000 rom.

The other pxe roms are loaded by the drivers individual drivers now.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

kill dead nic unplug code.

Cleanup on unplug happens via qdev->exit() callback now.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

zap DeviceState->nd

No users left.
Also cleanup obsolete helper functions.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

musicpal: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

stellaris_enet: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

xilinx_ethlite: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

smc91c111: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

eepro100: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

virtio: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

rtl8139: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

pcnet: split away lance.c (sparc32 code).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

pcnet: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

e1000: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

ne2k_pci: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

prepare pci nic init path for qdev property configuration.

Initialization path will work with both converted and not-converted
drivers, so we can convert drivers one by one.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qdev: add qdev_prop_exists()

Function test whenever a driver has a specific property.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

ne2k_isa: use qdev properties for configuration.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qdev/net: common nic property bits

Add a new type for properties common to all nics.
Add helper functions and macros to deal with it.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qdev: add vlan property

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qdev: add netdev property

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qdev: mac addr property fixups

Make the mac property use the newly added type for the mac address.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

net: add macaddr type.

Add new type for mac addresses.

Add function which sets the qemu default mac address if it finds the mac
address uninitialized (i.e. all zeros).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Send a RARP packet after migration.

Currently, after a migration qemu sends a broadcast packet to update
switches' MAC->port mappings.

Unfortunately, it picks a random (constant) ethertype and crosses its
fingers that no one else is using it.

This patch causes it to send a RARP packet instead.  RARP was chosen for
2 reasons.  One, it is always harmless, and will continue to be so even
as new ethertypes are allocated.  Two, it is what VMware ESX sends, so
people who write filtering rules for switches already know about it.

I also changed the code to send SELF_ANNOUNCE_ROUNDS packets, instead of
SELF_ANNOUNCE_ROUNDS + 1, and added a simple backoff scheme.

Signed-off-by: Nolan Leake <nolan <at> sigbus.net>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

multiboot.S patch for old as(1) (was: Re: [Qemu-devel] Some OpenBSD/amd64 build fixes)

On Wed, Sep 23, 2009 at 10:55:02PM +0200, Juergen Lock wrote:
> On Wed, Sep 23, 2009 at 09:31:16PM +0100, Laurence Tratt wrote:
>[...]
> >  then the above error doesn't occur, but
> > pc-bios/optionrom/multiboot.S dies as follows:
> >
> >   $
> >     AS    optionrom/multiboot.o
> >   multiboot.S: Assembler messages:
> >   multiboot.S:116: Error: `%es:-4(%edi)' is not a valid 16 bit base/index
> > expression
> >   $
> >
> > What little Intel assembler I ever knew has long since departed from my
> > brain, so I don't know why that error occurs, nor what a fix might be.
> >
>  It occurs because of too old binutils (as(1) in this case), on FreeBSD
> we now have a port for newer ones,
>  http://www.freshports.org/devel/binutils
> so I depend on that and have the optionrom Makefile use the new as
> like this: (the first change wrt CFLAGS is unrelated and has probably
> been fixed in the meantime; it caused gmake to complain about
> recursive use of CFLAGS.)
>
> Index: qemu/pc-bios/optionrom/Makefile
> @@ -9,10 +9,13 @@
>
>  CFLAGS = -Wall -Wstrict-prototypes -Werror -fomit-frame-pointer -fno-builtin
>  CFLAGS += -I$(SRC_PATH)
> -CFLAGS += $(call cc-option, $(CFLAGS), -fno-stack-protector,"")
> +CFLAGS := $(CFLAGS) $(call cc-option, $(CFLAGS), -fno-stack-protector,"")
>
>  build-all: multiboot.bin
>
> +%.o: %.S
> + $(CC) -E $(CFLAGS) -o - -c $< |${LOCALBASE}/bin/as -V -Qy -o $@
> +
>  %.img: %.o
>   $(call quiet-command,$(LD) -Ttext 0 -e _start -s -o $@ $<,"  Building $(TARGET_DIR)$@")
>

That patch didn't seem to help on OpenBSD so I now finally got around
making another one that just emits the bytes of the offending insn
instead so people can keep using old assemblers:

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

raw/linux-aio: Also initialize POSIX AIO

When using Linux AIO raw still falls back to POSIX AIO sometimes, so we should
initialize it.

Not initializing it happens to work if POSIX AIO is used by another drive, or
if the format is not specified (probing the format uses POSIX AIO) or by pure
luck (e.g. it doesn't seem to happen any more with qcow2 since we have re-added
synchronous qcow2 functions).

