all: support for Go 1.27 by deadprogram · Pull Request #5477 · tinygo-org/tinygo

deadprogram · 2026-06-25T20:27:18Z

This PR is to add support for Go 1.27

Current passes tests on 1.27-rc2

github-actions · 2026-06-25T20:43:34Z

Size difference with the dev branch:

Binary size difference

drivers/sizes-dev.txt has more commands than drivers/sizes-pr.txt
    tinygo build -size short -o ./build/test.hex -target=feather-rp2040 ./examples/adafruit4650
 flash                          ram
 before   after   diff          before   after   diff

deadprogram · 2026-06-30T18:41:06Z

Added another commit with support for the new //go:linknamestd directive that is new in Go 1.27

deadprogram · 2026-06-30T20:29:24Z

It got further. I think perhaps the new set of errors might have to do with the SSA changes for generics we were just discussing @jakebailey @aykevl

/home/runner/go/bin/tinygo test -target=wasip1 -skip='TestExtraMethods|TestParseAndBytesRoundTrip/P256/Generic|TestAsValidation' cmp compress/lzw compress/zlib container/heap container/list container/ring crypto/ecdsa crypto/elliptic crypto/md5 crypto/sha1 crypto/sha256 crypto/sha512 database/sql/driver debug/macho embed/internal/embedtest encoding encoding/ascii85 errors encoding/asn1 encoding/base32 encoding/base64 encoding/csv encoding/hex expvar go/ast go/format go/scanner go/token go/version hash hash/adler32 hash/crc64 hash/fnv html internal/itoa internal/profile math math/cmplx net/http/internal/ascii net/mail net/url os path reflect sync testing testing/iotest text/scanner unicode unicode/utf16 unicode/utf8 unique  ./tests/runtime_wasi
ok  	cmp	1.389s
ok  	compress/lzw	1.318s
panic: expected nil interface constant

goroutine 1702 [running]:
github.com/tinygo-org/tinygo/compiler.(*compilerContext).createConst(0x36af075d1320, 0x36aeff5398a0, 0x2dc853)
	/__w/tinygo/tinygo/compiler/compiler.go:3149 +0xf65
github.com/tinygo-org/tinygo/compiler.(*builder).getValue(0x36af07de0a80, {0xa0a3c8?, 0x36aeff5398a0}, 0x36af07d870e8?)
	/__w/tinygo/tinygo/compiler/compiler.go:2116 +0xb8
github.com/tinygo-org/tinygo/compiler.(*builder).createExpr(0x36af07de0a80, {0xa0a800, 0x36af07acee70})
	/__w/tinygo/tinygo/compiler/compiler.go:2202 +0x1a46
github.com/tinygo-org/tinygo/compiler.(*builder).createInstruction(0x36af07de0a80, {0xa0b388, 0x36af07acee70})
	/__w/tinygo/tinygo/compiler/compiler.go:1514 +0xd9f
github.com/tinygo-org/tinygo/compiler.(*builder).createFunction(0x36af07de0a80)
	/__w/tinygo/tinygo/compiler/compiler.go:1398 +0x4c5
github.com/tinygo-org/tinygo/compiler.(*compilerContext).createPackage(0x36af075d1320, {0x36aefb24f600?}, 0x36aef73e7d00)
	/__w/tinygo/tinygo/compiler/compiler.go:934 +0x945
github.com/tinygo-org/tinygo/compiler.CompilePackage({0x36af0baaeb00?, 0x5a?}, 0x36aef8f9ba40, 0x36aef73e7d00, {0x83f05cdfec64b02a?}, 0xdf8308?, 0xe4?)
	/__w/tinygo/tinygo/compiler/compiler.go:334 +0x445
github.com/tinygo-org/tinygo/builder.Build.func3(0x36af06bab680)
	/__w/tinygo/tinygo/builder/build.go:405 +0x1ea
github.com/tinygo-org/tinygo/builder.runJob(0x36af06bab680, 0x36aefd3efdc0)
	/__w/tinygo/tinygo/builder/jobs.go:212 +0x4d
created by github.com/tinygo-org/tinygo/builder.runJobs in goroutine 1582
	/__w/tinygo/tinygo/builder/jobs.go:113 +0x6c9
make: *** [GNUmakefile:533: tinygo-test-wasip1-fast] Error 2

