R11 is only used as a temporary by a very small set of instructions
(DIV, MOD, MULH and extended MVC/XC instructions). By marking these
instructions as clobbering R11 we can allocate R11 in the general
case.
Change-Id: I0d4ffe80e57c164d42a5ea5ef6308756a5b0f742
Reviewed-on: https://go-review.googlesource.com/110255
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Currently, each architecture lowers OpConvert to an arch-specific
OpXXXconvert. This is silly because OpConvert means the same thing on
all architectures and is logically a no-op that exists only to keep
track of conversions to and from unsafe.Pointer. Furthermore, lowering
it makes it harder to recognize in other analyses, particularly
liveness analysis.
This CL eliminates the lowering of OpConvert, leaving it as the
generic op until code generation time.
The main complexity here is that we still need to register-allocate
OpConvert operations. Currently, each arch's lowered OpConvert
specifies all GP registers in its register mask. Ideally, OpConvert
wouldn't affect value homing at all, and we could just copy the home
of OpConvert's source, but this can potentially home an OpConvert in a
LocalSlot, which neither regalloc nor stackalloc expect. Rather than
try to disentangle this assumption from regalloc and stackalloc, we
continue to register-allocate OpConvert, but teach regalloc that
OpConvert can be allocated to any allocatable GP register.
For #24543.
Change-Id: I795a6aee5fd94d4444a7bafac3838a400c9f7bb6
Reviewed-on: https://go-review.googlesource.com/108496
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
The atomic add instructions modify the condition code and so need to
be marked as clobbering flags.
Fixes#24449.
Change-Id: Ic69c8d775fbdbfb2a56c5e0cfca7a49c0d7f6897
Reviewed-on: https://go-review.googlesource.com/101455
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Add a bool to opInfo to indicate if an Op never results in any
instructions. This is a conservative approximation: some operations,
like Copy, may or may not generate code depending on their arguments.
I built the list by reading each arch's ssaGenValue function. Hopefully
I got them all.
Change-Id: I130b251b65f18208294e129bb7ddc3f91d57d31d
Reviewed-on: https://go-review.googlesource.com/97957
Reviewed-by: Keith Randall <khr@golang.org>
Because a call may ultimately invoke runtime.setg, we have to assume
that g may be clobbered by any call. All of the other architectures
that use a g register already do this, but it was missing from the
s390x caller save clobber set.
Change-Id: Ia931638d42c44979839f20d71097acf31475f423
Reviewed-on: https://go-review.googlesource.com/92835
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
The CMPWUconst op (32-bit unsigned comparison with immediate) takes
an unsigned immediate value. In SSA this should be sign extended to
64-bits to match the Int32 type given in the op and then zero
extended when producing the final assembly. Before this CL we were
zero extending in SSA which caused ssacheck to fail.
While we are here also ensure other 32-bit immediates are sign
extended in SSA.
Passes toolstash -cmp on std on s390x.
Fixes#22611.
Change-Id: I5c061a76a710b10ecb0650c9c42efd9fa1c123cc
Reviewed-on: https://go-review.googlesource.com/76336
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Prior to this CL loads with sign extension could not be replaced with
indexed loads (only loads with zero extension).
This CL also prevents large offsets (more than 20-bits) from being
merged into indexed loads. It is better to keep such offsets
separate.
Gives a small improvement in binary size, ~1.5KB from .text in cmd/go.
Change-Id: Ib848ffc2b05de6660c5ce2394ae1d1d144273e29
Reviewed-on: https://go-review.googlesource.com/36845
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
This change adds three new instructions:
- LPDFR: load positive (math.Abs(x))
- LNDFR: load negative (-math.Abs(x))
- CPSDR: copy sign (math.Copysign(x, y))
By making use of GPR <-> FPR moves we can now compile math.Abs and
math.Copysign to these instructions using SSA rules.
