* Re: Changing the treatment of the MIPS HI and LO registers [not found] ` <87r6bm1ebd.fsf@firetop.home> @ 2008-05-29 2:01 ` Maciej W. Rozycki [not found] ` <878wxtvarg.fsf@firetop.home> 0 siblings, 1 reply; 7+ messages in thread From: Maciej W. Rozycki @ 2008-05-29 2:01 UTC (permalink / raw) To: Richard Sandiford; +Cc: Ralf Baechle, gcc-patches, linux-mips Richard, I do not feel like speaking for the whole MIPS/Linux community and certainly not the maintainer. I have thus added a couple of recipients to the cc list as I fear gcc-patches does not reach exactly the right people here. Others may disagree with my thoughts expressed. On Wed, 28 May 2008, Richard Sandiford wrote: > >> I'm afraid they'll just have to use "x". Although looking at it: > >> > >> - delay.h is just doing common-or-garden multiplication, albeit > >> with optional R4000 workaounds. Why not just do them in C? > >> Can't you rely on GCC to work around the R4000 errata, now that we > >> have -mfix-r4000? > > > > Which C type is correct to access the high 64 bits of the 128-bit result > > of the multiplication on a 64-bit platform then? Please note the piece of > > code in question does not care about what the low 64 bits are -- a C > > construct that ultimately expands to a multiplication followed by an > > "mfhi" would be perfectly fine here. > > > > The workaround only came in quite recently here and could be avoided if C > > code was used of course -- don't get bothered with it. > > After the patch, you can use: > > ---------------------------------------------------------------------- > typedef unsigned int uint128_t __attribute__((mode(TI))); > > uint64_t foo (uint64_t x, uint64_t y) > { > return ((uint128_t) x * y) >> 64; > } > ---------------------------------------------------------------------- > > This doesn't work before the patch due to the lack of a movti instruction. > We have *diti3_highpart patterns, but it looks like they haven't been > used since combine started checking rtx_costs. > > If the patch goes in, I'll add tests to make sure that this code > continues to work without recourse to a libgcc function. As a follow-on, > I'll add proper TImode costs to rtx_costs, so that we take full advantage > of the highpart instructions. (This ought to avoid unnecessary moves.) OK, it is good to have a way to make use of some TImode functionality 64-bit hardware provides, but it still does not provide a clean solution to the compatibility problem seen here. While a construct like: #if (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 4) [new wonderful TImode stuff] #else [old compatibility asm crap] #endif will certainly work, it is as ugly as you can get. I am not sure I would be happy to see it in Linux, though obviously Ralf and the others may disagree. > > Is there absolutely no way to preserve the current asm semantics without > > complicating the internals horribly? I could imagine GCC could use some > > other constraints internally that would refer to the HI and LO registers > > explicitly and "h" and "l" could only be defined for asms -- I am not > > enough of a GCC expert here, so I don't know if it makes sense, but it > > would seem a reasonable idea to me. > > I just don't think there is, sorry. If you have: > > asm (... : "=h" (word1), "=l" (word2) : ...); > > then there needs to be at least two separately-allocatable registers, > each of which can hold word-sized values. And if we have two such > registers, the allocators will think they can use them for other things, > including spill space. Which brings us back to where we are now. I have just been wondering whether these can be defined as aliases to halves of the "x" register or something like that. Not as real registers seen by the allocator. Maciej ^ permalink raw reply [flat|nested] 7+ messages in thread
[parent not found: <878wxtvarg.fsf@firetop.home>]
* Re: Changing the treatment of the MIPS HI and LO registers [not found] ` <878wxtvarg.fsf@firetop.home> @ 2008-06-01 13:48 ` Richard Sandiford 2008-06-09 21:13 ` Richard Sandiford 0 siblings, 1 reply; 7+ messages in thread From: Richard Sandiford @ 2008-06-01 13:48 UTC (permalink / raw) To: Maciej W. Rozycki; +Cc: Ralf Baechle, gcc-patches, linux-mips This thread seems to have stagnated (if a thread can stagnate). Maciej said: > OK, it is good to have a way to make use of some TImode functionality > 64-bit hardware provides, but it still does not provide a clean solution > to the compatibility problem seen here. While a construct like: > > #if (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 4) > [new wonderful TImode stuff] > #else > [old compatibility asm crap] > #endif > > will certainly work, it is as ugly as you can get. I am not sure I would > be happy to see it in Linux, though obviously Ralf and the others may > disagree. but TBH, I don't see why it's so ugly. I imagine the definitions of int128_t and uint128_t (or whatever you decide to call them) would use the condition above. Then there'd be a macro to say whether the TImode stuff is available. I.e.: #if (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 4) typedef ... int128_t ...; typedef ... uint128_t ...; #define WE_HAVE_TIMODE 1 /* replace with a good name */ #endif So there'd only be one copy of the __GNUC__ check. The uses would simply be: #if WE_HAVE_TIMODE ... #else ... #endif That seems no worse than the kind of preprocessor condition you find all over the Linux source. Of course, the other alternative is simply to put the mfhi in the asm itself. While using "h" should in theory produce slightly better code, the difference is probably marginal (or at least accidental) in practice. GCC doesn't know that the asm performs a multiplication or division, so it isn't necessarily going to defer the mfhi for as long as you might like. (Better scheduling is of course another reason why the TImode version ought to be better. A third reason is that it would allow the compiler to constant-fold the result for constant delays periods; as the comment says, this currently only happens for the initialisation of usecs, not the multiplication itself.) I've attached an updated patch below. Differences from the last time are: - I've fixed the movti patterns for MIPS16. - I've added -mfix-r4000 support to the new TImode patterns. Also, now that the highpart patterns store directly into a GPR, it's easy to extend them to -mfix-r4000 too. - I've added a load of -mfix-r4000 tests. - I've added the promised TImode testcases. They cover addition, subtraction and logic as well as multiplication: - timode-1.c makes sure that everything is open-coded - timode-2.c makes sure that execution succeeds - int-moves-[12].c shadow the movtf tests I added earlier today Tested on mips64el-linux-gnu and mipsisa64-elfoabi. I'll reinstate the 48 hour warning, mostly as a way of seeing whether the thread is still alive. I don't think my last message made it through to linux-mips@. If this one does, and you're wanting some context, the discussion started here: http://gcc.gnu.org/ml/gcc-patches/2008-05/msg01750.html Basically, I'm trying to fix a wrong-code problem that would become especially dangerous with GCC's new register allocator. The only acceptable way I can see of doing it involves removing support for the asm "h" constraint. ("hi" would still be supported in clobber lists.) Richard gcc/ * doc/md.texi: Synchronize with later constraints.md change. * longlong.h (umul_ppmm): Replace the MIPS asm implementation with a C implementation. * config/mips/mips.c (mips_legitimize_move): Remove MFHI and MFLO handling. (mips_subword): Assume TImode for CONST_INTs if TARGET_64BIT. (mips_split_doubleword_move): Use special MTHI and MFHI instructions when moving to and from MD_REGNUM. (mips_output_move): Don't handle moves from GPRs to HI_REGNUM. Handle moves from LO_REGNUM to GPRs using MFLO, MACC or DMACC. Handle byte and halfword moves. (mips_hard_regno_mode_ok_p): Handle MD_REGS and DSP_ACC_REGS separately. * config/mips/constraints.md (h): Turn into NO_REGS. (l, x): Update documentation. * config/mips/mips.md (UNSPEC_MFHILO): Delete. (UNSPEC_MFHI, UNSPEC_MTHI, UNSPEC_SET_HILO): New. (UNSPEC_TLS_LDM, UNSPEC_TLS_GET_TP): Renumber. (HILO): New mode iterator. (MOVE128): Add TI. (any_div): New code iterator. (u): Extend code attribute to div and udiv. (*add<mode>3_mips16, *movdi_64bit_mips16, *movsi_mips16): Use d_operand in the splitters. Remove redundant CONST_INT checks. (mulsi3_mult3, mul<mode>3_internal, mul<mode>3_r4000, *mul_acc_si) (*macc, *msac, *msac_using_macc, *macc2, *msac2, *mul_sub_si) (*muls): Remove "=h" clobbers. Adjust peephole2s and define_splits accordingly, using normal moves instead of unspecs to move LO into a GPR. Use d_operand and lo_operand instead of *_REG_P checks. (<u>mulsidi3): Handle expansion in C code. (<u>mulsidi3_32bit_internal): Rename to... (<u>mulsidi3_32bit): ...this. (<u>mulsidi3_32bit_r4000): Fix insn separator. (*<u>mulsidi3_64bit): Rename to... (<u>mulsidi3_64bit): ...this. Combine DImode "=h" and "=l" clobbers into a TImode "=x" clobber. In the split, use an UNSPEC_SET_HILO to set LO and HI to the multiplication result. Use a normal move for MFLO and an unspec for MFHI. (*<u>mulsidi3_64bit_parts): Replace with... (<u>mulsidi3_64bit_hilo): ...this new instruction. (<su>mulsi3_highpart): Extend to TARGET_FIX_R4000. (<su>mulsi3_highpart_internal): Turn into a define_insn_and_split and extend it to TARGET_FIX_R4000. Store the destination in a GPR instead of HI. Split the instruction into a separate multiplication and MFHI if !TARGET_FIX_R4000. (<su>muldi3_highpart): Likewise. (<su>mulsi3_highpart_mulhi_internal): Remove the first alternative and the "=h" clobber. (*<su>mulsi3_highpart_neg_mulhi_internal): Likewise. (<u>mulditi3): New expander. (<u>mulditi3_internal, <u>mulditi3_r4000): New patterns. (madsi): Remove "=h" clobber. (divmod<mode>4, udivmod<mode>4): Turn into define_insn_and_splits. Force the modulus result to be a GPR and split the instruction into a division followed by an MFHI after reload. (<u>divmod<GPR:mode>4_hilo_<HILO:mode>): New instruction. (*lea_high64): Use d_operand in the define_peephole2. Likewise the MIPS16 HIGH define_split. (*movdi_32bit, *movdi_gp32_fp64, *movdi_32bit_mips16): Change type of acc<->gpr moves to "multi". (*movdi_64bit): Replace the single "x" alternative with alternatives for moving into and out of "a". (*movhi_internal, *movqi_internal): Likewise. Use mips_output_move. (*movsi_internal): Extend the "d<-A" alternative to "d<-a". (*movdi_64bit_mips16, *movsi_mips16): Add d<-a alternatives. Use d_operand in the splitters. Remove redundant CONST_INT checks. (*movhi_mips16, *movqi_mips16): Likewise. Use mips_output_move. (movti): New expander. (*movti, *movti_mips16): New insns. (mfhilo_<mode>, *mfhilo_<mode>, *mfhilo_<mode>_macc): Delete. (mfhi<GPR:mode>_<HILO:mode>): New pattern. (mthi<GPR:mode>_<HILO:mode>): Likewise. * config/mips/predicates.md (fpr_operand): Delete. (d_operand): New predicate. gcc/testsuite/ * gcc.dg/torture/mips-hilo-1.c: Delete. * gcc.target/mips/pr35232.c: Likewise. * gcc.target/mips/fix-vr4130-1.c: Use modulus to create an mfhi. * gcc.target/mips/fix-vr4130-3.c: Likewise. * gcc.target/mips/int-moves-1.c: New test. * gcc.target/mips/int-moves-2.c: Likewise. * gcc.target/mips/fix-r4000-1.c: Likewise. * gcc.target/mips/fix-r4000-2.c: Likewise. * gcc.target/mips/fix-r4000-3.c: Likewise. * gcc.target/mips/fix-r4000-4.c: Likewise. * gcc.target/mips/fix-r4000-5.c: Likewise. * gcc.target/mips/fix-r4000-6.c: Likewise. * gcc.target/mips/fix-r4000-7.c: Likewise. * gcc.target/mips/fix-r4000-8.c: Likewise. * gcc.target/mips/fix-r4000-9.c: Likewise. * gcc.target/mips/fix-r4000-10.c: Likewise. * gcc.target/mips/fix-r4000-11.c: Likewise. * gcc.target/mips/fix-r4000-12.c: Likewise. * gcc.target/mips/timode-1.c: Likewise. * gcc.target/mips/timode-2.c: Likewise. Index: gcc/doc/md.texi =================================================================== --- gcc/doc/md.texi 2008-06-01 10:23:31.000000000 +0100 +++ gcc/doc/md.texi 2008-06-01 14:08:13.000000000 +0100 @@ -2498,13 +2498,15 @@ generating MIPS16 code. A floating-point register (if available). @item h -The @code{hi} register. +Formerly the @code{hi} register. This constraint is no longer supported. @item l -The @code{lo} register. +The @code{lo} register. Use this register to store values that are +no bigger than a word. @item x -The @code{hi} and @code{lo} registers. +The concatenated @code{hi} and @code{lo} registers. Use this register +to store doubleword values. @item c A register suitable for use in an indirect jump. This will always be Index: gcc/longlong.h =================================================================== --- gcc/longlong.h 2008-06-01 10:23:31.000000000 +0100 +++ gcc/longlong.h 2008-06-01 14:06:06.000000000 +0100 @@ -623,12 +623,12 @@ #define UDIV_TIME 150 #endif /* __m88000__ */ #if defined (__mips__) && W_TYPE_SIZE == 32 -#define umul_ppmm(w1, w0, u, v) \ - __asm__ ("multu %2,%3" \ - : "=l" ((USItype) (w0)), \ - "=h" ((USItype) (w1)) \ - : "d" ((USItype) (u)), \ - "d" ((USItype) (v))) +#define umul_ppmm(w1, w0, u, v) \ + do { \ + UDItype __x = (UDItype) (USItype) (u) * (USItype) (v); \ + (w1) = (USItype) (__x >> 32); \ + (w0) = (USItype) (__x); \ + } while (0) #define UMUL_TIME 10 #define UDIV_TIME 100 Index: gcc/config/mips/mips.c =================================================================== --- gcc/config/mips/mips.c 2008-06-01 14:01:19.000000000 +0100 +++ gcc/config/mips/mips.c 2008-06-01 14:06:06.000000000 +0100 @@ -2659,23 +2659,6 @@ mips_legitimize_move (enum machine_mode return true; } - /* Check for individual, fully-reloaded mflo and mfhi instructions. */ - if (GET_MODE_SIZE (mode) <= UNITS_PER_WORD - && REG_P (src) && MD_REG_P (REGNO (src)) - && REG_P (dest) && GP_REG_P (REGNO (dest))) - { - int other_regno = REGNO (src) == HI_REGNUM ? LO_REGNUM : HI_REGNUM; - if (GET_MODE_SIZE (mode) <= 4) - emit_insn (gen_mfhilo_si (gen_lowpart (SImode, dest), - gen_lowpart (SImode, src), - gen_rtx_REG (SImode, other_regno))); - else - emit_insn (gen_mfhilo_di (gen_lowpart (DImode, dest), - gen_lowpart (DImode, src), - gen_rtx_REG (DImode, other_regno))); - return true; - } - /* We need to deal with constants that would be legitimate immediate_operands but aren't legitimate move_operands. */ if (CONSTANT_P (src) && !move_operand (src, mode)) @@ -3472,7 +3455,7 @@ mips_subword (rtx op, bool high_p) mode = GET_MODE (op); if (mode == VOIDmode) - mode = DImode; + mode = TARGET_64BIT ? TImode : DImode; if (TARGET_BIG_ENDIAN ? !high_p : high_p) byte = UNITS_PER_WORD; @@ -3523,6 +3506,8 @@ mips_split_64bit_move_p (rtx dest, rtx s void mips_split_doubleword_move (rtx dest, rtx src) { + rtx low_dest; + if (FP_REG_RTX_P (dest) || FP_REG_RTX_P (src)) { if (!TARGET_64BIT && GET_MODE (dest) == DImode) @@ -3536,12 +3521,27 @@ mips_split_doubleword_move (rtx dest, rt else gcc_unreachable (); } + else if (REG_P (dest) && REGNO (dest) == MD_REG_FIRST) + { + low_dest = mips_subword (dest, false); + mips_emit_move (low_dest, mips_subword (src, false)); + if (TARGET_64BIT) + emit_insn (gen_mthidi_ti (dest, mips_subword (src, true), low_dest)); + else + emit_insn (gen_mthisi_di (dest, mips_subword (src, true), low_dest)); + } + else if (REG_P (src) && REGNO (src) == MD_REG_FIRST) + { + mips_emit_move (mips_subword (dest, false), mips_subword (src, false)); + if (TARGET_64BIT) + emit_insn (gen_mfhidi_ti (mips_subword (dest, true), src)); + else + emit_insn (gen_mfhisi_di (mips_subword (dest, true), src)); + } else { /* The operation can be split into two normal moves. Decide in which order to do them. */ - rtx low_dest; - low_dest = mips_subword (dest, false); if (REG_P (low_dest) && reg_overlap_mentioned_p (low_dest, src)) @@ -3584,8 +3584,9 @@ mips_output_move (rtx dest, rtx src) if (GP_REG_P (REGNO (dest))) return "move\t%0,%z1"; - if (MD_REG_P (REGNO (dest))) - return "mt%0\t%z1"; + /* Moves to HI are handled by special .md insns. */ + if (REGNO (dest) == LO_REGNUM) + return "mtlo\t%z1"; if (DSP_ACC_REG_P (REGNO (dest))) { @@ -3608,14 +3609,29 @@ mips_output_move (rtx dest, rtx src) } } if (dest_code == MEM) - return dbl_p ? "sd\t%z1,%0" : "sw\t%z1,%0"; + switch (GET_MODE_SIZE (mode)) + { + case 1: return "sb\t%z1,%0"; + case 2: return "sh\t%z1,%0"; + case 4: return "sw\t%z1,%0"; + case 8: return "sd\t%z1,%0"; + } } if (dest_code == REG && GP_REG_P (REGNO (dest))) { if (src_code == REG) { - /* Handled by separate patterns. */ - gcc_assert (!MD_REG_P (REGNO (src))); + /* Moves from HI are handled by special .md insns. */ + if (REGNO (src) == LO_REGNUM) + { + /* When generating VR4120 or VR4130 code, we use MACC and + DMACC instead of MFLO. This avoids both the normal + MIPS III HI/LO hazards and the errata related to + -mfix-vr4130. */ + if (ISA_HAS_MACCHI) + return dbl_p ? "dmacc\t%0,%.,%." : "macc\t%0,%.,%."; + return "mflo\t%0"; + } if (DSP_ACC_REG_P (REGNO (src))) { @@ -3642,7 +3658,13 @@ mips_output_move (rtx dest, rtx src) } if (src_code == MEM) - return dbl_p ? "ld\t%0,%1" : "lw\t%0,%1"; + switch (GET_MODE_SIZE (mode)) + { + case 1: return "lbu\t%0,%1"; + case 2: return "lhu\t%0,%1"; + case 4: return "lw\t%0,%1"; + case 8: return "ld\t%0,%1"; + } if (src_code == CONST_INT) { @@ -8937,13 +8959,30 @@ mips_hard_regno_mode_ok_p (unsigned int if (ACC_REG_P (regno) && (INTEGRAL_MODE_P (mode) || ALL_FIXED_POINT_MODE_P (mode))) { - if (size <= UNITS_PER_WORD) - return true; + if (MD_REG_P (regno)) + { + /* After a multiplication or division, clobbering HI makes + the value of LO unpredictable, and vice versa. This means + that, for all interesting cases, HI and LO are effectively + a single register. - if (size <= UNITS_PER_WORD * 2) - return (DSP_ACC_REG_P (regno) - ? ((regno - DSP_ACC_REG_FIRST) & 1) == 0 - : regno == MD_REG_FIRST); + We model this by requiring that any value that uses HI + also uses LO. */ + if (size <= UNITS_PER_WORD * 2) + return regno == (size <= UNITS_PER_WORD ? LO_REGNUM : MD_REG_FIRST); + } + else + { + /* DSP accumulators do not have the same restrictions as + HI and LO, so we can treat them as normal doubleword + registers. */ + if (size <= UNITS_PER_WORD) + return true; + + if (size <= UNITS_PER_WORD * 2 + && ((regno - DSP_ACC_REG_FIRST) & 1) == 0) + return true; + } } if (ALL_COP_REG_P (regno)) Index: gcc/config/mips/constraints.md =================================================================== --- gcc/config/mips/constraints.md 2008-06-01 10:23:31.000000000 +0100 +++ gcc/config/mips/constraints.md 2008-06-01 14:07:37.000000000 +0100 @@ -29,14 +29,16 @@ (define_register_constraint "t" "T_REG" (define_register_constraint "f" "TARGET_HARD_FLOAT ? FP_REGS : NO_REGS" "A floating-point register (if available).") -(define_register_constraint "h" "TARGET_BIG_ENDIAN ? MD0_REG : MD1_REG" - "The @code{hi} register.") +(define_register_constraint "h" "NO_REGS" + "Formerly the @code{hi} register. This constraint is no longer supported.") (define_register_constraint "l" "TARGET_BIG_ENDIAN ? MD1_REG : MD0_REG" - "The @code{lo} register.") + "The @code{lo} register. Use this register to store values that are + no bigger than a word.") (define_register_constraint "x" "MD_REGS" - "The @code{hi} and @code{lo} registers.") + "The concatenated @code{hi} and @code{lo} registers. Use this register + to store doubleword values.") (define_register_constraint "b" "ALL_REGS" "@internal") Index: gcc/config/mips/mips.md =================================================================== --- gcc/config/mips/mips.md 2008-06-01 14:01:19.000000000 +0100 +++ gcc/config/mips/mips.md 2008-06-01 14:16:33.000000000 +0100 @@ -44,9 +44,11 @@ (define_constants (UNSPEC_LOAD_CALL 23) (UNSPEC_LOAD_GOT 24) (UNSPEC_GP 25) - (UNSPEC_MFHILO 26) - (UNSPEC_TLS_LDM 27) - (UNSPEC_TLS_GET_TP 28) + (UNSPEC_MFHI 26) + (UNSPEC_MTHI 27) + (UNSPEC_SET_HILO 28) + (UNSPEC_TLS_LDM 29) + (UNSPEC_TLS_GET_TP 30) (UNSPEC_MFHC1 31) (UNSPEC_MTHC1 32) (UNSPEC_CLEAR_HAZARD 33) @@ -484,6 +486,10 @@ (define_mode_iterator GPR [SI (DI "TARGE ;; modes. (define_mode_iterator GPR2 [SI (DI "TARGET_64BIT")]) +;; This mode iterator allows :HILO to be used as the mode of the +;; concatenated HI and LO registers. +(define_mode_iterator HILO [(DI "!TARGET_64BIT") (TI "TARGET_64BIT")]) + ;; This mode iterator allows :P to be used for patterns that operate on ;; pointer-sized quantities. Exactly one of the two alternatives will match. (define_mode_iterator P [(SI "Pmode == SImode") (DI "Pmode == DImode")]) @@ -497,7 +503,7 @@ (define_mode_iterator MOVE64 [DI DF (V2SF "TARGET_HARD_FLOAT && TARGET_PAIRED_SINGLE_FLOAT")]) ;; 128-bit modes for which we provide move patterns on 64-bit targets. -(define_mode_iterator MOVE128 [TF]) +(define_mode_iterator MOVE128 [TI TF]) ;; This mode iterator allows the QI and HI extension patterns to be ;; defined from the same template. @@ -613,6 +619,10 @@ (define_code_iterator any_extend [sign_e ;; from the same template. (define_code_iterator any_shift [ashift ashiftrt lshiftrt]) +;; This code iterator allows unsigned and signed division to be generated +;; from the same template. +(define_code_iterator any_div [div udiv]) + ;; This code iterator allows all native floating-point comparisons to be ;; generated from the same template. (define_code_iterator fcond [unordered uneq unlt unle eq lt le]) @@ -631,6 +641,7 @@ (define_code_iterator any_le [le leu]) ;; <u> expands to an empty string when doing a signed operation