deadprogram · 2026-07-01T08:22:32Z

Added another commit that updates x/tools to a newer version that handles the new generics changes. The consequence of this is that the new minimum version of Go for TinyGo is 1.25, which is something that we also discussed in the monthly meeting.

jakebailey · 2026-07-03T22:00:12Z

For Windows, seems like you need the chkstk.S change for more platforms.

I think TestBuild/EmulatedCortexM3 is unavoidable, new json v2 is just larger.

jakebailey · 2026-07-03T22:45:32Z

I think I actually have all of the other failures figured out, but, not sure what to do with the fixes.

One or two of them I can send to dev first. E.g. when I added recover support, I totally forgot to also enable FailNow and SkipNow, and that's what some of the failure is.

deadprogram · 2026-07-11T11:33:46Z

@jakebailey I just overwrote your commits that you had already added to this branch by force puching without having pulled your changes to my local branch first. Sorry about that, my bad :(

Can you help get them back in please?

jakebailey · 2026-07-11T11:37:25Z

I don't have the ability to for a few days but you can download these and git apply them

deadprogram · 2026-07-11T11:44:07Z

@jakebailey thanks! Just ammended this PR with those 2 commits...

deadprogram · 2026-07-11T12:08:52Z

Looks like the test failures so far are perhaps related to crypto/ecdsa.

jakebailey · 2026-07-11T12:17:40Z

Yeah, I couldn't figure those out as they did not reproduce locally

deadprogram · 2026-07-11T12:51:45Z

Actually it seems like it was compress/flate. Skipping for now to see if other tests pass.

jakebailey · 2026-07-11T13:54:25Z

That last commit reminds me of #5508

deadprogram · 2026-07-11T14:16:15Z

That last commit reminds me of #5508

Indeed, it looks like #5508 is a better version of the same idea.

deadprogram · 2026-07-11T15:42:40Z

#5508 was merged, then I removed my commit from this branch and rebased in the latest dev

deadprogram · 2026-07-11T16:29:38Z

Looks like I needed a version of that commit after all to handle the float32/float64 difference.

deadprogram · 2026-07-12T10:52:37Z

I have moved 2 of the commits in the PR into #5514 and #5515 so they can be merged sooner and to cleanup the commits in this PR.

deadprogram · 2026-07-12T12:23:54Z

Cleaned up commits a bit and also rebased against dev.

#5514 was merged, #5515 is still pending feedback so I left it in this PR still so tests could pass.

deadprogram · 2026-07-12T22:48:07Z

I removed the commit from #5515 and replaced with the commits from #5519 since it passed for me locally. Now running CI...

deadprogram · 2026-07-13T08:11:48Z

Rebased now that #5519 has been merged into dev

deadprogram · 2026-07-13T16:17:59Z

I think we should open an issue for each of the tests that we needed to skip so we do not lose track of them, and then we should squash/merge the rest of this PR since 1.27 is nearly here.

Any thoughts on this?

jakebailey · 2026-07-13T16:31:02Z

I'm already looking into some of them, but issues would be nice of course; they're definitely not things that are blockers given they don't affect 1.26

deadprogram · 2026-07-13T21:50:06Z

@jakebailey does that mean you are in favor of merge? Other people, also please provide feedback.

jakebailey · 2026-07-14T07:32:57Z

Yes; the issue with compress/flate is a large value type, but that's not a new tinygo problem, just something new in Go 1.27's code (https://go.dev/cl/707355). Fixing that is turning into quite a horrorshow. I have not looked into the crypto/ecdsa Wasm problem yet.