This CL also adds new rules to merge address generation into combined
load operations. This makes GPR <-> FPR move matching more reliable.
name old time/op new time/op delta
Copysign 1.85ns ± 0% 1.40ns ± 1% -24.65% (p=0.000 n=8+10)
Abs 1.58ns ± 1% 0.73ns ± 1% -53.64% (p=0.000 n=10+10)
The geo mean improvement for all math package benchmarks was 4.6%.
Change-Id: I0cec35c5c1b3fb45243bf666b56b57faca981bc9
Reviewed-on: https://go-review.googlesource.com/73950
Run-TryBot: Michael Munday <mike.munday@ibm.com>
Reviewed-by: Keith Randall <khr@golang.org>
Add a compiler intrinsic for getcallersp. So we are able to get
rid of the argument (not done in this CL).
Change-Id: Ic38fda1c694f918328659ab44654198fb116668d
Reviewed-on: https://go-review.googlesource.com/69350
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: David Chase <drchase@google.com>
Rematerializable ops can be inserted after the flagalloc phase,
they must therefore not clobber flags. This CL adds a check to
ensure this doesn't happen and fixes the instances where it
does currently.
amd64: ADDQconst and ADDLconst were recently changed to be
rematerializable in CL 54393 (only in tip, not 1.9). That change
has been reverted.
s390x: MOVDaddr could clobber flags when using dynamic linking due
to a ADD with immediate instruction. Change the code generation to
use LA/LAY instead.
Fixes#21080.
Change-Id: Ia85c882afa2a820a309e93775354b3169ec6d034
Reviewed-on: https://go-review.googlesource.com/63030
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Ilya Tocar <ilya.tocar@intel.com>
Ceil, Floor and Trunc are pre-existing intrinsics. Round is a new
function and has been added as an intrinsic in this CL. All of the
functions can be implemented as a single 'LOAD FP INTEGER'
instruction, FIDBR, on s390x.
name old time/op new time/op delta
Ceil 2.34ns ± 0% 0.85ns ± 0% -63.74% (p=0.000 n=5+4)
Floor 2.33ns ± 0% 0.85ns ± 1% -63.35% (p=0.008 n=5+5)
Round 4.23ns ± 0% 0.85ns ± 0% -79.89% (p=0.000 n=5+4)
Trunc 2.35ns ± 0% 0.85ns ± 0% -63.83% (p=0.029 n=4+4)
Change-Id: Idee7ba24a2899d12bf9afee4eedd6b4aaad3c510
Reviewed-on: https://go-review.googlesource.com/63890
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Enhance the one-live-memory-at-a-time check to run during many
more phases of the SSA backend. Also make it work in an interblock
fashion.
Change types.IsMemory to return true for tuples containing a memory type.
Fix trim pass to build the merged phi correctly. Doesn't affect
code but allows the check to pass after trim runs.
Switch the AddTuple* ops to take the memory-containing tuple argument second.
Update #20335
Change-Id: I5b03ef3606b75a9e4f765276bb8b183cdc172b43
Reviewed-on: https://go-review.googlesource.com/43495
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
This CL modifies how MOV[DWHB] instructions that store a constant to
memory are assembled to avoid them clobbering the condition code
(flags). It also modifies zeroAuto to use MOVD instructions instead of
CLEAR (which is assembled as XC).
MOV[DWHB]storeconst ops also no longer clobbers flags.
Note: this CL modifies the assembler so that it can no longer handle
immediates outside the range of an int16 or offsets from SB, which
reflects what the machine instructions support. The compiler doesn't
need this capability any more and I don't think this affects any existing
assembly, but it is easy to workaround if it does.
Fixes#20187.
Change-Id: Ie54947ff38367bd6a19962bf1a6d0296a4accffb
Reviewed-on: https://go-review.googlesource.com/42179
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The experiment "clobberdead" clobbers all pointer fields that the
compiler thinks are dead, just before and after every safepoint.
Useful for debugging the generation of live pointer bitmaps.
Helped find the following issues:
Update #15936
Update #16026
Update #16095
Update #18860
Change-Id: Id1d12f86845e3d93bae903d968b1eac61fc461f9
Reviewed-on: https://go-review.googlesource.com/23924
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Note that this is a redo of an undo of the original buggy CL 38666.