and ;; "u" when doing an unsigned operation. (define_code_attr u [(sign_extend "") (zero_extend "u") + (div "") (udiv "u") (gt "") (gtu "u") (ge "") (geu "u") (lt "") (ltu "u") @@ -864,13 +875,10 @@ (define_insn "*add<mode>3_mips16" ;; simply adding a constant to a register. (define_split - [(set (match_operand:SI 0 "register_operand") + [(set (match_operand:SI 0 "d_operand") (plus:SI (match_dup 0) (match_operand:SI 1 "const_int_operand")))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) > 0x7f && INTVAL (operands[1]) <= 0x7f + 0x7f) || (INTVAL (operands[1]) < - 0x80 @@ -893,16 +901,11 @@ (define_split }) (define_split - [(set (match_operand:SI 0 "register_operand") - (plus:SI (match_operand:SI 1 "register_operand") + [(set (match_operand:SI 0 "d_operand") + (plus:SI (match_operand:SI 1 "d_operand") (match_operand:SI 2 "const_int_operand")))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && REG_P (operands[1]) - && M16_REG_P (REGNO (operands[1])) && REGNO (operands[0]) != REGNO (operands[1]) - && GET_CODE (operands[2]) == CONST_INT && ((INTVAL (operands[2]) > 0x7 && INTVAL (operands[2]) <= 0x7 + 0x7f) || (INTVAL (operands[2]) < - 0x8 @@ -925,13 +928,10 @@ (define_split }) (define_split - [(set (match_operand:DI 0 "register_operand") + [(set (match_operand:DI 0 "d_operand") (plus:DI (match_dup 0) (match_operand:DI 1 "const_int_operand")))] "TARGET_MIPS16 && TARGET_64BIT && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) > 0xf && INTVAL (operands[1]) <= 0xf + 0xf) || (INTVAL (operands[1]) < - 0x10 @@ -954,16 +954,11 @@ (define_split }) (define_split - [(set (match_operand:DI 0 "register_operand") - (plus:DI (match_operand:DI 1 "register_operand") + [(set (match_operand:DI 0 "d_operand") + (plus:DI (match_operand:DI 1 "d_operand") (match_operand:DI 2 "const_int_operand")))] "TARGET_MIPS16 && TARGET_64BIT && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && REG_P (operands[1]) - && M16_REG_P (REGNO (operands[1])) && REGNO (operands[0]) != REGNO (operands[1]) - && GET_CODE (operands[2]) == CONST_INT && ((INTVAL (operands[2]) > 0x7 && INTVAL (operands[2]) <= 0x7 + 0xf) || (INTVAL (operands[2]) < - 0x8 @@ -1174,8 +1169,7 @@ (define_insn "mulsi3_mult3" [(set (match_operand:SI 0 "register_operand" "=d,l") (mult:SI (match_operand:SI 1 "register_operand" "d,d") (match_operand:SI 2 "register_operand" "d,d"))) - (clobber (match_scratch:SI 3 "=h,h")) - (clobber (match_scratch:SI 4 "=l,X"))] + (clobber (match_scratch:SI 3 "=l,X"))] "ISA_HAS_MUL3" { if (which_alternative == 1) @@ -1194,30 +1188,26 @@ (define_insn "mulsi3_mult3" ;; Operand 0: LO ;; Operand 1: GPR (1st multiplication operand) ;; Operand 2: GPR (2nd multiplication operand) -;; Operand 3: HI -;; Operand 4: GPR (destination) +;; Operand 3: GPR (destination) (define_peephole2 [(parallel - [(set (match_operand:SI 0 "register_operand") - (mult:SI (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand"))) - (clobber (match_operand:SI 3 "register_operand")) + [(set (match_operand:SI 0 "lo_operand") + (mult:SI (match_operand:SI 1 "d_operand") + (match_operand:SI 2 "d_operand"))) (clobber (scratch:SI))]) - (set (match_operand:SI 4 "register_operand") - (unspec [(match_dup 0) (match_dup 3)] UNSPEC_MFHILO))] + (set (match_operand:SI 3 "d_operand") + (match_dup 0))] "ISA_HAS_MUL3 && peep2_reg_dead_p (2, operands[0])" [(parallel - [(set (match_dup 4) + [(set (match_dup 3) (mult:SI (match_dup 1) (match_dup 2))) - (clobber (match_dup 3)) (clobber (match_dup 0))])]) (define_insn "mul<mode>3_internal" [(set (match_operand:GPR 0 "register_operand" "=l") (mult:GPR (match_operand:GPR 1 "register_operand" "d") - (match_operand:GPR 2 "register_operand" "d"))) - (clobber (match_scratch:GPR 3 "=h"))] + (match_operand:GPR 2 "register_operand" "d")))] "!TARGET_FIX_R4000" "<d>mult\t%1,%2" [(set_attr "type" "imul") @@ -1227,8 +1217,7 @@ (define_insn "mul<mode>3_r4000" [(set (match_operand:GPR 0 "register_operand" "=d") (mult:GPR (match_operand:GPR 1 "register_operand" "d") (match_operand:GPR 2 "register_operand" "d"))) - (clobber (match_scratch:GPR 3 "=h")) - (clobber (match_scratch:GPR 4 "=l"))] + (clobber (match_scratch:GPR 3 "=l"))] "TARGET_FIX_R4000" "<d>mult\t%1,%2\;mflo\t%0" [(set_attr "type" "imul") @@ -1242,16 +1231,13 @@ (define_insn "mul<mode>3_r4000" ;; Operand 0: LO ;; Operand 1: GPR (1st multiplication operand) ;; Operand 2: GPR (2nd multiplication operand) -;; Operand 3: HI -;; Operand 4: GPR (destination) +;; Operand 3: GPR (destination) (define_peephole2 - [(parallel - [(set (match_operand:SI 0 "register_operand") - (mult:SI (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand"))) - (clobber (match_operand:SI 3 "register_operand"))]) - (set (match_operand:SI 4 "register_operand") - (unspec:SI [(match_dup 0) (match_dup 3)] UNSPEC_MFHILO))] + [(set (match_operand:SI 0 "lo_operand") + (mult:SI (match_operand:SI 1 "d_operand") + (match_operand:SI 2 "d_operand"))) + (set (match_operand:SI 3 "d_operand") + (match_dup 0))] "ISA_HAS_MACC && !ISA_HAS_MUL3" [(set (match_dup 0) (const_int 0)) @@ -1260,11 +1246,10 @@ (define_peephole2 (plus:SI (mult:SI (match_dup 1) (match_dup 2)) (match_dup 0))) - (set (match_dup 4) + (set (match_dup 3) (plus:SI (mult:SI (match_dup 1) (match_dup 2)) - (match_dup 0))) - (clobber (match_dup 3))])]) + (match_dup 0)))])]) ;; Multiply-accumulate patterns @@ -1283,9 +1268,8 @@ (define_insn "*mul_acc_si" (plus:SI (mult:SI (match_operand:SI 1 "register_operand" "d,d,d") (match_operand:SI 2 "register_operand" "d,d,d")) (match_operand:SI 3 "register_operand" "0,l,*d"))) - (clobber (match_scratch:SI 4 "=h,h,h")) - (clobber (match_scratch:SI 5 "=X,3,l")) - (clobber (match_scratch:SI 6 "=X,X,&d"))] + (clobber (match_scratch:SI 4 "=X,3,l")) + (clobber (match_scratch:SI 5 "=X,X,&d"))] "(TARGET_MIPS3900 || GENERATE_MADD_MSUB) && !TARGET_MIPS16" @@ -1301,53 +1285,45 @@ (define_insn "*mul_acc_si" (set_attr "mode" "SI") (set_attr "length" "4,4,8")]) -;; Split the above insn if we failed to get LO allocated. +;; Split *mul_acc_si if both the source and destination accumulator +;; values are GPRs. (define_split - [(set (match_operand:SI 0 "register_operand") - (plus:SI (mult:SI (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand")) - (match_operand:SI 3 "register_operand"))) - (clobber (match_scratch:SI 4)) - (clobber (match_scratch:SI 5)) - (clobber (match_scratch:SI 6))] - "reload_completed && !TARGET_DEBUG_D_MODE - && GP_REG_P (true_regnum (operands[0])) - && GP_REG_P (true_regnum (operands[3]))" - [(parallel [(set (match_dup 6) + [(set (match_operand:SI 0 "d_operand") + (plus:SI (mult:SI (match_operand:SI 1 "d_operand") + (match_operand:SI 2 "d_operand")) + (match_operand:SI 3 "d_operand"))) + (clobber (match_operand:SI 4 "lo_operand")) + (clobber (match_operand:SI 5 "d_operand"))] + "reload_completed && !TARGET_DEBUG_D_MODE" + [(parallel [(set (match_dup 5) (mult:SI (match_dup 1) (match_dup 2))) - (clobber (match_dup 4)) - (clobber (match_dup 5))]) - (set (match_dup 0) (plus:SI (match_dup 6) (match_dup 3)))] + (clobber (match_dup 4))]) + (set (match_dup 0) (plus:SI (match_dup 5) (match_dup 3)))] "") -;; Splitter to copy result of MADD to a general register +;; Split *mul_acc_si if the destination accumulator value is in a GPR +;; and the source accumulator value is in LO. (define_split - [(set (match_operand:SI 0 "register_operand") - (plus:SI (mult:SI (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand")) - (match_operand:SI 3 "register_operand"))) - (clobber (match_scratch:SI 4)) - (clobber (match_scratch:SI 5)) - (clobber (match_scratch:SI 6))] - "reload_completed && !TARGET_DEBUG_D_MODE - && GP_REG_P (true_regnum (operands[0])) - && true_regnum (operands[3]) == LO_REGNUM" + [(set (match_operand:SI 0 "d_operand") + (plus:SI (mult:SI (match_operand:SI 1 "d_operand") + (match_operand:SI 2 "d_operand")) + (match_operand:SI 3 "lo_operand"))) + (clobber (match_dup 3)) + (clobber (scratch:SI))] + "reload_completed && !TARGET_DEBUG_D_MODE" [(parallel [(set (match_dup 3) (plus:SI (mult:SI (match_dup 1) (match_dup 2)) (match_dup 3))) - (clobber (match_dup 4)) - (clobber (match_dup 5)) - (clobber (match_dup 6))]) - (set (match_dup 0) (unspec:SI [(match_dup 5) (match_dup 4)] UNSPEC_MFHILO))] - "") + (clobber (scratch:SI)) + (clobber (scratch:SI))]) + (set (match_dup 0) (match_dup 3))]) (define_insn "*macc" [(set (match_operand:SI 0 "register_operand" "=l,d") (plus:SI (mult:SI (match_operand:SI 1 "register_operand" "d,d") (match_operand:SI 2 "register_operand" "d,d")) (match_operand:SI 3 "register_operand" "0,l"))) - (clobber (match_scratch:SI 4 "=h,h")) - (clobber (match_scratch:SI 5 "=X,3"))] + (clobber (match_scratch:SI 4 "=X,3"))] "ISA_HAS_MACC" { if (which_alternative == 1) @@ -1368,8 +1344,7 @@ (define_insn "*msac" (minus:SI (match_operand:SI 1 "register_operand" "0,l") (mult:SI (match_operand:SI 2 "register_operand" "d,d") (match_operand:SI 3 "register_operand" "d,d")))) - (clobber (match_scratch:SI 4 "=h,h")) - (clobber (match_scratch:SI 5 "=X,1"))] + (clobber (match_scratch:SI 4 "=X,1"))] "ISA_HAS_MSAC" { if (which_alternative == 1) @@ -1388,21 +1363,19 @@ (define_insn_and_split "*msac_using_macc (minus:SI (match_operand:SI 1 "register_operand" "0,l") (mult:SI (match_operand:SI 2 "register_operand" "d,d") (match_operand:SI 3 "register_operand" "d,d")))) - (clobber (match_scratch:SI 4 "=h,h")) - (clobber (match_scratch:SI 5 "=X,1")) - (clobber (match_scratch:SI 6 "=d,d"))] + (clobber (match_scratch:SI 4 "=X,1")) + (clobber (match_scratch:SI 5 "=d,d"))] "ISA_HAS_MACC && !ISA_HAS_MSAC" "#" "&& reload_completed" - [(set (match_dup 6) + [(set (match_dup 5) (neg:SI (match_dup 3))) (parallel [(set (match_dup 0) (plus:SI (mult:SI (match_dup 2) - (match_dup 6)) + (match_dup 5)) (match_dup 1))) - (clobber (match_dup 4)) - (clobber (match_dup 5))])] + (clobber (match_dup 4))])] "" [(set_attr "type" "imadd") (set_attr "length" "8")]) @@ -1417,8 +1390,7 @@ (define_insn "*macc2" (set (match_operand:SI 3 "register_operand" "=d") (plus:SI (mult:SI (match_dup 1) (match_dup 2)) - (match_dup 0))) - (clobber (match_scratch:SI 4 "=h"))] + (match_dup 0)))] "ISA_HAS_MACC && reload_completed" "macc\t%3,%1,%2" [(set_attr "type" "imadd") @@ -1432,8 +1404,7 @@ (define_insn "*msac2" (set (match_operand:SI 3 "register_operand" "=d") (minus:SI (match_dup 0) (mult:SI (match_dup 1) - (match_dup 2)))) - (clobber (match_scratch:SI 4 "=h"))] + (match_dup 2))))] "ISA_HAS_MSAC && reload_completed" "msac\t%3,%1,%2" [(set_attr "type" "imadd") @@ -1444,23 +1415,19 @@ (define_insn "*msac2" ;; ;; Operand 0: LO ;; Operand 1: macc/msac -;; Operand 2: HI -;; Operand 3: GPR (destination) +;; Operand 2: GPR (destination) (define_peephole2 [(parallel - [(set (match_operand:SI 0 "register_operand") + [(set (match_operand:SI 0 "lo_operand") (match_operand:SI 1 "macc_msac_operand")) - (clobber (match_operand:SI 2 "register_operand")) (clobber (scratch:SI))]) - (set (match_operand:SI 3 "register_operand") - (unspec:SI [(match_dup 0) (match_dup 2)] UNSPEC_MFHILO))] + (set (match_operand:SI 2 "d_operand") + (match_dup 0))] "" [(parallel [(set (match_dup 0) (match_dup 1)) - (set (match_dup 3) - (match_dup 1)) - (clobber (match_dup 2))])] - "") + (set (match_dup 2) + (match_dup 1))])]) ;; When we have a three-address multiplication instruction, it should ;; be faster to do a separate multiply and add, rather than moving @@ -1474,32 +1441,26 @@ (define_peephole2 ;; Operand 2: GPR (addend) ;; Operand 3: GPR (destination) ;; Operand 4: macc/msac -;; Operand 5: HI -;; Operand 6: new multiplication -;; Operand 7: new addition/subtraction +;; Operand 5: new multiplication +;; Operand 6: new addition/subtraction (define_peephole2 [(match_scratch:SI 0 "d") - (set (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand")) + (set (match_operand:SI 1 "lo_operand") + (match_operand:SI 2 "d_operand")) (match_dup 0) (parallel - [(set (match_operand:SI 3 "register_operand") + [(set (match_operand:SI 3 "d_operand") (match_operand:SI 4 "macc_msac_operand")) - (clobber (match_operand:SI 5 "register_operand")) (clobber (match_dup 1))])] - "ISA_HAS_MUL3 - && true_regnum (operands[1]) == LO_REGNUM - && peep2_reg_dead_p (2, operands[1]) - && GP_REG_P (true_regnum (operands[3]))" + "ISA_HAS_MUL3 && peep2_reg_dead_p (2, operands[1])" [(parallel [(set (match_dup 0) - (match_dup 6)) - (clobber (match_dup 5)) + (match_dup 5)) (clobber (match_dup 1))]) (set (match_dup 3) - (match_dup 7))] + (match_dup 6))] { - operands[6] = XEXP (operands[4], GET_CODE (operands[4]) == PLUS ? 0 : 1); - operands[7] = gen_rtx_fmt_ee (GET_CODE (operands[4]), SImode, + operands[5] = XEXP (operands[4], GET_CODE (operands[4]) == PLUS ? 0 : 1); + operands[6] = gen_rtx_fmt_ee (GET_CODE (operands[4]), SImode, operands[2], operands[0]); }) @@ -1509,33 +1470,30 @@ (define_peephole2 ;; Operand 1: LO ;; Operand 2: GPR (addend) ;; Operand 3: macc/msac -;; Operand 4: HI -;; Operand 5: GPR (destination) -;; Operand 6: new multiplication -;; Operand 7: new addition/subtraction +;; Operand 4: GPR (destination) +;; Operand 5: new multiplication +;; Operand 6: new addition/subtraction (define_peephole2 [(match_scratch:SI 0 "d") - (set (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand")) + (set (match_operand:SI 1 "lo_operand") + (match_operand:SI 2 "d_operand")) (match_dup 0) (parallel [(set (match_dup 1) (match_operand:SI 3 "macc_msac_operand")) - (clobber (match_operand:SI 4 "register_operand")) (clobber (scratch:SI))]) (match_dup 0) - (set (match_operand:SI 5 "register_operand") - (unspec:SI [(match_dup 1) (match_dup 4)] UNSPEC_MFHILO))] + (set (match_operand:SI 4 "d_operand") + (match_dup 1))] "ISA_HAS_MUL3 && peep2_reg_dead_p (3, operands[1])" [(parallel [(set (match_dup 0) - (match_dup 6)) - (clobber (match_dup 4)) + (match_dup 5)) (clobber (match_dup 1))]) - (set (match_dup 5) - (match_dup 7))] + (set (match_dup 4) + (match_dup 6))] { - operands[6] = XEXP (operands[4], GET_CODE (operands[4]) == PLUS ? 0 : 1); - operands[7] = gen_rtx_fmt_ee (GET_CODE (operands[4]), SImode, + operands[5] = XEXP (operands[3], GET_CODE (operands[3]) == PLUS ? 0 : 1); + operands[6] = gen_rtx_fmt_ee (GET_CODE (operands[3]), SImode, operands[2], operands[0]); }) @@ -1544,9 +1502,8 @@ (define_insn "*mul_sub_si" (minus:SI (match_operand:SI 1 "register_operand" "0,l,*d") (mult:SI (match_operand:SI 2 "register_operand" "d,d,d") (match_operand:SI 3 "register_operand" "d,d,d")))) - (clobber (match_scratch:SI 4 "=h,h,h")) - (clobber (match_scratch:SI 5 "=X,1,l")) - (clobber (match_scratch:SI 6 "=X,X,&d"))] + (clobber (match_scratch:SI 4 "=X,1,l")) + (clobber (match_scratch:SI 5 "=X,X,&d"))] "GENERATE_MADD_MSUB" "@ msub\t%2,%3 @@ -1556,52 +1513,45 @@ (define_insn "*mul_sub_si" (set_attr "mode" "SI") (set_attr "length" "4,8,8")]) -;; Split the above insn if we failed to get LO allocated. +;; Split *mul_sub_si if both the source and destination accumulator +;; values are GPRs. (define_split - [(set (match_operand:SI 0 "register_operand") - (minus:SI (match_operand:SI 1 "register_operand") - (mult:SI (match_operand:SI 2 "register_operand") - (match_operand:SI 3 "register_operand")))) - (clobber (match_scratch:SI 4)) - (clobber (match_scratch:SI 5)) - (clobber (match_scratch:SI 6))] - "reload_completed && !TARGET_DEBUG_D_MODE - && GP_REG_P (true_regnum (operands[0])) - && GP_REG_P (true_regnum (operands[1]))" - [(parallel [(set (match_dup 6) + [(set (match_operand:SI 0 "d_operand") + (minus:SI (match_operand:SI 1 "d_operand") + (mult:SI (match_operand:SI 2 "d_operand") + (match_operand:SI 3 "d_operand")))) + (clobber (match_operand:SI 4 "lo_operand")) + (clobber (match_operand:SI 5 "d_operand"))] + "reload_completed && !TARGET_DEBUG_D_MODE" + [(parallel [(set (match_dup 5) (mult:SI (match_dup 2) (match_dup 3))) - (clobber (match_dup 4)) - (clobber (match_dup 5))]) - (set (match_dup 0) (minus:SI (match_dup 1) (match_dup 6)))] + (clobber (match_dup 4))]) + (set (match_dup 0) (minus:SI (match_dup 1) (match_dup 5)))] "") -;; Splitter to copy result of MSUB to a general register +;; Split *mul_acc_si if the destination accumulator value is in a GPR +;; and the source accumulator value is in LO. (define_split - [(set (match_operand:SI 0 "register_operand") - (minus:SI (match_operand:SI 1 "register_operand") - (mult:SI (match_operand:SI 2 "register_operand") - (match_operand:SI 3 "register_operand")))) - (clobber (match_scratch:SI 4)) - (clobber (match_scratch:SI 5)) - (clobber (match_scratch:SI 6))] - "reload_completed && !TARGET_DEBUG_D_MODE - && GP_REG_P (true_regnum (operands[0])) - && true_regnum (operands[1]) == LO_REGNUM" + [(set (match_operand:SI 0 "d_operand") + (minus:SI (match_operand:SI 1 "lo_operand") + (mult:SI (match_operand:SI 2 "d_operand") + (match_operand:SI 3 "d_operand")))) + (clobber (match_dup 1)) + (clobber (scratch:SI))] + "reload_completed && !TARGET_DEBUG_D_MODE" [(parallel [(set (match_dup 1) (minus:SI (match_dup 1) (mult:SI (match_dup 2) (match_dup 3)))) - (clobber (match_dup 4)) - (clobber (match_dup 5)) - (clobber (match_dup 6))]) - (set (match_dup 0) (unspec:SI [(match_dup 5) (match_dup 4)] UNSPEC_MFHILO))] + (clobber (scratch:SI)) + (clobber (scratch:SI))]) + (set (match_dup 0) (match_dup 1))] "") (define_insn "*muls" - [(set (match_operand:SI 0 "register_operand" "=l,d") + [(set (match_operand:SI 0 "register_operand" "=l,d") (neg:SI (mult:SI (match_operand:SI 1 "register_operand" "d,d") (match_operand:SI 2 "register_operand" "d,d")))) - (clobber (match_scratch:SI 3 "=h,h")) - (clobber (match_scratch:SI 4 "=X,l"))] + (clobber (match_scratch:SI 3 "=X,l"))] "ISA_HAS_MULS" "@ muls\t$0,%1,%2 @@ -1609,31 +1559,23 @@ (define_insn "*muls" [(set_attr "type" "imul,imul3") (set_attr "mode" "SI")]) -;; ??? We could define a mulditi3 pattern when TARGET_64BIT. - (define_expand "<u>mulsidi3" - [(parallel - [(set (match_operand:DI 0 "register_operand") - (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand")) - (any_extend:DI (match_operand:SI 2 "register_operand")))) - (clobber (scratch:DI)) - (clobber (scratch:DI)) - (clobber (scratch:DI))])] + [(set (match_operand:DI 0 "register_operand") + (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand")) + (any_extend:DI (match_operand:SI 2 "register_operand"))))] "!TARGET_64BIT || !TARGET_FIX_R4000" { - if (!TARGET_64BIT) - { - if (!TARGET_FIX_R4000) - emit_insn (gen_<u>mulsidi3_32bit_internal (operands[0], operands[1], - operands[2])); - else - emit_insn (gen_<u>mulsidi3_32bit_r4000 (operands[0], operands[1], - operands[2])); - DONE; - } + if (TARGET_64BIT) + emit_insn (gen_<u>mulsidi3_64bit (operands[0], operands[1], operands[2])); + else if (TARGET_FIX_R4000) + emit_insn (gen_<u>mulsidi3_32bit_r4000 (operands[0], operands[1], + operands[2])); + else + emit_insn (gen_<u>mulsidi3_32bit (operands[0], operands[1], operands[2])); + DONE; }) -(define_insn "<u>mulsidi3_32bit_internal" +(define_insn "<u>mulsidi3_32bit" [(set (match_operand:DI 0 "register_operand" "=x") (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand" "d")) (any_extend:DI (match_operand:SI 2 "register_operand" "d"))))] @@ -1648,42 +1590,35 @@ (define_insn "<u>mulsidi3_32bit_r4000" (any_extend:DI (match_operand:SI 2 "register_operand" "d")))) (clobber (match_scratch:DI 3 "=x"))] "!TARGET_64BIT && TARGET_FIX_R4000" - "mult<u>\t%1,%2\;mflo\t%L0;mfhi\t%M0" + "mult<u>\t%1,%2\;mflo\t%L0\;mfhi\t%M0" [(set_attr "type" "imul") (set_attr "mode" "SI") (set_attr "length" "12")]) -(define_insn_and_split "*<u>mulsidi3_64bit" +(define_insn_and_split "<u>mulsidi3_64bit" [(set (match_operand:DI 0 "register_operand" "=d") (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand" "d")) (any_extend:DI (match_operand:SI 2 "register_operand" "d")))) - (clobber (match_scratch:DI 3 "=l")) - (clobber (match_scratch:DI 4 "=h")) - (clobber (match_scratch:DI 5 "=d"))] + (clobber (match_scratch:TI 3 "=x")) + (clobber (match_scratch:DI 4 "=d"))] "TARGET_64BIT && !TARGET_FIX_R4000" "#" "&& reload_completed" - [(parallel - [(set (match_dup 3) - (sign_extend:DI - (mult:SI (match_dup 1) - (match_dup 2)))) - (set (match_dup 4) - (ashiftrt:DI - (mult:DI (any_extend:DI (match_dup 1)) - (any_extend:DI (match_dup 2))) - (const_int 32)))]) - - ;; OP5 <- LO, OP0 <- HI - (set (match_dup 5) (unspec:DI [(match_dup 3) (match_dup 4)] UNSPEC_MFHILO)) - (set (match_dup 0) (unspec:DI [(match_dup 4) (match_dup 3)] UNSPEC_MFHILO)) - - ;; Zero-extend OP5. - (set (match_dup 5) - (ashift:DI (match_dup 5) + [(set (match_dup 3) + (unspec:TI [(mult:DI (any_extend:DI (match_dup 1)) + (any_extend:DI (match_dup 2)))] + UNSPEC_SET_HILO)) + + ;; OP4 <- LO, OP0 <- HI + (set (match_dup 4) (match_dup 5)) + (set (match_dup 0) (unspec:DI [(match_dup 3)] UNSPEC_MFHI)) + + ;; Zero-extend OP4. + (set (match_dup 4) + (ashift:DI (match_dup 4) (const_int 32))) - (set (match_dup 5) - (lshiftrt:DI (match_dup 5) + (set (match_dup 4) + (lshiftrt:DI (match_dup 4) (const_int 32))) ;; Shift OP0 into place. @@ -1694,24 +1629,21 @@ (define_insn_and_split "*<u>mulsidi3_64b ;; OR the two halves together (set (match_dup 0) (ior:DI (match_dup 0) - (match_dup 5)))] - "" + (match_dup 4)))] + { operands[5] = gen_rtx_REG (DImode, LO_REGNUM); } [(set_attr "type" "imul") (set_attr "mode" "SI") (set_attr "length" "24")]) -(define_insn "*<u>mulsidi3_64bit_parts" - [(set (match_operand:DI 0 "register_operand" "=l") - (sign_extend:DI - (mult:SI (match_operand:SI 2 "register_operand" "d") - (match_operand:SI 3 "register_operand" "d")))) - (set (match_operand:DI 1 "register_operand" "=h") - (ashiftrt:DI - (mult:DI (any_extend:DI (match_dup 2)) - (any_extend:DI (match_dup 3))) - (const_int 32)))] +(define_insn "<u>mulsidi3_64bit_hilo" + [(set (match_operand:TI 0 "register_operand" "=x") + (unspec:TI + [(mult:DI + (any_extend:DI (match_operand:SI 1 "register_operand" "d")) + (any_extend:DI (match_operand:SI 2 "register_operand" "d")))] + UNSPEC_SET_HILO))] "TARGET_64BIT && !TARGET_FIX_R4000" - "mult<u>\t%2,%3" + "mult<u>\t%1,%2" [(set_attr "type" "imul") (set_attr "mode" "SI")]) @@ -1755,7 +1687,7 @@ (define_expand "<su>mulsi3_highpart" (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand")) (any_extend:DI (match_operand:SI 2 "register_operand"))) (const_int 32))))] - "ISA_HAS_MULHI || !TARGET_FIX_R4000" + "ISA_HAS_MULHI" { if (ISA_HAS_MULHI) emit_insn (gen_<su>mulsi3_highpart_mulhi_internal (operands[0], @@ -1767,72 +1699,133 @@ (define_expand "<su>mulsi3_highpart" DONE; }) -(define_insn "<su>mulsi3_highpart_internal" - [(set (match_operand:SI 0 "register_operand" "=h") +(define_insn_and_split "<su>mulsi3_highpart_internal" + [(set (match_operand:SI 0 "register_operand" "=d") (truncate:SI (lshiftrt:DI (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand" "d")) (any_extend:DI (match_operand:SI 2 "register_operand" "d"))) (const_int 32)))) (clobber (match_scratch:SI 3 "=l"))] - "!ISA_HAS_MULHI && !TARGET_FIX_R4000" - "mult<u>\t%1,%2" + "!ISA_HAS_MULHI" + { return TARGET_FIX_R4000 ? "mult<u>\t%1,%2\n\tmfhi\t%0" : "#"; } + "&& reload_completed && !TARGET_FIX_R4000" + [(const_int 0)] +{ + rtx hilo; + + if (TARGET_64BIT) + { + hilo = gen_rtx_REG (TImode, MD_REG_FIRST); + emit_insn (gen_<u>mulsidi3_64bit_hilo (hilo, operands[1], operands[2])); + emit_insn (gen_mfhisi_ti (operands[0], hilo)); + } + else + { + hilo = gen_rtx_REG (DImode, MD_REG_FIRST); + emit_insn (gen_<u>mulsidi3_32bit (hilo, operands[1], operands[2])); + emit_insn (gen_mfhisi_di (operands[0], hilo)); + } + DONE; +} [(set_attr "type" "imul") - (set_attr "mode" "SI")]) + (set_attr "mode" "SI") + (set_attr "length" "8")]) (define_insn "<su>mulsi3_highpart_mulhi_internal" - [(set (match_operand:SI 0 "register_operand" "=h,d") + [(set (match_operand:SI 0 "register_operand" "=d") (truncate:SI (lshiftrt:DI (mult:DI - (any_extend:DI (match_operand:SI 1 "register_operand" "d,d")) - (any_extend:DI (match_operand:SI 2 "register_operand" "d,d"))) + (any_extend:DI (match_operand:SI 1 "register_operand" "d")) + (any_extend:DI (match_operand:SI 2 "register_operand" "d"))) (const_int 32)))) - (clobber (match_scratch:SI 3 "=l,l")) - (clobber (match_scratch:SI 4 "=X,h"))] + (clobber (match_scratch:SI 3 "=l"))] "ISA_HAS_MULHI" - "@ - mult<u>\t%1,%2 - mulhi<u>\t%0,%1,%2" - [(set_attr "type" "imul,imul3") + "mulhi<u>\t%0,%1,%2" + [(set_attr "type" "imul3") (set_attr "mode" "SI")]) (define_insn "*<su>mulsi3_highpart_neg_mulhi_internal" - [(set (match_operand:SI 0 "register_operand" "=h,d") + [(set (match_operand:SI 0 "register_operand" "=d") (truncate:SI (lshiftrt:DI (neg:DI (mult:DI - (any_extend:DI (match_operand:SI 1 "register_operand" "d,d")) - (any_extend:DI (match_operand:SI 2 "register_operand" "d,d")))) + (any_extend:DI (match_operand:SI 1 "register_operand" "d")) + (any_extend:DI (match_operand:SI 2 "register_operand" "d")))) (const_int 32)))) - (clobber (match_scratch:SI 3 "=l,l")) - (clobber (match_scratch:SI 4 "=X,h"))] + (clobber (match_scratch:SI 3 "=l"))] "ISA_HAS_MULHI" - "@ - mulshi<u>\t%.,%1,%2 - mulshi<u>\t%0,%1,%2" - [(set_attr "type" "imul,imul3") + "mulshi<u>\t%0,%1,%2" + [(set_attr "type" "imul3") (set_attr "mode" "SI")]) ;; Disable unsigned multiplication for -mfix-vr4120. This is for VR4120 ;; errata MD(0), which says that dmultu does not always produce the ;; correct result. -(define_insn "<su>muldi3_highpart" - [(set (match_operand:DI 0 "register_operand" "=h") +(define_insn_and_split "<su>muldi3_highpart" + [(set (match_operand:DI 0 "register_operand" "=d") (truncate:DI (lshiftrt:TI - (mult:TI - (any_extend:TI (match_operand:DI 1 "register_operand" "d")) - (any_extend:TI (match_operand:DI 2 "register_operand" "d"))) + (mult:TI (any_extend:TI (match_operand:DI 1 "register_operand" "d")) + (any_extend:TI (match_operand:DI 2 "register_operand" "d"))) (const_int 64)))) (clobber (match_scratch:DI 3 "=l"))] - "TARGET_64BIT && !TARGET_FIX_R4000 + "TARGET_64BIT && !(<CODE> == ZERO_EXTEND && TARGET_FIX_VR4120)" + { return TARGET_FIX_R4000 ? "dmult<u>\t%1,%2\n\tmfhi\t%0" : "#"; } + "&& reload_completed && !TARGET_FIX_R4000" + [(const_int 0)] +{ + rtx hilo; + + hilo = gen_rtx_REG (TImode, MD_REG_FIRST); + emit_insn (gen_<u>mulditi3_internal (hilo, operands[1], operands[2])); + emit_insn (gen_mfhidi_ti (operands[0], hilo)); + DONE; +} + [(set_attr "type" "imul") + (set_attr "mode" "DI") + (set_attr "length" "8")]) + +(define_expand "<u>mulditi3" + [(set (match_operand:TI 0 "register_operand") + (mult:TI (any_extend:TI (match_operand:DI 1 "register_operand")) + (any_extend:TI (match_operand:DI 2 "register_operand"))))] + "TARGET_64BIT && !(<CODE> == ZERO_EXTEND && TARGET_FIX_VR4120)" +{ + if (TARGET_FIX_R4000) + emit_insn (gen_<u>mulditi3_r4000 (operands[0], operands[1], operands[2])); + else + emit_insn (gen_<u>mulditi3_internal (operands[0], operands[1], + operands[2])); + DONE; +}) + +(define_insn "<u>mulditi3_internal" + [(set (match_operand:TI 0 "register_operand" "=x") + (mult:TI (any_extend:TI (match_operand:DI 1 "register_operand" "d")) + (any_extend:TI (match_operand:DI 2 "register_operand" "d"))))] + "TARGET_64BIT + && !TARGET_FIX_R4000 && !(<CODE> == ZERO_EXTEND && TARGET_FIX_VR4120)" "dmult<u>\t%1,%2" [(set_attr "type" "imul") (set_attr "mode" "DI")]) +(define_insn "<u>mulditi3_r4000" + [(set (match_operand:TI 0 "register_operand" "=d") + (mult:TI (any_extend:TI (match_operand:DI 1 "register_operand" "d")) + (any_extend:TI (match_operand:DI 2 "register_operand" "d")))) + (clobber (match_scratch:TI 3 "=x"))] + "TARGET_64BIT + && TARGET_FIX_R4000 + && !(<CODE> == ZERO_EXTEND && TARGET_FIX_VR4120)" + "dmult<u>\t%1,%2\;mflo\t%L0\;mfhi\t%M0" + [(set_attr "type" "imul") + (set_attr "mode" "DI") + (set_attr "length" "12")]) + ;; The R4650 supports a 32-bit multiply/ 64-bit accumulate ;; instruction. The HI/LO registers are used as a 64-bit accumulator. @@ -1840,8 +1833,7 @@ (define_insn "madsi" [(set (match_operand:SI 0 "register_operand" "+l") (plus:SI (mult:SI (match_operand:SI 1 "register_operand" "d") (match_operand:SI 2 "register_operand" "d")) - (match_dup 0))) - (clobber (match_scratch:SI 3 "=h"))] + (match_dup 0)))] "TARGET_MAD" "mad\t%1,%2" [(set_attr "type" "imadd") @@ -2016,29 +2008,74 @@ (define_insn "*recip<mode>3" ;; VR4120 errata MD(A1): signed division instructions do not work correctly ;; with negative operands. We use special libgcc functions instead. -(define_insn "divmod<mode>4" +(define_insn_and_split "divmod<mode>4" [(set (match_operand:GPR 0 "register_operand" "=l") (div:GPR (match_operand:GPR 1 "register_operand" "d") (match_operand:GPR 2 "register_operand" "d"))) - (set (match_operand:GPR 3 "register_operand" "=h") + (set (match_operand:GPR 3 "register_operand" "=d") (mod:GPR (match_dup 1) (match_dup 2)))] "!TARGET_FIX_VR4120" - { return mips_output_division ("<d>div\t$0,%1,%2", operands); } - [(set_attr "type" "idiv") - (set_attr "mode" "<MODE>")]) + "#" + "&& reload_completed" + [(const_int 0)] +{ + rtx hilo; + + if (TARGET_64BIT) + { + hilo = gen_rtx_REG (TImode, MD_REG_FIRST); + emit_insn (gen_divmod<mode>4_hilo_ti (hilo, operands[1], operands[2])); + emit_insn (gen_mfhi<mode>_ti (operands[3], hilo)); + } + else + { + hilo = gen_rtx_REG (DImode, MD_REG_FIRST); + emit_insn (gen_divmod<mode>4_hilo_di (hilo, operands[1], operands[2])); + emit_insn (gen_mfhi<mode>_di (operands[3], hilo)); + } + DONE; +}) -(define_insn "udivmod<mode>4" +(define_insn_and_split "udivmod<mode>4" [(set (match_operand:GPR 0 "register_operand" "=l") (udiv:GPR (match_operand:GPR 1 "register_operand" "d") (match_operand:GPR 2 "register_operand" "d"))) - (set (match_operand:GPR 3 "register_operand" "=h") + (set (match_operand:GPR 3 "register_operand" "=d") (umod:GPR (match_dup 1) (match_dup 2)))] "" - { return mips_output_division ("<d>divu\t$0,%1,%2", operands); } + "#" + "reload_completed" + [(const_int 0)] +{ + rtx hilo; + + if (TARGET_64BIT) + { + hilo = gen_rtx_REG (TImode, MD_REG_FIRST); + emit_insn (gen_udivmod<mode>4_hilo_ti (hilo, operands[1], operands[2])); + emit_insn (gen_mfhi<mode>_ti (operands[3], hilo)); + } + else + { + hilo = gen_rtx_REG (DImode, MD_REG_FIRST); + emit_insn (gen_udivmod<mode>4_hilo_di (hilo, operands[1], operands[2])); + emit_insn (gen_mfhi<mode>_di (operands[3], hilo)); + } + DONE; +}) + +(define_insn "<u>divmod<GPR:mode>4_hilo_<HILO:mode>" + [(set (match_operand:HILO 0 "register_operand" "=x") + (unspec:HILO + [(any_div:GPR (match_operand:GPR 1 "register_operand" "d") + (match_operand:GPR 2 "register_operand" "d"))] + UNSPEC_SET_HILO))] + "" + { return mips_output_division ("<GPR:d>div<u>\t%.,%1,%2", operands); } [(set_attr "type" "idiv") - (set_attr "mode" "<MODE>")]) + (set_attr "mode" "<GPR:MODE>")]) \f ;; ;; .................... @@ -3240,7 +3277,7 @@ (define_insn_and_split "*lea_high64" ;; dsll32 op1,op1,0 ;; daddu op1,op1,op0 (define_peephole2 - [(set (match_operand:DI 1 "register_operand") + [(set (match_operand:DI 1 "d_operand") (high:DI (match_operand:DI 2 "absolute_symbolic_operand"))) (match_scratch:DI 0 "d")] "TARGET_EXPLICIT_RELOCS && ABI_HAS_64BIT_SYMBOLS" @@ -3293,8 +3330,8 @@ (define_insn_and_split "*lea64" ;; ;; on MIPS16 targets. (define_split - [(set (match_operand:SI 0 "register_operand" "=d") - (high:SI (match_operand:SI 1 "absolute_symbolic_operand" "")))] + [(set (match_operand:SI 0 "d_operand") + (high:SI (match_operand:SI 1 "absolute_symbolic_operand")))] "TARGET_MIPS16 && reload_completed" [(set (match_dup 0) (match_dup 2)) (set (match_dup 0) (ashift:SI (match_dup 0) (const_int 16)))] @@ -3463,7 +3500,7 @@ (define_insn "*movdi_32bit" && (register_operand (operands[0], DImode) || reg_or_0_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "multi,multi,load,store,mthilo,mfhilo,mtc,load,mfc,store") + [(set_attr "type" "multi,multi,load,store,multi,multi,mtc,load,mfc,store") (set_attr "mode" "DI") (set_attr "length" "8,16,*,*,8,8,8,*,8,*")]) @@ -3474,7 +3511,7 @@ (define_insn "*movdi_gp32_fp64" && (register_operand (operands[0], DImode) || reg_or_0_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "multi,multi,load,store,mthilo,mfhilo,mtc,fpload,mfc,fpstore") + [(set_attr "type" "multi,multi,load,store,multi,multi,mtc,fpload,mfc,fpstore") (set_attr "mode" "DI") (set_attr "length" "8,16,*,*,8,8,8,*,8,*")]) @@ -3485,29 +3522,29 @@ (define_insn "*movdi_32bit_mips16" && (register_operand (operands[0], DImode) || register_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "multi,multi,multi,multi,multi,load,store,mfhilo") + [(set_attr "type" "multi,multi,multi,multi,multi,load,store,multi") (set_attr "mode" "DI") (set_attr "length" "8,8,8,8,12,*,*,8")]) (define_insn "*movdi_64bit" - [(set (match_operand:DI 0 "nonimmediate_operand" "=d,d,e,d,m,*f,*f,*d,*m,*x,*B*C*D,*B*C*D,*d,*m") - (match_operand:DI 1 "move_operand" "d,U,T,m,dJ,*d*J,*m,*f,*f,*J*d,*d,*m,*B*C*D,*B*C*D"))] + [(set (match_operand:DI 0 "nonimmediate_operand" "=d,d,e,d,m,*f,*f,*d,*m,*a,*d,*B*C*D,*B*C*D,*d,*m") + (match_operand:DI 1 "move_operand" "d,U,T,m,dJ,*d*J,*m,*f,*f,*J*d,*a,*d,*m,*B*C*D,*B*C*D"))] "TARGET_64BIT && !TARGET_MIPS16 && (register_operand (operands[0], DImode) || reg_or_0_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "move,const,const,load,store,mtc,fpload,mfc,fpstore,mthilo,mtc,load,mfc,store") + [(set_attr "type" "move,const,const,load,store,mtc,fpload,mfc,fpstore,mthilo,mfhilo,mtc,load,mfc,store") (set_attr "mode" "DI") - (set_attr "length" "4,*,*,*,*,4,*,4,*,4,8,*,8,*")]) + (set_attr "length" "4,*,*,*,*,4,*,4,*,4,4,8,*,8,*")]) (define_insn "*movdi_64bit_mips16" - [(set (match_operand:DI 0 "nonimmediate_operand" "=d,y,d,d,d,d,d,d,m") - (match_operand:DI 1 "move_operand" "d,d,y,K,N,kf,U,m,d"))] + [(set (match_operand:DI 0 "nonimmediate_operand" "=d,y,d,d,d,d,d,d,m,*d") + (match_operand:DI 1 "move_operand" "d,d,y,K,N,kf,U,m,d,*a"))] "TARGET_64BIT && TARGET_MIPS16 && (register_operand (operands[0], DImode) || register_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "move,move,move,arith,arith,load,const,load,store") + [(set_attr "type" "move,move,move,arith,arith,load,const,load,store,mfhilo") (set_attr "mode" "DI") (set_attr_alternative "length" [(const_int 4) @@ -3522,7 +3559,8 @@ (define_insn "*movdi_64bit_mips16" (const_int 8) (const_string "*") (const_string "*") - (const_string "*")])]) + (const_string "*") + (const_int 4)])]) ;; On the mips16, we can split ld $r,N($r) into an add and a load, @@ -3530,14 +3568,11 @@ (define_insn "*movdi_64bit_mips16" ;; load are 2 2 byte instructions. (define_split - [(set (match_operand:DI 0 "register_operand") + [(set (match_operand:DI 0 "d_operand") (mem:DI (plus:DI (match_dup 0) (match_operand:DI 1 "const_int_operand"))))] "TARGET_64BIT && TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) < 0 && INTVAL (operands[1]) >= -0x10) || (INTVAL (operands[1]) >= 32 * 8 @@ -3588,7 +3623,7 @@ (define_expand "movsi" (define_insn "*movsi_internal" [(set (match_operand:SI 0 "nonimmediate_operand" "=d,d,e,d,m,*f,*f,*d,*m,*d,*z,*a,*d,*B*C*D,*B*C*D,*d,*m") - (match_operand:SI 1 "move_operand" "d,U,T,m,dJ,*d*J,*m,*f,*f,*z,*d,*J*d,*A,*d,*m,*B*C*D,*B*C*D"))] + (match_operand:SI 1 "move_operand" "d,U,T,m,dJ,*d*J,*m,*f,*f,*z,*d,*J*d,*a,*d,*m,*B*C*D,*B*C*D"))] "!TARGET_MIPS16 && (register_operand (operands[0], SImode) || reg_or_0_operand (operands[1], SImode))" @@ -3598,13 +3633,13 @@ (define_insn "*movsi_internal" (set_attr "length" "4,*,*,*,*,4,*,4,*,4,4,4,4,4,*,4,*")]) (define_insn "*movsi_mips16" - [(set (match_operand:SI 0 "nonimmediate_operand" "=d,y,d,d,d,d,d,d,m") - (match_operand:SI 1 "move_operand" "d,d,y,K,N,kf,U,m,d"))] + [(set (match_operand:SI 0 "nonimmediate_operand" "=d,y,d,d,d,d,d,d,m,*d") + (match_operand:SI 1 "move_operand" "d,d,y,K,N,kf,U,m,d,*a"))] "TARGET_MIPS16 && (register_operand (operands[0], SImode) || register_operand (operands[1], SImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "move,move,move,arith,arith,load,const,load,store") + [(set_attr "type" "move,move,move,arith,arith,load,const,load,store,mfhilo") (set_attr "mode" "SI") (set_attr_alternative "length" [(const_int 4) @@ -3619,20 +3654,18 @@ (define_insn "*movsi_mips16" (const_int 8) (const_string "*") (const_string "*") - (const_string "*")])]) + (const_string "*") + (const_int 4)])]) ;; On the mips16, we can split lw $r,N($r) into an add and a load, ;; when the original load is a 4 byte instruction but the add and the ;; load are 2 2 byte instructions. (define_split - [(set (match_operand:SI 0 "register_operand") + [(set (match_operand:SI 0 "d_operand") (mem:SI (plus:SI (match_dup 0) (match_operand:SI 1 "const_int_operand"))))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) < 0 && INTVAL (operands[1]) >= -0x80) || (INTVAL (operands[1]) >= 32 * 4 @@ -3668,12 +3701,9 @@ (define_split ;; instructions. (define_split - [(set (match_operand:SI 0 "register_operand") + [(set (match_operand:SI 0 "d_operand") (match_operand:SI 1 "const_int_operand"))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && INTVAL (operands[1]) >= 0x100 && INTVAL (operands[1]) <= 0xff + 0x7f" [(set (match_dup 0) (match_dup 1)) @@ -3796,36 +3826,24 @@ (define_expand "movhi" }) (define_insn "*movhi_internal" - [(set (match_operand:HI 0 "nonimmediate_operand" "=d,d,d,m,*x") - (match_operand:HI 1 "move_operand" "d,I,m,dJ,*d"))] + [(set (match_operand:HI 0 "nonimmediate_operand" "=d,d,d,m,*a,*d") + (match_operand:HI 1 "move_operand" "d,I,m,dJ,*d*J,*a"))] "!TARGET_MIPS16 && (register_operand (operands[0], HImode) || reg_or_0_operand (operands[1], HImode))" - "@ - move\t%0,%1 - li\t%0,%1 - lhu\t%0,%1 - sh\t%z1,%0 - mt%0\t%1" - [(set_attr "type" "move,arith,load,store,mthilo") + { return mips_output_move (operands[0], operands[1]); } + [(set_attr "type" "move,arith,load,store,mthilo,mfhilo") (set_attr "mode" "HI") - (set_attr "length" "4,4,*,*,4")]) + (set_attr "length" "4,4,*,*,4,4")]) (define_insn "*movhi_mips16" - [(set (match_operand:HI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m") - (match_operand:HI 1 "move_operand" "d,d,y,K,N,m,d"))] + [(set (match_operand:HI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m,*d") + (match_operand:HI 1 "move_operand" "d,d,y,K,N,m,d,*a"))] "TARGET_MIPS16 && (register_operand (operands[0], HImode) || register_operand (operands[1], HImode))" - "@ - move\t%0,%1 - move\t%0,%1 - move\t%0,%1 - li\t%0,%1 - # - lhu\t%0,%1 - sh\t%1,%0" - [(set_attr "type" "move,move,move,arith,arith,load,store") + { return mips_output_move (operands[0], operands[1]); } + [(set_attr "type" "move,move,move,arith,arith,load,store,mfhilo") (set_attr "mode" "HI") (set_attr_alternative "length" [(const_int 4) @@ -3838,6 +3856,7 @@ (define_insn "*movhi_mips16" (const_int 8) (const_int 12)) (const_string "*") + (const_string "*") (const_string "*")])]) @@ -3846,13 +3865,10 @@ (define_insn "*movhi_mips16" ;; load are 2 2 byte instructions. (define_split - [(set (match_operand:HI 0 "register_operand") + [(set (match_operand:HI 0 "d_operand") (mem:HI (plus:SI (match_dup 0) (match_operand:SI 1 "const_int_operand"))))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) < 0 && INTVAL (operands[1]) >= -0x80) || (INTVAL (operands[1]) >= 32 * 2 @@ -3900,51 +3916,36 @@ (define_expand "movqi" }) (define_insn "*movqi_internal" - [(set (match_operand:QI 0 "nonimmediate_operand" "=d,d,d,m,*x") - (match_operand:QI 1 "move_operand" "d,I,m,dJ,*d"))] + [(set (match_operand:QI 0 "nonimmediate_operand" "=d,d,d,m,*a,*d") + (match_operand:QI 1 "move_operand" "d,I,m,dJ,*d*J,*a"))] "!TARGET_MIPS16 && (register_operand (operands[0], QImode) || reg_or_0_operand (operands[1], QImode))" - "@ - move\t%0,%1 - li\t%0,%1 - lbu\t%0,%1 - sb\t%z1,%0 - mt%0\t%1" - [(set_attr "type" "move,arith,load,store,mthilo") + { return mips_output_move (operands[0], operands[1]); } + [(set_attr "type" "move,arith,load,store,mthilo,mfhilo") (set_attr "mode" "QI") - (set_attr "length" "4,4,*,*,4")]) + (set_attr "length" "4,4,*,*,4,4")]) (define_insn "*movqi_mips16" - [(set (match_operand:QI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m") - (match_operand:QI 1 "move_operand" "d,d,y,K,N,m,d"))] + [(set (match_operand:QI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m,*d") + (match_operand:QI 1 "move_operand" "d,d,y,K,N,m,d,*a"))] "TARGET_MIPS16 && (register_operand (operands[0], QImode) || register_operand (operands[1], QImode))" - "@ - move\t%0,%1 - move\t%0,%1 - move\t%0,%1 - li\t%0,%1 - # - lbu\t%0,%1 - sb\t%1,%0" - [(set_attr "type" "move,move,move,arith,arith,load,store") + { return mips_output_move (operands[0], operands[1]); } + [(set_attr "type" "move,move,move,arith,arith,load,store,mfhilo") (set_attr "mode" "QI") - (set_attr "length" "4,4,4,4,8,*,*")]) + (set_attr "length" "4,4,4,4,8,*,*,4")]) ;; On the mips16, we can split lb $r,N($r) into an add and a load, ;; when the original load is a 4 byte instruction but the add and the ;; load are 2 2 byte instructions. (define_split - [(set (match_operand:QI 0 "register_operand") + [(set (match_operand:QI 0 "d_operand") (mem:QI (plus:SI (match_dup 0) (match_operand:SI 1 "const_int_operand"))))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) < 0 && INTVAL (operands[1]) >= -0x80) || (INTVAL (operands[1]) >= 32 @@ -4064,6 +4065,39 @@ (define_insn "*movdf_mips16" (set_attr "mode" "DF") (set_attr "length" "8,8,8,*,*")]) +;; 128-bit integer moves + +(define_expand "movti" + [(set (match_operand:TI 0) + (match_operand:TI 1))] + "TARGET_64BIT" +{ + if (mips_legitimize_move (TImode, operands[0], operands[1])) + DONE; +}) + +(define_insn "*movti" + [(set (match_operand:TI 0 "nonimmediate_operand" "=d,d,m,*a,*d") + (match_operand:TI 1 "move_operand" "di,m,dJ,*d*J,*a"))] + "TARGET_64BIT + && !TARGET_MIPS16 + && (register_operand (operands[0], TImode) + || reg_or_0_operand (operands[1], TImode))" + "#" + [(set_attr "type" "multi,load,store,multi,multi") + (set_attr "length" "8,*,*,8,8")]) + +(define_insn "*movti_mips16" + [(set (match_operand:TI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m,*d") + (match_operand:TI 1 "move_operand" "d,d,y,K,N,m,d,*a"))] + "TARGET_64BIT + && TARGET_MIPS16 + && (register_operand (operands[0], TImode) + || register_operand (operands[1], TImode))" + "#" + [(set_attr "type" "multi,multi,multi,multi,multi,load,store,multi") + (set_attr "length" "8,8,8,12,16,*,*,8")]) + ;; 128-bit floating point moves (define_expand "movtf" @@ -4122,7 +4156,7 @@ (define_split ;; When generating mips16 code, split moves of negative constants into ;; a positive "li" followed by a negation. (define_split - [(set (match_operand 0 "register_operand") + [(set (match_operand 0 "d_operand") (match_operand 1 "const_int_operand"))] "TARGET_MIPS16 && reload_completed && INTVAL (operands[1]) < 0" [(set (match_dup 2) @@ -4171,44 +4205,27 @@ (define_insn "movv2sf_hardfloat_32bit" (set_attr "mode" "SF") (set_attr "length" "4,8,*,*,*,8,8,8,*,*")]) -;; The HI and LO registers are not truly independent. If we move an mthi -;; instruction before an mflo instruction, it will make the result of the -;; mflo unpredictable. The same goes for mtlo and mfhi. -;; -;; We cope with this by making the mflo and mfhi patterns use both HI and