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qcow2: Fix grow_refcount_table error handling

In case of failure, we haven't increased the refcount for the newly allocated
cluster yet. Therefore we must not free the cluster or its refcount will become
negative (and endless recursion is possible).

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

usb-linux: return USB_RET_STALL on -EPIPE

0) This is an attempt to get an issue in usb-linux.c, for which a patch
was posted about a year ago, finally fixed.

1) Mark Burkley submitted a "EHCI emulation module" for review in in
October 2008 (see:
http://lists.gnu.org/archive/html/qemu-devel/2008-10/msg01326.html). No
EHCI emulation module was ever committed to qemu.

2) Part of that (large) patch was a fix for a separate issue in
usb-linux.c. Max Krasnyansky has ACK'ed that fix (see:
http://lists.gnu.org/archive/html/qemu-devel/2008-11/msg00032.html).

3) I already asked whether this fix was ready to be committed in last
April (see:
http://lists.gnu.org/archive/html/qemu-devel/2009-04/msg01763.html)

4) Maybe submitting this fix as a separate patch (with a really long
commit message but without a Signed-off-by) and cc-ing everbody involved
will help if actually getting this issue fixed.

Paul Bolle

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

Makefile: Change make to be quiet again when doing nothing

This patch makes make quiet again.

There is already a similar patch from Juan Quintela,
but maybe this shorter form is preferred.

Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

eepro100: Restructure code

This patch only moves about 150 lines of code from
function eepro100_cu_command to a new function action_command.

A goto statement was replaced by a for loop.

There are no functional changes. Nor did I change comments
starting with // (they will be removed by future patches).

Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

target-arm: use native tcg-ops for ror/bic/vorn

Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: fix neon shift helper functions

Current code is broken at least on recent compilers, comparison
between signed and unsigned types yield incorrect code and render
the neon shift helper functions defunct. This is the third revision
of this patch, casting all comparisons with the sizeof operator to
signed ssize_t type to force comparisons to be between signed integral
types.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: fix neon vsri, vshl and vsli ops

Shift by immediate value is incorrectly overwritten by a temporary
variable in the processing of NEON vsri, vshl and vsli instructions.
This patch has been revised to also include a fix for the special
case where the code would previously try to shift an integer value
over 31 bits left/right.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: allow modifying vfp fpexc en bit only

All other bits except for the EN in the VFP FPEXC register are defined
as subarchitecture specific and real functionality for any of the
other bits has not been implemented in QEMU. However, current code
allows modifying all bits in the VFP FPEXC register leading to
problems when guest code is writing 1's to the subarchitecture
specific bits and checking whether the bits stay up to verify the
existence of functionality which in fact does not exist in QEMU.
This patch has been revised to include the same behavior change in
the gdb register write function.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: add support for neon vld1.64/vst1.64 instructions

Add support for NEON vld1.64 and vst1.64 instructions. This patch is
revised to follow more closely the specification and raises
undefined exception if 64bit element size is used for vld2/vst2 or
vld4/vst4 instructions.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: fix neon vshrn/vrshrn ops

In the existing code shift value is clobbered during the pass loop.
This patch changes the code so that it stores the intermediate
result in the target neon register directly and eliminates the need
to use a temporary to hold the intermediate value thus leaving the
shift value in the temporary variable intact. This is a new patch
in this version of the patch series.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: fix incorrect temporary variable freeing

tmp4 and tmp5 temporary variables are allocated using tcg_const_i32
but incorrectly released using dead_tmp which will cause resource
leak tracking to report false leaks.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

sparc32: tcx: remove unused include directive

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

sparc32: convert sbi to VMState, vmsd and vmstate reset

Also remove unused include directive.

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

sparc32: convert Sun4c interrupt controller to reset + vmsd

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

sparc32: convert interrupt controller to reset + vmsd

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

sparc32: convert IOMMU to reset + vmsd

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

sparc32: convert DMA controller to reset + vmsd, fix reset on init

Add a missing call to reset on device init.

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

m48t59: convert to vmstate reset

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

sparc32: convert slavio_timer to reset + vmsd

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

fdc: convert to reset + vmsd

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

esp: convert to reset + vmsd

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

sparc32: convert cs4231 to VMState, vmsd and reset

Also remove unused include directive.

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

escc: add chipset docs

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

escc: convert to VMState, vmsd and reset

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

sparc32: add chipset docs for eccmemctl

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

sparc32: convert slavio_misc to reset + vmsd

Also remove unused include directive.

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

sparc32: convert eccmemctl to reset + vmsd

Also remove unused include directive.