We have lots of rewrite rules that vary only in the fact that
we have 2 versions for the 2 different orderings of various
commuting ops. For example:
(ADDL x (MOVLconst [c])) -> (ADDLconst [c] x)
(ADDL (MOVLconst [c]) x) -> (ADDLconst [c] x)
It can get unwieldly quickly, especially when there is more than
one commuting op in a rule.
Our existing "fix" for this problem is to have rules that
canonicalize the operations first. For example:
(Eq64 x (Const64 <t> [c])) && x.Op != OpConst64 -> (Eq64 (Const64 <t> [c]) x)
Subsequent rules can then assume if there is a constant arg to Eq64,
it will be the first one. This fix kinda works, but it is fragile and
only works when we remember to include the required extra rules.
The fundamental problem is that the rule matcher doesn't
know anything about commuting ops. This CL fixes that fact.
We already have information about which ops commute. (The register
allocator takes advantage of commutivity.) The rule generator now
automatically generates multiple rules for a single source rule when
there are commutative ops in the rule. We can now drop all of our
almost-duplicate source-level rules and the canonicalization rules.
I have some CLs in progress that will be a lot less verbose when
the rule generator handles commutivity for me.
I had to reorganize the load-combining rules a bit. The 8-way OR rules
generated 128 different reorderings, which was causing the generator
to put too much code in the rewrite*.go files (the big ones were going
from 25K lines to 132K lines). Instead I reorganized the rules to
combine pairs of loads at a time. The generated rule files are now
actually a bit (5%) smaller.
Make.bash times are ~unchanged.
Compiler benchmarks are not observably different. Probably because
we don't spend much compiler time in rule matching anyway.
I've also done a pass over all of our ops adding commutative markings
for ops which hadn't had them previously.
Fixes#18292
Change-Id: Ic1c0e43fbf579539f459971625f69690c9ab8805
Reviewed-on: https://go-review.googlesource.com/38801
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
During the review of CL 38801 it was noted that it would be nice
to have a bit more clarity on how-and-why SB addressing is handled
strangely on s390x. This additional comment should hopefully help.
In general SB is handled differently because not all instructions
have variants that use relative addressing.
Change-Id: I3379012ae3f167478c191c435939c3b876c645ed
Reviewed-on: https://go-review.googlesource.com/38952
Reviewed-by: Keith Randall <khr@golang.org>
This reverts commit 041ecb697f.
Reason for revert: Not working on S390x and some 386 archs.
I have a guess why the S390x is failing. No clue on the 386 yet.
Revert until I can figure it out.
Change-Id: I64f1ce78fa6d1037ebe7ee2a8a8107cb4c1db70c
Reviewed-on: https://go-review.googlesource.com/38790
Reviewed-by: Keith Randall <khr@golang.org>
We have lots of rewrite rules that vary only in the fact that
we have 2 versions for the 2 different orderings of various
commuting ops. For example:
(ADDL x (MOVLconst [c])) -> (ADDLconst [c] x)
(ADDL (MOVLconst [c]) x) -> (ADDLconst [c] x)
It can get unwieldly quickly, especially when there is more than
one commuting op in a rule.
Our existing "fix" for this problem is to have rules that
canonicalize the operations first. For example:
(Eq64 x (Const64 <t> [c])) && x.Op != OpConst64 -> (Eq64 (Const64 <t> [c]) x)
Subsequent rules can then assume if there is a constant arg to Eq64,
it will be the first one. This fix kinda works, but it is fragile and
only works when we remember to include the required extra rules.
The fundamental problem is that the rule matcher doesn't
know anything about commuting ops. This CL fixes that fact.
We already have information about which ops commute. (The register
allocator takes advantage of commutivity.) The rule generator now
automatically generates multiple rules for a single source rule when
there are commutative ops in the rule. We can now drop all of our
almost-duplicate source-level rules and the canonicalization rules.
I have some CLs in progress that will be a lot less verbose when
the rule generator handles commutivity for me.