LO. -;; Operand 1 is the register we want, operand 2 is the other one. -;; -;; When generating VR4120 or VR4130 code, we use macc{,hi} and -;; dmacc{,hi} instead of mfhi and mflo. This avoids both the normal -;; MIPS III hi/lo hazards and the errata related to -mfix-vr4130. - -(define_expand "mfhilo_<mode>" - [(set (match_operand:GPR 0 "register_operand") - (unspec:GPR [(match_operand:GPR 1 "register_operand") - (match_operand:GPR 2 "register_operand")] - UNSPEC_MFHILO))]) - -(define_insn "*mfhilo_<mode>" - [(set (match_operand:GPR 0 "register_operand" "=d,d") - (unspec:GPR [(match_operand:GPR 1 "register_operand" "h,l") - (match_operand:GPR 2 "register_operand" "l,h")] - UNSPEC_MFHILO))] - "!ISA_HAS_MACCHI" - "mf%1\t%0" +;; When generating VR4120 or VR4130 code, we use MACCHI and DMACCHI +;; instead of MFHI. This avoids both the normal MIPS III hi/lo hazards +;; and the errata related to -mfix-vr4130. +(define_insn "mfhi<GPR:mode>_<HILO:mode>" + [(set (match_operand:GPR 0 "register_operand" "=d") + (unspec:GPR [(match_operand:HILO 1 "register_operand" "x")] + UNSPEC_MFHI))] + "" + { return ISA_HAS_MACCHI ? "<GPR:d>macchi\t%0,%.,%." : "mfhi\t%0"; } [(set_attr "type" "mfhilo") - (set_attr "mode" "<MODE>")]) + (set_attr "mode" "<GPR:MODE>")]) -(define_insn "*mfhilo_<mode>_macc" - [(set (match_operand:GPR 0 "register_operand" "=d,d") - (unspec:GPR [(match_operand:GPR 1 "register_operand" "h,l") - (match_operand:GPR 2 "register_operand" "l,h")] - UNSPEC_MFHILO))] - "ISA_HAS_MACCHI" - "@ - <d>macchi\t%0,%.,%. - <d>macc\t%0,%.,%." - [(set_attr "type" "mfhilo") - (set_attr "mode" "<MODE>")]) +(define_insn "mthi<GPR:mode>_<HILO:mode>" + [(set (match_operand:HILO 0 "register_operand" "=x") + (unspec:HILO [(match_operand:GPR 1 "reg_or_0_operand" "dJ") + (match_operand:GPR 2 "register_operand" "l")] + UNSPEC_MTHI))] + "" + "mthi\t%z1" + [(set_attr "type" "mthilo") + (set_attr "mode" "SI")]) ;; Emit a doubleword move in which exactly one of the operands is ;; a floating-point register. We can't just emit two normal moves @@ -5120,11 +5137,10 @@ (define_insn "*lshrdi3_mips16" ;; On the mips16, we can split a 4 byte shift into 2 2 byte shifts. (define_split - [(set (match_operand:GPR 0 "register_operand") - (any_shift:GPR (match_operand:GPR 1 "register_operand") + [(set (match_operand:GPR 0 "d_operand") + (any_shift:GPR (match_operand:GPR 1 "d_operand") (match_operand:GPR 2 "const_int_operand")))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && GET_CODE (operands[2]) == CONST_INT && INTVAL (operands[2]) > 8 && INTVAL (operands[2]) <= 16" [(set (match_dup 0) (any_shift:GPR (match_dup 1) (const_int 8))) Index: gcc/config/mips/predicates.md =================================================================== --- gcc/config/mips/predicates.md 2008-06-01 10:23:31.000000000 +0100 +++ gcc/config/mips/predicates.md 2008-06-01 14:06:06.000000000 +0100 @@ -76,9 +76,11 @@ (define_predicate "const_0_or_1_operand" (ior (match_test "op == CONST0_RTX (GET_MODE (op))") (match_test "op == CONST1_RTX (GET_MODE (op))")))) -(define_predicate "fpr_operand" +(define_predicate "d_operand" (and (match_code "reg") - (match_test "FP_REG_P (REGNO (op))"))) + (match_test "TARGET_MIPS16 + ? M16_REG_P (REGNO (op)) + : GP_REG_P (REGNO (op))"))) (define_predicate "lo_operand" (and (match_code "reg") Index: gcc/testsuite/gcc.dg/torture/mips-hilo-1.c =================================================================== --- gcc/testsuite/gcc.dg/torture/mips-hilo-1.c 2008-06-01 10:23:31.000000000 +0100 +++ /dev/null 2008-05-31 10:25:53.548096750 +0100 @@ -1,73 +0,0 @@ -/* f1 checks that an mtlo is not moved before an mfhi. f2 does the same - for an mthi and an mflo. */ -/* { dg-do run { target mips*-*-* } } */ -/* { dg-options "-mtune=rm7000" } */ - -extern void abort (void); -extern void exit (int); - -#define DECLARE(TYPE) \ - TYPE __attribute__ ((noinline)) __attribute__ ((nomips16)) \ - f1##TYPE (TYPE x1, TYPE x2, TYPE x3) \ - { \ - TYPE t1, t2; \ - \ - asm ("mult\t%1,%2" : "=h" (t1) : "d" (x1), "d" (x2) : "lo"); \ - asm ("mflo\t%0" : "=r" (t2) : "l" (x3) : "hi"); \ - return t1 + t2; \ - } \ - \ - TYPE __attribute__ ((noinline)) __attribute__ ((nomips16)) \ - f2##TYPE (TYPE x1, TYPE x2, TYPE x3) \ - { \ - TYPE t1, t2; \ - \ - asm ("mult\t%1,%2" : "=l" (t1) : "d" (x1), "d" (x2) : "hi"); \ - asm ("mfhi\t%0" : "=r" (t2) : "h" (x3) : "lo"); \ - return t1 + t2; \ - } - -#define TEST(TYPE) \ - if (f1##TYPE (1, 2, 10) != 10) \ - abort (); \ - if (f2##TYPE (1, 2, 40) != 42) \ - abort () - -typedef char c; -typedef signed char sc; -typedef unsigned char uc; -typedef short s; -typedef unsigned short us; -typedef int i; -typedef unsigned int ui; -typedef long long ll; -typedef unsigned long long ull; - -DECLARE (c) -DECLARE (sc) -DECLARE (uc) -DECLARE (s) -DECLARE (us) -DECLARE (i) -DECLARE (ui) -#if defined (__mips64) -DECLARE (ll) -DECLARE (ull) -#endif - -int -main () -{ - TEST (c); - TEST (sc); - TEST (uc); - TEST (s); - TEST (us); - TEST (i); - TEST (ui); -#if defined (__mips64) - TEST (ll); - TEST (ull); -#endif - exit (0); -} Index: gcc/testsuite/gcc.target/mips/pr35232.c =================================================================== --- gcc/testsuite/gcc.target/mips/pr35232.c 2008-06-01 10:23:31.000000000 +0100 +++ /dev/null 2008-05-31 10:25:53.548096750 +0100 @@ -1,17 +0,0 @@ -/* { dg-do run } */ -/* { dg-mips-options "-O" } */ - -NOMIPS16 unsigned int -f1 (unsigned long long x) -{ - unsigned int r; - asm ("# %0" : "=a" (r) : "0" (x)); - asm ("# %0" : "=h" (r) : "0" (r)); - return r; -} - -int -main (void) -{ - return f1 (4) != 4; -} Index: gcc/testsuite/gcc.target/mips/fix-vr4130-1.c =================================================================== --- gcc/testsuite/gcc.target/mips/fix-vr4130-1.c 2008-06-01 10:23:31.000000000 +0100 +++ gcc/testsuite/gcc.target/mips/fix-vr4130-1.c 2008-06-01 14:06:06.000000000 +0100 @@ -1,4 +1,8 @@ /* { dg-do compile } */ /* { dg-mips-options "-march=vr4130 -mfix-vr4130" } */ -NOMIPS16 int foo (void) { int r; asm ("# foo" : "=h" (r)); return r; } +NOMIPS16 unsigned int +foo (unsigned int x, unsigned int y) +{ + return x % y; +} /* { dg-final { scan-assembler "\tmacchi\t" } } */ Index: gcc/testsuite/gcc.target/mips/fix-vr4130-3.c =================================================================== --- gcc/testsuite/gcc.target/mips/fix-vr4130-3.c 2008-06-01 10:23:31.000000000 +0100 +++ gcc/testsuite/gcc.target/mips/fix-vr4130-3.c 2008-06-01 14:06:06.000000000 +0100 @@ -1,10 +1,8 @@ /* { dg-do compile } */ /* { dg-mips-options "-march=vr4130 -mgp64 -mfix-vr4130" } */ -NOMIPS16 long long -foo (void) +NOMIPS16 unsigned long long +foo (unsigned long long x, unsigned long long y) { - long long r; - asm ("# foo" : "=h" (r)); - return r; + return x % y; } /* { dg-final { scan-assembler "\tdmacchi\t" } } */ Index: gcc/testsuite/gcc.target/mips/int-moves-1.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/int-moves-1.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,40 @@ +/* { dg-do compile { target mips16_attribute } } */ +/* { dg-mips-options "-mgp64 -msoft-float -O2 -EL" } */ +/* { dg-add-options mips16_attribute } */ + +typedef unsigned uint128_t __attribute__((mode(TI))); + +extern uint128_t g[16]; +extern unsigned char gstuff[0x10000]; + +NOMIPS16 uint128_t +foo (uint128_t i1, uint128_t i2, uint128_t i3, uint128_t i4, + uint128_t *x, unsigned char *lstuff) +{ + g[0] = i1; + g[1] = i2; + g[2] = i3; + g[3] = i4; + x[0] = x[4]; + x[1] = 0; + x[2] = ((uint128_t) 0x123456789abcdefULL << 64) | 0xaabbccddeeff1122ULL; + x[3] = g[4]; + x[4] = *(uint128_t *) (lstuff + 0x7fff); + return *(uint128_t *) (gstuff + 0x7fff); +} + +MIPS16 uint128_t +bar (uint128_t i1, uint128_t i2, uint128_t i3, uint128_t i4, + uint128_t *x, unsigned char *lstuff) +{ + g[0] = i1; + g[1] = i2; + g[2] = i3; + g[3] = i4; + x[0] = x[4]; + x[1] = 0; + x[2] = ((uint128_t) 0x123456789abcdefULL << 64) | 0xaabbccddeeff1122ULL; + x[3] = g[4]; + x[4] = *(uint128_t *) (lstuff + 0x7fff); + return *(uint128_t *) (gstuff + 0x7fff); +} Index: gcc/testsuite/gcc.target/mips/int-moves-2.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/int-moves-2.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,40 @@ +/* { dg-do compile { target mips16_attribute } } */ +/* { dg-mips-options "-mgp64 -msoft-float -O2 -EB" } */ +/* { dg-add-options mips16_attribute } */ + +typedef unsigned uint128_t __attribute__((mode(TI))); + +extern uint128_t g[16]; +extern unsigned char gstuff[0x10000]; + +NOMIPS16 uint128_t +foo (uint128_t i1, uint128_t i2, uint128_t i3, uint128_t i4, + uint128_t *x, unsigned char *lstuff) +{ + g[0] = i1; + g[1] = i2; + g[2] = i3; + g[3] = i4; + x[0] = x[4]; + x[1] = 0; + x[2] = ((uint128_t) 0x123456789abcdefULL << 64) | 0xaabbccddeeff1122ULL; + x[3] = g[4]; + x[4] = *(uint128_t *) (lstuff + 0x7fff); + return *(uint128_t *) (gstuff + 0x7fff); +} + +MIPS16 uint128_t +bar (uint128_t i1, uint128_t i2, uint128_t i3, uint128_t i4, + uint128_t *x, unsigned char *lstuff) +{ + g[0] = i1; + g[1] = i2; + g[2] = i3; + g[3] = i4; + x[0] = x[4]; + x[1] = 0; + x[2] = ((uint128_t) 0x123456789abcdefULL << 64) | 0xaabbccddeeff1122ULL; + x[3] = g[4]; + x[4] = *(uint128_t *) (lstuff + 0x7fff); + return *(uint128_t *) (gstuff + 0x7fff); +} Index: gcc/testsuite/gcc.target/mips/fix-r4000-1.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-1.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,6 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -O2 -dp" } */ +typedef int int32_t; +typedef int uint32_t; +int32_t foo (int32_t x, int32_t y) { return x * y; } +uint32_t bar (uint32_t x, uint32_t y) { return x * y; } +/* { dg-final { scan-assembler-times "[concat {\tmult\t\$[45],\$[45][^\n]+mulsi3_r4000[^\n]+\n\tmflo\t\$2\n}]" 2 } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-2.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-2.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,7 @@ +/* { dg-mips-options "-mips1 -mfix-r4000 -O2 -dp -EB" } */ +typedef int int32_t; +typedef long long int64_t; +int32_t foo (int32_t x, int32_t y) { return ((int64_t) x * y) >> 32; } +/* ??? A highpart pattern would be a better choice, but we currently + don't use them. */ +/* { dg-final { scan-assembler "[concat {\tmult\t\$[45],\$[45][^\n]+mulsidi3_32bit_r4000[^\n]+\n\tmflo\t\$3\n\tmfhi\t\$2\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-3.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-3.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,7 @@ +/* { dg-mips-options "-mips1 -mfix-r4000 -O2 -dp -EB" } */ +typedef unsigned int uint32_t; +typedef unsigned long long uint64_t; +uint32_t foo (uint32_t x, uint32_t y) { return ((uint64_t) x * y) >> 32; } +/* ??? A highpart pattern would be a better choice, but we currently + don't use them. */ +/* { dg-final { scan-assembler "[concat {\tmultu\t\$[45],\$[45][^\n]+umulsidi3_32bit_r4000[^\n]+\n\tmflo\t\$3\n\tmfhi\t\$2\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-4.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-4.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,8 @@ +/* ??? At the moment, lower-subreg.c decomposes the copy of the multiplication + result to $2, which prevents the register allocators from storing the + multiplication result in $2. */ +/* { dg-mips-options "-mips1 -mfix-r4000 -O2 -fno-split-wide-types -dp -EL" } */ +typedef int int32_t; +typedef long long int64_t; +int64_t foo (int32_t x, int32_t y) { return (int64_t) x * y; } +/* { dg-final { scan-assembler "[concat {\tmult\t\$[45],\$[45][^\n]+mulsidi3_32bit_r4000[^\n]+\n\tmflo\t\$2\n\tmfhi\t\$3\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-5.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-5.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,8 @@ +/* ??? At the moment, lower-subreg.c decomposes the copy of the multiplication + result to $2, which prevents the register allocators from storing the + multiplication result in $2. */ +/* { dg-mips-options "-mips1 -mfix-r4000 -O2 -fno-split-wide-types -dp -EL" } */ +typedef unsigned int uint32_t; +typedef unsigned long long uint64_t; +uint64_t foo (uint32_t x, uint32_t y) { return (uint64_t) x * y; } +/* { dg-final { scan-assembler "[concat {\tmultu\t\$[45],\$[45][^\n]+umulsidi3_32bit_r4000[^\n]+\n\tmflo\t\$2\n\tmfhi\t\$3\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-6.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-6.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,6 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -mgp64 -O2 -dp" } */ +typedef long long int64_t; +typedef unsigned long long uint64_t; +int64_t foo (int64_t x, int64_t y) { return x * y; } +uint64_t bar (uint64_t x, uint64_t y) { return x * y; } +/* { dg-final { scan-assembler-times "[concat {\tdmult\t\$[45],\$[45][^\n]+muldi3_r4000[^\n]+\n\tmflo\t\$2\n}]" 2 } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-7.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-7.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,7 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -O2 -mgp64 -dp -EB" } */ +typedef long long int64_t; +typedef int int128_t __attribute__((mode(TI))); +int64_t foo (int64_t x, int64_t y) { return ((int128_t) x * y) >> 64; } +/* ??? A highpart pattern would be a better choice, but we currently + don't use them. */ +/* { dg-final { scan-assembler "[concat {\tdmult\t\$[45],\$[45][^\n]+mulditi3[^\n]+\n\tmflo\t\$3\n\tmfhi\t\$2\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-8.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-8.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,7 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -O2 -mgp64 -dp -EB" } */ +typedef unsigned long long uint64_t; +typedef unsigned int uint128_t __attribute__((mode(TI))); +uint64_t foo (uint64_t x, uint64_t y) { return ((uint128_t) x * y) >> 64; } +/* ??? A highpart pattern would be a better choice, but we currently + don't use them. */ +/* { dg-final { scan-assembler "[concat {\tdmultu\t\$[45],\$[45][^\n]+umulditi3[^\n]+\n\tmflo\t\$3\n\tmfhi\t\$2\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-9.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-9.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,8 @@ +/* ??? At the moment, lower-subreg.c decomposes the copy of the multiplication + result to $2, which prevents the register allocators from storing the + multiplication result in $2. */ +/* { dg-mips-options "-mips3 -mfix-r4000 -mgp64 -O2 -fno-split-wide-types -dp -EL" } */ +typedef long long int64_t; +typedef int int128_t __attribute__((mode(TI))); +int128_t foo (int64_t x, int64_t y) { return (int128_t) x * y; } +/* { dg-final { scan-assembler "[concat {\tdmult\t\$[45],\$[45][^\n]+mulditi3_r4000[^\n]+\n\tmflo\t\$2\n\tmfhi\t\$3\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-10.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-10.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,8 @@ +/* ??? At the moment, lower-subreg.c decomposes the copy of the multiplication + result to $2, which prevents the register allocators from storing the + multiplication result in $2. */ +/* { dg-mips-options "-mips3 -mfix-r4000 -mgp64 -O2 -fno-split-wide-types -dp -EL" } */ +typedef unsigned long long uint64_t; +typedef unsigned int uint128_t __attribute__((mode(TI))); +uint128_t foo (uint64_t x, uint64_t y) { return (uint128_t) x * y; } +/* { dg-final { scan-assembler "[concat {\tdmultu\t\$[45],\$[45][^\n]+umulditi3_r4000[^\n]+\n\tmflo\t\$2\n\tmfhi\t\$3\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-11.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-11.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,4 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -mgp64 -O2 -dp" } */ +typedef long long int64_t; +int64_t foo (int64_t x) { return x / 11993; } +/* { dg-final { scan-assembler "[concat {\tdmult\t\$4,\$[0-9]+[^\n]+smuldi3_highpart[^\n]+\n\tmfhi\t\$[0-9]+\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-12.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-12.c 2008-06-01 14:06:06.000000000 +0100 @@ -0,0 +1,4 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -mgp64 -O2 -dp" } */ +typedef unsigned long long uint64_t; +uint64_t foo (uint64_t x) { return x / 11993; } +/* { dg-final { scan-assembler "[concat {\tdmultu\t\$4,\$[0-9]+[^\n]+umuldi3_highpart[^\n]+\n\tmfhi\t\$[0-9]+\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/timode-1.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/timode-1.c 2008-06-01 14:21:51.000000000 +0100 @@ -0,0 +1,65 @@ +/* { dg-mips-options "-mgp64" } */ +typedef int int128_t __attribute__((mode(TI))); +typedef unsigned int uint128_t __attribute__((mode(TI))); + +#define UINT128_CONST(A, B) \ + (((uint128_t) (0x ## A ## ULL) << 64) | (0x ## B ## ULL)) + +volatile uint128_t a = UINT128_CONST (1111111111111111, a222222222222222); +volatile uint128_t b = UINT128_CONST (0000000000000005, 0000000000000003); +volatile uint128_t c = UINT128_CONST (5dddddddddddddde, e666666666666666); +volatile uint128_t d = UINT128_CONST (e612340000000000, 5000000000234500); +volatile uint128_t e = UINT128_CONST (43f011dddddddddf, 366666666689ab66); +volatile uint128_t f = UINT128_CONST (4210100000000000, 1000000000010100); +volatile uint128_t g = UINT128_CONST (a5e225dddddddddf, 6666666666aaee66); +volatile uint128_t h = UINT128_CONST (e7f235dddddddddf, 7666666666abef66); +volatile uint128_t i = UINT128_CONST (5e225dddddddddf6, 666666666aaee660); +volatile uint128_t j = UINT128_CONST (0a5e225ddddddddd, f6666666666aaee6); +volatile uint128_t k = UINT128_CONST (fa5e225ddddddddd, f6666666666aaee6); + +volatile int amount = 4; + +volatile uint128_t result; + +int +main (void) +{ + result = a * b; + if (result != c) + return 1; + + result = c + d; + if (result != e) + return 1; + + result = e - d; + if (result != c) + return 1; + + result = d & e; + if (result != f) + return 1; + + result = d ^ e; + if (result != g) + return 1; + + result = d | e; + if (result != h) + return 1; + + result = g << amount; + if (result != i) + return 1; + + result = g >> amount; + if (result != j) + return 1; + + result = (int128_t) g >> amount; + if (result != k) + return 1; + + return 0; +} +/* { dg-final { scan-assembler-not "\tjal" } } */ Index: gcc/testsuite/gcc.target/mips/timode-2.c =================================================================== --- /dev/null 2008-05-31 10:25:53.548096750 +0100 +++ gcc/testsuite/gcc.target/mips/timode-2.c 2008-06-01 14:21:36.000000000 +0100 @@ -0,0 +1,64 @@ +/* { dg-do run { target mips64 } } */ +typedef int int128_t __attribute__((mode(TI))); +typedef unsigned int uint128_t __attribute__((mode(TI))); + +#define UINT128_CONST(A, B) \ + (((uint128_t) (0x ## A ## ULL) << 64) | (0x ## B ## ULL)) + +volatile uint128_t a = UINT128_CONST (1111111111111111, a222222222222222); +volatile uint128_t b = UINT128_CONST (0000000000000005, 0000000000000003); +volatile uint128_t c = UINT128_CONST (5dddddddddddddde, e666666666666666); +volatile uint128_t d = UINT128_CONST (e612340000000000, 5000000000234500); +volatile uint128_t e = UINT128_CONST (43f011dddddddddf, 366666666689ab66); +volatile uint128_t f = UINT128_CONST (4210100000000000, 1000000000010100); +volatile uint128_t g = UINT128_CONST (a5e225dddddddddf, 6666666666aaee66); +volatile uint128_t h = UINT128_CONST (e7f235dddddddddf, 7666666666abef66); +volatile uint128_t i = UINT128_CONST (5e225dddddddddf6, 666666666aaee660); +volatile uint128_t j = UINT128_CONST (0a5e225ddddddddd, f6666666666aaee6); +volatile uint128_t k = UINT128_CONST (fa5e225ddddddddd, f6666666666aaee6); + +volatile int amount = 4; + +volatile uint128_t result; + +int +main (void) +{ + result = a * b; + if (result != c) + return 1; + + result = c + d; + if (result != e) + return 1; + + result = e - d; + if (result != c) + return 1; + + result = d & e; + if (result != f) + return 1; + + result = d ^ e; + if (result != g) + return 1; + + result = d | e; + if (result != h) + return 1; + + result = g << amount; + if (result != i) + return 1; + + result = g >> amount; + if (result != j) + return 1; + + result = (int128_t) g >> amount; + if (result != k) + return 1; + + return 0; +} ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: Changing the treatment of the MIPS HI and LO registers 2008-06-01 13:48 ` Richard Sandiford @ 2008-06-09 21:13 ` Richard Sandiford 2008-06-09 21:54 ` Maciej W. Rozycki 2008-06-11 18:41 ` Maxim Kuvyrkov 0 siblings, 2 replies; 7+ messages in thread From: Richard Sandiford @ 2008-06-09 21:13 UTC (permalink / raw) To: Maciej W. Rozycki; +Cc: Ralf Baechle, gcc-patches, linux-mips Richard Sandiford <rdsandiford@googlemail.com> writes: > I'll reinstate the 48 hour warning, mostly as a way of seeing > whether the thread is still alive. Here's the final version of the patch. I fixed the <su>mulsi3_highpart condition and added a couple of missing attributes. Applied after retesting. I also made the following changes to the web page: Index: htdocs/gcc-4.4/changes.html =================================================================== RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-4.4/changes.html,v retrieving revision 1.11 diff -u -p -r1.11 changes.html --- htdocs/gcc-4.4/changes.html 9 Jun 2008 20:41:49 -0000 1.11 +++ htdocs/gcc-4.4/changes.html 9 Jun 2008 21:10:25 -0000 @@ -28,6 +28,28 @@ when <code>-Wdeprecated</code> or <code>-pedantic</code> is used. This extension has been deprecated for many years, but never warned about.