But in any case, I don't think that this means this PR is incomplete.

Signed-off-by: deadprogram <ron@hybridgroup.com>

Go 1.27 adds generic methods (golang/go#77273). golang.org/x/tools go/ssa v0.42.0 does not instantiate them: objectMethod only applies receiverTypeArgs and ignores method-level type parameters, so a call such as (*math/rand/v2.Rand).N resolves to the abstract generic method and its body reaches createConst with a type-parameter-typed zero constant, triggering "panic: expected nil interface constant". Upgrade golang.org/x/tools to v0.47.0, whose go/ssa returns nil from MethodValue for generic methods and instantiates method-level type parameters. No compiler changes are required. Since x/tools v0.47.0 requires Go 1.25, raise the minimum supported Go version from 1.24 to 1.25. Signed-off-by: deadprogram <ron@hybridgroup.com>

Go 1.27 packages can emit stack probes on Windows amd64 and arm64. Link the compiler-rt stack probe builtins and use compiler-rt for those targets so these packages build.

Go 1.27 jsonv2 pushes testdata/json.go beyond the LM3S6965 flash budget. Keep coverage on larger emulated targets while skipping only the too-small Cortex-M QEMU configuration.

Signed-off-by: deadprogram <ron@hybridgroup.com>

deadprogram · 2026-07-14T07:54:02Z

I have rebased the commits into the most atomic set I can detemine. Leaving the commits to skip tests to provide a more specific trail for us to restore those tests when we are able instead of just creating issues.

Thank you very much for all your help to get this done @jakebailey ⭐

Waiting for CI to complete, then will merge...

jakebailey · 2026-07-14T07:55:05Z

The Wasi issue is just Goexit not working on Wasm, so that just has to be skipped until Wasm in general supports panic/recover properly.

deadprogram · 2026-07-14T07:58:46Z

The Wasi issue is just Goexit not working on Wasm, so that just has to be skipped until Wasm in general supports panic/recover properly.

OK, that makes sense. There is already #4380 for when that occurs.

deadprogram force-pushed the go1.27-ci branch from 76dafd0 to 4f4320b Compare June 30, 2026 18:40

deadprogram force-pushed the go1.27-ci branch from 1ea4589 to cbc6503 Compare July 3, 2026 20:39

deadprogram mentioned this pull request Jul 5, 2026

apply compiler patch for fix of #5496 #5497

Open

aykevl reviewed Jul 5, 2026

View reviewed changes

Comment thread src/runtime/runtime_windows_go126.go Outdated

deadprogram force-pushed the go1.27-ci branch from cbc6503 to 68840b9 Compare July 8, 2026 07:14

deadprogram changed the title ~~all: build/test using Go 1.27-rc1~~ all: build/test using Go 1.27-rc2 Jul 8, 2026

jakebailey reviewed Jul 8, 2026

View reviewed changes

Comment thread builder/builtins.go Outdated

jakebailey force-pushed the go1.27-ci branch 3 times, most recently from cfcf9cd to db17b48 Compare July 8, 2026 17:47

deadprogram force-pushed the go1.27-ci branch from db17b48 to 2218634 Compare July 11, 2026 11:22

deadprogram force-pushed the go1.27-ci branch from 09db5df to dc2f031 Compare July 11, 2026 11:51

deadprogram force-pushed the go1.27-ci branch from 0be8f04 to 348b1b4 Compare July 11, 2026 15:41

deadprogram marked this pull request as ready for review July 11, 2026 20:29

deadprogram changed the title ~~all: support for Go 1.27-rc2~~ all: support for Go 1.27 Jul 12, 2026

deadprogram force-pushed the go1.27-ci branch from b840c5f to 452c68b Compare July 12, 2026 12:22

jakebailey mentioned this pull request Jul 12, 2026

compiler: further canonicalize function / signature names #5519

Merged

deadprogram force-pushed the go1.27-ci branch from 452c68b to 2cd3f82 Compare July 12, 2026 22:46

