]>
git.ipfire.org Git - thirdparty/valgrind.git/log
Julian Seward [Fri, 20 Jun 2014 14:27:27 +0000 (14:27 +0000)]
Fix an enum type confusion, PPCAvFpOp vs PPCAvOp, as excellently
detected by Clang. Gcc, are you paying attention?
git-svn-id: svn://svn.valgrind.org/vex/trunk@2882
Julian Seward [Fri, 20 Jun 2014 08:30:21 +0000 (08:30 +0000)]
Increase the number of vector registers available for allocation from
3 to 5.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2881
Julian Seward [Thu, 19 Jun 2014 22:20:47 +0000 (22:20 +0000)]
Implement: dup_{d_d[], s_s[], h_h[], b_b[]}, ext
git-svn-id: svn://svn.valgrind.org/vex/trunk@2880
Julian Seward [Thu, 19 Jun 2014 14:21:37 +0000 (14:21 +0000)]
Implement: orr_{8h,4h}_imm8_shifted, orr_{4s,2s}_imm8_shifted,
bic_{8h,4h}_imm8_shifted, bic_{4s,2s}_imm8_shifted, cls_std6_std6,
cm{eq,ge,gt,hi,hs,tst}_d_d_d, cm{ge,gt,le,lt}_d_d_zero,
cnt_{16,8}b_{16,8}b
git-svn-id: svn://svn.valgrind.org/vex/trunk@2879
Julian Seward [Sun, 15 Jun 2014 21:55:33 +0000 (21:55 +0000)]
arm64: implement: addp std7_std7_std7, addv vector, addp d_2d
git-svn-id: svn://svn.valgrind.org/vex/trunk@2878
Julian Seward [Sun, 15 Jun 2014 19:36:29 +0000 (19:36 +0000)]
arm64: implement: abs d_d, neg d_d, abs std7_std7, addhn, subhn, raddhn, rsubhn
git-svn-id: svn://svn.valgrind.org/vex/trunk@2877
Julian Seward [Sun, 15 Jun 2014 08:17:35 +0000 (08:17 +0000)]
Remove temporary front end scaffolding for Cat{Even,Odd}Lanes
and Interleave{LO,HI} operations, and instead generate real
UZP1/UZP2/ZIP1/ZIP2 instructions in the back end.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2876
Julian Seward [Sat, 14 Jun 2014 18:05:30 +0000 (18:05 +0000)]
Implement LD1R (single structure, replicate).
git-svn-id: svn://svn.valgrind.org/vex/trunk@2875
Julian Seward [Thu, 12 Jun 2014 13:16:01 +0000 (13:16 +0000)]
Implement FMUL 2d_2d_d[], 4s_4s_s[], 2s_2s_s[].
git-svn-id: svn://svn.valgrind.org/vex/trunk@2874
Julian Seward [Thu, 12 Jun 2014 10:15:46 +0000 (10:15 +0000)]
Remove the old SIMD decoder entirely.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2873
Julian Seward [Wed, 11 Jun 2014 20:57:23 +0000 (20:57 +0000)]
Move remaining implemented SIMD instructions into the new SIMD/FP
decoding framework.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2872
Julian Seward [Tue, 10 Jun 2014 22:52:05 +0000 (22:52 +0000)]
Reimplement the SIMD and FP instruction decoder, so as to avoid huge
amounts of duplicated decode, and to follow the documentation more
closely.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2871
Dejan Jevtic [Mon, 9 Jun 2014 10:54:49 +0000 (10:54 +0000)]
mips: Fix non mips compiler warning.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2870
Julian Seward [Wed, 4 Jun 2014 13:09:44 +0000 (13:09 +0000)]
Support ADC/ADCS/SBC/SBCS. Fixes #335496. (dimitry@google.com)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2868
Julian Seward [Wed, 4 Jun 2014 11:44:45 +0000 (11:44 +0000)]
Support the "ishst" variant of "dmb". Fixes #335263. (dimitry@google.com)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2867
Julian Seward [Wed, 4 Jun 2014 11:36:54 +0000 (11:36 +0000)]
Support movi_{16b,8b}_#imm8. Fixes #335262. (dimitry@google.com)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2866
Dejan Jevtic [Wed, 4 Jun 2014 11:28:07 +0000 (11:28 +0000)]
mips64: Support for Cavium MIPS Octeon Atomic and Count Instructions.
