The existing code for recover from deferrangefunc was broken in
several ways.
1. the code following a deferrangefunc call did not check the return
value for an out-of-band value indicating "return now" (i.e., recover
was called)
2. the returned value was delivered using a bespoke ABI that happened
to match on register-ABI platforms, but not on older stack-based
ABI.
3. the returned value was the wrong width (1 word versus 2) and
type/value(integer 1, not a pointer to anything) for deferrangefunc's
any-typed return value (in practice, the OOB value check could catch
this, but still, it's sketchy).
This -- using the deferreturn lookup method already in place for
open-coded defers -- turned out to be a much-less-ugly way of
obtaining the desired transfer of control for recover().
TODO: we also could do this for regular defer, and delete some code.
Fixes#71840
Change-Id: If7d7ea789ad4320821aab3b443759a7d71647ff0
Reviewed-on: https://go-review.googlesource.com/c/go/+/650476
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-on: https://go-review.googlesource.com/c/go/+/651497
Handles a lot more cases where constant ranges can eliminate
various (mostly bounds failure) paths.
Fixes#66826Fixes#66692Fixes#48213
Update #57959
TODO: remove constant logic from poset code, no longer needed.
Change-Id: Id196436fcd8a0c84c7d59c04f93bd92e26a0fd7e
Reviewed-on: https://go-review.googlesource.com/c/go/+/599096
Reviewed-by: David Chase <drchase@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
CL 586975 added support to the compiler back end to emit a synthetic
".closureptr" variable in range func bodies, plus code to spill the
incoming context pointer to that variable's location on the stack.
This patch fixes up the code in the back end that generates DWARF
location lists for incoming parameters (which sometimes arrive in
registers) in the "-l -N" no-optimization case to also create a
correct DWARF location list for ".closureptr", a two-piece list
reflecting the fact that its value arrives in a register and then is
spilled to the stack in the prolog.
Fixes#67918.
Change-Id: I029305b5248b8140253fdeb6821b877916fbb87a
Reviewed-on: https://go-review.googlesource.com/c/go/+/591595
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Alessandro Arzilli <alessandro.arzilli@gmail.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
This appears to be useful only on amd64, and was specifically
benchmarked on Apple Silicon and did not produce any benefit there.
This CL adds the assembly instruction `PCALIGNMAX align,amount`
which aligns to `align` if that can be achieved with `amount`
or fewer bytes of padding. (0 means never, but will align the
enclosing function.)
Specifically, if low-order-address-bits + amount are
greater than or equal to align; thus, `PCALIGNMAX 64,63` is
the same as `PCALIGN 64` and `PCALIGNMAX 64,0` will never
emit any alignment, but will still cause the function itself
to be aligned to (at least) 64 bytes.
Change-Id: Id51a056f1672f8095e8f755e01f72836c9686aa3
Reviewed-on: https://go-review.googlesource.com/c/go/+/577935
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
[This is a partial roll-forward of CL 553055, the main change here
is that the stack slot overlap operation is flagged off by default
(can be enabled by hand with -gcflags=-d=mergelocals=1) ]
Preliminary compiler support for merging/overlapping stack slots of
local variables whose access patterns are disjoint.
This patch includes changes in AllocFrame to do the actual
merging/overlapping based on information returned from a new
liveness.MergeLocals helper. The MergeLocals helper identifies
candidates by looking for sets of AUTO variables that either A) have
the same size and GC shape (if types contain pointers), or B) have the
same size (but potentially different types as long as those types have
no pointers). Variables must be greater than (3*types.PtrSize) in size
to be considered for merging.
After forming candidates, MergeLocals collects variables into "can be
overlapped" equivalence classes or partitions; this process is driven
by an additional liveness analysis pass. Ideally it would be nice to
move the existing stackmap liveness pass up before AllocFrame
and "widen" it to include merge candidates so that we can do just a
single liveness as opposed to two passes, however this may be difficult
given that the merge-locals liveness has to take into account
writes corresponding to dead stores.
This patch also required a change to the way ssa.OpVarDef pseudo-ops
are generated; prior to this point they would only be created for
variables whose type included pointers; if stack slot merging is
enabled then the ssagen code creates OpVarDef ops for all auto vars
that are merge candidates.