I had to reorganize the load-combining rules a bit. The 8-way OR rules
generated 128 different reorderings, which was causing the generator
to put too much code in the rewrite*.go files (the big ones were going
from 25K lines to 132K lines). Instead I reorganized the rules to
combine pairs of loads at a time. The generated rule files are now
actually a bit (5%) smaller.
[Note to reviewers: check these carefully. Most of the other rule
changes are trivial.]
Make.bash times are ~unchanged.
Compiler benchmarks are not observably different. Probably because
we don't spend much compiler time in rule matching anyway.
I've also done a pass over all of our ops adding commutative markings
for ops which hadn't had them previously.
Fixes#18292
Change-Id: I999b1307272e91965b66754576019dedcbe7527a
Reviewed-on: https://go-review.googlesource.com/38666
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Changes to ${GOARCH}Ops.go files were mechanically produced using
github.com/mdempsky/ssa-symops, a one-off tool that inserts
"SymEffect: X" elements by pattern matching against the Op names.
Change-Id: Ibf3e481ffd588647f2a31662d72114b740ccbfcf
Reviewed-on: https://go-review.googlesource.com/38084
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Explcitly block fused multiply-add pattern matching when a cast is used
after the multiplication, for example:
- (a * b) + c // can emit fused multiply-add
- float64(a * b) + c // cannot emit fused multiply-add
float{32,64} and complex{64,128} casts of matching types are now kept
as OCONV operations rather than being replaced with OCONVNOP operations
because they now imply a rounding operation (and therefore aren't a
no-op anymore).
Operations (for example, multiplication) on complex types may utilize
fused multiply-add and -subtract instructions internally. There is no
way to disable this behavior at the moment.
Improves the performance of the floating point implementation of
poly1305:
name old speed new speed delta
64 246MB/s ± 0% 275MB/s ± 0% +11.48% (p=0.000 n=10+8)
1K 312MB/s ± 0% 357MB/s ± 0% +14.41% (p=0.000 n=10+10)
64Unaligned 246MB/s ± 0% 274MB/s ± 0% +11.43% (p=0.000 n=10+10)
1KUnaligned 312MB/s ± 0% 357MB/s ± 0% +14.39% (p=0.000 n=10+8)
Updates #17895.
Change-Id: Ia771d275bb9150d1a598f8cc773444663de5ce16
Reviewed-on: https://go-review.googlesource.com/36963
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Added a flag to generic and various architectures' atomic
operations that are judged to have observable side effects
and thus cannot be dead-code-eliminated.
Test requires GOMAXPROCS > 1 without preemption in loop.
Fixes#19182.
Change-Id: Id2230031abd2cca0bbb32fd68fc8a58fb912070f
Reviewed-on: https://go-review.googlesource.com/37333
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
This CL fixes two issues:
1. Load ops were initially always lowered to unsigned loads, even
for signed types. This was fine by itself however LoadReg ops
(used to re-load spilled values) were lowered to signed loads
for signed types. This meant that spills could invalidate
optimizations that assumed the original unsigned load.
2. Types were not always being maintained correctly through rules
designed to eliminate unnecessary zero and sign extensions.
Fixes#18906.
Change-Id: I95785dcadba03f7e3e94524677e7d8d3d3b9b737
Reviewed-on: https://go-review.googlesource.com/36256
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Adds the new canMergeLoad function which can be used by rules to
decide whether a load can be merged into an operation. The function
ensures that the merge will not reorder the load relative to memory
operations (for example, stores) in such a way that the block can no
longer be scheduled.
This new function enables transformations such as:
MOVD 0(R1), R2
ADD R2, R3
to:
ADD 0(R1), R3
The two-operand form of the following instructions can now read a
single memory operand:
- ADD
- ADDC
- ADDW
- MULLD
- MULLW
- SUB
- SUBC
- SUBE
- SUBW
- AND
- ANDW
- OR
- ORW
- XOR
- XORW
Improves SHA3 performance by 6-8%.
Updates #15054.
