OSDN Git Service

qmiga/qemu.git
10 years agotarget-arm: use c13_context field for CONTEXTIDR
Sergey Fedorov [Fri, 20 Dec 2013 06:33:11 +0000 (10:33 +0400)]
target-arm: use c13_context field for CONTEXTIDR

Use c13_context field instead of c13_fcse for CONTEXTIDR register
definition.

Signed-off-by: Sergey Fedorov <s.fedorov@samsung.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1387521191-15350-1-git-send-email-s.fedorov@samsung.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Implement Tx flow control
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:40 +0000 (10:16 +0000)]
char/cadence_uart: Implement Tx flow control

If the UART back-end blocks, buffer in the Tx FIFO to try again later.
This stops the IO-thread busy waiting on char back-ends (which causes
all sorts of performance problems).

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 4bea048b3ab38425701d82ccc1ab92545c26b79c.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Delete redundant rx rst logic
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:40 +0000 (10:16 +0000)]
char/cadence_uart: Delete redundant rx rst logic

uart_rx_reset() called immediately above already does this. Remove.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 05e30826496cf2579084ed801ac0b2c0d0a3071f.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Use the TX fifo for transmission
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:40 +0000 (10:16 +0000)]
char/cadence_uart: Use the TX fifo for transmission

Populate the TxFIFO with the Tx data before sending. Prepares
support for proper Tx flow control implementation.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: bdf7f8af2ef02839bea18665701bc2612f7baa6f.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Fix can_receive logic
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:39 +0000 (10:16 +0000)]
char/cadence_uart: Fix can_receive logic

The can_receive logic was only taking into account the RxFIFO
occupancy. RxFIFO population is only used for the echo and normal modes
however. Improve the logic to correctly return the true number of
receivable characters based on the current mode:

Normal mode: RxFIFO vacancy.
Remote loopback: TxFIFO vacancy.
Echo mode: The min of the TxFIFO and RxFIFO vacancies.
Local Loopback: Return non-zero (to implement droppage)

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 36a58440c9ca5080151e95765c2c81342de8a8df.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Remove TX timer & add TX FIFO state
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:39 +0000 (10:16 +0000)]
char/cadence_uart: Remove TX timer & add TX FIFO state

This tx timer implementation is flawed. Despite the controller
attempting to time the guest visable assertion of the TX-empty status
bit (and corresponding interrupt) the controller is still transmitting
characters instantaneously. There is also no sense of multiple character
delay.

The only side effect of this timer is assertion of tx-empty status. So
just remove the timer completely and hold tx-empty as permanently
asserted (its reset status). This matches the actual behaviour of
instantaneous transmission.

While we are VMSD version bumping, add the tx_fifo as device state to
prepare for upcomming TxFIFO flow control. Implement the interrupt
generation logic for the TxFIFO occupancy.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 7a208a7eb8d79d6429fe28b1396c3104371807b2.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Define Missing SR/ISR fields
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:39 +0000 (10:16 +0000)]
char/cadence_uart: Define Missing SR/ISR fields

Some (interrupt) status register bits relating to the TxFIFO path were
not defined. Define them. This prepares support for proper Tx data path
flow control.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 2068b963f0af8cc834c353944e9fa816d950b163.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Simplify status generation
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:38 +0000 (10:16 +0000)]
char/cadence_uart: Simplify status generation

The status register bits are always pure functions of other device
state. Move the generation of these bits to the update_status()
function to simplify. Makes developing much easier as theres now no need
to recheck status bits on all the changes to rx/tx fifo state.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 321994929f789096975104f99c55732774be4cae.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: s/r_fifo/rx_fifo
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:38 +0000 (10:16 +0000)]
char/cadence_uart: s/r_fifo/rx_fifo

Rename this field to match the many other uses of "rx". Xilinx
docmentation (UG585) also refers to this as "RxFIFO".

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 7386d7cee0ea175f7e53ed5ff045265528d34e32.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Fix reset.
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:38 +0000 (10:16 +0000)]
char/cadence_uart: Fix reset.

Don't reset the uart as an init step. Register the reset function as a
proper reset fn instead.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: d82cd2e65e5a6f8b6deeecb6cced61f0bf3f8c89.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Add missing uart_update_state
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:37 +0000 (10:16 +0000)]
char/cadence_uart: Add missing uart_update_state

This should be rechecked on bus write accesses as such accesses may
change the underlying state that generates the interrupt. Particular
relevant for when the guest touches the interrupt status or mask.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 1c250cd61b7b8de492fbc8b79b8370958a56d83b.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agochar/cadence_uart: Mark struct fields as public/private
Peter Crosthwaite [Mon, 6 Jan 2014 10:16:37 +0000 (10:16 +0000)]
char/cadence_uart: Mark struct fields as public/private

As per current QOM conventions.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: a1e31bd62e9709ffb9b3efc6c120f83f30b7a660.1388626249.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Give the FPSCR rounding modes names
Alexander Graf [Sat, 4 Jan 2014 22:15:51 +0000 (22:15 +0000)]
target-arm: Give the FPSCR rounding modes names

When setting rounding modes we currently just hardcode the numeric values
for rounding modes in a big switch statement.

With AArch64 support coming, we will need to refer to these rounding modes
at different places throughout the code though, so let's better give them
names so we don't get confused by accident.

Signed-off-by: Alexander Graf <agraf@suse.de>
[WN: Commit message tweak, use names from ARM ARM.]
Signed-off-by: Will Newton <will.newton@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add support for floating point cond select
Claudio Fontana [Sat, 4 Jan 2014 22:15:51 +0000 (22:15 +0000)]
target-arm: A64: Add support for floating point cond select

This adds decoding support for C3.6.24 FP conditional select.

Signed-off-by: Claudio Fontana <claudio.fontana@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add support for floating point conditional compare
Claudio Fontana [Sat, 4 Jan 2014 22:15:51 +0000 (22:15 +0000)]
target-arm: A64: Add support for floating point conditional compare

This adds decoding support for C3.6.23 FP Conditional Compare.

Signed-off-by: Claudio Fontana <claudio.fontana@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add support for floating point compare
Claudio Fontana [Sat, 4 Jan 2014 22:15:50 +0000 (22:15 +0000)]
target-arm: A64: Add support for floating point compare

Add decoding support for C3.6.22 Floating-point compare.