Note that some temporaries created late in the compilation process
(e.g. during ssa backend) are difficult to reason about, especially in
cases where we take the address of a temp and pass it to the runtime.
For the time being we mark most of the vars created post-ssagen as
"not a merge candidate".
Stack slot merging for locals/autos is enabled by default if "-N" is
not in effect, and can be disabled via "-gcflags=-d=mergelocals=0".
Fixmes/todos/restrictions:
- try lowering size restrictions
- re-evaluate the various skips that happen in SSA-created autotmps
Updates #62737.
Updates #65532.
Updates #65495.
Change-Id: Ifda26bc48cde5667de245c8a9671b3f0a30bb45d
Reviewed-on: https://go-review.googlesource.com/c/go/+/575415
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
This reverts CL 553055.
Reason for revert: causes crypto/ecdsa failures on linux ppc64/s390x builders
Change-Id: I9266b030693a5b6b1e667a009de89d613755b048
Reviewed-on: https://go-review.googlesource.com/c/go/+/575236
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Auto-Submit: Than McIntosh <thanm@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Preliminary compiler support for merging/overlapping stack
slots of local variables whose access patterns are disjoint.
This patch includes changes in AllocFrame to do the actual
merging/overlapping based on information returned from a new
liveness.MergeLocals helper. The MergeLocals helper identifies
candidates by looking for sets of AUTO variables that either A) have
the same size and GC shape (if types contain pointers), or B) have the
same size (but potentially different types as long as those types have
no pointers). Variables must be greater than (3*types.PtrSize) in size
to be considered for merging.
After forming candidates, MergeLocals collects variables into "can be
overlapped" equivalence classes or partitions; this process is driven
by an additional liveness analysis pass. Ideally it would be nice to
move the existing stackmap liveness pass up before AllocFrame
and "widen" it to include merge candidates so that we can do just a
single liveness as opposed to two passes, however this may be difficult
given that the merge-locals liveness has to take into account
writes corresponding to dead stores.
This patch also required a change to the way ssa.OpVarDef pseudo-ops
are generated; prior to this point they would only be created for
variables whose type included pointers; if stack slot merging is
enabled then the ssagen code creates OpVarDef ops for all auto vars
that are merge candidates.
Note that some temporaries created late in the compilation process
(e.g. during ssa backend) are difficult to reason about, especially in
cases where we take the address of a temp and pass it to the runtime.
For the time being we mark most of the vars created post-ssagen as
"not a merge candidate".
Stack slot merging for locals/autos is enabled by default if "-N" is
not in effect, and can be disabled via "-gcflags=-d=mergelocals=0".
Fixmes/todos/restrictions:
- try lowering size restrictions
- re-evaluate the various skips that happen in SSA-created autotmps
Fixes#62737.
Updates #65532.
Updates #65495.
Cq-Include-Trybots: luci.golang.try:gotip-linux-amd64-longtest
Change-Id: Ibc22e8a76c87e47bc9fafe4959804d9ea923623d
Reviewed-on: https://go-review.googlesource.com/c/go/+/553055
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
This commit is aimed at improving the readability and consistency
of the code base. Extraneous newline characters were present after
some return statements, creating unnecessary separation in the code.
Fixes#64610
Change-Id: Ic1b05bf11761c4dff22691c2f1c3755f66d341f7
Reviewed-on: https://go-review.googlesource.com/c/go/+/548316
Auto-Submit: Dmitri Shuralyov <dmitshur@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Uses ,ABI instead of <ABI> because of problems with shell escaping
and windows file names, however if someone goes to all the trouble
of escaping the linker syntax and uses that instead, that works too.
