target-arm: Remove ad-hoc leak checking code
This commit removes the ad-hoc resource leak checking code fromtarget-arm. This includes replacing all uses of new_tmp() withtcg_temp_new_i32() and all uses of dead_tmp() withtcg_temp_free_i32().
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>...
target-arm: Implement cp15 VA->PA translation
Implement VA->PA translations by cp15-c7 that went through unchangedpreviously.
Signed-off-by: Adam Lackorzynski <adam@os.inf.tu-dresden.de>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm: Set carry flag correctly for Thumb2 ORNS
The code for Thumb2 ORNS (or negated and set flags) was trashinga TCG input register which was needed later for use in calculatingflags, with the effect that the carry flag was always set withthe wrong sense. Fix this by using the TCG orc op instead of...
target-arm: Handle VMOV between two core and VFP single regs
Fix two bugs in the translation of the instructions VMOV sa,sb,rx,ry andVMOV rx,ry,sa,sb (which copy between a pair of ARM core registers and apair of VFP single precision registers):
target-arm: Don't decode old cp15 WFI instructions on v7 cores
In v7 of the ARM architecture, WFI (wait for interrupt) is a first-classinstruction, but in previous versions this functionality was providedvia a cp15 coprocessor register. Add correct feature checks to the...
target-arm: Introduce float64_256 and float64_512 constants.
These two constants will be used by helper functions such as recpe_f32and rsqrte_f32.
Signed-off-by: Christophe Lyon <christophe.lyon@st.com>Reviewed-by: Peter Maydell <peter.maydell@linaro.org>...
target-arm: fix support for VRECPE.
Now use the same algorithm as described in the ARM ARM.
Signed-off-by: Christophe Lyon <christophe.lyon@st.com>Reviewed-by: Peter Maydell <peter.maydell@linaro.org>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm: fix support for VRSQRTE.
target-arm: Fix shift by immediate and narrow where src, dest overlap
For Neon shifts by immediate and narrow, correctly handle the casewhere the source registers and the destination registers overlap(the second pass should use the original register contents, not the...
target-arm: Refactor to pull narrowing decode into separate function
Pull the code which decodes narrowing operations as being eithersigned/unsigned saturate or plain out into its own function.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm: Fix rounding constant addition for Neon shifts
Handle cases where adding the rounding constant could overflow in Neonshift instructions: VRSHR, VRSRA, VQRSHRN, VQRSHRUN, VRSHRN.
Signed-off-by: Christophe Lyon <christophe.lyon@st.com>[peter.maydell@linaro.org: fix handling of large shifts in rshl_s32,...
target-arm: Fix signed VRSHL by large shift counts
Correctly handle VRSHL of signed values by a shift count of thewidth of the data type or larger, which must be special-cased in thershl_s* helper functions.
target-arm: Fix unsigned VRSHL.s8 and .s16 right shifts by type width
Fix handling of unsigned VRSHL.s8 and .s16 right shifts by the typewidth.
target-arm: fix unsigned 64 bit right shifts.
Fix range of shift amounts which always give 0 as result.
target-arm: Fix saturated values for Neon right shifts
Fix value returned by signed 8 and 16 bit qrshl helperswhen the result has saturated.
target-arm: fix Neon VQSHRN and VSHRN.
Call the normal shift helpers instead of the rounding ones.
target-arm: fix decoding of Neon 64 bit shifts.
Fix decoding of 64 bits variants of VSHRN, VRSHRN, VQSHRN, VQSHRUN,VQRSHRN, VQRSHRUN, taking into account whether inputs are unsignedor not.
target-arm: Fix signed VQRSHL by large shift counts
Handle the case of signed VQRSHL by a shift count of the width of thedata type or larger, which must be special cased in the qrshl_s*helper functions.
target-arm: Fix unsigned VQRSHL by large shift counts
Correctly handle VQRSHL of unsigned values by a shift count of thewidth of the data type or larger, which must be special-cased in theqrshl_u* helper functions.