Signed-off-by: Claudio Fontana <claudio.fontana@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add fmov (scalar, immediate) instruction
Alexander Graf [Sat, 4 Jan 2014 22:15:50 +0000 (22:15 +0000)]
target-arm: A64: Add fmov (scalar, immediate) instruction

This patch adds emulation for the fmov instruction working on scalars
with an immediate payload.

Signed-off-by: Alexander Graf <agraf@suse.de>
[WN: Commit message tweak, rebase and use new infrastructure.]
Signed-off-by: Will Newton <will.newton@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add "Floating-point data-processing (3 source)" insns
Alexander Graf [Sat, 4 Jan 2014 22:15:50 +0000 (22:15 +0000)]
target-arm: A64: Add "Floating-point data-processing (3 source)" insns

This patch adds emulation for the "Floating-point data-processing (3 source)"
group of instructions.

Signed-off-by: Alexander Graf <agraf@suse.de>
[WN: Commit message tweak, merged single and double precision patches.
 Implement using muladd as suggested by Richard Henderson.]
Signed-off-by: Will Newton <will.newton@linaro.org>
[PMM: pull field decode up a level, use register accessors]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add "Floating-point data-processing (2 source)" insns
Alexander Graf [Sat, 4 Jan 2014 22:15:50 +0000 (22:15 +0000)]
target-arm: A64: Add "Floating-point data-processing (2 source)" insns

This patch adds emulation for the "Floating-point data-processing (2 source)"
group of instructions.

Signed-off-by: Alexander Graf <agraf@suse.de>
[WN: Commit message tweak, merge single and double precision patches. Rebase
 and update to new infrastructure. Incorporate FMIN/FMAX support patch by
 Michael Matz.]
Signed-off-by: Will Newton <will.newton@linaro.org>
[PMM:
 * added convenience accessors for FP s and d regs
 * pulled the field decode and opcode validity check up a level]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: Use VFP_BINOP macro for min, max, minnum, maxnum
Peter Maydell [Sat, 4 Jan 2014 22:15:49 +0000 (22:15 +0000)]
target-arm: Use VFP_BINOP macro for min, max, minnum, maxnum

Use the VFP_BINOP macro to provide helpers for min, max, minnum
and maxnum, rather than hand-rolling them. (The float64 max
version is not used by A32 but will be needed for A64.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Fix vector register access on bigendian hosts
Peter Maydell [Sat, 4 Jan 2014 22:15:49 +0000 (22:15 +0000)]
target-arm: A64: Fix vector register access on bigendian hosts

The A64 128 bit vector registers are stored as a pair of
uint64_t values in the register array. This means that if
we're directly loading or storing a value of size less than
64 bits we must adjust the offset appropriately to account
for whether the host is bigendian or not. Provide utility
functions to abstract away the offsetof() calculations for
the FP registers.

For do_fp_st() we can sidestep most of the issues for 64 bit
and smaller reg-to-mem transfers by always doing a 64 bit
load from the register and writing just the piece we need
to memory.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add support for dumping AArch64 VFP register state
Alexander Graf [Sat, 4 Jan 2014 22:15:49 +0000 (22:15 +0000)]
target-arm: A64: Add support for dumping AArch64 VFP register state

When dumping the current CPU state, we can also get a request
to dump the FPU state along with the CPU's integer state.

Add support to dump the VFP state when that flag is set, so that
we can properly debug code that modifies floating point registers.

Signed-off-by: Alexander Graf <agraf@suse.de>
[WN: Commit message tweak, rebased. Output all registers, two per-line.]
Signed-off-by: Will Newton <will.newton@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agodefault-configs: Add config for aarch64-linux-user
Peter Maydell [Sat, 4 Jan 2014 22:15:48 +0000 (22:15 +0000)]
default-configs: Add config for aarch64-linux-user

Add a config for aarch64-linux-user, thereby enabling it as
a valid target.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years ago.travis.yml: Add aarch64-* targets
Alex Bennée [Sat, 4 Jan 2014 22:15:48 +0000 (22:15 +0000)]
.travis.yml: Add aarch64-* targets

Now the AArch64 targets are in mainline we can include them in our
Travis test matrix.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agolinux-user: AArch64: Use correct values for FPSR/FPCR in sigcontext
Will Newton [Sat, 4 Jan 2014 22:15:48 +0000 (22:15 +0000)]
linux-user: AArch64: Use correct values for FPSR/FPCR in sigcontext

Use the helpers provided for getting the correct FPSR and FPCR
values for the signal context.

Signed-off-by: Will Newton <will.newton@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agolinux-user: AArch64: define TARGET_CLONE_BACKWARDS
Claudio Fontana [Sat, 4 Jan 2014 22:15:47 +0000 (22:15 +0000)]
linux-user: AArch64: define TARGET_CLONE_BACKWARDS

The AArch64 linux-user support was written before but merged after
commit 4ce6243dc621 which cleaned up the handling of the clone()
syscall argument order, so we failed to notice that AArch64 also needs
TARGET_CLONE_BACKWARDS to be defined. Add this define so that clone
and fork syscalls work correctly.

Signed-off-by: Claudio Fontana <claudio.fontana@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: support for ld/st/cl exclusive
Michael Matz [Sat, 4 Jan 2014 22:15:47 +0000 (22:15 +0000)]
target-arm: A64: support for ld/st/cl exclusive

This implement exclusive loads/stores for aarch64 along the lines of
arm32 and ppc implementations. The exclusive load remembers the address
and loaded value. The exclusive store throws an an exception which uses
those values to check for equality in a proper exclusive region.

This is not actually the architecture mandated semantics (for either
AArch32 or AArch64) but it is close enough for typical guest code
sequences to work correctly, and saves us from having to monitor all
guest stores. It's fairly easy to come up with test cases where we
don't behave like hardware - we don't for example model cache line
behaviour. However in the common patterns this works, and the existing
32 bit ARM exclusive access implementation has the same limitations.

AArch64 also implements new acquire/release loads/stores (which may be
either exclusive or non-exclusive). These imposes extra ordering
constraints on memory operations (ie they act as if they have an implicit
barrier built into them). As TCG is single-threaded all our barriers
are no-ops, so these just behave like normal loads and stores.