Examples:
```
GOSSAFUNC=runtime.exitsyscall go build main.go
\# runtime
dumped SSA for exitsyscall,0 to ../../src/loopvar/ssa.html
dumped SSA for exitsyscall,1 to ../../src/loopvar/ssa.html
GOSSADIR=`pwd` GOSSAFUNC=runtime.exitsyscall go build main.go
\# runtime
dumped SSA for exitsyscall,0 to ../../src/loopvar/runtime.exitsyscall,0.html
dumped SSA for exitsyscall,1 to ../../src/loopvar/runtime.exitsyscall,1.html
GOSSAFUNC=runtime.exitsyscall,0 go build main.go
\# runtime
dumped SSA for exitsyscall,0 to ../../src/loopvar/ssa.html
GOSSAFUNC=runtime.exitsyscall\<1\> go build main.go
\# runtime
dumped SSA for exitsyscall,1 to ../../src/loopvar/ssa.html
```
Change-Id: Ia1138b61c797d0de49dbfae702dc306b9650a7f8
Reviewed-on: https://go-review.googlesource.com/c/go/+/532475
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: David Chase <drchase@google.com>
Change-Id: I0858568d225daba1c318842dc0c9b5e652dff612
Reviewed-on: https://go-review.googlesource.com/c/go/+/526519
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
This method is only used to find the path of the function being
compiled for hash debugging, but it was instead returning the path of
the package being compiled. These are typically the same, but can be
different for certain functions compiled across package boundaries
(e.g., method value wrappers and generic functions).
It's redundant either with f.fe.Func().Sym().Pkg.Path (package path of
the function being compiled) or f.Config.ctxt.Pkgpath (package path of
the compilation unit), so just remove it instead.
Change-Id: I1daae09055043d0ecb1fcc874a0b0006a6f8bddf
Reviewed-on: https://go-review.googlesource.com/c/go/+/526516
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Add Config and Cache as params rather than documenting that the caller
has to set them manually.
Change-Id: I8d530be695a0c94bcc4211b496d6e57ec2fff029
Reviewed-on: https://go-review.googlesource.com/c/go/+/526515
Auto-Submit: Matthew Dempsky <mdempsky@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Using more of internal/bisect gives us more that will be deleted
from base/hashdebug.go when we have updated the tools that
need the old protocol. It is also cheaper: there is no allocation to
make a decision about whether to enable, and no locking unless
printing is needed.
Change-Id: I43ec398461205a1a9e988512a134ed6b3a3b1587
Reviewed-on: https://go-review.googlesource.com/c/go/+/493736
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
HashDebugPos function/method included a parameter that was always
the same, and a variable in the same package as the hashdebug code.
So remove it.
(I wrote that code, there was no reason for it to be that way).
Also corrects a stale comment in the loopvar code.
Change-Id: Id3da69cfe6dadeb31d5de62fb76d15103a5d6152
Reviewed-on: https://go-review.googlesource.com/c/go/+/482816
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Instead of keeping track of in which blocks write barriers complete,
introduce a new op that marks the exact memory state where the
write barrier completes.
For future use. This allows us to move some of the write barrier code
to between the start of the merging block and the WBend marker.
Change-Id: If3809b260292667d91bf0ee18d7b4d0eb1e929f0
Reviewed-on: https://go-review.googlesource.com/c/go/+/447777
Reviewed-by: Keith Randall <khr@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
Change-Id: I4cff6b2a1fed6acdf754539c3c53a61eaa3b3f84
Reviewed-on: https://go-review.googlesource.com/c/go/+/450176
Auto-Submit: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@google.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Martin Möhrmann <moehrmann@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Temporary registers are sometimes needed for an architecture backend
which needs to use several machine instructions to implement a single
SSA instruction.
Mark such instructions so that regalloc can reserve the temporary register
for it. That way we don't have to reserve a fixed register like we do now.
Convert the temp-register-using instructions on amd64 to use this
new mechanism. Other archs can follow as needed.
Change-Id: I1d0c8588afdad5cd18b4398eb5a0f755be5dead7
Reviewed-on: https://go-review.googlesource.com/c/go/+/398556
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: David Chase <drchase@google.com>
Modified the fmahash gc debug flag to use this, and modified the
test to check for a hash match that includes inlining. Also
made the test non-short to ensure portability.
Note fma.go has been enhanced into an FMA test that requires
two separate FMAs in order to "fail"; if either one is 2-rounding,
then it "passes". (It neither passes nor fails here; its role
is to demonstrate that the FMAs are correctly reported; the
enhanced failure mode was discovered while testing the search
tool.)
Change-Id: I4e328e3654f442d498eac982135420abb59c5434
Reviewed-on: https://go-review.googlesource.com/c/go/+/448358
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Run-TryBot: David Chase <drchase@google.com>
This adds a -d debug flag "fmahash" for hashcode search for
floating point architecture-dependent problems. This variable has no
effect on architectures w/o fused-multiply-add.