Implement Cavium MIPS specific instructions:
baddu, pop, dpop, saa, saad, laa, laad, lai, laid, lad, ladd, law, lawd,
las, lasd, lac, lacd
Fixes BZ #327223.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2865
Julian Seward [Wed, 21 May 2014 14:42:04 +0000 (14:42 +0000)]
Implement PCMPxSTRx cases 0x0E, 0x34, 0x14, and reformat some of the
associated switch statements. Fixes #326469, #327639, #328878
respectively.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2864
Julian Seward [Fri, 16 May 2014 11:20:07 +0000 (11:20 +0000)]
Implement SHL_d_d_#imm.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2863
Julian Seward [Thu, 15 May 2014 16:49:21 +0000 (16:49 +0000)]
Initial front-end fixings needed to handle code generated by gcc-4.9
on arm64-linux.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2862
Julian Seward [Wed, 14 May 2014 23:38:23 +0000 (23:38 +0000)]
Implement VFPv4 VFMA and VFMS (F32 and F64 versions). Fixes #331057.
Patch from Janne Hellsten (jjhellst@gmail.com) with algebraic
rearrangement for the VFMS cases so as to make result signs match with
the hardware when some of the inputs are infinities.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2861
Julian Seward [Tue, 13 May 2014 15:54:14 +0000 (15:54 +0000)]
Thumb encoding: fix assertion failure caused by
"ldr.w pc, [reg, #imm]". Fixes #333428. (dimitry@google.com)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2860
Julian Seward [Tue, 13 May 2014 14:44:21 +0000 (14:44 +0000)]
Thumb encoding: correctly deal with misaligned loads of the form
LD Rt, [Rn +/- #imm12] when Rn == PC
Fixes #333145. (dimitry@google.com)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2859
Mark Wielaard [Fri, 9 May 2014 11:41:06 +0000 (11:41 +0000)]
Recognize MPX instructions and bnd prefix. Bug #333666.
Recognize and parse operands of new MPX instructions BNDMK, BNDCL,
BNDCU, BNDCN, BNDMOV, BNDLDX and BNDSTX. Also recognize bnd (F2) prefix
for CALL (E8,FF/2), RET (C2,C3), JMP (EB,E9,FF/4) and Jcc (70-7F,0F 80-8F).
All new MPX instructions are currently NOPs and the bnd prefix is ignored.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2858
Julian Seward [Wed, 7 May 2014 11:09:28 +0000 (11:09 +0000)]
Handle "blr lr" correctly -- read the destination register
_before_ writing the return address in LR.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2857
Julian Seward [Wed, 7 May 2014 09:41:40 +0000 (09:41 +0000)]
Enable 'smulh'.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2856
Julian Seward [Wed, 7 May 2014 09:20:59 +0000 (09:20 +0000)]
Handle IRStmt::STle of type F32.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2855
Julian Seward [Wed, 7 May 2014 09:20:11 +0000 (09:20 +0000)]
Allow early-writeback for the cases
stp d, d, [sp,#-imm]!
stp s, s, [sp,#-imm]!
as well as for the existing case
stp q, q, [sp,#-imm]!