target-arm: Move Neon VZIP to helper functions
Move the implementation of the Neon VUZP unzip instruction from inlinecode to helper functions. (At 50+ TCG ops it was well over therecommended limit for coding inline.) The helper implementations alsogive the correct answers where the inline implementation did not....
target-arm: Move Neon VUZP to helper functions
Move the implementation of the Neon VUZP unzip instruction from inlinecode to helper functions. (At 50+ TCG ops it was well over therecommended limit for coding inline.) The helper implementations alsofix the handling of the quadword version of the instruction....
target-arm: Correct conversion of Thumb Neon dp encodings into ARM
We handle Thumb Neon data processing instructions by converting theminto the equivalent ARM encoding, as the two are very close. Howeverthe ARM encoding should have bit 28 set, not clear. This wasn't causing...
target-arm: Fix Neon VQDMLSL instruction
For VQDMLSL, negation has to occur after saturation, not before.
target-arm: Refactor handling of VQDMULL
Refactor the handling of VQDMULL so that it is dealt with inits own if() case rather than together with the accumulatinginstructions.
target-arm: Implement VMULL.P8
Implement VMULL.P8 (the 32x32->64 version of the polynomial multiplyinstruction).
arm: drop unused irq-related part of CPUARMState
These two fields were added as a part of ARMv7 support patch (back in2007), were never used by any code, so can be dropped.
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>Reviewed-by: Peter Maydell <peter.maydell@linaro.org>...
target-arm: Remove stray #include from middle of neon_helper.c
Remove a stray #include <stdio.h> from the middle of neon_helper.c:it was harmless but pointless since we include stdio.h at the topof the file anyway.
target-arm: Silence NaNs resulting from half-precision conversions
Silence the NaNs that may result from half-precision conversion,as we do for the other conversions.
target-arm: Use standard FPSCR for Neon half-precision operations
The Neon half-precision conversion operations (VCVT.F16.F32 andVCVT.F32.F16) use ARM standard floating-point arithmetic, unlikethe VFP versions (VCVTB and VCVTT).
softfloat: Add float16 type and float16 NaN handling functions
Add a float16 type to softfloat, rather than using bits16 directly.Also add the missing functions float16_is_quiet_nan(),float16_is_signaling_nan() and float16_maybe_silence_nan(),which are needed for the float16 conversion routines....
target-arm: implement vsli.64, vsri.64
target-arm: fix VSHLL Neon instruction.
Fix bit mask used when widening the result of shift on narrow input.
target-arm: Fix 32 bit signed saturating narrow
The returned value when doing saturating signed 64->32 bitconversion of a negative number was incorrect due to a missing cast.
target-arm: Fix VQMOVUN Neon instruction.
VQMOVUN does a signed-to-unsigned saturating conversion. This isdifferent from both the signed-to-signed and unsigned-to-unsignedconversions already implemented, so we need a new set of helperfunctions (neon_unarrow_sat*)....
target-arm: Clean up handling of MPIDR
The ARM cp15 register 0,c0,c0,5 is standardised in the v7 architectureas the MPIDR. Clean up its implementation to remove A9 specific handling.
This commit includes fixing an error in the value returned for theMPIDR on A9, where we were erroneously claiming a cluster ID of 9....
target-arm: Fix decoding of preload and memory hint space
Correct the decoding of the ARM preload and memory hint space,by adding decoding of PLI, PLDW and the v7MP unallocated hintspace. This commit also corrects a slightly overexuberantdecoding of PLD which was not checking that bit 4...
target-arm: Fix decoding of Thumb preload and hint space
Refine the decoding of the Thumb preload and hint space, so weUNDEF on the patterns that are supposed to UNDEF rather than NOP.We also move the tests for this space earlier, so we don't emitharmless but unnecessary address generation code for preload...
target-arm: Add CPU feature flag for v7MP
Add a CPU feature flag for v7MP (the multiprocessing extensions); someinstructions exist only for v7MP and not for the base v7 architecture.
Set the right overflow bit for neon 32 and 64 bit saturating add/sub.
target-arm: Fix Neon vsra instructions.