This was rebased onto the GOSSAHASH renovation so that this could have
its own dedicated environment variable, and so that it would be
cheap (a nil check) to check it in the normal case.
Includes a basic test of the trigger plumbing.
Sample use (on arm64, ppc64le, s390x):
% GOCOMPILEDEBUG=fmahash=001110110 \
go build -o foo cmd/compile/internal/ssa/testdata/fma.go
fmahash triggered main.main:24 101111101101111001110110
GOFMAHASH triggered main.main:20 010111010000101110111011
1.0000000000000002 1.0000000000000004 -2.220446049250313e-16
exit status 1
The intended use is in conjunction with github.com/dr2chase/gossahash,
which will probably acquire a flag "-fma" to streamline its use. This
tool+use was inspired by an ad hoc use of this technique "in anger"
to debug this very problem. This is also a dry-run for using this
same technique to identify code sensitive to loop variable
lifetime/capture, should we make that change.
Example intended use, with current search tool (using old environment
variable), for a test example:
gossahash -e GOFMAHASH GOMAGIC=GOFMAHASH go run fma.go
Trying go args=[...], env=[GOFMAHASH=1 GOMAGIC=GOFMAHASH]
go failed (81 distinct triggers): exit status 1
Trying go args=[...], env=[GOFMAHASH=11 GOMAGIC=GOFMAHASH]
go failed (39 distinct triggers): exit status 1
Trying go args=[...], env=[GOFMAHASH=011 GOMAGIC=GOFMAHASH]
go failed (18 distinct triggers): exit status 1
Trying go args=[...], env=[GOFMAHASH=0011 GOMAGIC=GOFMAHASH]
Trying go args=[...], env=[GOFMAHASH=1011 GOMAGIC=GOFMAHASH]
...
Trying go args=[...], env=[GOFMAHASH=0110111011 GOMAGIC=GOFMAHASH]
Trying go args=[...], env=[GOFMAHASH=1110111011 GOMAGIC=GOFMAHASH]
go failed (2 distinct triggers): exit status 1
Trigger string is 'GOFMAHASH triggered math.qzero:427 111111101010011110111011', repeated 6 times
Trigger string is 'GOFMAHASH triggered main.main:20 010111010000101110111011', repeated 1 times
Trying go args=[...], env=[GOFMAHASH=01110111011 GOMAGIC=GOFMAHASH]
go failed (1 distinct triggers): exit status 1
Trigger string is 'GOFMAHASH triggered main.main:20 010111010000101110111011', repeated 1 times
Review GSHS_LAST_FAIL.0.log for failing run
FINISHED, suggest this command line for debugging:
GOSSAFUNC='main.main:20 010111010000101110111011' \
GOFMAHASH=01110111011 GOMAGIC=GOFMAHASH go run fma.go
Change-Id: Ifa22dd8f1c37c18fc8a4f7c396345a364bc367d5
Reviewed-on: https://go-review.googlesource.com/c/go/+/394754
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: David Chase <drchase@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Randomized feature enable/disable might be something we use to
help users debug any problems with changed loop variable capture,
and there's another CL that would like to use it to help in
locating places where "fused" multiply add instructions change
program behavior.
This CL:
- adds the ability to include an integer parameter (e.g. line number)
- replumbed the environment variable into a flag to simplify go build cache management
- but added an environment variable to allow flag setting through the environment
- which adds the possibility of switching on a different variable
(if there's one built-in for variable capture, it shouldn't be GOSSAHASH)
- cleaned up the checking code
- adds tests for all the intended behavior
- removes the case for GSHS_LOGFILE; TBD whether we'll need to put that back
or if there is another way.
Change-Id: I8503e1bb3dbc4a743aea696e04411ea7ab884787
Reviewed-on: https://go-review.googlesource.com/c/go/+/443063
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
Run-TryBot: David Chase <drchase@google.com>
We kind of have this mechanism already, just normalizing it and
using it in a bunch of places. Previously a bunch of places cached
slices only for the duration of a single function compilation. Now
we can reuse slices across a whole compiler run.
Use a sync.Pool of powers-of-two sizes. This lets us use not
too much memory, and avoid holding onto memory we're no longer
using when a GC happens.
There's a few different types we need, so generate the code for it.
Generics would be useful here, but we can't use generics in the
compiler because of bootstrapping.