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

sparc64: remove unused variables

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

target-arm: optimize thumb 32-bit multiply

Current implementation of thumb mul instruction is implemented as a
32x32->64 multiply which then uses only 32 least significant bits of
the result. Replace that with a simple 32x32->32 multiply.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: cleanup internal resource leaks

Revised patch for getting rid of tcg temporary variable leaks in
target-arm/translate.c. This version also includes the leak patch for
gen_set_cpsr macro, now converted as a static inline function, which I
sent earlier as a separate patch on top of this patch.

Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-i386: implement lzcnt emulation

lzcnt is a AMD Phenom/Barcelona added instruction returning the
number of leading zero bits in a word.
As this is similar to the "bsr" instruction, reuse the existing
code. There need to be some more changes, though, as lzcnt always
returns a valid value (in opposite to bsr, which has a special
case when the operand is 0).
lzcnt is guarded by the ABM CPUID bit (Fn8000_0001:ECX_5).

Signed-off-by: Andre Przywara <andre.przywara@amd.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-ppc: move often used CPU fields at the top of the structure

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: fix sdiv helper

(INT32_MIN / -1) triggers an overflow, and the result depends on the
host architecture (INT32_MIN on arm, -1 on ppc, SIGFPE on x86). Use a
test to output the correct value.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>

target-arm: use clz32() instead of a for loop

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>

Merge commit 'linux-user/linux-user-for-upstream' into staging

qdev: add string property.

Patchworks-ID: 35755
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

char: emit the OPENED event only when a new char connection is opened

The OPENED event gets sent also when qemu resets its state initially.
The consumers of the event aren't interested in receiving this event
on reset.

Patchworks-ID: 35288
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

char: rename CHR_EVENT_RESET to CHR_EVENT_OPENED

The char event RESET is emitted when a char device is opened.
Give it a better name.

Patchworks-ID: 35287
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

char: check for initial_reset_issued unnecessary

At init, qemu_chr_reset is always called with initial_reset_issued set to 1.
So checking for it to be set is not necessary.

Patchworks-ID: 35286
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

bsd-user: fix breakage by 78cfb07fe0dc556cae662a0fab5fe1bd33daabdb

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

Suppress warnings about 'warn_unused_result' attribute directive

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

bsd-user: FreeBSD update

basic FreeBSD sysarch(2) handling
fixed syscall errno return

Signed-off-by: Juergen Lock <nox@jelal.kn-bremen.de>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

x86: add 'static' to please Sparse

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

target-ppc: simpler definitions for microcontrollers based on e300

No need to alias e300 core for each CPU package.
Differences between microcontrollers have to be implemented in a higher layer
than translate_init.c

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-ppc: add declarations of microcontrollers based on e300

Add CPU declarations of MPC8343, MPC8343E, MPC8347 and MPC8347E.

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-ppc: better support of e300 CPU core

Declare HID2 register.

Use high BATs for e300 (8 instead of 4).

Fix index of high BATs registers.
Before the fix, IBAT4-7 were overwriting IBAT0-3.

Signed-off-by: François Armand <francois.armand@os4i.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: fix bugs introduced by 1b2b1e547bd912b7d3c4863d0a0f75f6f38330ed

Use load_reg_var() instead of accessing cpu_R[rn] directly to generate
correct code when rn = 15.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: fix bugs introduced by 3174f8e91fecf8756e861d1febb049f3c619a2c7

Use load_reg_var() instead of accessing cpu_R[rn] directly to generate
correct code when rn = 15.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: remove T0 and T1

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: remove cpu_T for ARM once and for all

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: convert VFP not to use cpu_T

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: convert disas_iwmmxt_insn not to use cpu_T

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: convert disas_dsp_insn not use cpu_T

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: convert disas_neon_ls_insn not to use cpu_T

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: convert disas_neon_data_insn and helpers not to use cpu_T

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: convert rest of disas_arm_insn / disas_thumb2_insn not to use cpu_T

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: replace thumb usage of cpu_T registers by proper register allocations

The goal is eventually to get rid of all cpu_T register usage and to use
just short-lived tmp/tmp2 registers. This patch converts all the places where
cpu_T was used in the Thumb code and replaces it with explicit TCG register
allocation.

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: fix TANDC and TORC instructions

Uninitialized register was used instead of proper TCG variable.

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-arm: convert NEON VZIP/VUZP/VTRN helper functions to pure TCG

The neon_trn_u8, neon_trn_u16, neon_unzip_u8, neon_zip_u8 and neon_zip_u16
helpers used fixed registers to return values. This patch replaces that with
TCG code, so T0/T1 is no longer directly used by the helper functions.

Bugs in the gen_neon_unzip register load code were also fixed.

Signed-off-by: Filip Navara <filip.navara@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>