git-svn-id: svn://svn.valgrind.org/vex/trunk@2854
Julian Seward [Mon, 5 May 2014 10:03:56 +0000 (10:03 +0000)]
Fix assertion failures resulting from change of arity of
Iop_{Add,Sub,Mul}32Fx4 introduced in r2809, in which said IROps
acquired a rounding-mode argument.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2853
Julian Seward [Sun, 4 May 2014 10:52:11 +0000 (10:52 +0000)]
Renaming only (no functional change): rename IR artefacts to do
with i-cache invalidation to be more consistent with new d-cache
invalidation functionality:
Ijk_TInval -> Ijk_InvalICache
TISTART -> CMSTART (CM == "Cache Management")
TILEN -> CMLEN
VEX_TRC_JMP_TINVAL -> VEX_TRC_JMP_INVALICACHE
git-svn-id: svn://svn.valgrind.org/vex/trunk@2852
Julian Seward [Sat, 3 May 2014 21:20:56 +0000 (21:20 +0000)]
ARM64: add support for cache management instructions (VEX side):
dc cvau, regX
ic ivau, regX
mrs regX, ctr_el0
Fixes #333228 and #333230.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2851
Julian Seward [Wed, 30 Apr 2014 22:50:34 +0000 (22:50 +0000)]
x87 instructions FSIN, FCOS, FSINCOS and FPTAN: handle out-of-range
arguments correctly. Mozilla bug 995564.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2850
Julian Seward [Sun, 27 Apr 2014 12:02:12 +0000 (12:02 +0000)]
Finish off vector integer comparison instructions, and
vector shift-by-immediates (Shr/Shl/Sar) instructions.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2849
Julian Seward [Tue, 8 Apr 2014 15:24:15 +0000 (15:24 +0000)]
Handle Iop_Max32U, so as to make origin tracking in Memcheck work.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2848
Julian Seward [Tue, 8 Apr 2014 15:23:42 +0000 (15:23 +0000)]
{FMOV,MOVI} (vector, immediate): fix incorrect DIP format string
git-svn-id: svn://svn.valgrind.org/vex/trunk@2847
Julian Seward [Thu, 3 Apr 2014 23:03:32 +0000 (23:03 +0000)]
Implement
LD2/ST2 (multiple structures, post index) (some cases)
LD1/ST1 (multiple structures, no offset) (some cases)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2846
Julian Seward [Thu, 3 Apr 2014 13:48:54 +0000 (13:48 +0000)]
Implement TBL and TBX instructions.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2845
Julian Seward [Thu, 3 Apr 2014 13:48:21 +0000 (13:48 +0000)]
Add a couple more constant folding rules for vectors.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2844
Julian Seward [Tue, 1 Apr 2014 11:00:36 +0000 (11:00 +0000)]
Bug 332658 - ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
(dimitry@google.com)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2843
Julian Seward [Thu, 27 Mar 2014 18:59:00 +0000 (18:59 +0000)]
Implement FCM{EQ,GE,GT}, FAC{GE,GT} (vector).
git-svn-id: svn://svn.valgrind.org/vex/trunk@2842
Dejan Jevtic [Wed, 19 Mar 2014 11:10:51 +0000 (11:10 +0000)]
mips32: Avoid compiler warnings.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2841
Julian Seward [Sat, 15 Mar 2014 11:41:39 +0000 (11:41 +0000)]
Un-break the arm32 compilation pipeline following the change of
arity of Iop_Mul32Fx4, Iop_Sub32Fx4, Iop_Add32Fx4 in r2809.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2840
Julian Seward [Sat, 15 Mar 2014 08:33:06 +0000 (08:33 +0000)]
LDRD/STRD reg+/-#imm8: allow PC as the base register in the
case "ldrd Rt, Rt2, [PC, #+/-imm8]". n-i-bz.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2839
Julian Seward [Sat, 15 Mar 2014 08:14:06 +0000 (08:14 +0000)]
Correctly handle add(hi) when the destination register is the PC.
Fixes #332037.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2838
Julian Seward [Mon, 10 Mar 2014 10:40:48 +0000 (10:40 +0000)]
Back-end handling of Iop_CmpNEZ32x4, Iop_CmpNEZ16x8, Iop_CmpNEZ8x16,
needed for Memchecking of SIMD arm64 code.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2837
Julian Seward [Sun, 9 Mar 2014 09:41:56 +0000 (09:41 +0000)]
Implement a couple of backend artefacts needed by Memcheck on large
applications:
Iop_CmpNEZ64x2 expressions
Ijk_NoRedir block terminators
git-svn-id: svn://svn.valgrind.org/vex/trunk@2836
Julian Seward [Sun, 9 Mar 2014 09:40:23 +0000 (09:40 +0000)]
Do early writeback of the base register for the following instruction
forms, to stop Memcheck complaining about writes below the stack
pointer:
str x3, [sp,#-16]!
stp q0, q1, [sp,#-512]!
git-svn-id: svn://svn.valgrind.org/vex/trunk@2835
Julian Seward [Sat, 8 Mar 2014 13:08:17 +0000 (13:08 +0000)]
* iselIntExpr_AMode_wrk: generate correct code for the case
"Sub64(expr,simm9)."