Signed-off-by: Michael Matz <matz@suse.de>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: Widen exclusive-access support struct fields to 64 bits
Peter Maydell [Sat, 4 Jan 2014 22:15:47 +0000 (22:15 +0000)]
target-arm: Widen exclusive-access support struct fields to 64 bits

In preparation for adding support for A64 load/store exclusive instructions,
widen the fields in the CPU state struct that deal with address and data values
for exclusives from 32 to 64 bits. Although in practice AArch64 and AArch32
exclusive accesses will be generally separate there are some odd theoretical
corner cases (eg you should be able to do the exclusive load in AArch32, take
an exception to AArch64 and successfully do the store exclusive there), and it's
also easier to reason about.

The changes in semantics for the variables are:
 exclusive_addr  -> extended to 64 bits; -1ULL for "monitor lost",
   otherwise always < 2^32 for AArch32
 exclusive_val   -> extended to 64 bits. 64 bit exclusives in AArch32 now
   use the high half of exclusive_val instead of a separate exclusive_high
 exclusive_high  -> is no longer used in AArch32; extended to 64 bits as
   it will be needed for AArch64's pair-of-64-bit-values exclusives.
 exclusive_test  -> extended to 64 bits, as it is an address. Since this is
   a linux-user-only field, in arm-linux-user it will always have the top
   32 bits zero.
 exclusive_info  -> stays 32 bits, as it is neither data nor address, but
   simply holds register indexes etc. AArch64 will be able to fit all its
   information into 32 bits as well.

Note that the refactoring of gen_store_exclusive() coincidentally fixes
a minor bug where ldrexd would incorrectly update the first CPU register
even if the load for the second register faulted.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: aarch64: add support for ld lit
Alexander Graf [Sat, 4 Jan 2014 22:15:46 +0000 (22:15 +0000)]
target-arm: aarch64: add support for ld lit

Adds support for Load Register (literal), both normal
and SIMD/FP forms.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for conditional compare insns
Claudio Fontana [Sat, 4 Jan 2014 22:15:46 +0000 (22:15 +0000)]
target-arm: A64: add support for conditional compare insns

this patch adds support for C3.5.4 - C3.5.5
Conditional compare (both immediate and register)

Signed-off-by: Claudio Fontana <claudio.fontana@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for add/sub with carry
Claudio Fontana [Sat, 4 Jan 2014 22:15:46 +0000 (22:15 +0000)]
target-arm: A64: add support for add/sub with carry

This patch adds support for C3.5.3 Add/subtract (with carry):
instructions ADC, ADCS, SBC, SBCS.

Signed-off-by: Claudio Fontana <claudio.fontana@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: Widen thread-local register state fields to 64 bits
Peter Maydell [Sat, 4 Jan 2014 22:15:45 +0000 (22:15 +0000)]
target-arm: Widen thread-local register state fields to 64 bits

The common pattern for system registers in a 64-bit capable ARM
CPU is that when in AArch32 the cp15 register is a view of the
bottom 32 bits of the 64-bit AArch64 system register; writes in
AArch32 leave the top half unchanged. The most natural way to
model this is to have the state field in the CPU struct be a
64 bit value, and simply have the AArch32 TCG code operate on
a pointer to its lower half.

For aarch64-linux-user the only registers we need to share like
this are the thread-local-storage ones. Widen their fields to
64 bits and provide the 64 bit reginfo struct to make them
visible in AArch64 state. Note that minor cleanup of the AArch64
system register encoding space means We can share the TPIDR_EL1
reginfo but need split encodings for TPIDR_EL0 and TPIDRRO_EL0.

Since we're touching almost every line in QEMU that uses the
c13_tls* fields in this patch anyway, we take the opportunity
to rename them in line with the standard ARM architectural names
for these registers.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Implement minimal set of EL0-visible sysregs
Peter Maydell [Sat, 4 Jan 2014 22:15:45 +0000 (22:15 +0000)]
target-arm: A64: Implement minimal set of EL0-visible sysregs

Implement an initial minimal set of EL0-visible system registers:
 * NZCV
 * FPCR
 * FPSR
 * CTR_EL0
 * DCZID_EL0

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
10 years agotarget-arm: A64: Implement MRS/MSR/SYS/SYSL
Peter Maydell [Sat, 4 Jan 2014 22:15:45 +0000 (22:15 +0000)]
target-arm: A64: Implement MRS/MSR/SYS/SYSL

The AArch64 equivalent of the traditional AArch32
cp15 coprocessor registers is the set of instructions
MRS/MSR/SYS/SYSL, which cover between them both true
system registers and the "operations with side effects"
such as cache maintenance which in AArch32 are mixed
in with other cp15 registers. Implement these instructions
to look in the cpregs hashtable for the register or
operation.

Since we don't yet populate the cpregs hashtable with
any registers with the "AA64" bit set, everything will
still UNDEF at this point.

MSR/MRS is the first user of is_jmp = DISAS_UPDATE, so
fix an infelicity in its handling where the main loop
was requiring the caller to do the update of PC rather
than just doing it itself.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: Remove ARMCPU/CPUARMState from cpregs APIs used by decoder
Peter Maydell [Sat, 4 Jan 2014 22:15:44 +0000 (22:15 +0000)]
target-arm: Remove ARMCPU/CPUARMState from cpregs APIs used by decoder

The cpregs APIs used by the decoder (get_arm_cp_reginfo() and
cp_access_ok()) currently take either a CPUARMState* or an ARMCPU*.
This is problematic for the A64 decoder, which doesn't pass the
environment pointer around everywhere the way the 32 bit decoder
does. Adjust the parameters these functions take so that we can
copy only the relevant info from the CPUARMState into the
DisasContext and then use that.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: Update generic cpreg code for AArch64
Peter Maydell [Sat, 4 Jan 2014 22:15:44 +0000 (22:15 +0000)]
target-arm: Update generic cpreg code for AArch64