</li> + + <li>The MIPS port no longer recognizes the <code>h</code> + <code>asm</code> constraint. It was necessary to remove + this constraint in order to avoid generating unpredictable + code sequences. + + <p>One of the main uses of the <code>h</code> constraint + was to extract the high part of a multiplication on + 64-bit targets. For example:</p> + <pre> + asm ("dmultu\t%1,%2" : "=h" (result) : "r" (x), "r" (y));</pre> + <p>You can now achieve the same effect using 128-bit types:</p> + <pre> + typedef unsigned int uint128_t __attribute__((mode(TI))); + result = ((uint128_t) x * y) >> 64;</pre> + <p>The second sequence is better in many ways. For example, + if <code>x</code> and <code>y</code> are constants, the + compiler can perform the multiplication at compile time. + If <code>x</code> and <code>y</code> are not constants, + the compiler can schedule the runtime multiplication + better than it can schedule an <code>asm</code> statement.</p> + </li> </ul> <h2>General Optimizer Improvements</h2> @@ -75,6 +97,8 @@ <ul> <li>Support for RMI's XLR processor is now available through the <code>-march=xlr</code> and <code>-mtune=xlr</code> options.</li> + <li>64-bit targets can now perform 128-bit multiplications inline, + instead of relying on a <code>libgcc</code> function.</li> </ul> <h2>Documentation improvements</h2> Please shout if you find anything wrong with either the patch or the documentation changes. Richard gcc/ * doc/md.texi: Synchronize with later constraints.md change. * longlong.h (umul_ppmm): Replace the MIPS asm implementation with a C implementation. * config/mips/mips.c (mips_legitimize_move): Remove MFHI and MFLO handling. (mips_subword): Assume TImode for CONST_INTs if TARGET_64BIT. (mips_split_doubleword_move): Use special MTHI and MFHI instructions when moving to and from MD_REGNUM. (mips_output_move): Don't handle moves from GPRs to HI_REGNUM. Handle moves from LO_REGNUM to GPRs using MFLO, MACC or DMACC. Handle byte and halfword moves. (mips_hard_regno_mode_ok_p): Handle MD_REGS and DSP_ACC_REGS separately. * config/mips/constraints.md (h): Turn into NO_REGS. (l, x): Update documentation. * config/mips/mips.md (UNSPEC_MFHILO): Delete. (UNSPEC_MFHI, UNSPEC_MTHI, UNSPEC_SET_HILO): New. (UNSPEC_TLS_LDM, UNSPEC_TLS_GET_TP): Renumber. (HILO): New mode iterator. (MOVE128): Add TI. (any_div): New code iterator. (u): Extend code attribute to div and udiv. (*add<mode>3_mips16, *movdi_64bit_mips16, *movsi_mips16): Use d_operand in the splitters. Remove redundant CONST_INT checks. (mulsi3_mult3, mul<mode>3_internal, mul<mode>3_r4000, *mul_acc_si) (*macc, *msac, *msac_using_macc, *macc2, *msac2, *mul_sub_si) (*muls): Remove "=h" clobbers. Adjust peephole2s and define_splits accordingly, using normal moves instead of unspecs to move LO into a GPR. Use d_operand and lo_operand instead of *_REG_P checks. (<u>mulsidi3): Handle expansion in C code. (<u>mulsidi3_32bit_internal): Rename to... (<u>mulsidi3_32bit): ...this. (<u>mulsidi3_32bit_r4000): Fix insn separator. (*<u>mulsidi3_64bit): Rename to... (<u>mulsidi3_64bit): ...this. Combine DImode "=h" and "=l" clobbers into a TImode "=x" clobber. In the split, use an UNSPEC_SET_HILO to set LO and HI to the multiplication result. Use a normal move for MFLO and an unspec for MFHI. (*<u>mulsidi3_64bit_parts): Replace with... (<u>mulsidi3_64bit_hilo): ...this new instruction. (<su>mulsi3_highpart): Extend to TARGET_FIX_R4000. (<su>mulsi3_highpart_internal): Turn into a define_insn_and_split and extend it to TARGET_FIX_R4000. Store the destination in a GPR instead of HI. Split the instruction into a separate multiplication and MFHI if !TARGET_FIX_R4000. (<su>muldi3_highpart): Likewise. (<su>mulsi3_highpart_mulhi_internal): Remove the first alternative and the "=h" clobber. (*<su>mulsi3_highpart_neg_mulhi_internal): Likewise. (<u>mulditi3): New expander. (<u>mulditi3_internal, <u>mulditi3_r4000): New patterns. (madsi): Remove "=h" clobber. (divmod<mode>4, udivmod<mode>4): Turn into define_insn_and_splits. Force the modulus result to be a GPR and split the instruction into a division followed by an MFHI after reload. (<u>divmod<GPR:mode>4_hilo_<HILO:mode>): New instruction. (*lea_high64): Use d_operand in the define_peephole2. Likewise the MIPS16 HIGH define_split. (*movdi_32bit, *movdi_gp32_fp64, *movdi_32bit_mips16): Change type of acc<->gpr moves to "multi". (*movdi_64bit): Replace the single "x" alternative with alternatives for moving into and out of "a". (*movhi_internal, *movqi_internal): Likewise. Use mips_output_move. (*movsi_internal): Extend the "d<-A" alternative to "d<-a". (*movdi_64bit_mips16, *movsi_mips16): Add d<-a alternatives. Use d_operand in the splitters. Remove redundant CONST_INT checks. (*movhi_mips16, *movqi_mips16): Likewise. Use mips_output_move. (movti): New expander. (*movti, *movti_mips16): New insns. (mfhilo_<mode>, *mfhilo_<mode>, *mfhilo_<mode>_macc): Delete. (mfhi<GPR:mode>_<HILO:mode>): New pattern. (mthi<GPR:mode>_<HILO:mode>): Likewise. * config/mips/predicates.md (fpr_operand): Delete. (d_operand): New predicate. gcc/testsuite/ * gcc.dg/torture/mips-hilo-1.c: Delete. * gcc.target/mips/pr35232.c: Likewise. * gcc.target/mips/fix-vr4130-1.c: Use modulus to create an mfhi. * gcc.target/mips/fix-vr4130-3.c: Likewise. * gcc.target/mips/int-moves-1.c: New test. * gcc.target/mips/int-moves-2.c: Likewise. * gcc.target/mips/fix-r4000-1.c: Likewise. * gcc.target/mips/fix-r4000-2.c: Likewise. * gcc.target/mips/fix-r4000-3.c: Likewise. * gcc.target/mips/fix-r4000-4.c: Likewise. * gcc.target/mips/fix-r4000-5.c: Likewise. * gcc.target/mips/fix-r4000-6.c: Likewise. * gcc.target/mips/fix-r4000-7.c: Likewise. * gcc.target/mips/fix-r4000-8.c: Likewise. * gcc.target/mips/fix-r4000-9.c: Likewise. * gcc.target/mips/fix-r4000-10.c: Likewise. * gcc.target/mips/fix-r4000-11.c: Likewise. * gcc.target/mips/fix-r4000-12.c: Likewise. * gcc.target/mips/timode-1.c: Likewise. * gcc.target/mips/timode-2.c: Likewise. Index: gcc/doc/md.texi =================================================================== --- gcc/doc/md.texi 2008-06-09 21:42:44.000000000 +0100 +++ gcc/doc/md.texi 2008-06-09 21:45:18.000000000 +0100 @@ -2469,13 +2469,15 @@ generating MIPS16 code. A floating-point register (if available). @item h -The @code{hi} register. +Formerly the @code{hi} register. This constraint is no longer supported. @item l -The @code{lo} register. +The @code{lo} register. Use this register to store values that are +no bigger than a word. @item x -The @code{hi} and @code{lo} registers. +The concatenated @code{hi} and @code{lo} registers. Use this register +to store doubleword values. @item c A register suitable for use in an indirect jump. This will always be Index: gcc/longlong.h =================================================================== --- gcc/longlong.h 2008-06-09 21:42:44.000000000 +0100 +++ gcc/longlong.h 2008-06-09 21:45:18.000000000 +0100 @@ -623,12 +623,12 @@ #define UDIV_TIME 150 #endif /* __m88000__ */ #if defined (__mips__) && W_TYPE_SIZE == 32 -#define umul_ppmm(w1, w0, u, v) \ - __asm__ ("multu %2,%3" \ - : "=l" ((USItype) (w0)), \ - "=h" ((USItype) (w1)) \ - : "d" ((USItype) (u)), \ - "d" ((USItype) (v))) +#define umul_ppmm(w1, w0, u, v) \ + do { \ + UDItype __x = (UDItype) (USItype) (u) * (USItype) (v); \ + (w1) = (USItype) (__x >> 32); \ + (w0) = (USItype) (__x); \ + } while (0) #define UMUL_TIME 10 #define UDIV_TIME 100 Index: gcc/config/mips/mips.c =================================================================== --- gcc/config/mips/mips.c 2008-06-09 21:42:44.000000000 +0100 +++ gcc/config/mips/mips.c 2008-06-09 21:45:18.000000000 +0100 @@ -2675,23 +2675,6 @@ mips_legitimize_move (enum machine_mode return true; } - /* Check for individual, fully-reloaded mflo and mfhi instructions. */ - if (GET_MODE_SIZE (mode) <= UNITS_PER_WORD - && REG_P (src) && MD_REG_P (REGNO (src)) - && REG_P (dest) && GP_REG_P (REGNO (dest))) - { - int other_regno = REGNO (src) == HI_REGNUM ? LO_REGNUM : HI_REGNUM; - if (GET_MODE_SIZE (mode) <= 4) - emit_insn (gen_mfhilo_si (gen_lowpart (SImode, dest), - gen_lowpart (SImode, src), - gen_rtx_REG (SImode, other_regno))); - else - emit_insn (gen_mfhilo_di (gen_lowpart (DImode, dest), - gen_lowpart (DImode, src), - gen_rtx_REG (DImode, other_regno))); - return true; - } - /* We need to deal with constants that would be legitimate immediate_operands but aren't legitimate move_operands. */ if (CONSTANT_P (src) && !move_operand (src, mode)) @@ -3488,7 +3471,7 @@ mips_subword (rtx op, bool high_p) mode = GET_MODE (op); if (mode == VOIDmode) - mode = DImode; + mode = TARGET_64BIT ? TImode : DImode; if (TARGET_BIG_ENDIAN ? !high_p : high_p) byte = UNITS_PER_WORD; @@ -3539,6 +3522,8 @@ mips_split_64bit_move_p (rtx dest, rtx s void mips_split_doubleword_move (rtx dest, rtx src) { + rtx low_dest; + if (FP_REG_RTX_P (dest) || FP_REG_RTX_P (src)) { if (!TARGET_64BIT && GET_MODE (dest) == DImode) @@ -3552,12 +3537,27 @@ mips_split_doubleword_move (rtx dest, rt else gcc_unreachable (); } + else if (REG_P (dest) && REGNO (dest) == MD_REG_FIRST) + { + low_dest = mips_subword (dest, false); + mips_emit_move (low_dest, mips_subword (src, false)); + if (TARGET_64BIT) + emit_insn (gen_mthidi_ti (dest, mips_subword (src, true), low_dest)); + else + emit_insn (gen_mthisi_di (dest, mips_subword (src, true), low_dest)); + } + else if (REG_P (src) && REGNO (src) == MD_REG_FIRST) + { + mips_emit_move (mips_subword (dest, false), mips_subword (src, false)); + if (TARGET_64BIT) + emit_insn (gen_mfhidi_ti (mips_subword (dest, true), src)); + else + emit_insn (gen_mfhisi_di (mips_subword (dest, true), src)); + } else { /* The operation can be split into two normal moves. Decide in which order to do them. */ - rtx low_dest; - low_dest = mips_subword (dest, false); if (REG_P (low_dest) && reg_overlap_mentioned_p (low_dest, src)) @@ -3600,8 +3600,9 @@ mips_output_move (rtx dest, rtx src) if (GP_REG_P (REGNO (dest))) return "move\t%0,%z1"; - if (MD_REG_P (REGNO (dest))) - return "mt%0\t%z1"; + /* Moves to HI are handled by special .md insns. */ + if (REGNO (dest) == LO_REGNUM) + return "mtlo\t%z1"; if (DSP_ACC_REG_P (REGNO (dest))) { @@ -3624,14 +3625,29 @@ mips_output_move (rtx dest, rtx src) } } if (dest_code == MEM) - return dbl_p ? "sd\t%z1,%0" : "sw\t%z1,%0"; + switch (GET_MODE_SIZE (mode)) + { + case 1: return "sb\t%z1,%0"; + case 2: return "sh\t%z1,%0"; + case 4: return "sw\t%z1,%0"; + case 8: return "sd\t%z1,%0"; + } } if (dest_code == REG && GP_REG_P (REGNO (dest))) { if (src_code == REG) { - /* Handled by separate patterns. */ - gcc_assert (!MD_REG_P (REGNO (src))); + /* Moves from HI are handled by special .md insns. */ + if (REGNO (src) == LO_REGNUM) + { + /* When generating VR4120 or VR4130 code, we use MACC and + DMACC instead of MFLO. This avoids both the normal + MIPS III HI/LO hazards and the errata related to + -mfix-vr4130. */ + if (ISA_HAS_MACCHI) + return dbl_p ? "dmacc\t%0,%.,%." : "macc\t%0,%.,%."; + return "mflo\t%0"; + } if (DSP_ACC_REG_P (REGNO (src))) { @@ -3658,7 +3674,13 @@ mips_output_move (rtx dest, rtx src) } if (src_code == MEM) - return dbl_p ? "ld\t%0,%1" : "lw\t%0,%1"; + switch (GET_MODE_SIZE (mode)) + { + case 1: return "lbu\t%0,%1"; + case 2: return "lhu\t%0,%1"; + case 4: return "lw\t%0,%1"; + case 8: return "ld\t%0,%1"; + } if (src_code == CONST_INT) { @@ -8954,13 +8976,30 @@ mips_hard_regno_mode_ok_p (unsigned int if (ACC_REG_P (regno) && (INTEGRAL_MODE_P (mode) || ALL_FIXED_POINT_MODE_P (mode))) { - if (size <= UNITS_PER_WORD) - return true; + if (MD_REG_P (regno)) + { + /* After a multiplication or division, clobbering HI makes + the value of LO unpredictable, and vice versa. This means + that, for all interesting cases, HI and LO are effectively + a single register. - if (size <= UNITS_PER_WORD * 2) - return (DSP_ACC_REG_P (regno) - ? ((regno - DSP_ACC_REG_FIRST) & 1) == 0 - : regno == MD_REG_FIRST); + We model this by requiring that any value that uses HI + also uses LO. */ + if (size <= UNITS_PER_WORD * 2) + return regno == (size <= UNITS_PER_WORD ? LO_REGNUM : MD_REG_FIRST); + } + else + { + /* DSP accumulators do not have the same restrictions as + HI and LO, so we can treat them as normal doubleword + registers. */ + if (size <= UNITS_PER_WORD) + return true; + + if (size <= UNITS_PER_WORD * 2 + && ((regno - DSP_ACC_REG_FIRST) & 1) == 0) + return true; + } } if (ALL_COP_REG_P (regno)) Index: gcc/config/mips/constraints.md =================================================================== --- gcc/config/mips/constraints.md 2008-06-09 21:42:44.000000000 +0100 +++ gcc/config/mips/constraints.md 2008-06-09 21:45:18.000000000 +0100 @@ -29,14 +29,16 @@ (define_register_constraint "t" "T_REG" (define_register_constraint "f" "TARGET_HARD_FLOAT ? FP_REGS : NO_REGS" "A floating-point register (if available).") -(define_register_constraint "h" "TARGET_BIG_ENDIAN ? MD0_REG : MD1_REG" - "The @code{hi} register.") +(define_register_constraint "h" "NO_REGS" + "Formerly the @code{hi} register. This constraint is no longer supported.") (define_register_constraint "l" "TARGET_BIG_ENDIAN ? MD1_REG : MD0_REG" - "The @code{lo} register.") + "The @code{lo} register. Use this register to store values that are + no bigger than a word.") (define_register_constraint "x" "MD_REGS" - "The @code{hi} and @code{lo} registers.") + "The concatenated @code{hi} and @code{lo} registers. Use this register + to store doubleword values.") (define_register_constraint "b" "ALL_REGS" "@internal") Index: gcc/config/mips/mips.md =================================================================== --- gcc/config/mips/mips.md 2008-06-09 21:42:44.000000000 +0100 +++ gcc/config/mips/mips.md 2008-06-09 21:45:18.000000000 +0100 @@ -44,9 +44,11 @@ (define_constants (UNSPEC_LOAD_CALL 23) (UNSPEC_LOAD_GOT 24) (UNSPEC_GP 25) - (UNSPEC_MFHILO 26) - (UNSPEC_TLS_LDM 27) - (UNSPEC_TLS_GET_TP 28) + (UNSPEC_MFHI 26) + (UNSPEC_MTHI 27) + (UNSPEC_SET_HILO 28) + (UNSPEC_TLS_LDM 29) + (UNSPEC_TLS_GET_TP 30) (UNSPEC_MFHC1 31) (UNSPEC_MTHC1 32) (UNSPEC_CLEAR_HAZARD 33) @@ -484,6 +486,10 @@ (define_mode_iterator GPR [SI (DI "TARGE ;; modes. (define_mode_iterator GPR2 [SI (DI "TARGET_64BIT")]) +;; This mode iterator allows :HILO to be used as the mode of the +;; concatenated HI and LO registers. +(define_mode_iterator HILO [(DI "!TARGET_64BIT") (TI "TARGET_64BIT")]) + ;; This mode iterator allows :P to be used for patterns that operate on ;; pointer-sized quantities. Exactly one of the two alternatives will match. (define_mode_iterator P [(SI "Pmode == SImode") (DI "Pmode == DImode")]) @@ -497,7 +503,7 @@ (define_mode_iterator MOVE64 [DI DF (V2SF "TARGET_HARD_FLOAT && TARGET_PAIRED_SINGLE_FLOAT")]) ;; 128-bit modes for which we provide move patterns on 64-bit targets. -(define_mode_iterator MOVE128 [TF]) +(define_mode_iterator MOVE128 [TI TF]) ;; This mode iterator allows the QI and HI extension patterns to be ;; defined from the same template. @@ -613,6 +619,10 @@ (define_code_iterator any_extend [sign_e ;; from the same template. (define_code_iterator any_shift [ashift ashiftrt lshiftrt]) +;; This code iterator allows unsigned and signed division to be generated +;; from the same template. +(define_code_iterator any_div [div udiv]) + ;; This code iterator allows all native floating-point comparisons to be ;; generated from the same template. (define_code_iterator fcond [unordered uneq unlt unle eq lt le]) @@ -631,6 +641,7 @@ (define_code_iterator any_le [le leu]) ;; <u> expands to an empty string when doing a signed operation and ;; "u" when doing an unsigned operation. (define_code_attr u [(sign_extend "") (zero_extend "u") + (div "") (udiv "u") (gt "") (gtu "u") (ge "") (geu "u") (lt "") (ltu "u") @@ -865,13 +876,10 @@ (define_insn "*add<mode>3_mips16" ;; simply adding a constant to a register. (define_split - [(set (match_operand:SI 0 "register_operand") + [(set (match_operand:SI 0 "d_operand") (plus:SI (match_dup 0) (match_operand:SI 1 "const_int_operand")))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) > 0x7f && INTVAL (operands[1]) <= 0x7f + 0x7f) || (INTVAL (operands[1]) < - 0x80 @@ -894,16 +902,11 @@ (define_split }) (define_split - [(set (match_operand:SI 0 "register_operand") - (plus:SI (match_operand:SI 1 "register_operand") + [(set (match_operand:SI 0 "d_operand") + (plus:SI (match_operand:SI 1 "d_operand") (match_operand:SI 2 "const_int_operand")))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && REG_P (operands[1]) - && M16_REG_P (REGNO (operands[1])) && REGNO (operands[0]) != REGNO (operands[1]) - && GET_CODE (operands[2]) == CONST_INT && ((INTVAL (operands[2]) > 0x7 && INTVAL (operands[2]) <= 0x7 + 0x7f) || (INTVAL (operands[2]) < - 0x8 @@ -926,13 +929,10 @@ (define_split }) (define_split - [(set (match_operand:DI 0 "register_operand") + [(set (match_operand:DI 0 "d_operand") (plus:DI (match_dup 0) (match_operand:DI 1 "const_int_operand")))] "TARGET_MIPS16 && TARGET_64BIT && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) > 0xf && INTVAL (operands[1]) <= 0xf + 0xf) || (INTVAL (operands[1]) < - 0x10 @@ -955,16 +955,11 @@ (define_split }) (define_split - [(set (match_operand:DI 0 "register_operand") - (plus:DI (match_operand:DI 1 "register_operand") + [(set (match_operand:DI 0 "d_operand") + (plus:DI (match_operand:DI 1 "d_operand") (match_operand:DI 2 "const_int_operand")))] "TARGET_MIPS16 && TARGET_64BIT && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && REG_P (operands[1]) - && M16_REG_P (REGNO (operands[1])) && REGNO (operands[0]) != REGNO (operands[1]) - && GET_CODE (operands[2]) == CONST_INT && ((INTVAL (operands[2]) > 0x7 && INTVAL (operands[2]) <= 0x7 + 0xf) || (INTVAL (operands[2]) < - 0x8 @@ -1175,8 +1170,7 @@ (define_insn "mulsi3_mult3" [(set (match_operand:SI 0 "register_operand" "=d,l") (mult:SI (match_operand:SI 1 "register_operand" "d,d") (match_operand:SI 2 "register_operand" "d,d"))) - (clobber (match_scratch:SI 3 "=h,h")) - (clobber (match_scratch:SI 4 "=l,X"))] + (clobber (match_scratch:SI 3 "=l,X"))] "ISA_HAS_MUL3" { if (which_alternative == 1) @@ -1195,30 +1189,26 @@ (define_insn "mulsi3_mult3" ;; Operand 0: LO ;; Operand 1: GPR (1st multiplication operand) ;; Operand 2: GPR (2nd multiplication operand) -;; Operand 3: HI -;; Operand 4: GPR (destination) +;; Operand 3: GPR (destination) (define_peephole2 [(parallel - [(set (match_operand:SI 0 "register_operand") - (mult:SI (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand"))) - (clobber (match_operand:SI 3 "register_operand")) + [(set (match_operand:SI 0 "lo_operand") + (mult:SI (match_operand:SI 1 "d_operand") + (match_operand:SI 2 "d_operand"))) (clobber (scratch:SI))]) - (set (match_operand:SI 4 "register_operand") - (unspec [(match_dup 0) (match_dup 3)] UNSPEC_MFHILO))] + (set (match_operand:SI 3 "d_operand") + (match_dup 0))] "ISA_HAS_MUL3 && peep2_reg_dead_p (2, operands[0])" [(parallel - [(set (match_dup 4) + [(set (match_dup 3) (mult:SI (match_dup 1) (match_dup 2))) - (clobber (match_dup 3)) (clobber (match_dup 0))])]) (define_insn "mul<mode>3_internal" [(set (match_operand:GPR 0 "register_operand" "=l") (mult:GPR (match_operand:GPR 1 "register_operand" "d") - (match_operand:GPR 2 "register_operand" "d"))) - (clobber (match_scratch:GPR 3 "=h"))] + (match_operand:GPR 2 "register_operand" "d")))] "!TARGET_FIX_R4000" "<d>mult\t%1,%2" [(set_attr "type" "imul") @@ -1228,8 +1218,7 @@ (define_insn "mul<mode>3_r4000" [(set (match_operand:GPR 0 "register_operand" "=d") (mult:GPR (match_operand:GPR 1 "register_operand" "d") (match_operand:GPR 2 "register_operand" "d"))) - (clobber (match_scratch:GPR 3 "=h")) - (clobber (match_scratch:GPR 4 "=l"))] + (clobber (match_scratch:GPR 3 "=l"))] "TARGET_FIX_R4000" "<d>mult\t%1,%2\;mflo\t%0" [(set_attr "type" "imul") @@ -1243,16 +1232,13 @@ (define_insn "mul<mode>3_r4000" ;; Operand 0: LO ;; Operand 1: GPR (1st multiplication operand) ;; Operand 2: GPR (2nd multiplication operand) -;; Operand 3: HI -;; Operand 4: GPR (destination) +;; Operand 3: GPR (destination) (define_peephole2 - [(parallel - [(set (match_operand:SI 0 "register_operand") - (mult:SI (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand"))) - (clobber (match_operand:SI 3 "register_operand"))]) - (set (match_operand:SI 4 "register_operand") - (unspec:SI [(match_dup 0) (match_dup 3)] UNSPEC_MFHILO))] + [(set (match_operand:SI 0 "lo_operand") + (mult:SI (match_operand:SI 1 "d_operand") + (match_operand:SI 2 "d_operand"))) + (set (match_operand:SI 3 "d_operand") + (match_dup 0))] "ISA_HAS_MACC && !ISA_HAS_MUL3" [(set (match_dup 0) (const_int 0)) @@ -1261,11 +1247,10 @@ (define_peephole2 (plus:SI (mult:SI (match_dup 1) (match_dup 2)) (match_dup 0))) - (set (match_dup 4) + (set (match_dup 3) (plus:SI (mult:SI (match_dup 1) (match_dup 2)) - (match_dup 0))) - (clobber (match_dup 3))])]) + (match_dup 0)))])]) ;; Multiply-accumulate patterns @@ -1284,9 +1269,8 @@ (define_insn "*mul_acc_si" (plus:SI (mult:SI (match_operand:SI 1 "register_operand" "d,d,d") (match_operand:SI 2 "register_operand" "d,d,d")) (match_operand:SI 3 "register_operand" "0,l,*d"))) - (clobber (match_scratch:SI 4 "=h,h,h")) - (clobber (match_scratch:SI 5 "=X,3,l")) - (clobber (match_scratch:SI 6 "=X,X,&d"))] + (clobber (match_scratch:SI 4 "=X,3,l")) + (clobber (match_scratch:SI 5 "=X,X,&d"))] "(TARGET_MIPS3900 || GENERATE_MADD_MSUB) && !TARGET_MIPS16" @@ -1302,53 +1286,45 @@ (define_insn "*mul_acc_si" (set_attr "mode" "SI") (set_attr "length" "4,4,8")]) -;; Split the