* handle 1Uto64(arbitrary-expression)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2834
Julian Seward [Fri, 7 Mar 2014 22:52:19 +0000 (22:52 +0000)]
Support extra instruction bits and pieces, enough to get Firefox started:
* more scalar int <-> FP conversions
* more vector integer narrowing
* a few more vector shift by imm cases
* FCVTAS (kludged)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2833
Dejan Jevtic [Mon, 3 Mar 2014 14:13:37 +0000 (14:13 +0000)]
mips32: Fix the problem with reading the guest_FCSR register from the wrong guest state.
When Valgrind isn't executed in mode64, register fcsr need to read from the VexGuestMIPS32State.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2832
Julian Seward [Mon, 3 Mar 2014 08:42:16 +0000 (08:42 +0000)]
Fix error in 64-bit and smaller load versions of
LDR/STR (immediate, SIMD&FP, unsigned offset)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2831
Julian Seward [Sun, 2 Mar 2014 12:47:18 +0000 (12:47 +0000)]
Implement REV16, REV32, FCVTN, SHL (vector, immediate), NEG (vector)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2830
Julian Seward [Sat, 1 Mar 2014 11:19:45 +0000 (11:19 +0000)]
Remove redundant FMOV (vector, immediate) case.
Minor comment fixes.
Fix bugs in {U,S}{MIN,MAX}V, {U,S}{MIN,MAX}, {S,U}SSHL
git-svn-id: svn://svn.valgrind.org/vex/trunk@2829
Julian Seward [Sat, 1 Mar 2014 11:16:57 +0000 (11:16 +0000)]
Select and emit insns for
Iop_ZeroHI64ofV128 Iop_Max8Sx16 Iop_Min8Sx16
git-svn-id: svn://svn.valgrind.org/vex/trunk@2828
Dejan Jevtic [Thu, 27 Feb 2014 14:17:19 +0000 (14:17 +0000)]
mips32: Fpu guest registers are ULong and the initial values need to be
extended.
Because we are supporting both big and little endian mips32 we need to
make sure that the initial values for the fpu registers are the same for both
endian.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2827
Julian Seward [Thu, 27 Feb 2014 11:10:19 +0000 (11:10 +0000)]
Implement a few more integer instructions:
NOP
LDA{R,RH,RB}
STL{R,RH,RB}
RBIT
git-svn-id: svn://svn.valgrind.org/vex/trunk@2826
Dejan Jevtic [Tue, 25 Feb 2014 15:25:49 +0000 (15:25 +0000)]
mips32: Fix the problem with the floating point compare instruction on mips32.
This patch is fixing the problem with emitting Iop_CmpF64.
Problem was introduced while running Valgrind for mips with v8 javascript engine.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2825
Julian Seward [Fri, 21 Feb 2014 14:49:44 +0000 (14:49 +0000)]
* add a kludgey fix for "mrs rT, dczid_el0"
* make ISB and DSB really generate memory barriers
git-svn-id: svn://svn.valgrind.org/vex/trunk@2824
Julian Seward [Thu, 20 Feb 2014 17:43:38 +0000 (17:43 +0000)]
First pass at implementation of load/store exclusive and
load/store exclusive w/ read-acquire/store-release:
LD{,A}X{R,RH,RB}
ST{,L}X{R,RH,RB}
git-svn-id: svn://svn.valgrind.org/vex/trunk@2823
Julian Seward [Wed, 19 Feb 2014 17:42:59 +0000 (17:42 +0000)]
Implement unchainXDirect_ARM64.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2822
Dejan Jevtic [Wed, 19 Feb 2014 11:56:29 +0000 (11:56 +0000)]
mips32: VEX Support for 64bit FPU on MIPS32 platforms.
This patch is adding support for mips32 with 64bit FPU.
Assume that floating-point registers are 64 bits wide.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2821
Julian Seward [Mon, 17 Feb 2014 11:00:53 +0000 (11:00 +0000)]
Implement more aarch64 vector insns:
CM{EQ,HI,HS,GE,GT,TST,LE,LT} (vector)
{EOR,BSL,BIT,BIF} (vector)
{USHR,SSHR} (vector, immediate)
{U,S}SHLL{,2}
INS (general)
FABD Vd,Vn,Vm
git-svn-id: svn://svn.valgrind.org/vex/trunk@2820
Petar Jovanovic [Fri, 14 Feb 2014 17:28:15 +0000 (17:28 +0000)]
mips64: add support for load indexed instructions from DSP ASE
Handling lwx, ldx and lbux for MIPS-Cavium processors.
Patch by Zahid Anwar, with some changes.