Update the generic cpreg support code to also handle AArch64:
AArch64-visible registers coexist in the same hash table with
AArch32-visible ones, with a bit in the hash key distinguishing
them.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
10 years agotarget-arm: Pull "add one cpreg to hashtable" into its own function
Peter Maydell [Sun, 22 Dec 2013 22:32:30 +0000 (22:32 +0000)]
target-arm: Pull "add one cpreg to hashtable" into its own function

define_one_arm_cp_reg_with_opaque() has a set of nested loops which
insert a cpreg entry into the hashtable for each of the possible
opc/crn/crm values allowed by wildcard specifications. We're about
to add an extra loop to this nesting, so pull the core of the loop
(which adds a single entry to the hashtable) out into its own
function for clarity.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: implement FMOV
Peter Maydell [Mon, 23 Dec 2013 23:27:31 +0000 (23:27 +0000)]
target-arm: A64: implement FMOV

Implement FMOV, ie non-converting moves between general purpose
registers and floating point registers. This is a subtype of
the floating point <-> integer instruction class.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add decoder skeleton for FP instructions
Peter Maydell [Mon, 23 Dec 2013 23:27:30 +0000 (23:27 +0000)]
target-arm: A64: Add decoder skeleton for FP instructions

Add a top level decoder skeleton for FP instructions.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: implement SVC, BRK
Alexander Graf [Mon, 23 Dec 2013 23:27:30 +0000 (23:27 +0000)]
target-arm: A64: implement SVC, BRK

Add decoding for the exception generating instructions, and implement
SVC (syscalls) and BRK (software breakpoint).

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for 3 src data proc insns
Alexander Graf [Mon, 23 Dec 2013 23:27:30 +0000 (23:27 +0000)]
target-arm: A64: add support for 3 src data proc insns

This patch adds emulation for the "Data-processing (3 source)"
family of instructions, namely MADD, MSUB, SMADDL, SMSUBL, SMULH,
UMADDL, UMSUBL, UMULH.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for move wide instructions
Alex Bennée [Mon, 23 Dec 2013 23:27:29 +0000 (23:27 +0000)]
target-arm: A64: add support for move wide instructions

This patch adds emulation for the mov wide instructions
(MOVN, MOVZ, MOVK).

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for add, addi, sub, subi
Alex Bennée [Mon, 23 Dec 2013 23:27:29 +0000 (23:27 +0000)]
target-arm: A64: add support for add, addi, sub, subi

Implement the non-carry forms of addition and subtraction
(immediate, extended register and shifted register).
This includes the code to calculate NZCV if the instruction
calls for setting the flags.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for ld/st with index
Alex Bennée [Mon, 23 Dec 2013 23:27:29 +0000 (23:27 +0000)]
target-arm: A64: add support for ld/st with index

This adds support for the pre/post-index ld/st forms with immediate
offsets as well as the un-scaled immediate form (which are all
variations on the same 9-bit immediate instruction form).

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for ld/st with reg offset
Alex Bennée [Mon, 23 Dec 2013 23:27:29 +0000 (23:27 +0000)]
target-arm: A64: add support for ld/st with reg offset

This adds support for the load/store forms using a register offset.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for ld/st unsigned imm
Alex Bennée [Mon, 23 Dec 2013 23:27:28 +0000 (23:27 +0000)]
target-arm: A64: add support for ld/st unsigned imm

This adds support for the forms of ld/st with a 12 bit
unsigned immediate offset.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: add support for ld/st pair
Peter Maydell [Sun, 22 Dec 2013 22:32:27 +0000 (22:32 +0000)]
target-arm: A64: add support for ld/st pair

This patch support the basic load and store pair instructions and
includes the generic helper functions:

  * do_gpr_st()
  * do_fp_st()
  * do_gpr_ld()
  * do_fp_ld()
  * read_cpu_reg_sp()
  * gen_check_sp_alignment()

The last function gen_check_sp_alignment() is a NULL op currently but
put in place to make it easy to add SP alignment checking later.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agoPPC: Fix compilation with TCG debug
Alexander Graf [Fri, 20 Dec 2013 10:01:50 +0000 (11:01 +0100)]
PPC: Fix compilation with TCG debug

The recent VSX patches broken compilation of QEMU when configurated
with --enable-debug, as it was treating "target long" TCG variables
as "i64" which is not true for 32bit targets.

This patch fixes all the places that the compiler has found to use
the correct variable type and if necessary manually cast.

Reported-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
10 years agoMerge tag 'signed-s390-for-upstream' of git://github.com/agraf/qemu
Aurelien Jarno [Sat, 21 Dec 2013 15:46:07 +0000 (16:46 +0100)]
Merge tag 'signed-s390-for-upstream' of git://github.com/agraf/qemu

Patch queue for s390 - 2013-12-18

This covers mostly minor bug fixes and implements the SIGP START
hypercall which allows to start a remote CPU without changing its
state.

Cornelia Huck (1):
      s390x/kvm: Fix diagnose handling.

Thomas Huth (7):
      s390x/kvm: Removed duplicated SIGP defines
      s390x/kvm: Removed s390_store_status stub
      s390x/kvm: Fix coding style in handle_sigp()
      s390x/kvm: Implemented SIGP START
      s390x/kvm: Simplified the calculation of the SIGP order code
      s390x/kvm: Fixed condition code for unknown SIGP orders
      s390x/ioinst: CHSC has to set a condition code

* tag 'signed-s390-for-upstream' of git://github.com/agraf/qemu:
  s390x/ioinst: CHSC has to set a condition code
  s390x/kvm: Fixed condition code for unknown SIGP orders
  s390x/kvm: Simplified the calculation of the SIGP order code
  s390x/kvm: Implemented SIGP START
  s390x/kvm: Fix coding style in handle_sigp()
  s390x/kvm: Removed s390_store_status stub
  s390x/kvm: Removed duplicated SIGP defines
  s390x/kvm: Fix diagnose handling.

10 years agotarget-sh4: Use new qemu_ld/st opcodes
Aurelien Jarno [Wed, 11 Dec 2013 06:56:47 +0000 (07:56 +0100)]
target-sh4: Use new qemu_ld/st opcodes

Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
10 years agotarget-mips: Use new qemu_ld/st opcodes
Aurelien Jarno [Wed, 11 Dec 2013 07:35:27 +0000 (08:35 +0100)]
target-mips: Use new qemu_ld/st opcodes

Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
10 years agotcg/i386: fix a comment
Aurelien Jarno [Tue, 10 Dec 2013 09:35:28 +0000 (10:35 +0100)]
tcg/i386: fix a comment

The comments apply to 8-bit stores, not 8-byte stores.

Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
10 years agoMerge remote-tracking branch 'agraf/tags/signed-ppc-for-upstream' into staging
Anthony Liguori [Fri, 20 Dec 2013 01:03:17 +0000 (17:03 -0800)]
Merge remote-tracking branch 'agraf/tags/signed-ppc-for-upstream' into staging

Patch queue for ppc - 2013-12-20

Alexander Graf (3):
      PPC: Use default pci bus name for grackle and heathrow
      roms: Flush icache when writing roms to guest memory
      PPC: Add VSX to hflags

Alexey Kardashevskiy (5):
      powerpc: add PVR mask support
      target-ppc: move POWER7+ to a separate family
      spapr-rtas: replace return code constants with macros
      spapr-rtas: add ibm, (get|set)-system-parameter
      spapr: make sure RMA is in first mode of first memory node

Greg Kurz (1):
      target-ppc: add stubs for KVM breakpoints

Paolo Bonzini (1):
      spapr: tie spapr-nvram to -pflash

Paul Mackerras (1):
      spapr: limit numa memory regions by ram size

Peter Crosthwaite (2):
      device_tree: s/qemu_devtree/qemu_fdt globally
      device_tree: qemu_fdt_setprop: Rename val_array arg

Tom Musta (19):
      Declare and Enable VSX
      Add MSR VSX and Associated Exception
      Add VSX Instruction Decoders
      Add VSR to Global Registers
      Add lxvd2x
      Add stxvd2x
      Add xxpermdi
      Add lxsdx
      Add lxvdsx
      Add lxvw4x
      Add stxsdx
      Add stxvw4x
      Add VSX Scalar Move Instructions
      Add VSX Vector Move Instructions
      Add Power7 VSX Logical Instructions
      Add xxmrgh/xxmrgl
      Add xxsel
      Add xxspltw
      Add xxsldwi

* agraf/tags/signed-ppc-for-upstream: (32 commits)
  spapr: limit numa memory regions by ram size
  spapr: make sure RMA is in first mode of first memory node
  device_tree: qemu_fdt_setprop: Rename val_array arg
  device_tree: s/qemu_devtree/qemu_fdt globally
  PPC: Add VSX to hflags
  Add xxsldwi
  Add xxspltw
  Add xxsel
  Add xxmrgh/xxmrgl
  Add Power7 VSX Logical Instructions
  Add VSX Vector Move Instructions
  Add VSX Scalar Move Instructions
  roms: Flush icache when writing roms to guest memory
  spapr: tie spapr-nvram to -pflash
  PPC: Use default pci bus name for grackle and heathrow
  spapr-rtas: add ibm, (get|set)-system-parameter
  spapr-rtas: replace return code constants with macros
  target-ppc: move POWER7+ to a separate family
  Add stxvw4x
  Add stxsdx
  ...

10 years agospapr: limit numa memory regions by ram size
Paul Mackerras [Mon, 25 Nov 2013 03:14:51 +0000 (14:14 +1100)]
spapr: limit numa memory regions by ram size

This makes sure that all NUMA memory blocks reside within RAM or
have zero length.

Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agospapr: make sure RMA is in first mode of first memory node
Alexey Kardashevskiy [Mon, 25 Nov 2013 03:14:50 +0000 (14:14 +1100)]
spapr: make sure RMA is in first mode of first memory node

The SPAPR specification says that the RMA starts at the LPAR's logical
address 0 and is the first logical memory block reported in
the LPAR’s device tree.

So SLOF only maps the first block and that block needs to span
the full RMA.

This makes sure that the RMA area is where SLOF expects it.

Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agodevice_tree: qemu_fdt_setprop: Rename val_array arg
Peter Crosthwaite [Mon, 11 Nov 2013 08:15:21 +0000 (18:15 +1000)]
device_tree: qemu_fdt_setprop: Rename val_array arg

Looking at the implementation, this doesn't really have a lot to do
with arrays. Its just a pointer to a buffer and is passed through
to the wrapped fn (qemu_fdt_setprop) unchanged. So rename to make it
consistent with libfdt, which in the wrapped function just calls it
"val".

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agodevice_tree: s/qemu_devtree/qemu_fdt globally
Peter Crosthwaite [Mon, 11 Nov 2013 08:14:41 +0000 (18:14 +1000)]
device_tree: s/qemu_devtree/qemu_fdt globally

The qemu_devtree API is a wrapper around the fdt_ set of APIs.
Rename accordingly.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
[agraf: also convert hw/arm/virt.c]
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoPPC: Add VSX to hflags
Alexander Graf [Wed, 18 Dec 2013 08:21:02 +0000 (09:21 +0100)]
PPC: Add VSX to hflags

We generate different code depending on whether MSR_VSX is set or
clear, so it needs to be part of our hflags too which indicate whether
we're still in the same translation block cache bucket.

Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd xxsldwi
Tom Musta [Fri, 1 Nov 2013 13:21:23 +0000 (08:21 -0500)]
Add xxsldwi

This patch adds the VSX Shift Left Double by Word Immediate
(xxsldwi) instruction.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd xxspltw
Tom Musta [Fri, 1 Nov 2013 13:21:22 +0000 (08:21 -0500)]
Add xxspltw

This patch adds the VSX Splat Word (xxsplatw) instruction.

This is the first instruction to use the UIM immediate field
and consequently a decoder is also added.

V2: reworked implementation per Richard Henderson's comments.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd xxsel
Tom Musta [Fri, 1 Nov 2013 13:21:21 +0000 (08:21 -0500)]
Add xxsel

This patch adds the VSX Select (xxsel) instruction.

The xxsel instruction has four VSR operands.  Thus the xC
instruction decoder is added.

The xxsel instruction is massively overloaded in the opcode
table since only bits 26 and 27 are opcode bits.  This
overloading is done in matrix fashion with two macros
(GEN_XXSEL_ROW and GEN_XX_SEL).