Change-Id: I6cf37e7b7b2e802882aaa723a0b29770511ccd82
Reviewed-on: https://go-review.googlesource.com/c/go/+/444820
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Make them a separate type, so the normal sparse maps don't
need the extra storage.
Change-Id: I3a0219487c35ea63723499723b0c742e321d97c4
Reviewed-on: https://go-review.googlesource.com/c/go/+/444819
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Change-Id: I4dad103d23121a21b04800ec157487fdf79f89a6
Reviewed-on: https://go-review.googlesource.com/c/go/+/424398
Run-TryBot: hopehook <hopehook@qq.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Ian Lance Taylor <iant@google.com>
Run-TryBot: Ian Lance Taylor <iant@google.com>
Reviewed-by: Ian Lance Taylor <iant@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
I've needed this more than once in the past, I hack it in,
then throw it away, seems sensible to make the change and
save it.
Fixes#53937.
Change-Id: I7fe886b1c93d73cbf553bed587f2c30f0f5d5a0b
Reviewed-on: https://go-review.googlesource.com/c/go/+/418015
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
When we add GOEXPERIMENT=boringcrypto, the bootstrap process
will not converge if the compiler itself depends on the boringcrypto
cgo-based implementations of sha1 and sha256.
Using notsha256 avoids boringcrypto and makes bootstrap converge.
Removing md5 is not strictly necessary but it seemed worthwhile to
be consistent.
For #51940.
Change-Id: Iba649507e0964d1a49a1d16e463dd23c4e348f14
Reviewed-on: https://go-review.googlesource.com/c/go/+/402595
Reviewed-by: Ian Lance Taylor <iant@google.com>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
[This CL is part of a sequence implementing the proposal #51082.
The design doc is at https://go.dev/s/godocfmt-design.]
Run the updated gofmt, which reformats doc comments,
on the main repository. Vendored files are excluded.
For #51082.
Change-Id: I7332f099b60f716295fb34719c98c04eb1a85407
Reviewed-on: https://go-review.googlesource.com/c/go/+/384268
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
A run of lines that are indented with any number of spaces or tabs
format as a <pre> block. This commit fixes various doc comments
that format badly according to that (standard) rule.
For example, consider:
// - List item.
// Second line.
// - Another item.
Because the - lines are unindented, this is actually two paragraphs
separated by a one-line <pre> block. This CL rewrites it to:
// - List item.
// Second line.
// - Another item.
Today, that will format as a single <pre> block.
In a future release, we hope to format it as a bulleted list.
Various other minor fixes as well, all in preparation for reformatting.
For #51082.
Change-Id: I95cf06040d4186830e571cd50148be3bf8daf189
Reviewed-on: https://go-review.googlesource.com/c/go/+/384257
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Extend the existing dump-to-file to also do assembly output
to make it easier to write debug-information tests that check
for line-numbering in particular orders.
Includes POC test (which is silent w/o -v):
go test -v -run TestDebugLines cmd/compile/internal/ssa
=== RUN TestDebugLines
Preserving temporary directory /var/folders/v6/xyzzy/T/debug_lines_test321
About to run (cd /var/folders/v6/xyzzy/T/debug_lines_test321; \
GOSSADIR=/var/folders/v6/xyzzy/T/debug_lines_test321 \
/Users/drchase/work/go/bin/go build -o foo.o \
'-gcflags=-N -l -d=ssa/genssa/dump=sayhi' \
/Users/drchase/work/go/src/cmd/compile/internal/ssa/testdata/sayhi.go )
Saw stmt# 8 for submatch '8' on dump line #7 = ' v107 00005 (+8) MOVQ AX, "".n(SP)'
Saw stmt# 9 for submatch '9' on dump line #9 = ' v87 00007 (+9) MOVUPS X15, ""..autotmp_2-32(SP)'
Saw stmt# 10 for submatch '10' on dump line #46 = ' v65 00044 (+10) MOVUPS X15, ""..autotmp_2-32(SP)'
Saw stmt# 11 for submatch '11' on dump line #83 = ' v131 00081 (+11) MOVQ "".wg+8(SP), AX'
--- PASS: TestDebugLines (4.95s)
PASS
ok cmd/compile/internal/ssa 5.685s
Includes a test to ensure that inlining information is printed correctly.
Updates #47880.