deadprogram force-pushed the go1.27-ci branch from 2cd3f82 to 18e4780 Compare July 13, 2026 08:11

jakebailey reviewed Jul 14, 2026

View reviewed changes

Comment thread src/runtime/os_darwin_go126.go Outdated

all: build/test using Go 1.27-rc2

213e93b

Signed-off-by: deadprogram <ron@hybridgroup.com>

deadprogram force-pushed the go1.27-ci branch from 18e4780 to 06b7812 Compare July 14, 2026 07:42

jakebailey approved these changes Jul 14, 2026

View reviewed changes

deadprogram and others added 6 commits July 14, 2026 09:51

builder: update max supported Go version to 1.27

7b5db7b

Signed-off-by: deadprogram <ron@hybridgroup.com>

builder: link stack probes on Windows amd64 and arm64

a53cfae

Go 1.27 packages can emit stack probes on Windows amd64 and arm64. Link the compiler-rt stack probe builtins and use compiler-rt for those targets so these packages build.

main_test: skip json on Cortex-M QEMU with Go 1.27

57e92c0

Go 1.27 jsonv2 pushes testdata/json.go beyond the LM3S6965 flash budget. Keep coverage on larger emulated targets while skipping only the too-small Cortex-M QEMU configuration.

test: skip compress/flate for now due to Go 1.27 changes

457acb9

Signed-off-by: deadprogram <ron@hybridgroup.com>

test: skip crypto/ecdsa on wasi for now, since it never finishes

7238da4

Signed-off-by: deadprogram <ron@hybridgroup.com>

deadprogram force-pushed the go1.27-ci branch from 06b7812 to 7238da4 Compare July 14, 2026 07:51

deadprogram merged commit 801bd48 into dev Jul 14, 2026
26 checks passed

deadprogram deleted the go1.27-ci branch July 14, 2026 09:53

jakebailey mentioned this pull request Jul 14, 2026

compiler: preemptively indirect large params/results instead of letting LLVM do it #5526

Open

Uh oh!

Uh oh!

Conversation

deadprogram commented Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

deadprogram commented Jun 30, 2026

Uh oh!

deadprogram commented Jun 30, 2026

Uh oh!

deadprogram commented Jul 1, 2026

Uh oh!

jakebailey commented Jul 3, 2026

Uh oh!

jakebailey commented Jul 3, 2026

Uh oh!

Uh oh!

Uh oh!

deadprogram commented Jul 11, 2026

Uh oh!

jakebailey commented Jul 11, 2026

Uh oh!

deadprogram commented Jul 11, 2026

Uh oh!

deadprogram commented Jul 11, 2026

Uh oh!

jakebailey commented Jul 11, 2026

Uh oh!

deadprogram commented Jul 11, 2026

Uh oh!

jakebailey commented Jul 11, 2026

Uh oh!

deadprogram commented Jul 11, 2026

Uh oh!

deadprogram commented Jul 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

deadprogram commented Jul 11, 2026

Uh oh!

deadprogram commented Jul 12, 2026

Uh oh!

deadprogram commented Jul 12, 2026

Uh oh!

deadprogram commented Jul 12, 2026

Uh oh!

deadprogram commented Jul 13, 2026

Uh oh!

deadprogram commented Jul 13, 2026

Uh oh!

jakebailey commented Jul 13, 2026

Uh oh!

deadprogram commented Jul 13, 2026

Uh oh!

Uh oh!

jakebailey commented Jul 14, 2026

Uh oh!

deadprogram commented Jul 14, 2026

Uh oh!

jakebailey commented Jul 14, 2026

Uh oh!

deadprogram commented Jul 14, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

deadprogram commented Jun 25, 2026 •

edited

Loading

github-actions Bot commented Jun 25, 2026 •

edited

Loading

deadprogram commented Jul 11, 2026 •

edited

Loading