diff options
Diffstat (limited to 'libc/string/ia64/memccpy.S')
| -rw-r--r-- | libc/string/ia64/memccpy.S | 102 |
1 files changed, 51 insertions, 51 deletions
diff --git a/libc/string/ia64/memccpy.S b/libc/string/ia64/memccpy.S index 1afba3637..259d680bc 100644 --- a/libc/string/ia64/memccpy.S +++ b/libc/string/ia64/memccpy.S @@ -23,7 +23,7 @@ Inputs: in0: dest in1: src - in2: char + in2: char in3: byte count This implementation assumes little endian mode (UM.be = 0). @@ -69,75 +69,75 @@ ENTRY(memccpy) .rotr r[MEMLAT + 7], tmp1[4], tmp2[4], val[4], tmp3[2], pos0[2] .rotp p[MEMLAT + 6 + 1] - mov ret0 = r0 // return NULL if no match + mov ret0 = r0 /* return NULL if no match */ .save pr, saved_pr - mov saved_pr = pr // save the predicate registers - mov dest = in0 // dest + mov saved_pr = pr /* save the predicate registers */ + mov dest = in0 /* dest */ .save ar.lc, saved_lc - mov saved_lc = ar.lc // save the loop counter - mov saved_ec = ar.ec // save the loop counter + mov saved_lc = ar.lc /* save the loop counter */ + mov saved_ec = ar.ec /* save the loop counter */ .body - mov src = in1 // src - extr.u char = in2, 0, 8 // char - mov len = in3 // len - sub tmp = r0, in0 // tmp = -dest - cmp.ne p7, p0 = r0, r0 // clear p7 + mov src = in1 /* src */ + extr.u char = in2, 0, 8 /* char */ + mov len = in3 /* len */ + sub tmp = r0, in0 /* tmp = -dest */ + cmp.ne p7, p0 = r0, r0 /* clear p7 */ ;; - and loopcnt = 7, tmp // loopcnt = -dest % 8 - cmp.ge p6, p0 = OP_T_THRES, len // is len <= OP_T_THRES - mov ar.ec = 0 // ec not guaranteed zero on entry -(p6) br.cond.spnt .cpyfew // copy byte by byte + and loopcnt = 7, tmp /* loopcnt = -dest % 8 */ + cmp.ge p6, p0 = OP_T_THRES, len /* is len <= OP_T_THRES */ + mov ar.ec = 0 /* ec not guaranteed zero on entry */ +(p6) br.cond.spnt .cpyfew /* copy byte by byte */ ;; cmp.eq p6, p0 = loopcnt, r0 mux1 charx8 = char, @brcst (p6) br.cond.sptk .dest_aligned - sub len = len, loopcnt // len -= -dest % 8 - adds loopcnt = -1, loopcnt // --loopcnt + sub len = len, loopcnt /* len -= -dest % 8 */ + adds loopcnt = -1, loopcnt /* --loopcnt */ ;; mov ar.lc = loopcnt -.l1: // copy -dest % 8 bytes - ld1 value = [src], 1 // value = *src++ +.l1: /* copy -dest % 8 bytes */ + ld1 value = [src], 1 /* value = *src++ */ ;; - st1 [dest] = value, 1 // *dest++ = value + st1 [dest] = value, 1 /* *dest++ = value */ cmp.eq p6, p0 = value, char (p6) br.cond.spnt .foundit br.cloop.dptk .l1 .dest_aligned: - and sh1 = 7, src // sh1 = src % 8 - and tmp = -8, len // tmp = len & -OPSIZ - and asrc = -8, src // asrc = src & -OPSIZ -- align src - shr.u loopcnt = len, 3 // loopcnt = len / 8 - and len = 7, len ;; // len = len % 8 - shl sh1 = sh1, 3 // sh1 = 8 * (src % 8) - adds loopcnt = -1, loopcnt // --loopcnt - mov pr.rot = 1 << 16 ;; // set rotating predicates - sub sh2 = 64, sh1 // sh2 = 64 - sh1 - mov ar.lc = loopcnt // set LC - cmp.eq p6, p0 = sh1, r0 // is the src aligned? + and sh1 = 7, src /* sh1 = src % 8 */ + and tmp = -8, len /* tmp = len & -OPSIZ */ + and asrc = -8, src /* asrc = src & -OPSIZ -- align src */ + shr.u loopcnt = len, 3 /* loopcnt = len / 8 */ + and len = 7, len ;; /* len = len % 8 */ + shl sh1 = sh1, 3 /* sh1 = 8 * (src % 8) */ + adds loopcnt = -1, loopcnt /* --loopcnt */ + mov pr.rot = 1 << 16 ;; /* set rotating predicates */ + sub sh2 = 64, sh1 /* sh2 = 64 - sh1 */ + mov ar.lc = loopcnt /* set LC */ + cmp.eq p6, p0 = sh1, r0 /* is the src aligned? */ (p6) br.cond.sptk .src_aligned ;; - add src = src, tmp // src += len & -OPSIZ - mov ar.ec = MEMLAT + 6 + 1 // six more passes needed - ld8 r[1] = [asrc], 8 // r[1] = w0 - cmp.ne p6, p0 = r0, r0 ;; // clear p6 + add src = src, tmp /* src += len & -OPSIZ */ + mov ar.ec = MEMLAT + 6 + 1 /* six more passes needed */ + ld8 r[1] = [asrc], 8 /* r[1] = w0 */ + cmp.ne p6, p0 = r0, r0 ;; /* clear p6 */ ALIGN(32) .l2: -(p[0]) ld8.s r[0] = [asrc], 8 // r[0] = w1 -(p[MEMLAT]) shr.u tmp1[0] = r[1 + MEMLAT], sh1 // tmp1 = w0 >> sh1 -(p[MEMLAT]) shl tmp2[0] = r[0 + MEMLAT], sh2 // tmp2 = w1 << sh2 +(p[0]) ld8.s r[0] = [asrc], 8 /* r[0] = w1 */ +(p[MEMLAT]) shr.u tmp1[0] = r[1 + MEMLAT], sh1 /* tmp1 = w0 >> sh1 */ +(p[MEMLAT]) shl tmp2[0] = r[0 + MEMLAT], sh2 /* tmp2 = w1 << sh2 */ (p[MEMLAT+4]) xor tmp3[0] = val[1], charx8 (p[MEMLAT+5]) czx1.r pos0[0] = tmp3[1] -(p[MEMLAT+6]) chk.s r[6 + MEMLAT], .recovery1 // our data isn't - // valid - rollback! +(p[MEMLAT+6]) chk.s r[6 + MEMLAT], .recovery1 /* our data isn't */ + /* valid - rollback! */ (p[MEMLAT+6]) cmp.ne p6, p0 = 8, pos0[1] (p6) br.cond.spnt .gotit -(p[MEMLAT+6]) st8 [dest] = val[3], 8 // store val to dest -(p[MEMLAT+3]) or val[0] = tmp1[3], tmp2[3] // val = tmp1 | tmp2 +(p[MEMLAT+6]) st8 [dest] = val[3], 8 /* store val to dest */ +(p[MEMLAT+3]) or val[0] = tmp1[3], tmp2[3] /* val = tmp1 | tmp2 */ br.ctop.sptk .l2 br.cond.sptk .cpyfew .src_aligned: - cmp.ne p6, p0 = r0, r0 // clear p6 - mov ar.ec = MEMLAT + 2 + 1 ;; // set EC + cmp.ne p6, p0 = r0, r0 /* clear p6 */ + mov ar.ec = MEMLAT + 2 + 1 ;; /* set EC */ .l3: (p[0]) ld8.s r[0] = [src], 8 (p[MEMLAT]) xor tmp3[0] = r[MEMLAT], charx8 @@ -149,8 +149,8 @@ ENTRY(memccpy) (p[MEMLAT+2]) st8 [dest] = r[MEMLAT+2], 8 br.ctop.dptk .l3 .cpyfew: - cmp.eq p6, p0 = len, r0 // is len == 0 ? - adds len = -1, len // --len; + cmp.eq p6, p0 = len, r0 /* is len == 0 ? */ + adds len = -1, len /* --len; */ (p6) br.cond.spnt .restore_and_exit ;; mov ar.lc = len .l4: @@ -163,14 +163,14 @@ ENTRY(memccpy) .foundit: (p6) mov ret0 = dest .restore_and_exit: - mov pr = saved_pr, -1 // restore the predicate registers - mov ar.lc = saved_lc // restore the loop counter - mov ar.ec = saved_ec ;; // restore the epilog counter + mov pr = saved_pr, -1 /* restore the predicate registers */ + mov ar.lc = saved_lc /* restore the loop counter */ + mov ar.ec = saved_ec ;; /* restore the epilog counter */ br.ret.sptk.many b0 .gotit: .pred.rel "mutex" p6, p7 -(p6) mov value = val[3] // if coming from l2 -(p7) mov value = r[MEMLAT+2] // if coming from l3 +(p6) mov value = val[3] /* if coming from l2 */ +(p7) mov value = r[MEMLAT+2] /* if coming from l3 */ mov ar.lc = pos0[1] ;; .l5: extr.u tmp = value, 0, 8 ;; |