Change-Id: I83b596476a88687d71d5b65dbb94641a576d747e
Reviewed-on: https://go-review.googlesource.com/c/go/+/348970
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
steals idea from CL 312093
further investigation revealed additional duplicate
slots (equivalent, but not equal), so delete those too.
Rearranged Func.Names to be addresses of slots,
create canonical addresses so that split slots
(which use those addresses to refer to their parent,
and split slots can be further split)
will preserve "equivalent slots are equal".
Removes duplicates, improves metrics for "args at entry".
Change-Id: I5bbdcb50bd33655abcab3d27ad8cdce25499faaf
Reviewed-on: https://go-review.googlesource.com/c/go/+/312292
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
former code only spilled those parameters mentioned in code
AT THE REGISTER LEVEL, this caused problems with liveness
sometimes (which worked on whole variables including
aggregates).
Updates #40724.
Change-Id: Ib9fdc50d95d1d2b1f1e405dd370540e88582ac71
Reviewed-on: https://go-review.googlesource.com/c/go/+/310690
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
When an SSA pass ICEs, it calls f.Fatalf, which terminates the
compiler. When GOSSAFUNC is set, the current pass is not written
to ssa.html. This CL makes it write ssa.html when it calls Fatalf,
for the ease of debugging.
Change-Id: I5d55e4258f0693d89c48c0a84984f2f893b0811d
Reviewed-on: https://go-review.googlesource.com/c/go/+/307509
Trust: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: David Chase <drchase@google.com>
Plumb abi information into ssa/ssagen for plain calls
and plain functions (not methods). Does not extend all the
way through the compiler (yet).
One test disabled because it extends far enough to break the test.
Normalized all the compiler's register args TODOs to
// TODO(register args) ...
For #40724.
Change-Id: I0173a4579f032ac3c9db3aef1749d40da5ea01ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/293389
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
By-hand rebase of earlier CL, because that was easier than
letting git try to figure things out.
This will naively insert self-moves; in the case that these
involve memory, the expander detects these and removes them
and their vardefs.
Change-Id: Icf72575eb7ae4a186b0de462bc8cf0bedc84d3e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/279519
Trust: David Chase <drchase@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
This is a selected copy from the register ABI experiment CL, focused
on the files and data structures that handle spilling around morestack.
Unnecessary code from the experiment was removed, other code was adapted.
Would it make sense to leave comments in the experiment as pieces are
brought over?
Experiment CL (for comparison purposes)
https://go-review.googlesource.com/c/go/+/28832
Change-Id: I92136f070351d4fcca1407b52ecf9b80898fed95
Reviewed-on: https://go-review.googlesource.com/c/go/+/279520
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
It's currently hard to automate refactorings around the Value.Aux
field, because we don't have any static typing information for it.
Adding a tag interface will make subsequent CLs easier and safer.
Passes buildall w/ toolstash -cmp.
Updates #42982.
Change-Id: I41ae8e411a66bda3195a0957b60c2fe8a8002893
Reviewed-on: https://go-review.googlesource.com/c/go/+/275756
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Still needs to generate the calls that will need lowering.
Change-Id: Ifd4e510193441a5e27c462c1f1d704f07bf6dec3
Reviewed-on: https://go-review.googlesource.com/c/go/+/242359
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
It was not necessarily consistent before, we were just lucky.
Change-Id: I3a92dc724e0af7b4d810a6a0b7b1d58844eb8f87
Reviewed-on: https://go-review.googlesource.com/c/go/+/251440
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Turns out if your failure is in a function with a name like "Reset()"
there will be a lot of hits on the same hashcode. Adding package sensitivity
solves this problem.
In additionm, it turned out that in the case that a logfile was specified
for the GOSSAHASH logging, that it was opened in create mode, which meant
that multiple compiler invocations would reset the file to zero length.
Opening in append mode works better; the automated harness
(github.com/dr2chase/gossahash) takes care of truncating the file before use.
Change-Id: I5601bc280faa94cbd507d302448831849db6c842
Reviewed-on: https://go-review.googlesource.com/c/go/+/246937
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
I noticed that there is a Todo comment here. This variable is only used for filename when dump a function's ssa passes result in details. It is no problem to print a function alone, but may be edited by not only one goroutine if dump multiple functions at the same time. Although it looks only dump one function's ssa passes now. As far as I am concerned this variable can be a member variable of the struct Func. I'm not sure if this change is necessary. Looking forward to your advices, thank you very much.