V2: (1) eliminated unecessary XXSEL macro  (2) tighter implementation
using tcg_gen_andc_i64.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd xxmrgh/xxmrgl
Tom Musta [Fri, 1 Nov 2013 13:21:20 +0000 (08:21 -0500)]
Add xxmrgh/xxmrgl

This patch adds the VSX Merge High Word and VSX Merge Low Word
instructions.

V2: Now implemented using deposit (per Richard Henderson's comment)

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd Power7 VSX Logical Instructions
Tom Musta [Fri, 1 Nov 2013 13:21:19 +0000 (08:21 -0500)]
Add Power7 VSX Logical Instructions

This patch adds the VSX logical instructions that are defined
by the Version 2.06 Power ISA (aka Power7):

   - xxland
   - xxlandc
   - xxlor
   - xxlxor
   - xxlnor

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd VSX Vector Move Instructions
Tom Musta [Fri, 1 Nov 2013 13:21:18 +0000 (08:21 -0500)]
Add VSX Vector Move Instructions

This patch adds the vector move instructions:

  - xvabsdp - Vector Absolute Value Double-Precision
  - xvnabsdp - Vector Negative Absolute Value Double-Precision
  - xvnegdp - Vector Negate Double-Precision
  - xvcpsgndp - Vector Copy Sign Double-Precision
  - xvabssp - Vector Absolute Value Single-Precision
  - xvnabssp - Vector Negative Absolute Value Single-Precision
  - xvnegsp - Vector Negate Single-Precision
  - xvcpsgnsp - Vector Copy Sign Single-Precision

V3: Per Paolo Bonzini's suggestion, used a temporary for the
sign mask and andc.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd VSX Scalar Move Instructions
Tom Musta [Fri, 1 Nov 2013 13:21:17 +0000 (08:21 -0500)]
Add VSX Scalar Move Instructions

This patch adds the VSX scalar move instructions:

  - xsabsdp (Scalar Absolute Value Double-Precision)
  - xsnabspd (Scalar Negative Absolute Value Double-Precision)
  - xsnegdp (Scalar Negate Double-Precision)
  - xscpsgndp (Scalar Copy Sign Double-Precision)

A common generator macro (VSX_SCALAR_MOVE) is added since these
instructions vary only slightly from each other.

Macros to support VSX XX2 and XX3 form opcodes are also added.
These macros handle the overloading of "opcode 2" space (instruction
bits 26:30) caused by AX and BX bits (29 and 30, respectively).

V3: Per feedback from Paolo Bonzini, moved the sign mask into a
temporary and used andc.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoroms: Flush icache when writing roms to guest memory
Alexander Graf [Wed, 11 Dec 2013 13:17:44 +0000 (14:17 +0100)]
roms: Flush icache when writing roms to guest memory

We use the rom infrastructure to write firmware and/or initial kernel
blobs into guest address space. So we're basically emulating the cache
off phase on very early system bootup.

That phase is usually responsible for clearing the instruction cache for
anything it writes into cachable memory, to ensure that after reboot we
don't happen to execute stale bits from the instruction cache.

So we need to invalidate the icache every time we write a rom into guest
address space. We do not need to do this for every DMA since the guest
expects it has to flush the icache manually in that case.

This fixes random reboot issues on e5500 (booke ppc) for me.

Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agospapr: tie spapr-nvram to -pflash
Paolo Bonzini [Fri, 22 Nov 2013 09:27:40 +0000 (10:27 +0100)]
spapr: tie spapr-nvram to -pflash

spapr-nvram's drive property is currently connected to a non-existent
"-machine nvram=<drivename>" option.  Instead, tie it to -pflash like
other non-volatile RAM devices.  This provides the following possibilities
for adding a backend for the sPAPR non-volatile RAM:

* -pflash filename

* -drive if=pflash,file=filename,format=raw,...

* -drive if=none,file=filename,format=raw,id=foo,... -global spapr-nvram.drive=foo

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoPPC: Use default pci bus name for grackle and heathrow
Alexander Graf [Wed, 4 Dec 2013 11:42:32 +0000 (12:42 +0100)]
PPC: Use default pci bus name for grackle and heathrow

There's no good reason to call our bus "pci" rather than let the default
bus name take over ("pci.0").

The big downside to calling it different from anyone else is that tools
that pass -device get confused. They are looking for a bus "pci.0" rather
than "pci".

To make life easier for everyone, let's just drop the name override.

Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agospapr-rtas: add ibm, (get|set)-system-parameter
Alexey Kardashevskiy [Tue, 19 Nov 2013 04:28:55 +0000 (15:28 +1100)]
spapr-rtas: add ibm, (get|set)-system-parameter

This adds very basic handlers for ibm,get-system-parameter and
ibm,set-system-parameter RTAS calls.

The only parameter handled at the moment is
"platform-processor-diagnostics-run-mode" which is always disabled and
does not support changing. This is expected to make
"ppc64_cpu --run-mode=1" happy.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
[agraf: s/papameter/parameter/g]
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agospapr-rtas: replace return code constants with macros
Alexey Kardashevskiy [Tue, 19 Nov 2013 04:28:54 +0000 (15:28 +1100)]
spapr-rtas: replace return code constants with macros

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agotarget-ppc: move POWER7+ to a separate family
Alexey Kardashevskiy [Tue, 19 Nov 2013 01:39:15 +0000 (12:39 +1100)]
target-ppc: move POWER7+ to a separate family

So far POWER7+ was a part of POWER7 family. However it has a different
PVR base value so in order to support PVR masks, it needs a separate
family class.

This adds a new family class, PVR base and mask values and moves
Power7+ v2.1 CPU to a new family. The class init function is copied
from the POWER7 family.

This defines a firmware name for the new family as "PowerPC,POWER7+"
instead of previously used "PowerPC,POWER7" from the POWER7 family.
The reason for that is that the Sapphire firmware (a h0st firmware)
uses "PowerPC,POWER7+" already and since no specification defines
exactly the CPU nodes naming in the device tree, we better stay
in sync with the host firmware.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd stxvw4x
Tom Musta [Fri, 1 Nov 2013 13:21:16 +0000 (08:21 -0500)]
Add stxvw4x

This patch adds the Store VSX Vector Word*4 Indexed (stxvw4x)
instruction.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd stxsdx
Tom Musta [Fri, 1 Nov 2013 13:21:15 +0000 (08:21 -0500)]
Add stxsdx

This patch adds the Store VSX Scalar Doubleword Indexed (stxsdx)
instruction.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd lxvw4x
Tom Musta [Fri, 1 Nov 2013 13:21:14 +0000 (08:21 -0500)]
Add lxvw4x

This patch adds the Load VSX Vector Word*4 Indexed (lxvw4x)
instruction.