above insn if we failed to get LO allocated. +;; Split *mul_acc_si if both the source and destination accumulator +;; values are GPRs. (define_split - [(set (match_operand:SI 0 "register_operand") - (plus:SI (mult:SI (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand")) - (match_operand:SI 3 "register_operand"))) - (clobber (match_scratch:SI 4)) - (clobber (match_scratch:SI 5)) - (clobber (match_scratch:SI 6))] - "reload_completed && !TARGET_DEBUG_D_MODE - && GP_REG_P (true_regnum (operands[0])) - && GP_REG_P (true_regnum (operands[3]))" - [(parallel [(set (match_dup 6) + [(set (match_operand:SI 0 "d_operand") + (plus:SI (mult:SI (match_operand:SI 1 "d_operand") + (match_operand:SI 2 "d_operand")) + (match_operand:SI 3 "d_operand"))) + (clobber (match_operand:SI 4 "lo_operand")) + (clobber (match_operand:SI 5 "d_operand"))] + "reload_completed && !TARGET_DEBUG_D_MODE" + [(parallel [(set (match_dup 5) (mult:SI (match_dup 1) (match_dup 2))) - (clobber (match_dup 4)) - (clobber (match_dup 5))]) - (set (match_dup 0) (plus:SI (match_dup 6) (match_dup 3)))] + (clobber (match_dup 4))]) + (set (match_dup 0) (plus:SI (match_dup 5) (match_dup 3)))] "") -;; Splitter to copy result of MADD to a general register +;; Split *mul_acc_si if the destination accumulator value is in a GPR +;; and the source accumulator value is in LO. (define_split - [(set (match_operand:SI 0 "register_operand") - (plus:SI (mult:SI (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand")) - (match_operand:SI 3 "register_operand"))) - (clobber (match_scratch:SI 4)) - (clobber (match_scratch:SI 5)) - (clobber (match_scratch:SI 6))] - "reload_completed && !TARGET_DEBUG_D_MODE - && GP_REG_P (true_regnum (operands[0])) - && true_regnum (operands[3]) == LO_REGNUM" + [(set (match_operand:SI 0 "d_operand") + (plus:SI (mult:SI (match_operand:SI 1 "d_operand") + (match_operand:SI 2 "d_operand")) + (match_operand:SI 3 "lo_operand"))) + (clobber (match_dup 3)) + (clobber (scratch:SI))] + "reload_completed && !TARGET_DEBUG_D_MODE" [(parallel [(set (match_dup 3) (plus:SI (mult:SI (match_dup 1) (match_dup 2)) (match_dup 3))) - (clobber (match_dup 4)) - (clobber (match_dup 5)) - (clobber (match_dup 6))]) - (set (match_dup 0) (unspec:SI [(match_dup 5) (match_dup 4)] UNSPEC_MFHILO))] - "") + (clobber (scratch:SI)) + (clobber (scratch:SI))]) + (set (match_dup 0) (match_dup 3))]) (define_insn "*macc" [(set (match_operand:SI 0 "register_operand" "=l,d") (plus:SI (mult:SI (match_operand:SI 1 "register_operand" "d,d") (match_operand:SI 2 "register_operand" "d,d")) (match_operand:SI 3 "register_operand" "0,l"))) - (clobber (match_scratch:SI 4 "=h,h")) - (clobber (match_scratch:SI 5 "=X,3"))] + (clobber (match_scratch:SI 4 "=X,3"))] "ISA_HAS_MACC" { if (which_alternative == 1) @@ -1369,8 +1345,7 @@ (define_insn "*msac" (minus:SI (match_operand:SI 1 "register_operand" "0,l") (mult:SI (match_operand:SI 2 "register_operand" "d,d") (match_operand:SI 3 "register_operand" "d,d")))) - (clobber (match_scratch:SI 4 "=h,h")) - (clobber (match_scratch:SI 5 "=X,1"))] + (clobber (match_scratch:SI 4 "=X,1"))] "ISA_HAS_MSAC" { if (which_alternative == 1) @@ -1389,21 +1364,19 @@ (define_insn_and_split "*msac_using_macc (minus:SI (match_operand:SI 1 "register_operand" "0,l") (mult:SI (match_operand:SI 2 "register_operand" "d,d") (match_operand:SI 3 "register_operand" "d,d")))) - (clobber (match_scratch:SI 4 "=h,h")) - (clobber (match_scratch:SI 5 "=X,1")) - (clobber (match_scratch:SI 6 "=d,d"))] + (clobber (match_scratch:SI 4 "=X,1")) + (clobber (match_scratch:SI 5 "=d,d"))] "ISA_HAS_MACC && !ISA_HAS_MSAC" "#" "&& reload_completed" - [(set (match_dup 6) + [(set (match_dup 5) (neg:SI (match_dup 3))) (parallel [(set (match_dup 0) (plus:SI (mult:SI (match_dup 2) - (match_dup 6)) + (match_dup 5)) (match_dup 1))) - (clobber (match_dup 4)) - (clobber (match_dup 5))])] + (clobber (match_dup 4))])] "" [(set_attr "type" "imadd") (set_attr "length" "8")]) @@ -1418,8 +1391,7 @@ (define_insn "*macc2" (set (match_operand:SI 3 "register_operand" "=d") (plus:SI (mult:SI (match_dup 1) (match_dup 2)) - (match_dup 0))) - (clobber (match_scratch:SI 4 "=h"))] + (match_dup 0)))] "ISA_HAS_MACC && reload_completed" "macc\t%3,%1,%2" [(set_attr "type" "imadd") @@ -1433,8 +1405,7 @@ (define_insn "*msac2" (set (match_operand:SI 3 "register_operand" "=d") (minus:SI (match_dup 0) (mult:SI (match_dup 1) - (match_dup 2)))) - (clobber (match_scratch:SI 4 "=h"))] + (match_dup 2))))] "ISA_HAS_MSAC && reload_completed" "msac\t%3,%1,%2" [(set_attr "type" "imadd") @@ -1445,23 +1416,19 @@ (define_insn "*msac2" ;; ;; Operand 0: LO ;; Operand 1: macc/msac -;; Operand 2: HI -;; Operand 3: GPR (destination) +;; Operand 2: GPR (destination) (define_peephole2 [(parallel - [(set (match_operand:SI 0 "register_operand") + [(set (match_operand:SI 0 "lo_operand") (match_operand:SI 1 "macc_msac_operand")) - (clobber (match_operand:SI 2 "register_operand")) (clobber (scratch:SI))]) - (set (match_operand:SI 3 "register_operand") - (unspec:SI [(match_dup 0) (match_dup 2)] UNSPEC_MFHILO))] + (set (match_operand:SI 2 "d_operand") + (match_dup 0))] "" [(parallel [(set (match_dup 0) (match_dup 1)) - (set (match_dup 3) - (match_dup 1)) - (clobber (match_dup 2))])] - "") + (set (match_dup 2) + (match_dup 1))])]) ;; When we have a three-address multiplication instruction, it should ;; be faster to do a separate multiply and add, rather than moving @@ -1475,32 +1442,26 @@ (define_peephole2 ;; Operand 2: GPR (addend) ;; Operand 3: GPR (destination) ;; Operand 4: macc/msac -;; Operand 5: HI -;; Operand 6: new multiplication -;; Operand 7: new addition/subtraction +;; Operand 5: new multiplication +;; Operand 6: new addition/subtraction (define_peephole2 [(match_scratch:SI 0 "d") - (set (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand")) + (set (match_operand:SI 1 "lo_operand") + (match_operand:SI 2 "d_operand")) (match_dup 0) (parallel - [(set (match_operand:SI 3 "register_operand") + [(set (match_operand:SI 3 "d_operand") (match_operand:SI 4 "macc_msac_operand")) - (clobber (match_operand:SI 5 "register_operand")) (clobber (match_dup 1))])] - "ISA_HAS_MUL3 - && true_regnum (operands[1]) == LO_REGNUM - && peep2_reg_dead_p (2, operands[1]) - && GP_REG_P (true_regnum (operands[3]))" + "ISA_HAS_MUL3 && peep2_reg_dead_p (2, operands[1])" [(parallel [(set (match_dup 0) - (match_dup 6)) - (clobber (match_dup 5)) + (match_dup 5)) (clobber (match_dup 1))]) (set (match_dup 3) - (match_dup 7))] + (match_dup 6))] { - operands[6] = XEXP (operands[4], GET_CODE (operands[4]) == PLUS ? 0 : 1); - operands[7] = gen_rtx_fmt_ee (GET_CODE (operands[4]), SImode, + operands[5] = XEXP (operands[4], GET_CODE (operands[4]) == PLUS ? 0 : 1); + operands[6] = gen_rtx_fmt_ee (GET_CODE (operands[4]), SImode, operands[2], operands[0]); }) @@ -1510,33 +1471,30 @@ (define_peephole2 ;; Operand 1: LO ;; Operand 2: GPR (addend) ;; Operand 3: macc/msac -;; Operand 4: HI -;; Operand 5: GPR (destination) -;; Operand 6: new multiplication -;; Operand 7: new addition/subtraction +;; Operand 4: GPR (destination) +;; Operand 5: new multiplication +;; Operand 6: new addition/subtraction (define_peephole2 [(match_scratch:SI 0 "d") - (set (match_operand:SI 1 "register_operand") - (match_operand:SI 2 "register_operand")) + (set (match_operand:SI 1 "lo_operand") + (match_operand:SI 2 "d_operand")) (match_dup 0) (parallel [(set (match_dup 1) (match_operand:SI 3 "macc_msac_operand")) - (clobber (match_operand:SI 4 "register_operand")) (clobber (scratch:SI))]) (match_dup 0) - (set (match_operand:SI 5 "register_operand") - (unspec:SI [(match_dup 1) (match_dup 4)] UNSPEC_MFHILO))] + (set (match_operand:SI 4 "d_operand") + (match_dup 1))] "ISA_HAS_MUL3 && peep2_reg_dead_p (3, operands[1])" [(parallel [(set (match_dup 0) - (match_dup 6)) - (clobber (match_dup 4)) + (match_dup 5)) (clobber (match_dup 1))]) - (set (match_dup 5) - (match_dup 7))] + (set (match_dup 4) + (match_dup 6))] { - operands[6] = XEXP (operands[4], GET_CODE (operands[4]) == PLUS ? 0 : 1); - operands[7] = gen_rtx_fmt_ee (GET_CODE (operands[4]), SImode, + operands[5] = XEXP (operands[3], GET_CODE (operands[3]) == PLUS ? 0 : 1); + operands[6] = gen_rtx_fmt_ee (GET_CODE (operands[3]), SImode, operands[2], operands[0]); }) @@ -1545,9 +1503,8 @@ (define_insn "*mul_sub_si" (minus:SI (match_operand:SI 1 "register_operand" "0,l,*d") (mult:SI (match_operand:SI 2 "register_operand" "d,d,d") (match_operand:SI 3 "register_operand" "d,d,d")))) - (clobber (match_scratch:SI 4 "=h,h,h")) - (clobber (match_scratch:SI 5 "=X,1,l")) - (clobber (match_scratch:SI 6 "=X,X,&d"))] + (clobber (match_scratch:SI 4 "=X,1,l")) + (clobber (match_scratch:SI 5 "=X,X,&d"))] "GENERATE_MADD_MSUB" "@ msub\t%2,%3 @@ -1557,52 +1514,45 @@ (define_insn "*mul_sub_si" (set_attr "mode" "SI") (set_attr "length" "4,8,8")]) -;; Split the above insn if we failed to get LO allocated. +;; Split *mul_sub_si if both the source and destination accumulator +;; values are GPRs. (define_split - [(set (match_operand:SI 0 "register_operand") - (minus:SI (match_operand:SI 1 "register_operand") - (mult:SI (match_operand:SI 2 "register_operand") - (match_operand:SI 3 "register_operand")))) - (clobber (match_scratch:SI 4)) - (clobber (match_scratch:SI 5)) - (clobber (match_scratch:SI 6))] - "reload_completed && !TARGET_DEBUG_D_MODE - && GP_REG_P (true_regnum (operands[0])) - && GP_REG_P (true_regnum (operands[1]))" - [(parallel [(set (match_dup 6) + [(set (match_operand:SI 0 "d_operand") + (minus:SI (match_operand:SI 1 "d_operand") + (mult:SI (match_operand:SI 2 "d_operand") + (match_operand:SI 3 "d_operand")))) + (clobber (match_operand:SI 4 "lo_operand")) + (clobber (match_operand:SI 5 "d_operand"))] + "reload_completed && !TARGET_DEBUG_D_MODE" + [(parallel [(set (match_dup 5) (mult:SI (match_dup 2) (match_dup 3))) - (clobber (match_dup 4)) - (clobber (match_dup 5))]) - (set (match_dup 0) (minus:SI (match_dup 1) (match_dup 6)))] + (clobber (match_dup 4))]) + (set (match_dup 0) (minus:SI (match_dup 1) (match_dup 5)))] "") -;; Splitter to copy result of MSUB to a general register +;; Split *mul_acc_si if the destination accumulator value is in a GPR +;; and the source accumulator value is in LO. (define_split - [(set (match_operand:SI 0 "register_operand") - (minus:SI (match_operand:SI 1 "register_operand") - (mult:SI (match_operand:SI 2 "register_operand") - (match_operand:SI 3 "register_operand")))) - (clobber (match_scratch:SI 4)) - (clobber (match_scratch:SI 5)) - (clobber (match_scratch:SI 6))] - "reload_completed && !TARGET_DEBUG_D_MODE - && GP_REG_P (true_regnum (operands[0])) - && true_regnum (operands[1]) == LO_REGNUM" + [(set (match_operand:SI 0 "d_operand") + (minus:SI (match_operand:SI 1 "lo_operand") + (mult:SI (match_operand:SI 2 "d_operand") + (match_operand:SI 3 "d_operand")))) + (clobber (match_dup 1)) + (clobber (scratch:SI))] + "reload_completed && !TARGET_DEBUG_D_MODE" [(parallel [(set (match_dup 1) (minus:SI (match_dup 1) (mult:SI (match_dup 2) (match_dup 3)))) - (clobber (match_dup 4)) - (clobber (match_dup 5)) - (clobber (match_dup 6))]) - (set (match_dup 0) (unspec:SI [(match_dup 5) (match_dup 4)] UNSPEC_MFHILO))] + (clobber (scratch:SI)) + (clobber (scratch:SI))]) + (set (match_dup 0) (match_dup 1))] "") (define_insn "*muls" - [(set (match_operand:SI 0 "register_operand" "=l,d") + [(set (match_operand:SI 0 "register_operand" "=l,d") (neg:SI (mult:SI (match_operand:SI 1 "register_operand" "d,d") (match_operand:SI 2 "register_operand" "d,d")))) - (clobber (match_scratch:SI 3 "=h,h")) - (clobber (match_scratch:SI 4 "=X,l"))] + (clobber (match_scratch:SI 3 "=X,l"))] "ISA_HAS_MULS" "@ muls\t$0,%1,%2 @@ -1610,31 +1560,23 @@ (define_insn "*muls" [(set_attr "type" "imul,imul3") (set_attr "mode" "SI")]) -;; ??? We could define a mulditi3 pattern when TARGET_64BIT. - (define_expand "<u>mulsidi3" - [(parallel - [(set (match_operand:DI 0 "register_operand") - (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand")) - (any_extend:DI (match_operand:SI 2 "register_operand")))) - (clobber (scratch:DI)) - (clobber (scratch:DI)) - (clobber (scratch:DI))])] + [(set (match_operand:DI 0 "register_operand") + (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand")) + (any_extend:DI (match_operand:SI 2 "register_operand"))))] "!TARGET_64BIT || !TARGET_FIX_R4000" { - if (!TARGET_64BIT) - { - if (!TARGET_FIX_R4000) - emit_insn (gen_<u>mulsidi3_32bit_internal (operands[0], operands[1], - operands[2])); - else - emit_insn (gen_<u>mulsidi3_32bit_r4000 (operands[0], operands[1], - operands[2])); - DONE; - } + if (TARGET_64BIT) + emit_insn (gen_<u>mulsidi3_64bit (operands[0], operands[1], operands[2])); + else if (TARGET_FIX_R4000) + emit_insn (gen_<u>mulsidi3_32bit_r4000 (operands[0], operands[1], + operands[2])); + else + emit_insn (gen_<u>mulsidi3_32bit (operands[0], operands[1], operands[2])); + DONE; }) -(define_insn "<u>mulsidi3_32bit_internal" +(define_insn "<u>mulsidi3_32bit" [(set (match_operand:DI 0 "register_operand" "=x") (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand" "d")) (any_extend:DI (match_operand:SI 2 "register_operand" "d"))))] @@ -1649,42 +1591,35 @@ (define_insn "<u>mulsidi3_32bit_r4000" (any_extend:DI (match_operand:SI 2 "register_operand" "d")))) (clobber (match_scratch:DI 3 "=x"))] "!TARGET_64BIT && TARGET_FIX_R4000" - "mult<u>\t%1,%2\;mflo\t%L0;mfhi\t%M0" + "mult<u>\t%1,%2\;mflo\t%L0\;mfhi\t%M0" [(set_attr "type" "imul") (set_attr "mode" "SI") (set_attr "length" "12")]) -(define_insn_and_split "*<u>mulsidi3_64bit" +(define_insn_and_split "<u>mulsidi3_64bit" [(set (match_operand:DI 0 "register_operand" "=d") (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand" "d")) (any_extend:DI (match_operand:SI 2 "register_operand" "d")))) - (clobber (match_scratch:DI 3 "=l")) - (clobber (match_scratch:DI 4 "=h")) - (clobber (match_scratch:DI 5 "=d"))] + (clobber (match_scratch:TI 3 "=x")) + (clobber (match_scratch:DI 4 "=d"))] "TARGET_64BIT && !TARGET_FIX_R4000" "#" "&& reload_completed" - [(parallel - [(set (match_dup 3) - (sign_extend:DI - (mult:SI (match_dup 1) - (match_dup 2)))) - (set (match_dup 4) - (ashiftrt:DI - (mult:DI (any_extend:DI (match_dup 1)) - (any_extend:DI (match_dup 2))) - (const_int 32)))]) - - ;; OP5 <- LO, OP0 <- HI - (set (match_dup 5) (unspec:DI [(match_dup 3) (match_dup 4)] UNSPEC_MFHILO)) - (set (match_dup 0) (unspec:DI [(match_dup 4) (match_dup 3)] UNSPEC_MFHILO)) - - ;; Zero-extend OP5. - (set (match_dup 5) - (ashift:DI (match_dup 5) + [(set (match_dup 3) + (unspec:TI [(mult:DI (any_extend:DI (match_dup 1)) + (any_extend:DI (match_dup 2)))] + UNSPEC_SET_HILO)) + + ;; OP4 <- LO, OP0 <- HI + (set (match_dup 4) (match_dup 5)) + (set (match_dup 0) (unspec:DI [(match_dup 3)] UNSPEC_MFHI)) + + ;; Zero-extend OP4. + (set (match_dup 4) + (ashift:DI (match_dup 4) (const_int 32))) - (set (match_dup 5) - (lshiftrt:DI (match_dup 5) + (set (match_dup 4) + (lshiftrt:DI (match_dup 4) (const_int 32))) ;; Shift OP0 into place. @@ -1695,24 +1630,21 @@ (define_insn_and_split "*<u>mulsidi3_64b ;; OR the two halves together (set (match_dup 0) (ior:DI (match_dup 0) - (match_dup 5)))] - "" + (match_dup 4)))] + { operands[5] = gen_rtx_REG (DImode, LO_REGNUM); } [(set_attr "type" "imul") (set_attr "mode" "SI") (set_attr "length" "24")]) -(define_insn "*<u>mulsidi3_64bit_parts" - [(set (match_operand:DI 0 "register_operand" "=l") - (sign_extend:DI - (mult:SI (match_operand:SI 2 "register_operand" "d") - (match_operand:SI 3 "register_operand" "d")))) - (set (match_operand:DI 1 "register_operand" "=h") - (ashiftrt:DI - (mult:DI (any_extend:DI (match_dup 2)) - (any_extend:DI (match_dup 3))) - (const_int 32)))] +(define_insn "<u>mulsidi3_64bit_hilo" + [(set (match_operand:TI 0 "register_operand" "=x") + (unspec:TI + [(mult:DI + (any_extend:DI (match_operand:SI 1 "register_operand" "d")) + (any_extend:DI (match_operand:SI 2 "register_operand" "d")))] + UNSPEC_SET_HILO))] "TARGET_64BIT && !TARGET_FIX_R4000" - "mult<u>\t%2,%3" + "mult<u>\t%1,%2" [(set_attr "type" "imul") (set_attr "mode" "SI")]) @@ -1756,7 +1688,7 @@ (define_expand "<su>mulsi3_highpart" (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand")) (any_extend:DI (match_operand:SI 2 "register_operand"))) (const_int 32))))] - "ISA_HAS_MULHI || !TARGET_FIX_R4000" + "" { if (ISA_HAS_MULHI) emit_insn (gen_<su>mulsi3_highpart_mulhi_internal (operands[0], @@ -1768,72 +1700,133 @@ (define_expand "<su>mulsi3_highpart" DONE; }) -(define_insn "<su>mulsi3_highpart_internal" - [(set (match_operand:SI 0 "register_operand" "=h") +(define_insn_and_split "<su>mulsi3_highpart_internal" + [(set (match_operand:SI 0 "register_operand" "=d") (truncate:SI (lshiftrt:DI (mult:DI (any_extend:DI (match_operand:SI 1 "register_operand" "d")) (any_extend:DI (match_operand:SI 2 "register_operand" "d"))) (const_int 32)))) (clobber (match_scratch:SI 3 "=l"))] - "!ISA_HAS_MULHI && !TARGET_FIX_R4000" - "mult<u>\t%1,%2" + "!ISA_HAS_MULHI" + { return TARGET_FIX_R4000 ? "mult<u>\t%1,%2\n\tmfhi\t%0" : "#"; } + "&& reload_completed && !TARGET_FIX_R4000" + [(const_int 0)] +{ + rtx hilo; + + if (TARGET_64BIT) + { + hilo = gen_rtx_REG (TImode, MD_REG_FIRST); + emit_insn (gen_<u>mulsidi3_64bit_hilo (hilo, operands[1], operands[2])); + emit_insn (gen_mfhisi_ti (operands[0], hilo)); + } + else + { + hilo = gen_rtx_REG (DImode, MD_REG_FIRST); + emit_insn (gen_<u>mulsidi3_32bit (hilo, operands[1], operands[2])); + emit_insn (gen_mfhisi_di (operands[0], hilo)); + } + DONE; +} [(set_attr "type" "imul") - (set_attr "mode" "SI")]) + (set_attr "mode" "SI") + (set_attr "length" "8")]) (define_insn "<su>mulsi3_highpart_mulhi_internal" - [(set (match_operand:SI 0 "register_operand" "=h,d") + [(set (match_operand:SI 0 "register_operand" "=d") (truncate:SI (lshiftrt:DI (mult:DI - (any_extend:DI (match_operand:SI 1 "register_operand" "d,d")) - (any_extend:DI (match_operand:SI 2 "register_operand" "d,d"))) + (any_extend:DI (match_operand:SI 1 "register_operand" "d")) + (any_extend:DI (match_operand:SI 2 "register_operand" "d"))) (const_int 32)))) - (clobber (match_scratch:SI 3 "=l,l")) - (clobber (match_scratch:SI 4 "=X,h"))] + (clobber (match_scratch:SI 3 "=l"))] "ISA_HAS_MULHI" - "@ - mult<u>\t%1,%2 - mulhi<u>\t%0,%1,%2" - [(set_attr "type" "imul,imul3") + "mulhi<u>\t%0,%1,%2" + [(set_attr "type" "imul3") (set_attr "mode" "SI")]) (define_insn "*<su>mulsi3_highpart_neg_mulhi_internal" - [(set (match_operand:SI 0 "register_operand" "=h,d") + [(set (match_operand:SI 0 "register_operand" "=d") (truncate:SI (lshiftrt:DI (neg:DI (mult:DI - (any_extend:DI (match_operand:SI 1 "register_operand" "d,d")) - (any_extend:DI (match_operand:SI 2 "register_operand" "d,d")))) + (any_extend:DI (match_operand:SI 1 "register_operand" "d")) + (any_extend:DI (match_operand:SI 2 "register_operand" "d")))) (const_int 32)))) - (clobber (match_scratch:SI 3 "=l,l")) - (clobber (match_scratch:SI 4 "=X,h"))] + (clobber (match_scratch:SI 3 "=l"))] "ISA_HAS_MULHI" - "@ - mulshi<u>\t%.,%1,%2 - mulshi<u>\t%0,%1,%2" - [(set_attr "type" "imul,imul3") + "mulshi<u>\t%0,%1,%2" + [(set_attr "type" "imul3") (set_attr "mode" "SI")]) ;; Disable unsigned multiplication for -mfix-vr4120. This is for VR4120 ;; errata MD(0), which says that dmultu does not always produce the ;; correct result. -(define_insn "<su>muldi3_highpart" - [(set (match_operand:DI 0 "register_operand" "=h") +(define_insn_and_split "<su>muldi3_highpart" + [(set (match_operand:DI 0 "register_operand" "=d") (truncate:DI (lshiftrt:TI - (mult:TI - (any_extend:TI (match_operand:DI 1 "register_operand" "d")) - (any_extend:TI (match_operand:DI 2 "register_operand" "d"))) + (mult:TI (any_extend:TI (match_operand:DI 1 "register_operand" "d")) + (any_extend:TI (match_operand:DI 2 "register_operand" "d"))) (const_int 64)))) (clobber (match_scratch:DI 3 "=l"))] - "TARGET_64BIT && !TARGET_FIX_R4000 + "TARGET_64BIT && !(<CODE> == ZERO_EXTEND && TARGET_FIX_VR4120)" + { return TARGET_FIX_R4000 ? "dmult<u>\t%1,%2\n\tmfhi\t%0" : "#"; } + "&& reload_completed && !TARGET_FIX_R4000" + [(const_int 0)] +{ + rtx hilo; + + hilo = gen_rtx_REG (TImode, MD_REG_FIRST); + emit_insn (gen_<u>mulditi3_internal (hilo, operands[1], operands[2])); + emit_insn (gen_mfhidi_ti (operands[0], hilo)); + DONE; +} + [(set_attr "type" "imul") + (set_attr "mode" "DI") + (set_attr "length" "8")]) + +(define_expand "<u>mulditi3" + [(set (match_operand:TI 0 "register_operand") + (mult:TI (any_extend:TI (match_operand:DI 1 "register_operand")) + (any_extend:TI (match_operand:DI 2 "register_operand"))))] + "TARGET_64BIT && !(<CODE> == ZERO_EXTEND && TARGET_FIX_VR4120)" +{ + if (TARGET_FIX_R4000) + emit_insn (gen_<u>mulditi3_r4000 (operands[0], operands[1], operands[2])); + else + emit_insn (gen_<u>mulditi3_internal (operands[0], operands[1], + operands[2])); + DONE; +}) + +(define_insn "<u>mulditi3_internal" + [(set (match_operand:TI 0 "register_operand" "=x") + (mult:TI (any_extend:TI (match_operand:DI 1 "register_operand" "d")) + (any_extend:TI (match_operand:DI 2 "register_operand" "d"))))] + "TARGET_64BIT + && !TARGET_FIX_R4000 && !(<CODE> == ZERO_EXTEND && TARGET_FIX_VR4120)" "dmult<u>\t%1,%2" [(set_attr "type" "imul") (set_attr "mode" "DI")]) +(define_insn "<u>mulditi3_r4000" + [(set (match_operand:TI 0 "register_operand" "=d") + (mult:TI (any_extend:TI (match_operand:DI 1 "register_operand" "d")) + (any_extend:TI (match_operand:DI 2 "register_operand" "d")))) + (clobber (match_scratch:TI 3 "=x"))] + "TARGET_64BIT + && TARGET_FIX_R4000 + && !