Related to Bugzilla issue 326444.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2819
Florian Krohm [Fri, 14 Feb 2014 08:55:32 +0000 (08:55 +0000)]
Fix comments and code snippets that were making incorrect claims about
the alignment requirement of the guest state, shadow areas, and register
spill area sizes.
The size of these areas ought to be a multiple of 16 bytes.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2818
Florian Krohm [Tue, 11 Feb 2014 09:23:01 +0000 (09:23 +0000)]
s390: Fix s390_amode_for_guest_state. In general the offset relative
to the guest state pointer may be more than the B12 addressing mode can
handle. Fall back and use a B20 addressing mode in those cases.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2817
Julian Seward [Mon, 10 Feb 2014 12:27:29 +0000 (12:27 +0000)]
Fix the ppc32 special-instruction magic sequence so it really does
preserve the value of r0, as claimed. Fixes #278808 (VEX side).
git-svn-id: svn://svn.valgrind.org/vex/trunk@2816
Julian Seward [Mon, 10 Feb 2014 10:28:13 +0000 (10:28 +0000)]
Implement more aarch64 vector insns:
{S,U}{MIN,MAX} Vd.T, Vn.T, Vm.T (8bitx16lane variants)
{S,U}{MIN,MAX}V Vd.T, Vn.T, Vm.T (8bitx16lane variants)
FMOV (vector, immediate)
MOVI (vector, immediate)
FABS (vector)
FNEG (vector)
FMLA (vector)
FMLS (vector)
{AND,BIC,ORR,ORN} (vector)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2815
Tom Hughes [Sun, 9 Feb 2014 11:40:20 +0000 (11:40 +0000)]
Add support for syscall on x86
Patch from Ivo Raisr via BZ#330939 also fixes BZ#308729
git-svn-id: svn://svn.valgrind.org/vex/trunk@2814
Julian Seward [Thu, 6 Feb 2014 12:57:58 +0000 (12:57 +0000)]
Implement a few more vector aarch64 insns:
LD1 {vT.8h}, [xN|SP], #16
LD1 {vT.16b}, [xN|SP], #16
ST1 {vT.4h}, [xN|SP], #8
MUL Vd.T, Vn.T, Vm.T
PMUL Vd.T, Vn.T, Vm.T (fe only)
MLA Vd.T, Vn.T, Vm.T
MLS Vd.T, Vn.T, Vm.T
UMOV Xd/Wd, Vn.Ts[index]
SMOV Xd/Wd, Vn.Ts[index]
git-svn-id: svn://svn.valgrind.org/vex/trunk@2813
Julian Seward [Wed, 5 Feb 2014 11:01:19 +0000 (11:01 +0000)]
Implement a few more vector aarch64 insns:
LD1 {vT.4s}, [xN|SP], #16
ADD Dd, Dn, Dm
SUB Dd, Dn, Dm
SMIN Vd.T, Vn.T, Vm.T
UMIN Vd.T, Vn.T, Vm.T
SMAX Vd.T, Vn.T, Vm.T
UMAX Vd.T, Vn.T, Vm.T
SMINV Vd.T, Vn.T, Vm.T
UMINV Vd.T, Vn.T, Vm.T
SMAXV Vd.T, Vn.T, Vm.T
UMAXV Vd.T, Vn.T, Vm.T
DUP Vd.T, Rn
FADD/FSUB/FMUL/FDIV32x4
git-svn-id: svn://svn.valgrind.org/vex/trunk@2812
Petar Jovanovic [Fri, 31 Jan 2014 12:14:20 +0000 (12:14 +0000)]
mips64: Support for Cavium-specific load indexed instructions
Support for lhux, lwux, lbx for Cavium.
Patch by Zahid Anwar, with style changes.
Related to Bugzilla issue 326444.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2811
Julian Seward [Sun, 26 Jan 2014 19:11:14 +0000 (19:11 +0000)]
Improve front and back end support for SIMD instructions on Arm64.