Change-Id: I35dd7247889e0cc7f19c0b400b597206592dee75
Reviewed-on: https://go-review.googlesource.com/c/go/+/244918
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Taking the live variable set from the last return point is problematic.
See #40629 for details, but there may not be a return point, or it may
be before the final defer.
Additionally, keeping track of the last call as a *Value doesn't quite
work. If it is dead-code eliminated, the storage for the Value is reused
for some other random instruction. Its live variable information,
if it is available at all, is wrong.
Instead, just mark all the open-defer argument slots as live
throughout the function. (They are already zero-initialized.)
Fixes#40629
Change-Id: Ie456c7db3082d0de57eaa5234a0f32525a1cce13
Reviewed-on: https://go-review.googlesource.com/c/go/+/247522
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Once defined, a stack slot holding an open-coded defer arg should always be marked
live, since it may be used at any time if there is a panic. These stack slots are
typically kept live naturally by the open-defer code inlined at each return/exit point.
However, we need to do extra work to make sure that they are kept live if a
function has an infinite loop or a panic exit.
For this fix, only in the case of a function that is using open-coded defers, we
compute the set of blocks (most often empty) that cannot reach a return or a
BlockExit (panic) because of an infinite loop. Then, for each block b which
cannot reach a return or BlockExit or is a BlockExit block, we mark each defer arg
slot as live, as long as the definition of the defer arg slot dominates block b.
For this change, had to export (*Func).sdom (-> Sdom) and SparseTree.isAncestorEq
(-> IsAncestorEq)
Updates #35277
Change-Id: I7b53c9bd38ba384a3794386dd0eb94e4cbde4eb1
Reviewed-on: https://go-review.googlesource.com/c/go/+/204802
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Generate inline code at defer time to save the args of defer calls to unique
(autotmp) stack slots, and generate inline code at exit time to check which defer
calls were made and make the associated function/method/interface calls. We
remember that a particular defer statement was reached by storing in the deferBits
variable (always stored on the stack). At exit time, we check the bits of the
deferBits variable to determine which defer function calls to make (in reverse
order). These low-cost defers are only used for functions where no defers
appear in loops. In addition, we don't do these low-cost defers if there are too
many defer statements or too many exits in a function (to limit code increase).
When a function uses open-coded defers, we produce extra
FUNCDATA_OpenCodedDeferInfo information that specifies the number of defers, and
for each defer, the stack slots where the closure and associated args have been
stored. The funcdata also includes the location of the deferBits variable.
Therefore, for panics, we can use this funcdata to determine exactly which defers
are active, and call the appropriate functions/methods/closures with the correct
arguments for each active defer.
In order to unwind the stack correctly after a recover(), we need to add an extra
code segment to functions with open-coded defers that simply calls deferreturn()
and returns. This segment is not reachable by the normal function, but is returned
to by the runtime during recovery. We set the liveness information of this
deferreturn() to be the same as the liveness at the first function call during the
last defer exit code (so all return values and all stack slots needed by the defer
calls will be live).
I needed to increase the stackguard constant from 880 to 896, because of a small
amount of new code in deferreturn().
The -N flag disables open-coded defers. '-d defer' prints out the kind of defer
being used at each defer statement (heap-allocated, stack-allocated, or
open-coded).
Cost of defer statement [ go test -run NONE -bench BenchmarkDefer$ runtime ]
With normal (stack-allocated) defers only: 35.4 ns/op
With open-coded defers: 5.6 ns/op
Cost of function call alone (remove defer keyword): 4.4 ns/op
Text size increase (including funcdata) for go binary without/with open-coded defers: 0.09%
The average size increase (including funcdata) for only the functions that use
open-coded defers is 1.1%.