V2: changed to use deposit_i64 per Richard Henderson's review.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd lxvdsx
Tom Musta [Fri, 1 Nov 2013 13:21:13 +0000 (08:21 -0500)]
Add lxvdsx

This patch adds the Load VSX Vector Doubleword & Splat Indexed
(lxvdsx) instruction.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd lxsdx
Tom Musta [Fri, 1 Nov 2013 13:21:12 +0000 (08:21 -0500)]
Add lxsdx

This patch adds the Load VSX Scalar Doubleowrd Indexed (lxsdx)
instruction.

The lower 8 bytes of the target register are undefined; this
implementation leaves those bytes unaltered.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd xxpermdi
Tom Musta [Tue, 22 Oct 2013 11:09:35 +0000 (22:09 +1100)]
Add xxpermdi

This patch adds the xxpermdi instruction.  The instruction
uses bits 22, 23, 29 and 30 for non-opcode fields (DM, AX
and BX).  This results in overloading of the opcode table
with aliases, which can be seen in the GEN_XX3FORM_DM
macro.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd stxvd2x
Tom Musta [Tue, 22 Oct 2013 11:09:00 +0000 (22:09 +1100)]
Add stxvd2x

This patch adds the stxvd2x instruction.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd lxvd2x
Tom Musta [Tue, 22 Oct 2013 11:08:32 +0000 (22:08 +1100)]
Add lxvd2x

This patch adds the lxvd2x instruction.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd VSR to Global Registers
Tom Musta [Tue, 22 Oct 2013 11:07:41 +0000 (22:07 +1100)]
Add VSR to Global Registers

This patch adds VSX VSRs to the the list of global register indices.
More specifically, it adds the lower halves of the first 32 VSRs to
the list of global register indices.  The upper halves of the first
32 VSRs are already defined via cpu_fpr[].  And the second 32 VSRs
are already defined via the cpu_avrh[] and cpu_avrl[] arrays.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd VSX Instruction Decoders
Tom Musta [Tue, 22 Oct 2013 11:06:46 +0000 (22:06 +1100)]
Add VSX Instruction Decoders

This patch adds decoders for the VSX fields XT, XS, XA, XB and
DM.  The first four are split fields and a general helper for
these types of fields is also added.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoAdd MSR VSX and Associated Exception
Tom Musta [Tue, 22 Oct 2013 11:06:17 +0000 (22:06 +1100)]
Add MSR VSX and Associated Exception

This patch adds support for the VSX bit of the PowerPC Machine
State Register (MSR) as well as the corresponding VSX Unavailable
exception.

The VSX bit is added to the defined bits masks of the Power7 and
Power8 CPU models.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoDeclare and Enable VSX
Tom Musta [Tue, 22 Oct 2013 11:05:46 +0000 (22:05 +1100)]
Declare and Enable VSX

This patch adds the flag POWERPC_FLAG_VSX to the list of defined
flags and also adds this flag to the list of supported features of
the Power7 and Power8 CPUs.  Additionally, the VSX instructions
are added to the list of TCG-enabled instruction.

Signed-off-by: Tom Musta <tommusta@gmail.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agopowerpc: add PVR mask support
Alexey Kardashevskiy [Fri, 27 Sep 2013 08:05:03 +0000 (18:05 +1000)]
powerpc: add PVR mask support

IBM POWERPC processors encode PVR as a CPU family in higher 16 bits and
a CPU version in lower 16 bits. Since there is no significant change
in behavior between versions, there is no point to add every single CPU
version in QEMU's CPU list. Also, new CPU versions of already supported
CPU won't break the existing code.

This adds PVR value/mask support for KVM, i.e. for -cpu host option.

As CPU family class name for POWER7 is "POWER7-family", there is no need
to touch aliases.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agotarget-ppc: add stubs for KVM breakpoints
Greg Kurz [Wed, 11 Dec 2013 13:15:34 +0000 (14:15 +0100)]
target-ppc: add stubs for KVM breakpoints

The latest update to v3.13-rc3 (bf63839f) breaks the
ppc build with KVM:

kvm-all.o: In function `kvm_update_guest_debug':
kvm-all.c:1910: undefined reference to `kvm_arch_update_guest_debug'
kvm-all.o: In function `kvm_insert_breakpoint':
kvm-all.c:1937: undefined reference to `kvm_arch_insert_sw_breakpoint'
kvm-all.c:1945: undefined reference to `kvm_arch_insert_hw_breakpoint'
kvm-all.o: In function `kvm_remove_breakpoint':
kvm-all.c:1977: undefined reference to `kvm_arch_remove_sw_breakpoint'
kvm-all.c:1985: undefined reference to `kvm_arch_remove_hw_breakpoint'
kvm-all.o: In function `kvm_remove_all_breakpoints':
kvm-all.c:2009: undefined reference to `kvm_arch_remove_sw_breakpoint'
kvm-all.c:2006: undefined reference to `kvm_arch_remove_sw_breakpoint'
kvm-all.c:2017: undefined reference to `kvm_arch_remove_all_hw_breakpoints'

We need stubs until something gets implemented.

Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Reviewed-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoMerge remote-tracking branch 'pmaydell/tags/pull-target-arm-20131217' into staging
Anthony Liguori [Thu, 19 Dec 2013 19:56:33 +0000 (11:56 -0800)]
Merge remote-tracking branch 'pmaydell/tags/pull-target-arm-20131217' into staging

target-arm queue:
 * AES instruction support for 32 bit ARM
 * pflash01: much better emulation of 2x16bit and similar configs
   where multiple flash devices are banked together
 * fixed CBAR handling on Zynq, Highbank
 * initial AArch64 KVM control support
 * first two chunks of patches for A64 instruction emulation
 * new board: canon-a1100 (Canon DIGIC SoC)
 * new board: cubieboard (Allwinner A10 SoC)

# gpg: Signature made Tue 17 Dec 2013 12:18:39 PM PST using RSA key ID 14360CDE
# gpg: Can't check signature: public key not found

# By Alexander Graf (14) and others
# Via Peter Maydell
* pmaydell/tags/pull-target-arm-20131217: (62 commits)
  MAINTAINERS: add myself to maintain allwinner-a10
  hw/arm: add cubieboard support
  hw/arm: add allwinner a10 SoC support
  hw/intc: add allwinner A10 interrupt controller
  hw/timer: add allwinner a10 timer
  vmstate: Add support for an array of ptimer_state *
  MAINTAINERS: Document 'Canon DIGIC' machine
  hw/arm/digic: add NOR ROM support
  hw/arm/digic: add UART support
  hw/arm/digic: add timer support
  hw/arm/digic: prepare DIGIC-based boards support
  hw/arm: add very initial support for Canon DIGIC SoC
  target-arm: A64: add support for logical (immediate) insns
  target-arm: A64: add support for 1-src CLS insn
  host-utils: add clrsb32/64 - count leading redundant sign bits
  target-arm: A64: add support for bitfield insns
  target-arm: A64: add support for 1-src REV insns
  target-arm: A64: add support for 1-src RBIT insn
  target-arm: A64: add support for 1-src data processing and CLZ
  target-arm: A64: add support for 2-src shift reg insns
  ...

Message-id: 1387312160-12318-1-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>
10 years agos390x/ioinst: CHSC has to set a condition code
Thomas Huth [Tue, 17 Dec 2013 18:50:01 +0000 (19:50 +0100)]
s390x/ioinst: CHSC has to set a condition code

I missed to set the CC in the CHSC instruction when I refactored
the CC setting in the IO instructions with the following commit:
5d9bf1c07c1369ab3506fc82cc65a10f4415d867
s390/ioinst: Moved the CC setting to the IO instruction handlers
This patch now restores the correct behaviour of CHSC by setting the
condition code 0 at the end of the instruction.

Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agos390x/kvm: Fixed condition code for unknown SIGP orders
Thomas Huth [Tue, 17 Dec 2013 13:22:09 +0000 (14:22 +0100)]
s390x/kvm: Fixed condition code for unknown SIGP orders

If SIGP is called with an unknown order code, it has to return CC1
instead of CC3 and set the "invalid order" bit in the return status.

Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agos390x/kvm: Simplified the calculation of the SIGP order code
Thomas Huth [Tue, 17 Dec 2013 13:22:08 +0000 (14:22 +0100)]
s390x/kvm: Simplified the calculation of the SIGP order code

We've already got a helper function for calculating the
base/displacement of RS formatted instructions, so we can
get rid of the manual calculation of the SIGP order code.

Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agos390x/kvm: Implemented SIGP START
Thomas Huth [Tue, 17 Dec 2013 13:22:07 +0000 (14:22 +0100)]
s390x/kvm: Implemented SIGP START

This patch adds the missing START order to the SIGP instruction handler.

Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agos390x/kvm: Fix coding style in handle_sigp()
Thomas Huth [Tue, 17 Dec 2013 13:22:06 +0000 (14:22 +0100)]
s390x/kvm: Fix coding style in handle_sigp()

To make scripts/checkpatch.pl happy for the following patches,
the coding style in handle_sigp() has to be fixed first.

Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agos390x/kvm: Removed s390_store_status stub
Thomas Huth [Tue, 17 Dec 2013 13:22:05 +0000 (14:22 +0100)]
s390x/kvm: Removed s390_store_status stub

The SIGP order STORE STATUS AT ADDRESS will be handled in
kernel space, so we do not need the stub in QEMU anymore.

Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agos390x/kvm: Removed duplicated SIGP defines
Thomas Huth [Tue, 17 Dec 2013 13:22:04 +0000 (14:22 +0100)]
s390x/kvm: Removed duplicated SIGP defines

The SIGP order defines are also available in cpu.h,
so there is no need to re-define them in kvm.c.

Signed-off-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agos390x/kvm: Fix diagnose handling.
Cornelia Huck [Tue, 17 Dec 2013 17:27:33 +0000 (18:27 +0100)]
s390x/kvm: Fix diagnose handling.

The instruction intercept handler for diagnose used only the displacement
when trying to calculate the function code. This is only correct for base
0, however; we need to perform a complete base/displacement address
calculation and use bits 48-63 as the function code.

Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
10 years agoMAINTAINERS: add myself to maintain allwinner-a10
liguang [Tue, 17 Dec 2013 19:42:38 +0000 (19:42 +0000)]
MAINTAINERS: add myself to maintain allwinner-a10

Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
Message-id: 1387159292-10436-7-git-send-email-lig.fnst@cn.fujitsu.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agohw/arm: add cubieboard support
liguang [Tue, 17 Dec 2013 19:42:38 +0000 (19:42 +0000)]
hw/arm: add cubieboard support

Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 1387159292-10436-6-git-send-email-lig.fnst@cn.fujitsu.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agohw/arm: add allwinner a10 SoC support
liguang [Tue, 17 Dec 2013 19:42:38 +0000 (19:42 +0000)]
hw/arm: add allwinner a10 SoC support

Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 1387159292-10436-5-git-send-email-lig.fnst@cn.fujitsu.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agohw/intc: add allwinner A10 interrupt controller
liguang [Tue, 17 Dec 2013 19:42:38 +0000 (19:42 +0000)]
hw/intc: add allwinner A10 interrupt controller

Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1387159292-10436-4-git-send-email-lig.fnst@cn.fujitsu.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agohw/timer: add allwinner a10 timer
liguang [Tue, 17 Dec 2013 19:42:37 +0000 (19:42 +0000)]
hw/timer: add allwinner a10 timer

Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 1387159292-10436-3-git-send-email-lig.fnst@cn.fujitsu.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agovmstate: Add support for an array of ptimer_state *
Peter Maydell [Tue, 17 Dec 2013 19:42:37 +0000 (19:42 +0000)]
vmstate: Add support for an array of ptimer_state *

Add support for defining a vmstate field which is an array
of pointers to structures, and use this to define a
VMSTATE_PTIMER_ARRAY() which allows an array of ptimer_state*
to be used by devices.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1387159292-10436-2-git-send-email-lig.fnst@cn.fujitsu.com