Implement the following instructions -- some but not necessarily
all laneage combinations:
LD1 {vT.2d}, [Xn|SP]
ST1 {vT.2d}, [Xn|SP]
LD1 {vT.4s}, [Xn|SP]
ST1 {vT.4s}, [Xn|SP]
LD1 {vT.8h}, [Xn|SP]
ST1 {vT.8h}, [Xn|SP]
LD1 {vT.16b}, [Xn|SP]
ST1 {vT.16b}, [Xn|SP]
LD1 {vT.1d}, [Xn|SP]
ST1 {vT.1d}, [Xn|SP]
LD1 {vT.2s}, [Xn|SP]
ST1 {vT.2s}, [Xn|SP]
LD1 {vT.4h}, [Xn|SP]
ST1 {vT.4h}, [Xn|SP]
LD1 {vT.8b}, [Xn|SP]
ST1 {vT.8b}, [Xn|SP]
ST1 {vT.2d}, [xN|SP], #16
LD1 {vT.2d}, [xN|SP], #16
ST1 {vT.4s}, [xN|SP], #16
ST1 {vT.8h}, [xN|SP], #16
ST1 {vT.2s}, [xN|SP], #8
SCVTF Vd, Vn
UCVTF Vd, Vn
FADD Vd,Vn,Vm 1
FSUB Vd,Vn,Vm 2
FMUL Vd,Vn,Vm 3
FDIV Vd,Vn,Vm 4
FMLA Vd,Vn,Vm 5
FMLS Vd,Vn,Vm 6
ADD Vd.T, Vn.T, Vm.T
SUB Vd.T, Vn.T, Vm.T
XTN {,2}
DUP Vd.T, Vn.Ts[index]
git-svn-id: svn://svn.valgrind.org/vex/trunk@2810
Julian Seward [Sun, 26 Jan 2014 18:34:23 +0000 (18:34 +0000)]
Make the following primops take a third (initial) argument to
indicate the rounding mode to use, like their scalar cousins do:
Iop_Add32Fx4 Iop_Sub32Fx4 Iop_Mul32Fx4 Iop_Div32Fx4
Iop_Add64Fx2 Iop_Sub64Fx2 Iop_Mul64Fx2 Iop_Div64Fx2
Iop_Add64Fx4 Iop_Sub64Fx4 Iop_Mul64Fx4 Iop_Div64Fx4
Iop_Add32Fx8 Iop_Sub32Fx8 Iop_Mul32Fx8 Iop_Div32Fx8
Fix up the x86 and amd64 front ends to add fake rounding modes
(Irrm_NEAREST) when generating expressions using these primops.
Fix up the x86 and amd64 back ends to accept these as triops
rather than as binops, and ignore the first arg.
Add three more ir_opt folding rules to remove memcheck
instrumentation arising from instrumentation of known-defined
rounding modes.
Overall functional and performance effects should be zero.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2809
Carl Love [Fri, 24 Jan 2014 16:42:26 +0000 (16:42 +0000)]
This patch by adrian.sendroiu@freescale.com fixes the lrmw and stmw
instructions.
The patch also adds ppc32 and ppc64 test cases for the instructions.
The patch is a fix for bugzilla 329956 "valgrind crashes when lmw/stmw instructions are used on ppc64".
git-svn-id: svn://svn.valgrind.org/vex/trunk@2808
Julian Seward [Tue, 21 Jan 2014 16:33:51 +0000 (16:33 +0000)]
Whitespace-only change: restrict to 80 col width.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2807
Dejan Jevtic [Wed, 15 Jan 2014 16:26:31 +0000 (16:26 +0000)]
mips64: Change the initial value of fpu registers.
Initial value of mips fpu registers should be nan instead of 0xffffffffffffffff.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2806
Julian Seward [Wed, 15 Jan 2014 10:25:21 +0000 (10:25 +0000)]
arm64: rename guest_SP to guest_XSP so as to avoid a name clash with
guest_SP from s390 world. Also back out the rename of guest_SP to
guest_s390_SP that caused s390 build breakage in r2803.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2805
Julian Seward [Sun, 12 Jan 2014 18:23:45 +0000 (18:23 +0000)]
Add missing ULLs to some 64-bit immediates.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2804
Julian Seward [Sun, 12 Jan 2014 12:49:10 +0000 (12:49 +0000)]
Add support for ARMv8 AArch64 (the 64 bit ARM instruction set):
integer and FP instructions.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2803
Julian Seward [Wed, 11 Dec 2013 16:47:59 +0000 (16:47 +0000)]
LibVEX_GuestAMD64_initialise(): give an initial value for
guest_ACFLAG. Kinda worrying that this was missing.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2802
Florian Krohm [Tue, 10 Dec 2013 16:51:15 +0000 (16:51 +0000)]
The result of rounding a 128-bit BFP/DFP value to 32/64 bit needs to
be stored in a register pair. This constraint was not observed previously
and the result was stored in any FPR that happened to be chosen. If the
selected FPR was not identifying a proper FPR pair, a SIGILL was delivered.