The cost of a panic followed by a recover got noticeably slower, since panic
processing now requires a scan of the stack for open-coded defer frames. This scan
is required, even if no frames are using open-coded defers:
Cost of panic and recover [ go test -run NONE -bench BenchmarkPanicRecover runtime ]
Without open-coded defers: 62.0 ns/op
With open-coded defers: 255 ns/op
A CGO Go-to-C-to-Go benchmark got noticeably faster because of open-coded defers:
CGO Go-to-C-to-Go benchmark [cd misc/cgo/test; go test -run NONE -bench BenchmarkCGoCallback ]
Without open-coded defers: 443 ns/op
With open-coded defers: 347 ns/op
Updates #14939 (defer performance)
Updates #34481 (design doc)
Change-Id: I63b1a60d1ebf28126f55ee9fd7ecffe9cb23d1ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/202340
Reviewed-by: Austin Clements <austin@google.com>
Generate inline code at defer time to save the args of defer calls to unique
(autotmp) stack slots, and generate inline code at exit time to check which defer
calls were made and make the associated function/method/interface calls. We
remember that a particular defer statement was reached by storing in the deferBits
variable (always stored on the stack). At exit time, we check the bits of the
deferBits variable to determine which defer function calls to make (in reverse
order). These low-cost defers are only used for functions where no defers
appear in loops. In addition, we don't do these low-cost defers if there are too
many defer statements or too many exits in a function (to limit code increase).
When a function uses open-coded defers, we produce extra
FUNCDATA_OpenCodedDeferInfo information that specifies the number of defers, and
for each defer, the stack slots where the closure and associated args have been
stored. The funcdata also includes the location of the deferBits variable.
Therefore, for panics, we can use this funcdata to determine exactly which defers
are active, and call the appropriate functions/methods/closures with the correct
arguments for each active defer.
In order to unwind the stack correctly after a recover(), we need to add an extra
code segment to functions with open-coded defers that simply calls deferreturn()
and returns. This segment is not reachable by the normal function, but is returned
to by the runtime during recovery. We set the liveness information of this
deferreturn() to be the same as the liveness at the first function call during the
last defer exit code (so all return values and all stack slots needed by the defer
calls will be live).
I needed to increase the stackguard constant from 880 to 896, because of a small
amount of new code in deferreturn().
The -N flag disables open-coded defers. '-d defer' prints out the kind of defer
being used at each defer statement (heap-allocated, stack-allocated, or
open-coded).
Cost of defer statement [ go test -run NONE -bench BenchmarkDefer$ runtime ]
With normal (stack-allocated) defers only: 35.4 ns/op
With open-coded defers: 5.6 ns/op
Cost of function call alone (remove defer keyword): 4.4 ns/op
Text size increase (including funcdata) for go cmd without/with open-coded defers: 0.09%
The average size increase (including funcdata) for only the functions that use
open-coded defers is 1.1%.
The cost of a panic followed by a recover got noticeably slower, since panic
processing now requires a scan of the stack for open-coded defer frames. This scan
is required, even if no frames are using open-coded defers:
Cost of panic and recover [ go test -run NONE -bench BenchmarkPanicRecover runtime ]
Without open-coded defers: 62.0 ns/op
With open-coded defers: 255 ns/op
A CGO Go-to-C-to-Go benchmark got noticeably faster because of open-coded defers:
CGO Go-to-C-to-Go benchmark [cd misc/cgo/test; go test -run NONE -bench BenchmarkCGoCallback ]
Without open-coded defers: 443 ns/op
With open-coded defers: 347 ns/op
Updates #14939 (defer performance)
Updates #34481 (design doc)
Change-Id: I51a389860b9676cfa1b84722f5fb84d3c4ee9e28
Reviewed-on: https://go-review.googlesource.com/c/go/+/190098
Reviewed-by: Austin Clements <austin@google.com>
This reduces allocations and also resolves some
lurking inliner/inlinee line-number match problems.
However, it does add about 1.5% to compile time.
This fixes compiler OOMs seen compiling some large protobuf-
derived inputs. For compiling the compiler itself,
compilebench -pkg cmd/compile/internal/ssa -memprofile withcl.prof
the numberlines-related memory consumption is reduced from 129MB
to 29MB (about a 5% overall reduction in allocation).
Additionally modified after going over changes with Austin
to remove unused code (nobody called size()) and correct
the cache-clearing code.
I've attempted to speed this up by not using maps, and have
not succeeded. I'd rather get correct code in now, speed it
up later if I can.
Updates #27739.
Fixes#29279.
Change-Id: I098005de4e45196a5f5b10c0886a49f88e9f8fd5
Reviewed-on: https://go-review.googlesource.com/c/go/+/154617
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