Change-Id: Ibcb9122126cd1a26f2c01c0dfdbb42fe5e7b5b94
Reviewed-on: https://go-review.googlesource.com/29272
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
We save and restore the link register in non-leaf functions because
it is clobbered by CALLs. It is therefore available for general
purpose use.
Only enabled on s390x currently. The RC4 benchmarks in particular
benefit from the extra register:
name old speed new speed delta
RC4_128 243MB/s ± 2% 341MB/s ± 2% +40.46% (p=0.008 n=5+5)
RC4_1K 267MB/s ± 0% 359MB/s ± 1% +34.32% (p=0.008 n=5+5)
RC4_8K 271MB/s ± 0% 362MB/s ± 0% +33.61% (p=0.008 n=5+5)
Change-Id: Id23bff95e771da9425353da2f32668b8e34ba09f
Reviewed-on: https://go-review.googlesource.com/30597
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
TESTB was implemented as AND $0xff, Rx, REGTMP. Unfortunately there
is no 3-operand AND-with-immediate instruction and so it was emulated
by the assembler using two instructions.
This CL uses CMPW instead of AND and also optimizes CMPW to use
the chi instruction where possible.
Overall this CL reduces the size of the .text section of the
bin/go binary by ~2%.
Change-Id: Ic335c29fc1129378fcbb1265bfb10f5b744a0f3f
Reviewed-on: https://go-review.googlesource.com/30690
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Adds the following instructions and uses them in the SSA backend:
- ANDW
- ORW
- XORW
The instruction encodings for 32-bit operations are typically shorter,
particularly when an immediate is used. For example, XORW $-1, R1
only requires one instruction, whereas XOR requires two.
Also removes some unused instructions (that were emulated):
- ANDN
- NAND
- ORN
- NOR
Change-Id: Iff2a16f52004ba498720034e354be9771b10cac4
Reviewed-on: https://go-review.googlesource.com/30291
Reviewed-by: Cherry Zhang <cherryyz@google.com>
This commit makes the process of load/store merging more incremental
for both big and little endian operations. It also adds support for
32-bit shifts (needed to merge 16- and 32-bit loads/stores).
In addition, the merging of little endian stores is now supported.
Little endian stores are now up to 30 times faster.
Change-Id: Iefdd81eda4a65b335f23c3ff222146540083ad9c
Reviewed-on: https://go-review.googlesource.com/29956
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Mark nil check operations as faulting if their arg is zero.
This lets the late nilcheck pass remove duplicates.
Fixes#17242.
Change-Id: I4c9938d8a5a1e43edd85b4a66f0b34004860bcd9
Reviewed-on: https://go-review.googlesource.com/29952
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Also adds the 'find leftmost one' instruction (FLOGR) and replaces the
WORD-encoded use of FLOGR in math/big with it.
Change-Id: I18e7cd19e75b8501a6ae8bd925471f7e37ded206
Reviewed-on: https://go-review.googlesource.com/29372
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Get rid of BlockCheck. Josh goaded me into it, and I went
down a rabbithole making it happen.
NilCheck now panics if the pointer is nil and returns void, as before.
BlockCheck is gone, and NilCheck is no longer a Control value for
any block. It just exists (and deadcode knows not to throw it away).
I rewrote the nilcheckelim pass to handle this case. In particular,
there can now be multiple NilCheck ops per block.
I moved all of the arch-dependent nil check elimination done as
part of ssaGenValue into its own proper pass, so we don't have to
duplicate that code for every architecture.
Making the arch-dependent nil check its own pass means I needed
to add a bunch of flags to the opcode table so I could write
the code without arch-dependent ops everywhere.
Change-Id: I419f891ac9b0de313033ff09115c374163416a9f
Reviewed-on: https://go-review.googlesource.com/29120
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
The new SSA backend modifies the ABI slightly: R0 is now a usable
general purpose register.
Fixes#16677.
Change-Id: I367435ce921e0c7e79e021c80cf8ef5d1d1466cf
Reviewed-on: https://go-review.googlesource.com/28978
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