Fixes BZ #328455.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2801
Mark Wielaard [Mon, 9 Dec 2013 12:54:06 +0000 (12:54 +0000)]
Bug 328100 - XABORT not implemented.
XABORT can be called even when there is no current transaction.
In such a case XABORT acts as a NOP. Implement xabort as nop.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2800
Dejan Jevtic [Thu, 14 Nov 2013 15:44:42 +0000 (15:44 +0000)]
mips32/64: Fixed the problem with fpu instructions.
Include the value of fcsr register when emitting
some fpu instructions.
Calculate a new value of the fcsr register after
some FPU instructions.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2799
Christian Borntraeger [Thu, 7 Nov 2013 21:37:28 +0000 (21:37 +0000)]
Fix Bug 327284. The condition code of risbg was not correct.
This instruction might be used by by gcc for masking out bits,
e.g. code like
n &= 3;
if (n == 0)
might result in
risbg %r4,%r4,62,128+63,0
je <target>
The old code set the condition code depending on the operand before
masking. Fix it. This patch also indicates that we need test suite
coverage for risbg and friends.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2798
Dejan Jevtic [Wed, 23 Oct 2013 14:05:15 +0000 (14:05 +0000)]
mips32: Fix problem with some mips32 dsp instructions.
This patch includes VEX optimizations for mips32 and
fixes for some mips32 dsp instructions.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2796
Dejan Jevtic [Tue, 22 Oct 2013 08:52:46 +0000 (08:52 +0000)]
mips32: Fixed the problem with FCSR register.
Until now, Valgrind has read two registers to calculate FCSR
value. From now on, Valgrind reads exact number of fpu
registers (one or two).
git-svn-id: svn://svn.valgrind.org/vex/trunk@2794
Julian Seward [Mon, 21 Oct 2013 10:05:33 +0000 (10:05 +0000)]
In 64 bit mode, allow 64 bit return values from clean helper calls.
This makes SMC checking calls work (even though they are irrelevant
on PPC targets). Fixes #309430.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2793
Julian Seward [Fri, 18 Oct 2013 14:12:58 +0000 (14:12 +0000)]
Update copyright dates (20XY-2012 ==> 20XY-2013)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2792
Julian Seward [Fri, 18 Oct 2013 10:45:21 +0000 (10:45 +0000)]
Handle PCMPxSTRx cases 0x30 and 0x40. Fixes #320998.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2791
Carl Love [Fri, 18 Oct 2013 01:19:06 +0000 (01:19 +0000)]
This commit adds support for the following instructions:
vaddcuq, vadduqm, vaddecuq, vaddeuqm,
vsubcuq, vsubuqm, vsubecuq, vsubeuqm,
vbpermq and vgbbd.
The vgbbd instruction required a new Iop -- Iop_PwBitMtxXpose64x2.
All other instructions were emulated using existing Iops.
The completes adding the Power ISA 2.07 support.
Bugzilla 325816
git-svn-id: svn://svn.valgrind.org/vex/trunk@2790
Carl Love [Tue, 15 Oct 2013 18:11:20 +0000 (18:11 +0000)]
Power 8 support, phase 5
This commit adds support for the following instructions for doing
various arithmetic, bit transformation, cipher, count, logical,
and SHA operations.
vpmsumb, vpmsumh, vpmsumw, vpmsumd, vpermxor, vcipher, vcipherlast,
vncipher, vncipherlast, vsbox,
vclzb, vclzw, vclzh, vclzd,
vpopcntb, vpopcnth, vpopcntw, vpopcntd,
vnand, vorc, veqv,
vshasigmaw, vshasigmad,
bcdadd, bcdsub
The following Iops were added to support the above instructions:
Iop_BCDAdd, Iop_BCDSub,
Iop_PolynomialMulAdd8x16, Iop_PolynomialMulAdd16x8,
Iop_PolynomialMulAdd32x4, Iop_PolynomialMulAdd64x2,
Iop_CipherV128, Iop_CipherLV128, Iop_CipherSV128,
Iop_NCipherV128, Iop_NCipherLV128,
Iop_SHA512, Iop_SHA256, Iop_Clz64x2
The patch is for Bugzilla 325628
git-svn-id: svn://svn.valgrind.org/vex/trunk@2789
Julian Seward [Tue, 15 Oct 2013 17:29:19 +0000 (17:29 +0000)]
Tidyup -- no functional change. Replace all "pfx & PFX_LOCK"
with haveLOCK(pfx).
