diff options
| author | Michael Munday <mike.munday@ibm.com> | 2019-08-15 21:53:37 +0100 |
|---|---|---|
| committer | Michael Munday <mike.munday@ibm.com> | 2019-10-09 15:04:59 +0000 |
| commit | 38c4a7370670e6b73cbae192c7d3079c90203832 (patch) | |
| tree | 0ed95e4710b6cb9d1f0d40a28b62b8d81eb4dd20 /src/cmd/compile/internal/s390x | |
| parent | ff86ce13b69f427248203493d669a730988f882c (diff) | |
| download | go-38c4a7370670e6b73cbae192c7d3079c90203832.tar.gz go-38c4a7370670e6b73cbae192c7d3079c90203832.zip | |
cmd/asm: add s390x branch-on-count instructions
The branch-on-count instructions on s390x decrement the input
register and then compare its value to 0. If not equal the branch
is taken.
These instructions are useful for implementing loops with a set
number of iterations (which might be in a register).
For example, this for loop:
for i := 0; i < n; i++ {
... // i is not used or modified in the loop
}
Could be implemented using this assembly:
MOVD Rn, Ri
loop:
...
BRCTG Ri, loop
Note that i will count down from n in the assembly whereas in the
original for loop it counted up to n which is why we can't use i
in the loop.
These instructions will only be used in hand-written codegen and
assembly for now since SSA blocks cannot currently modify values.
We could look into this in the future though.
Change-Id: Iaab93b8aa2699513b825439b8ea20d8fe2ea1ee6
Reviewed-on: https://go-review.googlesource.com/c/go/+/199977
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Diffstat (limited to 'src/cmd/compile/internal/s390x')
| -rw-r--r-- | src/cmd/compile/internal/s390x/ggen.go | 12 |
1 files changed, 4 insertions, 8 deletions
diff --git a/src/cmd/compile/internal/s390x/ggen.go b/src/cmd/compile/internal/s390x/ggen.go index ae9965c378..16af190b2f 100644 --- a/src/cmd/compile/internal/s390x/ggen.go +++ b/src/cmd/compile/internal/s390x/ggen.go @@ -38,18 +38,14 @@ func zerorange(pp *gc.Progs, p *obj.Prog, off, cnt int64, _ *uint32) *obj.Prog { // Generate a loop of large clears. if cnt > clearLoopCutoff { - n := cnt - (cnt % 256) - end := int16(s390x.REGRT2) - p = pp.Appendpp(p, s390x.AADD, obj.TYPE_CONST, 0, off+n, obj.TYPE_REG, end, 0) - p.Reg = reg + ireg := int16(s390x.REGRT2) // register holds number of remaining loop iterations + p = pp.Appendpp(p, s390x.AMOVD, obj.TYPE_CONST, 0, cnt/256, obj.TYPE_REG, ireg, 0) p = pp.Appendpp(p, s390x.ACLEAR, obj.TYPE_CONST, 0, 256, obj.TYPE_MEM, reg, off) pl := p p = pp.Appendpp(p, s390x.AADD, obj.TYPE_CONST, 0, 256, obj.TYPE_REG, reg, 0) - p = pp.Appendpp(p, s390x.ACMP, obj.TYPE_REG, reg, 0, obj.TYPE_REG, end, 0) - p = pp.Appendpp(p, s390x.ABNE, obj.TYPE_NONE, 0, 0, obj.TYPE_BRANCH, 0, 0) + p = pp.Appendpp(p, s390x.ABRCTG, obj.TYPE_REG, ireg, 0, obj.TYPE_BRANCH, 0, 0) gc.Patch(p, pl) - - cnt -= n + cnt = cnt % 256 } // Generate remaining clear instructions without a loop. |