This patch fixes the errors reported by my tests in VSRA.
target-arm: Fix Neon VQDMULH.S16 instructions
Correct an error in the implementation of the 16 bitforms of VQDMULH, bringing them into line with the32 bit implementation.
Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>Reviewed-by: Peter Maydell <peter.maydell@linaro.org>...
Support saturation with shift=0.
This patch fixes corner-case saturations, when the target range iszero. It merely removes the guard against (sh == 0), and makes:_ssat(0x87654321, 1) return 0xffffffff and set the saturation flag_usat(0x87654321, 0) return 0 and set the saturation flag...
target-arm: Fix garbage collection of temporaries in Neon emulation.
Fix garbage collection of temporaries in Neon emulation.
target-arm: Fix loading of scalar value for Neon multiply-by-scalar
Fix the register and part of register we get the scalar from inthe various "multiply vector by scalar" ops (VMUL by scalarand friends).
target-arm: Log instruction start in TCG code
Add support for logging the start of instructions in TCGcode debug dumps for ARM targets.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
target-arm: Translate with VFP-enabled from TB flags, not CPUState
When translating code, whether the VFP unit is enabled for this TBis stored in a bit in the TB flags. Use this rather than incorrectlyreading the FPEXC from the CPUState passed to translation....
target-arm: Translate with VFP len/stride from TB flags, not CPUState
When translating, the VFP vector length and stride for this TB are encodedin the TB flags; the CPUState copies may be different and must not be used.
target-arm: Translate with Thumb state from TB flags, not CPUState
The Thumb/ARM state for the TB being translated should come fromthe TB flags, not the CPUState.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>...
target-arm: Translate with condexec bits from TB flags, not CPUState
When translating, the condexec bits for the TB are in the TB flags;the CPUState condexec bits may be different.
This patch fixes https://bugs.launchpad.net/bugs/604872 where we mightsegfault if we took an exception in the middle of a TB with an IT...
target-arm: Set privileged bit in TB flags correctly for M profile
M profile ARM cores don't have a CPSR mode field. Set the bit in theTB flags that indicates non-user mode correctly for these cores.
target-arm: Translate with user-state from TB flags, not CPUState
When translating, get the user/priv state from the TB flags, notthe CPUState.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm: Remove redundant setting of IT bits before Thumb SWI
Remove a redundant call to gen_set_condexec() in the translation of Thumbmode SWI. (SWI and WFI generate "exceptions" which happen after theexecution of the instruction, ie when PC and IT bits have updated....
target-arm: Refactor translation of exception generating instructions
Create a new function which does the common sequence of gen_set_condexec,gen_set_pc_im, gen_exception, set is_jmp to DISAS_JUMP.
target-arm: Restore IT bits when resuming after an exception
We were not correctly restoring the IT bits when resuming executionafter taking an unexpected exception in the middle of an IT block.Fix this by tracking them along with PC changes and restoring in...
target-arm: Fix implementation of VRSQRTS
The implementation of the ARM VRSQRTS instruction (which calculates(3 - op1 * op2) / 2) was missing the division operation. It alsodid not handle the special cases of (0,inf) and (inf,0).
target-arm: Add support for 'Standard FPSCR Value' as used by Neon
Add support to the ARM helper routines for a second fp_status valuewhich should be used for operations which the ARM ARM indicates use"ARM standard floating-point arithmetic" rather than being controlled...
target-arm: Use the standard FPSCR value for VRSQRTS
VSQRTS always uses the standard FPSCR value as it is a Neon instruction.
target-arm: Don't generate code specific to current CPU mode for SRS
When translating the SRS instruction, handle the "store registersto stack of current mode" case in the helper function rather thaninline. This means the generated code does not make assumptions...
target-arm: Add symbolic constants for bitfields in TB flags
Add symbolic constants for the bitfields we use in the TB flags.
ARM: add neon helpers for VQSHLU
Add neon helper functions to implement VQSHLU, which is asigned-to-unsigned version of VQSHL available only as animmediate form.
Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>...
ARM: Fix decoding of VQSHL/VQSHLU immediate forms
Fix errors in the decoding of ARM VQSHL/VQSHLU immediate forms,including using the new VQSHLU helper functions where appropriate.
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>Signed-off-by: Peter Maydell <peter.maydell@linaro.org>...
target-arm: Set softfloat cumulative exc flags from correct FPSCR bits
When handling a write to the ARM FPSCR, set the softfloat cumulativeexception flags from the cumulative flags in the FPSCR, not theexception-enable bits. Also don't apply a mask: vfp_exceptbits_to_host...
target-arm: wire up the softfloat flush_input_to_zero flag
Wire up the new softfloat support for flushing input denormalsto zero on ARM. The FPSCR FZ bit enables flush-to-zero forboth inputs and outputs, but the reporting of when inputs areflushed to zero is via a separate IDC bit rather than the UFC...
target-arm: fix SMMLA/SMMLS instructions
SMMLA and SMMLS are broken on both in normal and thumb mode, that isboth (different) implementations are wrong. They try to avoid a 64-bitadd for the rounding, which is not trivial if you want to support bothSMMLA and SMMLS with the same code....
target-arm: fix UMAAL instruction
UMAAL should use unsigned multiply instead of signed.
This patch fixes this issue by handling UMAAL separately fromUMULL/UMLAL/SMULL/SMLAL as these instructions are differentenough. It also explicitly list instructions in case and catch...
target-arm: correct cp15 c1_sys reset value for cortex-a8
Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm: correct cp15 c1_sys reset value for arm1136 and cortex-a9
target-arm: fix vmsav6 access control
Override access control checks (including execute) for mmu translationtable descriptors assigned to manager domains.
target-arm: Correct result in saturating cases for VQSHL of s8/16/32
Where VQSHL of a signed 8/16/32 bit value saturated, the resultvalue was not being calculated correctly (it should be eitherthe minimum or maximum value for the size of the signed type)....
target-arm: remove pointless else clause in VQSHL of u64
Remove a pointless else clause in the neon_qshl_u64 helper.
Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>Reviewed-by: Peter Maydell <peter.maydell@linaro.org>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm: Fix VQSHL of signed 64 bit values by shift counts >= 64
VQSHL of a signed 64 bit non-zero value by a shift count >= 64 shouldsaturate; return the correct value in this case.
target-arm: Fix VQSHL of signed 64 bit values
Add a missing '-' which meant that we were misinterpreting the shiftargument for VQSHL of 64 bit signed values and treating almost everyshift value as if it were an extremely large right shift.
Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>...
target-arm: Fix arguments passed to VQSHL helpers
Correct the arguments passed when generating neon qshl_{u,s}64()helpers so that we use the correct registers.
target-arm: fix bug in translation of REVSH
The translation of REVSH shifted the low byte 8 steps left before performingan 8-bit sign extend, causing this part of the expression to alwas be 0.
Reported-by: Johan Bengtsson <teofrastius@gmail.com>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
ARM: fix ldrexd/strexd
Correct ldrexd and strexd code to always read and write thehigh word of the 64-bit value from addr+4.Also make ldrexd and strexd agree that for a 64 bit value theaddress in env->exclusive_addr is that of the low word.
This fixes the issues reported in...
ARM: Fix decoding of VFP forms of VCVT between float and int/fixed
Correct the decoding of source and destination registersfor the VFP forms of the VCVT instructions which convertbetween floating point and integer or fixed-point.
ARM: Fix decoding of Neon forms of VCVT between float and fixed point
Fix errors in the decoding of the Neon forms of fixed-point VCVT: * fixed-point VCVT is op 14 and 15, not 15 and 16 * the fbits immediate field was being misinterpreted * the sense of the to_fixed bit was inverted...