git-svn-id: svn://svn.valgrind.org/vex/trunk@2788
Julian Seward [Tue, 15 Oct 2013 17:21:42 +0000 (17:21 +0000)]
amd64 front end: accept XACQUIRE and XRELEASE on exactly the insns that
the Intel docs say can validly have them.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2787
Julian Seward [Mon, 14 Oct 2013 21:47:14 +0000 (21:47 +0000)]
Add support for an alternative encoding of 'PUSH reg', viz FF /6,
that is used by MSVC generated code. Fixes #324834.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2786
Julian Seward [Mon, 14 Oct 2013 11:40:24 +0000 (11:40 +0000)]
PPC32/64: Allow 16 byte icache lines.
Partial fix for #308135. (christophe.leroy@c-s.fr)
git-svn-id: svn://svn.valgrind.org/vex/trunk@2785
Carl Love [Wed, 9 Oct 2013 17:52:01 +0000 (17:52 +0000)]
Power PC, add the two privileged Transactional Memory instructions.
The initial Transactional Memory instruction patch did not include the two
privileged (OS) instructions. This patch adds support for the two
instructions, treclaim and trechkpt.
The patch if for Bugzilla 325751.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2784
Dejan Jevtic [Mon, 7 Oct 2013 10:28:56 +0000 (10:28 +0000)]
mips64: add extra Iop cases in VEX.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2783
Mark Wielaard [Fri, 4 Oct 2013 22:31:48 +0000 (22:31 +0000)]
Fix guest_amd64_toIR xbegin and xtest to match cpuid given for AVX hwcaps.
Otherwise valgrind none/tests/amd64/tm1.vgtest might fail on amd64-avx.
git-svn-id: svn://svn.valgrind.org/vex/trunk@2782
Carl Love [Thu, 3 Oct 2013 21:38:45 +0000 (21:38 +0000)]
Phase 4 support for IBM Power ISA 2.07
This patch adds support for the following instructions for doing
various arithmetic, logic, and load/store VSX operations:
xscvsxdsp xscvuxdsp xsaddsp xssubsp xsdivsp xsmaddasp xsmaddmsp
xsmsubasp xsmsubmsp xsnmaddasp xsnmaddmsp xsnmsubasp xsnmsubmsp
xsmulsp xssqrtsp xsresp xsrsqrtesp xsrsp xxlorc xxlnand xxleqv
lxsiwzx lxsiwax lxsspx stxsiwx stxsspx
Signed-off-by: Maynard Johnson <maynardj@us.ibm.com>
Bugzilla 325477
git-svn-id: svn://svn.valgrind.org/vex/trunk@2781
Carl Love [Wed, 2 Oct 2013 16:25:57 +0000 (16:25 +0000)]
Power PC, Approach 1, add Transactional Memory instruction support
The following Transactional Memory instructions are added:
tbegin., tend., tsr., tcheck., tabortwc.,
tabortdc., tabortwci., tabortdci., tabort.
The patch implements the first proposal by Julian on how to handle the
TM instructions. The proposal is as follows:
translate "XBEGIN fail-addr" as "goto fail-addr"; that is: push
simulated execution directly onto the failure path. This is simple
but will have poor performance, if (as is likely) the failure path
uses normal locking and is not tuned for speed.
The tbegin instruction on Power sets the condition code register to
indicate if the tbegin instruction suceeded or failed. The compiler
then generates a conditional branch instruction to take the success
or failure code path for the tbegin instruction. In order to fail the
tbegin instruction, the condition code register is updated to indicate
that the tbegin instruction failed. This patch assumes that there is
always an error handler for the tbegin instruction. The other TM
instructions are all treated as no ops as we shouldn't be executing the
sucess transactional code path.
Signed-off-by: Carl Love <cel@us.ibm.com>
Bugzilla 323803
git-svn-id: svn://svn.valgrind.org/vex/trunk@2780