(<CODE> == ZERO_EXTEND && TARGET_FIX_VR4120)" + "dmult<u>\t%1,%2\;mflo\t%L0\;mfhi\t%M0" + [(set_attr "type" "imul") + (set_attr "mode" "DI") + (set_attr "length" "12")]) + ;; The R4650 supports a 32-bit multiply/ 64-bit accumulate ;; instruction. The HI/LO registers are used as a 64-bit accumulator. @@ -1841,8 +1834,7 @@ (define_insn "madsi" [(set (match_operand:SI 0 "register_operand" "+l") (plus:SI (mult:SI (match_operand:SI 1 "register_operand" "d") (match_operand:SI 2 "register_operand" "d")) - (match_dup 0))) - (clobber (match_scratch:SI 3 "=h"))] + (match_dup 0)))] "TARGET_MAD" "mad\t%1,%2" [(set_attr "type" "imadd") @@ -2017,29 +2009,80 @@ (define_insn "*recip<mode>3" ;; VR4120 errata MD(A1): signed division instructions do not work correctly ;; with negative operands. We use special libgcc functions instead. -(define_insn "divmod<mode>4" +(define_insn_and_split "divmod<mode>4" [(set (match_operand:GPR 0 "register_operand" "=l") (div:GPR (match_operand:GPR 1 "register_operand" "d") (match_operand:GPR 2 "register_operand" "d"))) - (set (match_operand:GPR 3 "register_operand" "=h") + (set (match_operand:GPR 3 "register_operand" "=d") (mod:GPR (match_dup 1) (match_dup 2)))] "!TARGET_FIX_VR4120" - { return mips_output_division ("<d>div\t$0,%1,%2", operands); } - [(set_attr "type" "idiv") - (set_attr "mode" "<MODE>")]) + "#" + "&& reload_completed" + [(const_int 0)] +{ + rtx hilo; + + if (TARGET_64BIT) + { + hilo = gen_rtx_REG (TImode, MD_REG_FIRST); + emit_insn (gen_divmod<mode>4_hilo_ti (hilo, operands[1], operands[2])); + emit_insn (gen_mfhi<mode>_ti (operands[3], hilo)); + } + else + { + hilo = gen_rtx_REG (DImode, MD_REG_FIRST); + emit_insn (gen_divmod<mode>4_hilo_di (hilo, operands[1], operands[2])); + emit_insn (gen_mfhi<mode>_di (operands[3], hilo)); + } + DONE; +} + [(set_attr "type" "idiv") + (set_attr "mode" "<MODE>") + (set_attr "length" "8")]) -(define_insn "udivmod<mode>4" +(define_insn_and_split "udivmod<mode>4" [(set (match_operand:GPR 0 "register_operand" "=l") (udiv:GPR (match_operand:GPR 1 "register_operand" "d") (match_operand:GPR 2 "register_operand" "d"))) - (set (match_operand:GPR 3 "register_operand" "=h") + (set (match_operand:GPR 3 "register_operand" "=d") (umod:GPR (match_dup 1) (match_dup 2)))] "" - { return mips_output_division ("<d>divu\t$0,%1,%2", operands); } + "#" + "reload_completed" + [(const_int 0)] +{ + rtx hilo; + + if (TARGET_64BIT) + { + hilo = gen_rtx_REG (TImode, MD_REG_FIRST); + emit_insn (gen_udivmod<mode>4_hilo_ti (hilo, operands[1], operands[2])); + emit_insn (gen_mfhi<mode>_ti (operands[3], hilo)); + } + else + { + hilo = gen_rtx_REG (DImode, MD_REG_FIRST); + emit_insn (gen_udivmod<mode>4_hilo_di (hilo, operands[1], operands[2])); + emit_insn (gen_mfhi<mode>_di (operands[3], hilo)); + } + DONE; +} + [(set_attr "type" "idiv") + (set_attr "mode" "<MODE>") + (set_attr "length" "8")]) + +(define_insn "<u>divmod<GPR:mode>4_hilo_<HILO:mode>" + [(set (match_operand:HILO 0 "register_operand" "=x") + (unspec:HILO + [(any_div:GPR (match_operand:GPR 1 "register_operand" "d") + (match_operand:GPR 2 "register_operand" "d"))] + UNSPEC_SET_HILO))] + "" + { return mips_output_division ("<GPR:d>div<u>\t%.,%1,%2", operands); } [(set_attr "type" "idiv") - (set_attr "mode" "<MODE>")]) + (set_attr "mode" "<GPR:MODE>")]) \f ;; ;; .................... @@ -3241,7 +3284,7 @@ (define_insn_and_split "*lea_high64" ;; dsll32 op1,op1,0 ;; daddu op1,op1,op0 (define_peephole2 - [(set (match_operand:DI 1 "register_operand") + [(set (match_operand:DI 1 "d_operand") (high:DI (match_operand:DI 2 "absolute_symbolic_operand"))) (match_scratch:DI 0 "d")] "TARGET_EXPLICIT_RELOCS && ABI_HAS_64BIT_SYMBOLS" @@ -3294,8 +3337,8 @@ (define_insn_and_split "*lea64" ;; ;; on MIPS16 targets. (define_split - [(set (match_operand:SI 0 "register_operand" "=d") - (high:SI (match_operand:SI 1 "absolute_symbolic_operand" "")))] + [(set (match_operand:SI 0 "d_operand") + (high:SI (match_operand:SI 1 "absolute_symbolic_operand")))] "TARGET_MIPS16 && reload_completed" [(set (match_dup 0) (match_dup 2)) (set (match_dup 0) (ashift:SI (match_dup 0) (const_int 16)))] @@ -3464,7 +3507,7 @@ (define_insn "*movdi_32bit" && (register_operand (operands[0], DImode) || reg_or_0_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "multi,multi,load,store,mthilo,mfhilo,mtc,load,mfc,store") + [(set_attr "type" "multi,multi,load,store,multi,multi,mtc,load,mfc,store") (set_attr "mode" "DI") (set_attr "length" "8,16,*,*,8,8,8,*,8,*")]) @@ -3475,7 +3518,7 @@ (define_insn "*movdi_gp32_fp64" && (register_operand (operands[0], DImode) || reg_or_0_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "multi,multi,load,store,mthilo,mfhilo,mtc,fpload,mfc,fpstore") + [(set_attr "type" "multi,multi,load,store,multi,multi,mtc,fpload,mfc,fpstore") (set_attr "mode" "DI") (set_attr "length" "8,16,*,*,8,8,8,*,8,*")]) @@ -3486,29 +3529,29 @@ (define_insn "*movdi_32bit_mips16" && (register_operand (operands[0], DImode) || register_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "multi,multi,multi,multi,multi,load,store,mfhilo") + [(set_attr "type" "multi,multi,multi,multi,multi,load,store,multi") (set_attr "mode" "DI") (set_attr "length" "8,8,8,8,12,*,*,8")]) (define_insn "*movdi_64bit" - [(set (match_operand:DI 0 "nonimmediate_operand" "=d,d,e,d,m,*f,*f,*d,*m,*x,*B*C*D,*B*C*D,*d,*m") - (match_operand:DI 1 "move_operand" "d,U,T,m,dJ,*d*J,*m,*f,*f,*J*d,*d,*m,*B*C*D,*B*C*D"))] + [(set (match_operand:DI 0 "nonimmediate_operand" "=d,d,e,d,m,*f,*f,*d,*m,*a,*d,*B*C*D,*B*C*D,*d,*m") + (match_operand:DI 1 "move_operand" "d,U,T,m,dJ,*d*J,*m,*f,*f,*J*d,*a,*d,*m,*B*C*D,*B*C*D"))] "TARGET_64BIT && !TARGET_MIPS16 && (register_operand (operands[0], DImode) || reg_or_0_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "move,const,const,load,store,mtc,fpload,mfc,fpstore,mthilo,mtc,load,mfc,store") + [(set_attr "type" "move,const,const,load,store,mtc,fpload,mfc,fpstore,mthilo,mfhilo,mtc,load,mfc,store") (set_attr "mode" "DI") - (set_attr "length" "4,*,*,*,*,4,*,4,*,4,8,*,8,*")]) + (set_attr "length" "4,*,*,*,*,4,*,4,*,4,4,8,*,8,*")]) (define_insn "*movdi_64bit_mips16" - [(set (match_operand:DI 0 "nonimmediate_operand" "=d,y,d,d,d,d,d,d,m") - (match_operand:DI 1 "move_operand" "d,d,y,K,N,kf,U,m,d"))] + [(set (match_operand:DI 0 "nonimmediate_operand" "=d,y,d,d,d,d,d,d,m,*d") + (match_operand:DI 1 "move_operand" "d,d,y,K,N,kf,U,m,d,*a"))] "TARGET_64BIT && TARGET_MIPS16 && (register_operand (operands[0], DImode) || register_operand (operands[1], DImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "move,move,move,arith,arith,load,const,load,store") + [(set_attr "type" "move,move,move,arith,arith,load,const,load,store,mfhilo") (set_attr "mode" "DI") (set_attr_alternative "length" [(const_int 4) @@ -3523,7 +3566,8 @@ (define_insn "*movdi_64bit_mips16" (const_int 8) (const_string "*") (const_string "*") - (const_string "*")])]) + (const_string "*") + (const_int 4)])]) ;; On the mips16, we can split ld $r,N($r) into an add and a load, @@ -3531,14 +3575,11 @@ (define_insn "*movdi_64bit_mips16" ;; load are 2 2 byte instructions. (define_split - [(set (match_operand:DI 0 "register_operand") + [(set (match_operand:DI 0 "d_operand") (mem:DI (plus:DI (match_dup 0) (match_operand:DI 1 "const_int_operand"))))] "TARGET_64BIT && TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) < 0 && INTVAL (operands[1]) >= -0x10) || (INTVAL (operands[1]) >= 32 * 8 @@ -3589,7 +3630,7 @@ (define_expand "movsi" (define_insn "*movsi_internal" [(set (match_operand:SI 0 "nonimmediate_operand" "=d,d,e,d,m,*f,*f,*d,*m,*d,*z,*a,*d,*B*C*D,*B*C*D,*d,*m") - (match_operand:SI 1 "move_operand" "d,U,T,m,dJ,*d*J,*m,*f,*f,*z,*d,*J*d,*A,*d,*m,*B*C*D,*B*C*D"))] + (match_operand:SI 1 "move_operand" "d,U,T,m,dJ,*d*J,*m,*f,*f,*z,*d,*J*d,*a,*d,*m,*B*C*D,*B*C*D"))] "!TARGET_MIPS16 && (register_operand (operands[0], SImode) || reg_or_0_operand (operands[1], SImode))" @@ -3599,13 +3640,13 @@ (define_insn "*movsi_internal" (set_attr "length" "4,*,*,*,*,4,*,4,*,4,4,4,4,4,*,4,*")]) (define_insn "*movsi_mips16" - [(set (match_operand:SI 0 "nonimmediate_operand" "=d,y,d,d,d,d,d,d,m") - (match_operand:SI 1 "move_operand" "d,d,y,K,N,kf,U,m,d"))] + [(set (match_operand:SI 0 "nonimmediate_operand" "=d,y,d,d,d,d,d,d,m,*d") + (match_operand:SI 1 "move_operand" "d,d,y,K,N,kf,U,m,d,*a"))] "TARGET_MIPS16 && (register_operand (operands[0], SImode) || register_operand (operands[1], SImode))" { return mips_output_move (operands[0], operands[1]); } - [(set_attr "type" "move,move,move,arith,arith,load,const,load,store") + [(set_attr "type" "move,move,move,arith,arith,load,const,load,store,mfhilo") (set_attr "mode" "SI") (set_attr_alternative "length" [(const_int 4) @@ -3620,20 +3661,18 @@ (define_insn "*movsi_mips16" (const_int 8) (const_string "*") (const_string "*") - (const_string "*")])]) + (const_string "*") + (const_int 4)])]) ;; On the mips16, we can split lw $r,N($r) into an add and a load, ;; when the original load is a 4 byte instruction but the add and the ;; load are 2 2 byte instructions. (define_split - [(set (match_operand:SI 0 "register_operand") + [(set (match_operand:SI 0 "d_operand") (mem:SI (plus:SI (match_dup 0) (match_operand:SI 1 "const_int_operand"))))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) < 0 && INTVAL (operands[1]) >= -0x80) || (INTVAL (operands[1]) >= 32 * 4 @@ -3669,12 +3708,9 @@ (define_split ;; instructions. (define_split - [(set (match_operand:SI 0 "register_operand") + [(set (match_operand:SI 0 "d_operand") (match_operand:SI 1 "const_int_operand"))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && INTVAL (operands[1]) >= 0x100 && INTVAL (operands[1]) <= 0xff + 0x7f" [(set (match_dup 0) (match_dup 1)) @@ -3797,36 +3833,24 @@ (define_expand "movhi" }) (define_insn "*movhi_internal" - [(set (match_operand:HI 0 "nonimmediate_operand" "=d,d,d,m,*x") - (match_operand:HI 1 "move_operand" "d,I,m,dJ,*d"))] + [(set (match_operand:HI 0 "nonimmediate_operand" "=d,d,d,m,*a,*d") + (match_operand:HI 1 "move_operand" "d,I,m,dJ,*d*J,*a"))] "!TARGET_MIPS16 && (register_operand (operands[0], HImode) || reg_or_0_operand (operands[1], HImode))" - "@ - move\t%0,%1 - li\t%0,%1 - lhu\t%0,%1 - sh\t%z1,%0 - mt%0\t%1" - [(set_attr "type" "move,arith,load,store,mthilo") + { return mips_output_move (operands[0], operands[1]); } + [(set_attr "type" "move,arith,load,store,mthilo,mfhilo") (set_attr "mode" "HI") - (set_attr "length" "4,4,*,*,4")]) + (set_attr "length" "4,4,*,*,4,4")]) (define_insn "*movhi_mips16" - [(set (match_operand:HI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m") - (match_operand:HI 1 "move_operand" "d,d,y,K,N,m,d"))] + [(set (match_operand:HI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m,*d") + (match_operand:HI 1 "move_operand" "d,d,y,K,N,m,d,*a"))] "TARGET_MIPS16 && (register_operand (operands[0], HImode) || register_operand (operands[1], HImode))" - "@ - move\t%0,%1 - move\t%0,%1 - move\t%0,%1 - li\t%0,%1 - # - lhu\t%0,%1 - sh\t%1,%0" - [(set_attr "type" "move,move,move,arith,arith,load,store") + { return mips_output_move (operands[0], operands[1]); } + [(set_attr "type" "move,move,move,arith,arith,load,store,mfhilo") (set_attr "mode" "HI") (set_attr_alternative "length" [(const_int 4) @@ -3839,6 +3863,7 @@ (define_insn "*movhi_mips16" (const_int 8) (const_int 12)) (const_string "*") + (const_string "*") (const_string "*")])]) @@ -3847,13 +3872,10 @@ (define_insn "*movhi_mips16" ;; load are 2 2 byte instructions. (define_split - [(set (match_operand:HI 0 "register_operand") + [(set (match_operand:HI 0 "d_operand") (mem:HI (plus:SI (match_dup 0) (match_operand:SI 1 "const_int_operand"))))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) < 0 && INTVAL (operands[1]) >= -0x80) || (INTVAL (operands[1]) >= 32 * 2 @@ -3901,51 +3923,36 @@ (define_expand "movqi" }) (define_insn "*movqi_internal" - [(set (match_operand:QI 0 "nonimmediate_operand" "=d,d,d,m,*x") - (match_operand:QI 1 "move_operand" "d,I,m,dJ,*d"))] + [(set (match_operand:QI 0 "nonimmediate_operand" "=d,d,d,m,*a,*d") + (match_operand:QI 1 "move_operand" "d,I,m,dJ,*d*J,*a"))] "!TARGET_MIPS16 && (register_operand (operands[0], QImode) || reg_or_0_operand (operands[1], QImode))" - "@ - move\t%0,%1 - li\t%0,%1 - lbu\t%0,%1 - sb\t%z1,%0 - mt%0\t%1" - [(set_attr "type" "move,arith,load,store,mthilo") + { return mips_output_move (operands[0], operands[1]); } + [(set_attr "type" "move,arith,load,store,mthilo,mfhilo") (set_attr "mode" "QI") - (set_attr "length" "4,4,*,*,4")]) + (set_attr "length" "4,4,*,*,4,4")]) (define_insn "*movqi_mips16" - [(set (match_operand:QI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m") - (match_operand:QI 1 "move_operand" "d,d,y,K,N,m,d"))] + [(set (match_operand:QI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m,*d") + (match_operand:QI 1 "move_operand" "d,d,y,K,N,m,d,*a"))] "TARGET_MIPS16 && (register_operand (operands[0], QImode) || register_operand (operands[1], QImode))" - "@ - move\t%0,%1 - move\t%0,%1 - move\t%0,%1 - li\t%0,%1 - # - lbu\t%0,%1 - sb\t%1,%0" - [(set_attr "type" "move,move,move,arith,arith,load,store") + { return mips_output_move (operands[0], operands[1]); } + [(set_attr "type" "move,move,move,arith,arith,load,store,mfhilo") (set_attr "mode" "QI") - (set_attr "length" "4,4,4,4,8,*,*")]) + (set_attr "length" "4,4,4,4,8,*,*,4")]) ;; On the mips16, we can split lb $r,N($r) into an add and a load, ;; when the original load is a 4 byte instruction but the add and the ;; load are 2 2 byte instructions. (define_split - [(set (match_operand:QI 0 "register_operand") + [(set (match_operand:QI 0 "d_operand") (mem:QI (plus:SI (match_dup 0) (match_operand:SI 1 "const_int_operand"))))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && REG_P (operands[0]) - && M16_REG_P (REGNO (operands[0])) - && GET_CODE (operands[1]) == CONST_INT && ((INTVAL (operands[1]) < 0 && INTVAL (operands[1]) >= -0x80) || (INTVAL (operands[1]) >= 32 @@ -4065,6 +4072,39 @@ (define_insn "*movdf_mips16" (set_attr "mode" "DF") (set_attr "length" "8,8,8,*,*")]) +;; 128-bit integer moves + +(define_expand "movti" + [(set (match_operand:TI 0) + (match_operand:TI 1))] + "TARGET_64BIT" +{ + if (mips_legitimize_move (TImode, operands[0], operands[1])) + DONE; +}) + +(define_insn "*movti" + [(set (match_operand:TI 0 "nonimmediate_operand" "=d,d,m,*a,*d") + (match_operand:TI 1 "move_operand" "di,m,dJ,*d*J,*a"))] + "TARGET_64BIT + && !TARGET_MIPS16 + && (register_operand (operands[0], TImode) + || reg_or_0_operand (operands[1], TImode))" + "#" + [(set_attr "type" "multi,load,store,multi,multi") + (set_attr "length" "8,*,*,8,8")]) + +(define_insn "*movti_mips16" + [(set (match_operand:TI 0 "nonimmediate_operand" "=d,y,d,d,d,d,m,*d") + (match_operand:TI 1 "move_operand" "d,d,y,K,N,m,d,*a"))] + "TARGET_64BIT + && TARGET_MIPS16 + && (register_operand (operands[0], TImode) + || register_operand (operands[1], TImode))" + "#" + [(set_attr "type" "multi,multi,multi,multi,multi,load,store,multi") + (set_attr "length" "8,8,8,12,16,*,*,8")]) + ;; 128-bit floating point moves (define_expand "movtf" @@ -4123,7 +4163,7 @@ (define_split ;; When generating mips16 code, split moves of negative constants into ;; a positive "li" followed by a negation. (define_split - [(set (match_operand 0 "register_operand") + [(set (match_operand 0 "d_operand") (match_operand 1 "const_int_operand"))] "TARGET_MIPS16 && reload_completed && INTVAL (operands[1]) < 0" [(set (match_dup 2) @@ -4172,44 +4212,33 @@ (define_insn "movv2sf_hardfloat_32bit" (set_attr "mode" "SF") (set_attr "length" "4,8,*,*,*,8,8,8,*,*")]) -;; The HI and LO registers are not truly independent. If we move an mthi -;; instruction before an mflo instruction, it will make the result of the -;; mflo unpredictable. The same goes for mtlo and mfhi. -;; -;; We cope with this by making the mflo and mfhi patterns use both HI and LO. -;; Operand 1 is the register we want, operand 2 is the other one. -;; -;; When generating VR4120 or VR4130 code, we use macc{,hi} and -;; dmacc{,hi} instead of mfhi and mflo. This avoids both the normal -;; MIPS III hi/lo hazards and the errata related to -mfix-vr4130. - -(define_expand "mfhilo_<mode>" - [(set (match_operand:GPR 0 "register_operand") - (unspec:GPR [(match_operand:GPR 1 "register_operand") - (match_operand:GPR 2 "register_operand")] - UNSPEC_MFHILO))]) - -(define_insn "*mfhilo_<mode>" - [(set (match_operand:GPR 0 "register_operand" "=d,d") - (unspec:GPR [(match_operand:GPR 1 "register_operand" "h,l") - (match_operand:GPR 2 "register_operand" "l,h")] - UNSPEC_MFHILO))] - "!ISA_HAS_MACCHI" - "mf%1\t%0" +;; Extract the high part of a HI/LO value. See mips_hard_regno_mode_ok_p +;; for the reason why we can't just use (reg:GPR HI_REGNUM). +;; +;; When generating VR4120 or VR4130 code, we use MACCHI and DMACCHI +;; instead of MFHI. This avoids both the normal MIPS III hi/lo hazards +;; and the errata related to -mfix-vr4130. +(define_insn "mfhi<GPR:mode>_<HILO:mode>" + [(set (match_operand:GPR 0 "register_operand" "=d") + (unspec:GPR [(match_operand:HILO 1 "register_operand" "x")] + UNSPEC_MFHI))] + "" + { return ISA_HAS_MACCHI ? "<GPR:d>macchi\t%0,%.,%." : "mfhi\t%0"; } [(set_attr "type" "mfhilo") - (set_attr "mode" "<MODE>")]) + (set_attr "mode" "<GPR:MODE>")]) -(define_insn "*mfhilo_<mode>_macc" - [(set (match_operand:GPR 0 "register_operand" "=d,d") - (unspec:GPR [(match_operand:GPR 1 "register_operand" "h,l") - (match_operand:GPR 2 "register_operand" "l,h")] - UNSPEC_MFHILO))] - "ISA_HAS_MACCHI" - "@ - <d>macchi\t%0,%.,%. - <d>macc\t%0,%.,%." - [(set_attr "type" "mfhilo") - (set_attr "mode" "<MODE>")]) +;; Set the high part of a HI/LO value, given that the low part has +;; already been set. See mips_hard_regno_mode_ok_p for the reason +;; why we can't just use (reg:GPR HI_REGNUM). +(define_insn "mthi<GPR:mode>_<HILO:mode>" + [(set (match_operand:HILO 0 "register_operand" "=x") + (unspec:HILO [(match_operand:GPR 1 "reg_or_0_operand" "dJ") + (match_operand:GPR 2 "register_operand" "l")] + UNSPEC_MTHI))] + "" + "mthi\t%z1" + [(set_attr "type" "mthilo") + (set_attr "mode" "SI")]) ;; Emit a doubleword move in which exactly one of the operands is ;; a floating-point register. We can't just emit two normal moves @@ -5120,11 +5149,10 @@ (define_insn "*lshrdi3_mips16" ;; On the mips16, we can split a 4 byte shift into 2 2 byte shifts. (define_split - [(set (match_operand:GPR 0 "register_operand") - (any_shift:GPR (match_operand:GPR 1 "register_operand") + [(set (match_operand:GPR 0 "d_operand") + (any_shift:GPR (match_operand:GPR 1 "d_operand") (match_operand:GPR 2 "const_int_operand")))] "TARGET_MIPS16 && reload_completed && !TARGET_DEBUG_D_MODE - && GET_CODE (operands[2]) == CONST_INT && INTVAL (operands[2]) > 8 && INTVAL (operands[2]) <= 16" [(set (match_dup 0) (any_shift:GPR (match_dup 1) (const_int 8))) Index: gcc/config/mips/predicates.md =================================================================== --- gcc/config/mips/predicates.md 2008-06-09 21:42:44.000000000 +0100 +++ gcc/config/mips/predicates.md 2008-06-09 21:45:18.000000000 +0100 @@ -76,9 +76,11 @@ (define_predicate "const_0_or_1_operand" (ior (match_test "op == CONST0_RTX (GET_MODE (op))") (match_test "op == CONST1_RTX (GET_MODE (op))")))) -(define_predicate "fpr_operand" +(define_predicate "d_operand" (and (match_code "reg") - (match_test "FP_REG_P (REGNO (op))"))) + (match_test "TARGET_MIPS16 + ? M16_REG_P (REGNO (op)) + : GP_REG_P (REGNO (op))"))) (define_predicate "lo_operand" (and (match_code "reg") Index: gcc/testsuite/gcc.dg/torture/mips-hilo-1.c =================================================================== --- gcc/testsuite/gcc.dg/torture/mips-hilo-1.c 2008-06-09 21:42:44.000000000 +0100 +++ /dev/null 2008-06-08 10:32:14.544096500 +0100 @@ -1,73 +0,0 @@ -/* f1 checks that an mtlo is not moved before an mfhi. f2 does the same - for an mthi and an mflo. */ -/* { dg-do run { target mips*-*-* } } */ -/* { dg-options "-mtune=rm7000" } */ - -extern void abort (void); -extern void exit (int); - -#define DECLARE(TYPE) \ - TYPE __attribute__ ((noinline)) __attribute__ ((nomips16)) \ - f1##TYPE (TYPE x1, TYPE x2, TYPE x3) \ - { \ - TYPE t1, t2; \ - \ - asm ("mult\t%1,%2" : "=h" (t1) : "d" (x1), "d" (x2) : "lo"); \ - asm ("mflo\t%0" : "=r" (t2) : "l" (x3) : "hi"); \ - return t1 + t2; \ - } \ - \ - TYPE __attribute__ ((noinline)) __attribute__ ((nomips16)) \ - f2##TYPE (TYPE x1, TYPE x2, TYPE x3) \ - { \ - TYPE t1, t2; \ - \ - asm ("mult\t%1,%2" : "=l" (t1) : "d" (x1), "d" (x2) : "hi"); \ - asm ("mfhi\t%0" : "=r" (t2) : "h" (x3) : "lo"); \ - return t1 + t2; \ - } - -#define TEST(TYPE) \ - if (f1##TYPE (1, 2, 10) != 10) \ - abort (); \ - if (f2##TYPE (1, 2, 40) != 42) \ - abort () - -typedef char c; -typedef signed char sc; -typedef unsigned char uc; -typedef short s; -typedef unsigned short us; -typedef int i; -typedef unsigned int ui; -typedef long long ll; -typedef unsigned long long ull; - -DECLARE (c) -DECLARE (sc) -DECLARE (uc) -DECLARE (s) -DECLARE (us) -DECLARE (i) -DECLARE (ui) -#if defined (__mips64) -DECLARE (ll) -DECLARE (ull) -#endif - -int -main () -{ - TEST (c); - TEST (sc); - TEST (uc); - TEST (s); - TEST (us); - TEST (i); - TEST (ui); -#if defined (__mips64) - TEST (ll); - TEST (ull); -#endif - exit (0); -} Index: gcc/testsuite/gcc.target/mips/pr35232.c =================================================================== --- gcc/testsuite/gcc.target/mips/pr35232.c 2008-06-09 21:42:44.000000000 +0100 +++ /dev/null 2008-06-08 10:32:14.544096500 +0100 @@ -1,17 +0,0 @@ -/* { dg-do run } */ -/* { dg-mips-options "-O" } */ - -NOMIPS16 unsigned int -f1 (unsigned long long x) -{ - unsigned int r; - asm ("# %0" : "=a" (r) : "0" (x)); - asm ("# %0" : "=h" (r) : "0" (r)); - return r; -} - -int -main (void) -{ - return f1 (4) != 4; -} Index: gcc/testsuite/gcc.target/mips/fix-vr4130-1.c =================================================================== --- gcc/testsuite/gcc.target/mips/fix-vr4130-1.c 2008-06-09 21:42:44.000000000 +0100 +++ gcc/testsuite/gcc.target/mips/fix-vr4130-1.c 2008-06-09 21:45:18.000000000 +0100 @@ -1,4 +1,8 @@ /* { dg-do compile } */ /* { dg-mips-options "-march=vr4130 -mfix-vr4130" } */ -NOMIPS16 int foo (void) { int r; asm ("# foo" : "=h" (r)); return r; } +NOMIPS16 unsigned int +foo (unsigned int x, unsigned int y) +{ + return x % y; +} /* { dg-final { scan-assembler "\tmacchi\t" } } */ Index: gcc/testsuite/gcc.target/mips/fix-vr4130-3.c =================================================================== --- gcc/testsuite/gcc.target/mips/fix-vr4130-3.c 2008-06-09 21:42:44.000000000 +0100 +++ gcc/testsuite/gcc.target/mips/fix-vr4130-3.c 2008-06-09 21:45:18.000000000 +0100 @@ -1,10 +1,8 @@ /* { dg-do compile } */ /* { dg-mips-options "-march=vr4130 -mgp64 -mfix-vr4130" } */ -NOMIPS16 long long -foo (void) +NOMIPS16 unsigned long long +foo (unsigned long long x, unsigned long long y) { - long long r; - asm ("# foo" : "=h" (r)); - return r; + return x % y; } /* { dg-final { scan-assembler "\tdmacchi\t" } } */ Index: gcc/testsuite/gcc.target/mips/int-moves-1.