ARM: Fix sense of to_integer bit in Neon VCVT float/int conversion
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>Reviewed-by: Nathan Froyd <froydnj@codesourcery.com>
ARM: Return correct result for float-to-integer conversion of NaN
The ARM architecture mandates that converting a NaN value tointeger gives zero (if Invalid Operation FP exceptions arenot being trapped). This isn't the behaviour of the SoftFloatlibrary, so NaNs must be special-cased....
ARM: Return correct result for single<->double conversion of NaN
The ARM ARM defines that if the input to a single<->double conversionis a NaN then the output is always forced to be a quiet NaN by settingthe most significant bit of the fraction part.
ARM: Ignore top 16 bits when doing VCVT from 16 bit fixed point
VCVT of 16 bit fixed point to float should ignore the top 16 bitsof the source register. Cast to int16_t and friends rather thanint16 -- the former is guaranteed exactly 16 bits wide where the...
ARM: Implement VCVT to 16 bit integer using new softfloat routines
Use the softfloat conversion routines for conversion to 16 bitintegers, because just casting to a 16 bit type truncates thevalue rather than saturating it at 16-bit MAXINT/MININT.
target-arm: Add support for PKHxx in thumb2
The PKHxx instructions were not recognized by the thumb2 decoder. Thesolution provided in this changeset is identical to the arm-modeimplementation.
Signed-off-by: Johan Bengtsson <teofrastius@gmail.com>Signed-off-by: Peter Maydell <peter.maydell@linaro.org>...
target-arm: Fix mixup in decoding of saturating add and sub
The thumb2 decoder contained a mixup between the bit controllingdoubling and the bit controlling if the operation was an add or a sub.
target-arm: Handle 'smc' as an undefined instruction
Refine check on bkpt so that smc and undefined instruction encodings arehandled as an undefined instruction and trap.
Signed-off-by: Adam Lackorzynski <adam@os.inf.tu-dresden.de>Signed-off-by: Peter Maydell <peter.maydell@linaro.org>...
ARM: enable XScale/iWMMXT in linux-user mode
In linux-user mode, the XScale/iWMMXT coprocessors must be enabledat reset so that we can run code that uses these instructions.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
ARM: Expose vfp_get_fpscr() and vfp_set_fpscr() to C code
Expose the vfp_get_fpscr() and vfp_set_fpscr() functions to Ccode as well as generated code, so we can use them to read andwrite the FPSCR when saving and restoring VFP registers acrosssignal handlers in linux-user mode....
[PATCH] target-arm: remove unused functions cpu_lock(), cpu_unlock()
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
target-xxx: Use fprintf_function (format checking)
fprintf_function uses format checking with GCC_FMT_ATTR.
Cc: Blue Swirl <blauwirbel@gmail.com>Signed-off-by: Stefan Weil <weil@mail.berlios.de>...
remove exec-all.h inclusion from cpu.h
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
move cpu_pc_from_tb to target-*/exec.h
target-arm: fix addsub/subadd implementation
Signed-off-by: Chih-Min Chao <cmchao@gmail.com>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm : fix thumb2 parallel add/sub opcode decoding
target-arm : fix parallel saturated subtraction implementation
NEON vldN optimization
When combining multiple values as part of a NEON array load, do explcitshift/or rather than using gen_bfi. This voids redundant maskoperations.
Signed-off-by: Paul Brook <paul@codesourcery.com>
arm: fix arm kernel boot for non zero start addr
Booting an arm kernel has been broken a while when booting from non zero startaddress. This is due to the order of events: board init loads the kernel andsets register 15 to the start address and then qemu_system_reset reset the cpu...
arm: prevent coprocessor IO reset
This prevent coprocessor IO structure from being reset on cpu reset. This wasa problem for PXA which uses coprocessor 6 and 14.
Signed-off-by: Lars Munch <lars@segv.dk>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
arm: remove dead assignments, spotted by clang analyzer
Value stored is never read.
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
target-arm: fix neon vmon/vmvn with modified immediate
Signed-Off-By: Riku Voipio <riku.voipio@nokia.com>Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm: resource leak fixes for iwmmxt disassemble
This patch fixes few resource leaks in the iwmmxt disassemble.