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/int-moves-1.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,40 @@ +/* { dg-do compile { target mips16_attribute } } */ +/* { dg-mips-options "-mgp64 -msoft-float -O2 -EL" } */ +/* { dg-add-options mips16_attribute } */ + +typedef unsigned uint128_t __attribute__((mode(TI))); + +extern uint128_t g[16]; +extern unsigned char gstuff[0x10000]; + +NOMIPS16 uint128_t +foo (uint128_t i1, uint128_t i2, uint128_t i3, uint128_t i4, + uint128_t *x, unsigned char *lstuff) +{ + g[0] = i1; + g[1] = i2; + g[2] = i3; + g[3] = i4; + x[0] = x[4]; + x[1] = 0; + x[2] = ((uint128_t) 0x123456789abcdefULL << 64) | 0xaabbccddeeff1122ULL; + x[3] = g[4]; + x[4] = *(uint128_t *) (lstuff + 0x7fff); + return *(uint128_t *) (gstuff + 0x7fff); +} + +MIPS16 uint128_t +bar (uint128_t i1, uint128_t i2, uint128_t i3, uint128_t i4, + uint128_t *x, unsigned char *lstuff) +{ + g[0] = i1; + g[1] = i2; + g[2] = i3; + g[3] = i4; + x[0] = x[4]; + x[1] = 0; + x[2] = ((uint128_t) 0x123456789abcdefULL << 64) | 0xaabbccddeeff1122ULL; + x[3] = g[4]; + x[4] = *(uint128_t *) (lstuff + 0x7fff); + return *(uint128_t *) (gstuff + 0x7fff); +} Index: gcc/testsuite/gcc.target/mips/int-moves-2.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/int-moves-2.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,40 @@ +/* { dg-do compile { target mips16_attribute } } */ +/* { dg-mips-options "-mgp64 -msoft-float -O2 -EB" } */ +/* { dg-add-options mips16_attribute } */ + +typedef unsigned uint128_t __attribute__((mode(TI))); + +extern uint128_t g[16]; +extern unsigned char gstuff[0x10000]; + +NOMIPS16 uint128_t +foo (uint128_t i1, uint128_t i2, uint128_t i3, uint128_t i4, + uint128_t *x, unsigned char *lstuff) +{ + g[0] = i1; + g[1] = i2; + g[2] = i3; + g[3] = i4; + x[0] = x[4]; + x[1] = 0; + x[2] = ((uint128_t) 0x123456789abcdefULL << 64) | 0xaabbccddeeff1122ULL; + x[3] = g[4]; + x[4] = *(uint128_t *) (lstuff + 0x7fff); + return *(uint128_t *) (gstuff + 0x7fff); +} + +MIPS16 uint128_t +bar (uint128_t i1, uint128_t i2, uint128_t i3, uint128_t i4, + uint128_t *x, unsigned char *lstuff) +{ + g[0] = i1; + g[1] = i2; + g[2] = i3; + g[3] = i4; + x[0] = x[4]; + x[1] = 0; + x[2] = ((uint128_t) 0x123456789abcdefULL << 64) | 0xaabbccddeeff1122ULL; + x[3] = g[4]; + x[4] = *(uint128_t *) (lstuff + 0x7fff); + return *(uint128_t *) (gstuff + 0x7fff); +} Index: gcc/testsuite/gcc.target/mips/fix-r4000-1.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-1.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,6 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -O2 -dp" } */ +typedef int int32_t; +typedef int uint32_t; +int32_t foo (int32_t x, int32_t y) { return x * y; } +uint32_t bar (uint32_t x, uint32_t y) { return x * y; } +/* { dg-final { scan-assembler-times "[concat {\tmult\t\$[45],\$[45][^\n]+mulsi3_r4000[^\n]+\n\tmflo\t\$2\n}]" 2 } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-2.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-2.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,7 @@ +/* { dg-mips-options "-mips1 -mfix-r4000 -O2 -dp -EB" } */ +typedef int int32_t; +typedef long long int64_t; +int32_t foo (int32_t x, int32_t y) { return ((int64_t) x * y) >> 32; } +/* ??? A highpart pattern would be a better choice, but we currently + don't use them. */ +/* { dg-final { scan-assembler "[concat {\tmult\t\$[45],\$[45][^\n]+mulsidi3_32bit_r4000[^\n]+\n\tmflo\t\$3\n\tmfhi\t\$2\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-3.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-3.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,7 @@ +/* { dg-mips-options "-mips1 -mfix-r4000 -O2 -dp -EB" } */ +typedef unsigned int uint32_t; +typedef unsigned long long uint64_t; +uint32_t foo (uint32_t x, uint32_t y) { return ((uint64_t) x * y) >> 32; } +/* ??? A highpart pattern would be a better choice, but we currently + don't use them. */ +/* { dg-final { scan-assembler "[concat {\tmultu\t\$[45],\$[45][^\n]+umulsidi3_32bit_r4000[^\n]+\n\tmflo\t\$3\n\tmfhi\t\$2\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-4.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-4.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,8 @@ +/* ??? At the moment, lower-subreg.c decomposes the copy of the multiplication + result to $2, which prevents the register allocators from storing the + multiplication result in $2. */ +/* { dg-mips-options "-mips1 -mfix-r4000 -O2 -fno-split-wide-types -dp -EL" } */ +typedef int int32_t; +typedef long long int64_t; +int64_t foo (int32_t x, int32_t y) { return (int64_t) x * y; } +/* { dg-final { scan-assembler "[concat {\tmult\t\$[45],\$[45][^\n]+mulsidi3_32bit_r4000[^\n]+\n\tmflo\t\$2\n\tmfhi\t\$3\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-5.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-5.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,8 @@ +/* ??? At the moment, lower-subreg.c decomposes the copy of the multiplication + result to $2, which prevents the register allocators from storing the + multiplication result in $2. */ +/* { dg-mips-options "-mips1 -mfix-r4000 -O2 -fno-split-wide-types -dp -EL" } */ +typedef unsigned int uint32_t; +typedef unsigned long long uint64_t; +uint64_t foo (uint32_t x, uint32_t y) { return (uint64_t) x * y; } +/* { dg-final { scan-assembler "[concat {\tmultu\t\$[45],\$[45][^\n]+umulsidi3_32bit_r4000[^\n]+\n\tmflo\t\$2\n\tmfhi\t\$3\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-6.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-6.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,6 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -mgp64 -O2 -dp" } */ +typedef long long int64_t; +typedef unsigned long long uint64_t; +int64_t foo (int64_t x, int64_t y) { return x * y; } +uint64_t bar (uint64_t x, uint64_t y) { return x * y; } +/* { dg-final { scan-assembler-times "[concat {\tdmult\t\$[45],\$[45][^\n]+muldi3_r4000[^\n]+\n\tmflo\t\$2\n}]" 2 } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-7.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-7.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,7 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -O2 -mgp64 -dp -EB" } */ +typedef long long int64_t; +typedef int int128_t __attribute__((mode(TI))); +int64_t foo (int64_t x, int64_t y) { return ((int128_t) x * y) >> 64; } +/* ??? A highpart pattern would be a better choice, but we currently + don't use them. */ +/* { dg-final { scan-assembler "[concat {\tdmult\t\$[45],\$[45][^\n]+mulditi3[^\n]+\n\tmflo\t\$3\n\tmfhi\t\$2\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-8.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-8.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,7 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -O2 -mgp64 -dp -EB" } */ +typedef unsigned long long uint64_t; +typedef unsigned int uint128_t __attribute__((mode(TI))); +uint64_t foo (uint64_t x, uint64_t y) { return ((uint128_t) x * y) >> 64; } +/* ??? A highpart pattern would be a better choice, but we currently + don't use them. */ +/* { dg-final { scan-assembler "[concat {\tdmultu\t\$[45],\$[45][^\n]+umulditi3[^\n]+\n\tmflo\t\$3\n\tmfhi\t\$2\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-9.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-9.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,8 @@ +/* ??? At the moment, lower-subreg.c decomposes the copy of the multiplication + result to $2, which prevents the register allocators from storing the + multiplication result in $2. */ +/* { dg-mips-options "-mips3 -mfix-r4000 -mgp64 -O2 -fno-split-wide-types -dp -EL" } */ +typedef long long int64_t; +typedef int int128_t __attribute__((mode(TI))); +int128_t foo (int64_t x, int64_t y) { return (int128_t) x * y; } +/* { dg-final { scan-assembler "[concat {\tdmult\t\$[45],\$[45][^\n]+mulditi3_r4000[^\n]+\n\tmflo\t\$2\n\tmfhi\t\$3\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-10.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-10.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,8 @@ +/* ??? At the moment, lower-subreg.c decomposes the copy of the multiplication + result to $2, which prevents the register allocators from storing the + multiplication result in $2. */ +/* { dg-mips-options "-mips3 -mfix-r4000 -mgp64 -O2 -fno-split-wide-types -dp -EL" } */ +typedef unsigned long long uint64_t; +typedef unsigned int uint128_t __attribute__((mode(TI))); +uint128_t foo (uint64_t x, uint64_t y) { return (uint128_t) x * y; } +/* { dg-final { scan-assembler "[concat {\tdmultu\t\$[45],\$[45][^\n]+umulditi3_r4000[^\n]+\n\tmflo\t\$2\n\tmfhi\t\$3\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-11.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-11.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,4 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -mgp64 -O2 -dp" } */ +typedef long long int64_t; +int64_t foo (int64_t x) { return x / 11993; } +/* { dg-final { scan-assembler "[concat {\tdmult\t\$4,\$[0-9]+[^\n]+smuldi3_highpart[^\n]+\n\tmfhi\t\$[0-9]+\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/fix-r4000-12.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/fix-r4000-12.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,4 @@ +/* { dg-mips-options "-march=r4000 -mfix-r4000 -mgp64 -O2 -dp" } */ +typedef unsigned long long uint64_t; +uint64_t foo (uint64_t x) { return x / 11993; } +/* { dg-final { scan-assembler "[concat {\tdmultu\t\$4,\$[0-9]+[^\n]+umuldi3_highpart[^\n]+\n\tmfhi\t\$[0-9]+\n}]" } } */ Index: gcc/testsuite/gcc.target/mips/timode-1.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/timode-1.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,65 @@ +/* { dg-mips-options "-mgp64" } */ +typedef int int128_t __attribute__((mode(TI))); +typedef unsigned int uint128_t __attribute__((mode(TI))); + +#define UINT128_CONST(A, B) \ + (((uint128_t) (0x ## A ## ULL) << 64) | (0x ## B ## ULL)) + +volatile uint128_t a = UINT128_CONST (1111111111111111, a222222222222222); +volatile uint128_t b = UINT128_CONST (0000000000000005, 0000000000000003); +volatile uint128_t c = UINT128_CONST (5dddddddddddddde, e666666666666666); +volatile uint128_t d = UINT128_CONST (e612340000000000, 5000000000234500); +volatile uint128_t e = UINT128_CONST (43f011dddddddddf, 366666666689ab66); +volatile uint128_t f = UINT128_CONST (4210100000000000, 1000000000010100); +volatile uint128_t g = UINT128_CONST (a5e225dddddddddf, 6666666666aaee66); +volatile uint128_t h = UINT128_CONST (e7f235dddddddddf, 7666666666abef66); +volatile uint128_t i = UINT128_CONST (5e225dddddddddf6, 666666666aaee660); +volatile uint128_t j = UINT128_CONST (0a5e225ddddddddd, f6666666666aaee6); +volatile uint128_t k = UINT128_CONST (fa5e225ddddddddd, f6666666666aaee6); + +volatile int amount = 4; + +volatile uint128_t result; + +int +main (void) +{ + result = a * b; + if (result != c) + return 1; + + result = c + d; + if (result != e) + return 1; + + result = e - d; + if (result != c) + return 1; + + result = d & e; + if (result != f) + return 1; + + result = d ^ e; + if (result != g) + return 1; + + result = d | e; + if (result != h) + return 1; + + result = g << amount; + if (result != i) + return 1; + + result = g >> amount; + if (result != j) + return 1; + + result = (int128_t) g >> amount; + if (result != k) + return 1; + + return 0; +} +/* { dg-final { scan-assembler-not "\tjal" } } */ Index: gcc/testsuite/gcc.target/mips/timode-2.c =================================================================== --- /dev/null 2008-06-08 10:32:14.544096500 +0100 +++ gcc/testsuite/gcc.target/mips/timode-2.c 2008-06-09 21:45:18.000000000 +0100 @@ -0,0 +1,64 @@ +/* { dg-do run { target mips64 } } */ +typedef int int128_t __attribute__((mode(TI))); +typedef unsigned int uint128_t __attribute__((mode(TI))); + +#define UINT128_CONST(A, B) \ + (((uint128_t) (0x ## A ## ULL) << 64) | (0x ## B ## ULL)) + +volatile uint128_t a = UINT128_CONST (1111111111111111, a222222222222222); +volatile uint128_t b = UINT128_CONST (0000000000000005, 0000000000000003); +volatile uint128_t c = UINT128_CONST (5dddddddddddddde, e666666666666666); +volatile uint128_t d = UINT128_CONST (e612340000000000, 5000000000234500); +volatile uint128_t e = UINT128_CONST (43f011dddddddddf, 366666666689ab66); +volatile uint128_t f = UINT128_CONST (4210100000000000, 1000000000010100); +volatile uint128_t g = UINT128_CONST (a5e225dddddddddf, 6666666666aaee66); +volatile uint128_t h = UINT128_CONST (e7f235dddddddddf, 7666666666abef66); +volatile uint128_t i = UINT128_CONST (5e225dddddddddf6, 666666666aaee660); +volatile uint128_t j = UINT128_CONST (0a5e225ddddddddd, f6666666666aaee6); +volatile uint128_t k = UINT128_CONST (fa5e225ddddddddd, f6666666666aaee6); + +volatile int amount = 4; + +volatile uint128_t result; + +int +main (void) +{ + result = a * b; + if (result != c) + return 1; + + result = c + d; + if (result != e) + return 1; + + result = e - d; + if (result != c) + return 1; + + result = d & e; + if (result != f) + return 1; + + result = d ^ e; + if (result != g) + return 1; + + result = d | e; + if (result != h) + return 1; + + result = g << amount; + if (result != i) + return 1; + + result = g >> amount; + if (result != j) + return 1; + + result = (int128_t) g >> amount; + if (result != k) + return 1; + + return 0; +} ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: Changing the treatment of the MIPS HI and LO registers 2008-06-09 21:13 ` Richard Sandiford @ 2008-06-09 21:54 ` Maciej W. Rozycki 2008-06-11 18:41 ` Maxim Kuvyrkov 1 sibling, 0 replies; 7+ messages in thread From: Maciej W. Rozycki @ 2008-06-09 21:54 UTC (permalink / raw) To: Richard Sandiford; +Cc: Ralf Baechle, gcc-patches, linux-mips On Mon, 9 Jun 2008, Richard Sandiford wrote: > Please shout if you find anything wrong with either the patch or > the documentation changes. Lacking the time to review your changes or at least test at the run time (sorry!) I trust you have done a decent job as usually. Perhaps someone else can to that. Otherwise the results will come back and haunt you in some two years or suchlike when people start using this version of GCC to build the kernel. ;) To give you an indicator -- I am still using 4.1.2 and will keep doing so for another year or so. I may skip a version or two when I finally decide to upgrade though. Maciej ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: Changing the treatment of the MIPS HI and LO registers 2008-06-09 21:13 ` Richard Sandiford 2008-06-09 21:54 ` Maciej W. Rozycki @ 2008-06-11 18:41 ` Maxim Kuvyrkov 2008-06-11 18:54 ` Richard Sandiford 2008-06-11 23:24 ` Paolo Bonzini 1 sibling, 2 replies; 7+ messages in thread From: Maxim Kuvyrkov @ 2008-06-11 18:41 UTC (permalink / raw) To: Maciej W. Rozycki, Ralf Baechle, gcc-patches, linux-mips, rdsandiford Richard Sandiford wrote: ... > + <li>The MIPS port no longer recognizes the <code>h</code> > + <code>asm</code> constraint. It was necessary to remove > + this constraint in order to avoid generating unpredictable > + code sequences. > + > + <p>One of the main uses of the <code>h</code> constraint > + was to extract the high part of a multiplication on > + 64-bit targets. For example:</p> > + <pre> > + asm ("dmultu\t%1,%2" : "=h" (result) : "r" (x), "r" (y));</pre> > + <p>You can now achieve the same effect using 128-bit types:</p> > + <pre> > + typedef unsigned int uint128_t __attribute__((mode(TI))); > + result = ((uint128_t) x * y) >> 64;</pre> > + <p>The second sequence is better in many ways. For example, > + if <code>x</code> and <code>y</code> are constants, the > + compiler can perform the multiplication at compile time. > + If <code>x</code> and <code>y</code> are not constants, > + the compiler can schedule the runtime multiplication > + better than it can schedule an <code>asm</code> statement.</p> > + </li> > </ul> Hi, GLIBC contains the following code in stdlib/longlong.h: <snip> #if defined (__mips__) && W_TYPE_SIZE == 32 #define umul_ppmm(w1, w0, u, v) \ __asm__ ("multu %2,%3" \ : "=l" ((USItype) (w0)), \ "=h" ((USItype) (w1)) \ : "d" ((USItype) (u)), \ "d" ((USItype) (v))) #define UMUL_TIME 10 #define UDIV_TIME 100 #endif /* __mips__ */ </snip> What would be a correct fix in this case? Something like this: <snip> #define umul_ppmm(w1, w0, u, v) \ ({unsigned int __attribute__((mode(DI))) __xx; \ __xx = (unsigned int __attribute__((mode(DI)))) u * v; \ w0 = __xx & ((1 << 32) - 1); \ w1 = __xx >> 32;}) </snip> Or is there a better way? Thanks, Maxim ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: Changing the treatment of the MIPS HI and LO registers 2008-06-11 18:41 ` Maxim Kuvyrkov @ 2008-06-11 18:54 ` Richard Sandiford 2008-06-11 23:24 ` Paolo Bonzini 1 sibling, 0 replies; 7+ messages in thread From: Richard Sandiford @ 2008-06-11 18:54 UTC (permalink / raw) To: Maxim Kuvyrkov; +Cc: Maciej W. Rozycki, Ralf Baechle, gcc-patches, linux-mips Maxim Kuvyrkov <maxim@codesourcery.com> writes: > GLIBC contains the following code in stdlib/longlong.h: > <snip> > #if defined (__mips__) && W_TYPE_SIZE == 32 > #define umul_ppmm(w1, w0, u, v) \ > __asm__ ("multu %2,%3" \ > : "=l" ((USItype) (w0)), \ > "=h" ((USItype) (w1)) \ > : "d" ((USItype) (u)), \ > "d" ((USItype) (v))) > #define UMUL_TIME 10 > #define UDIV_TIME 100 > #endif /* __mips__ */ > </snip> > > What would be a correct fix in this case? Something like this: > <snip> > #define umul_ppmm(w1, w0, u, v) \ > ({unsigned int __attribute__((mode(DI))) __xx; \ > __xx = (unsigned int __attribute__((mode(DI)))) u * v; \ > w0 = __xx & ((1 << 32) - 1); \ > w1 = __xx >> 32;}) > </snip> > > Or is there a better way? All being well, you should just be able to do the same as I did for GCC's copy of longlong.h (included in the patch you responded to). Richard ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: Changing the treatment of the MIPS HI and LO registers 2008-06-11 18:41 ` Maxim Kuvyrkov 2008-06-11 18:54 ` Richard Sandiford @ 2008-06-11 23:24 ` Paolo Bonzini 1 sibling, 0 replies; 7+ messages in thread From: Paolo Bonzini @ 2008-06-11 23:24 UTC (permalink / raw) To: Maxim Kuvyrkov Cc: Maciej W. Rozycki, Ralf Baechle, gcc-patches, linux-mips, rdsandiford > GLIBC contains the following code in stdlib/longlong.h: > <snip> > #if defined (__mips__) && W_TYPE_SIZE == 32 > #define umul_ppmm(w1, w0, u, v) \ > __asm__ ("multu %2,%3" \ > : "=l" ((USItype) (w0)), \ > "=h" ((USItype) (w1)) \ > : "d" ((USItype) (u)), \ > "d" ((USItype) (v))) > #define UMUL_TIME 10 > #define UDIV_TIME 100 > #endif /* __mips__ */ > </snip> Actually, so does GCC itself. Can you prepare a patch? > What would be a correct fix in this case? Something like this: > <snip> > #define umul_ppmm(w1, w0, u, v) \ > ({unsigned int __attribute__((mode(DI))) __xx; \ > __xx = (unsigned int __attribute__((mode(DI)))) u * v; \ > w0 = __xx & ((1 << 32) - 1); \ > w1 = __xx >> 32;}) > </snip> > > Or is there a better way? Almost; this: #define umul_ppmm(w1, w0, u, v) \ ({UDWtype __xx; \ UWtype __u = (u), __v = (v); \ __xx = (UDWtype) __u * __v; \ w0 = (UWtype) __xx; \ w1 = __xx >> 32;}) should work. Paolo ^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2008-06-11 23:24 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <87tzgj4nh6.fsf@firetop.home>
[not found] ` <Pine.LNX.4.55.0805272134540.18833@cliff.in.clinika.pl>
[not found] ` <87abib4d9t.fsf@firetop.home>
[not found] ` <Pine.LNX.4.55.0805272357020.18833@cliff.in.clinika.pl>
[not found] ` <87r6bm1ebd.fsf@firetop.home>
2008-05-29 2:01 ` Changing the treatment of the MIPS HI and LO registers Maciej W. Rozycki
[not found] ` <878wxtvarg.fsf@firetop.home>
2008-06-01 13:48 ` Richard Sandiford
2008-06-09 21:13 ` Richard Sandiford
2008-06-09 21:54 ` Maciej W. Rozycki
2008-06-11 18:41 ` Maxim Kuvyrkov
2008-06-11 18:54 ` Richard Sandiford
2008-06-11 23:24 ` Paolo Bonzini
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox