Update musl to v1.2.6 by kleisauke · Pull Request #26860 · emscripten-core/emscripten

kleisauke · 2026-05-05T10:17:17Z

No description provided.

This is an automatic change generated with: ``` ./system/lib/update_musl.py ~/Downloads/musl-1.2.5 ```

In preparation for the next commit.

This is an automatic change generated with: ``` ./system/lib/update_musl.py ~/Downloads/musl-1.2.6 ```

This reverts commit 8afc084.

Done via: ``` ./tools/gen_struct_info.py ./tools/gen_struct_info.py --wasm64 ```

See: https://git.musl-libc.org/cgit/musl/commit/?id=ef7d0ae21240eac9fc1e8088112bfb0fac507578

Details: ``` wasm-ld: error: --shared-memory is disallowed by exit.o because it was not compiled with 'atomics' or 'bulk-memory' features. ```

This is an automatic change generated by tools/maint/rebaseline_tests.py. The following (25) test expectation files were updated by running the tests with `--rebaseline`: ``` codesize/test_codesize_cxx_ctors1.json: 151887 => 151940 [+53 bytes / +0.03%] codesize/test_codesize_cxx_ctors2.json: 151292 => 151346 [+54 bytes / +0.04%] codesize/test_codesize_cxx_except.json: 195766 => 195851 [+85 bytes / +0.04%] codesize/test_codesize_cxx_except_wasm.json: 167018 => 167109 [+91 bytes / +0.05%] codesize/test_codesize_cxx_except_wasm_legacy.json: 164898 => 164989 [+91 bytes / +0.06%] codesize/test_codesize_cxx_lto.json: 120587 => 120608 [+21 bytes / +0.02%] codesize/test_codesize_cxx_mangle.json: 262245 => 262330 [+85 bytes / +0.03%] codesize/test_codesize_cxx_noexcept.json: 153917 => 153956 [+39 bytes / +0.03%] codesize/test_codesize_cxx_wasmfs.json: 179773 => 179812 [+39 bytes / +0.02%] codesize/test_codesize_file_preload.json: 23855 => 23873 [+18 bytes / +0.08%] codesize/test_codesize_hello_O0.json: 39070 => 39335 [+265 bytes / +0.68%] codesize/test_codesize_hello_O1.json: 8875 => 8889 [+14 bytes / +0.16%] codesize/test_codesize_hello_O2.json: 6221 => 6235 [+14 bytes / +0.23%] codesize/test_codesize_hello_O3.json: 5913 => 5931 [+18 bytes / +0.30%] codesize/test_codesize_hello_Os.json: 5903 => 5919 [+16 bytes / +0.27%] codesize/test_codesize_hello_Oz.json: 5072 => 5088 [+16 bytes / +0.32%] codesize/test_codesize_hello_dylink.json: 43922 => 44112 [+190 bytes / +0.43%] codesize/test_codesize_hello_dylink_all.json: 822610 => 822562 [-48 bytes / -0.01%] codesize/test_codesize_hello_single_file.json: 5223 => 5241 [+18 bytes / +0.34%] codesize/test_codesize_hello_wasmfs.json: 5913 => 5931 [+18 bytes / +0.30%] codesize/test_codesize_minimal_pthreads.json: 26370 => 26347 [-23 bytes / -0.09%] codesize/test_codesize_minimal_pthreads_memgrowth.json: 26774 => 26751 [-23 bytes / -0.09%] codesize/test_minimal_runtime_code_size_random_printf_wasm.json: 10815 => 11057 [+242 bytes / +2.24%] codesize/test_minimal_runtime_code_size_random_printf_wasm2js.json: 17114 => 17417 [+303 bytes / +1.77%] codesize/test_unoptimized_code_size.json: 180144 => 180893 [+749 bytes / +0.42%] Average change: +0.31% (-0.09% - +2.24%) ```

sbc100 · 2026-05-05T15:33:05Z

Were you able to follow up process for updating musl in https://github.com/emscripten-core/emscripten/blob/main/docs/process.md#updating-musl?

I would expect to see some updates to https://github.com/emscripten-core/musl before the emscripten side change.

kleisauke · 2026-05-05T15:52:04Z

I skipped https://github.com/emscripten-core/musl because maintaining the code in two separate places feels unconventional (and I don't have contributor access there anyway). Instead, I followed a workflow similar to the one used when updating mimalloc:

Replace system/lib/libc/musl/ with original musl v1.2.5 via:
```
$ ./system/lib/update_musl.py ~/Downloads/musl-1.2.5
```
Commit those changes (see 8afc084).

Replace system/lib/libc/musl/ with musl v1.2.6 via:

$ ./system/lib/update_musl.py ~/Downloads/musl-1.2.6

Commit those changes (see fa6d187).
Revert the changes from step 2 (see a67d130).

sbc100 · 2026-05-05T16:03:04Z

If you think the instructions in https://github.com/emscripten-core/emscripten/blob/main/docs/process.md#updating-musl are too complex then perhaps we should update them?

I find it kind of convenient to be able to see the emscripten changes as a git diff, and it means we can follow the same pattern for llvm and musl.

Another option, if we are look into changing things up is what wasi-libc recently did suing git subtree: WebAssembly/wasi-libc#797.

But for this update would you mind updating the emscripten/musl fork? Even if we end up landing this PR, I'd still like the result reflected there.. unless we want to deprecate that repo.

kleisauke · 2026-05-05T16:16:20Z

But for this update would you mind updating the emscripten/musl fork?

Sure, I'll look into this tomorrow (as I'm in UTC+2).

kleisauke · 2026-05-06T10:02:06Z

See companion PR emscripten-core/musl#3 for a diff that includes only the musl-specific changes.

sbc100 · 2026-05-06T14:44:24Z

 typedef jmp_buf sigjmp_buf;
 /* XXX EMSCRIPTEN: No signals support, alias sigsetjmp and siglongjmp to their non-signals counterparts. */
-#if __EMSCRIPTEN__ && !defined(LLVM_LIBC)
+#if defined(__EMSCRIPTEN__) && !defined(LLVM_LIBC)


Would you mind splitting out a PR for all the places where we should have done ifdef rather then if like this?

Sure, reverted with commit a499fc2.

sbc100 · 2026-05-06T14:45:11Z

+#ifdef __EMSCRIPTEN__
 #define _POSIX_SPAWN            -1
+#else
+#define _POSIX_SPAWN            _POSIX_VERSION


Maybe you can include this kind of refactor in the split-out PR too?

Reverted with commit 1f05268.

sbc100 · 2026-05-06T14:45:43Z

 		 * that is not fdpic-aware. We can assume normal fixed-
 		 * displacement ELF loading was performed, but when ldso was
-		 * run as a command, finding the Ehdr is a heursitic: we
+		 * run as a command, finding the Ehdr is a heuristic: we


Is this a spelling fix from upstream of from you?

From upstream, see:
https://git.musl-libc.org/cgit/musl/commit?id=18289e5dea03f46dbb2315bf51a261d1f9ebe3ef

This reverts commit 2bd5f45.

This partially reverts commit 82cc2d3.

This change is the result of running ./system/lib/push_musl_changes.py

sbc100 · 2026-05-06T15:00:56Z

I created and emscripten branch update emscripten-core/musl#4.

Hopefully if we land this first it will make this PR more precise/concise?

In-line with: https://git.musl-libc.org/cgit/musl/commit?id=ffaaa6d230512f3a7f3d040b943517728f3dc3cf

This change was generated by running ./system/lib/push_musl_changes.py.

This change was generated by running ./system/lib/push_musl_changes.py within the Emscripten repo.

This is an automatic change generated by tools/maint/rebaseline_tests.py. The following (23) test expectation files were updated by running the tests with `--rebaseline`: ``` codesize/test_codesize_cxx_ctors1.json: 151949 => 152002 [+53 bytes / +0.03%] codesize/test_codesize_cxx_ctors2.json: 151352 => 151408 [+56 bytes / +0.04%] codesize/test_codesize_cxx_except.json: 195817 => 195902 [+85 bytes / +0.04%] codesize/test_codesize_cxx_except_wasm.json: 167044 => 167135 [+91 bytes / +0.05%] codesize/test_codesize_cxx_except_wasm_legacy.json: 164924 => 165015 [+91 bytes / +0.06%] codesize/test_codesize_cxx_lto.json: 120826 => 120848 [+22 bytes / +0.02%] codesize/test_codesize_cxx_mangle.json: 262296 => 262381 [+85 bytes / +0.03%] codesize/test_codesize_cxx_noexcept.json: 153965 => 154004 [+39 bytes / +0.03%] codesize/test_codesize_cxx_wasmfs.json: 179712 => 179751 [+39 bytes / +0.02%] codesize/test_codesize_file_preload.json: 23872 => 23890 [+18 bytes / +0.08%] codesize/test_codesize_hello_O0.json: 39104 => 39369 [+265 bytes / +0.68%] codesize/test_codesize_hello_O1.json: 8908 => 8922 [+14 bytes / +0.16%] codesize/test_codesize_hello_O2.json: 6242 => 6256 [+14 bytes / +0.22%] codesize/test_codesize_hello_O3.json: 5934 => 5952 [+18 bytes / +0.30%] codesize/test_codesize_hello_Os.json: 5924 => 5940 [+16 bytes / +0.27%] codesize/test_codesize_hello_Oz.json: 5093 => 5109 [+16 bytes / +0.31%] codesize/test_codesize_hello_dylink.json: 44104 => 44294 [+190 bytes / +0.43%] codesize/test_codesize_hello_dylink_all.json: 855794 => 855865 [+71 bytes / +0.01%] codesize/test_codesize_hello_single_file.json: 5244 => 5262 [+18 bytes / +0.34%] codesize/test_codesize_hello_wasmfs.json: 5934 => 5952 [+18 bytes / +0.30%] codesize/test_minimal_runtime_code_size_random_printf_wasm.json: 10815 => 11057 [+242 bytes / +2.24%] codesize/test_minimal_runtime_code_size_random_printf_wasm2js.json: 17114 => 17417 [+303 bytes / +1.77%] codesize/test_unoptimized_code_size.json: 180220 => 180969 [+749 bytes / +0.42%] Average change: +0.34% (+0.01% - +2.24%) ```

kleisauke · 2026-05-20T15:00:40Z

See companion PR emscripten-core/musl#3 for a diff that includes only the musl-specific changes.

PR emscripten-core/musl#7 includes only the Emscripten-specific changes (+ changes that were not synced yet).

This change was generated by running ./system/lib/push_musl_changes.py within the Emscripten repo.

This reverts commit 3d7b8f3.

Alternative to commit 3d7b8f3.

This change was generated by running ./system/lib/push_musl_changes.py within the Emscripten repo.

kleisauke · 2026-05-20T17:40:08Z

I'm not sure why browser64_4gb.test_pthread_hello_thread is failing on CI, it seems to work locally.

sbc100 · 2026-05-20T17:42:44Z

Thats a known flake that started showing up with recent chrome releases.

This is an automatic change generated by tools/maint/rebaseline_tests.py. The following (5) test expectation files were updated by running the tests with `--rebaseline`: ``` codesize/test_codesize_libcxxabi_message_O3_standalone.json: 3727 => 3817 [+90 bytes / +2.41%] codesize/test_codesize_mem_O3_grow_standalone.json: 9675 => 9768 [+93 bytes / +0.96%] codesize/test_codesize_mem_O3_standalone.json: 9532 => 9625 [+93 bytes / +0.98%] codesize/test_codesize_mem_O3_standalone_narg.json: 8860 => 8949 [+89 bytes / +1.00%] codesize/test_codesize_mem_O3_standalone_narg_flto.json: 7796 => 7880 [+84 bytes / +1.08%] Average change: +1.29% (+0.96% - +2.41%) ```

sbc100 · 2026-05-20T20:16:45Z

+#define PTRDIFF_MIN     INT64_MIN
+#define PTRDIFF_MAX     INT64_MAX
+#define SIZE_MAX        UINT64_MAX
+#endif


I modified stdint.h to not depend on bits/stdint.h at all, so maybe we can just skip this file? (i.e. delete it).

See #26689

Great! Removed with commit ae38f6f.

FWIW, I had to make update_musl.py more precise in commit 4289e41, otherwise, include/stdint.h would have been removed as well.

In preparation for the next commit.

It's no longer needed.

This is an automatic change generated by tools/maint/rebaseline_tests.py. The following (2) test expectation files were updated by running the tests with `--rebaseline`: ``` codesize/test_codesize_hello_dylink_all.json: 855671 => 855742 [+71 bytes / +0.01%] codesize/test_minimal_runtime_code_size_audio_worklet.json: 16226 => 16399 [+173 bytes / +1.07%] Average change: +0.54% (+0.01% - +1.07%) ```

sbc100 · 2026-05-21T16:48:51Z

      {{{ makeSetValue('buf', C_STRUCTS.stat.st_mtim.tv_nsec, '(mtime % 1000) * 1000 * 1000', SIZE_TYPE) }}};
      {{{ makeSetValue('buf', C_STRUCTS.stat.st_ctim.tv_sec, 'Math.floor(ctime / 1000)', 'i64') }}};
      {{{ makeSetValue('buf', C_STRUCTS.stat.st_ctim.tv_nsec, '(ctime % 1000) * 1000 * 1000', SIZE_TYPE) }}};
-      {{{ makeSetValue('buf', C_STRUCTS.stat.st_ino, 'stat.ino', 'i64') }}};


This looks like just a re-ordering? Revert this file?

It was re-ordered to match the stat struct, see commit 7378ee6.

... though I can revert this, I need to resolve the merge conflicts anyway.

Actually, let's revert this stat struct field reordering. I forgot that changing the layout breaks Rust compatibility, see for example:
https://github.com/rust-lang/libc/blob/0.2.186/src/unix/linux_like/linux/musl/b32/x86/mod.rs#L9-L48
versus:
https://github.com/rust-lang/libc/blob/0.2.186/src/unix/linux_like/emscripten/mod.rs#L217-L240

See also rust-lang/rust#131467, where a similar change caused a significant amount of churn.

The ordering of assignments in JS doesn't not need to match the ordering in native code necessarily.

I don't really care of we keep the old ordering on the native side but it seems very bad if rust depends on this ..

Reverted with commit d7aeb05 and 29fc8dd to be on the safe side.

sbc100 · 2026-05-21T16:49:05Z

          mtime: {{{ makeGetValue('statBuf', C_STRUCTS.stat.st_mtim.tv_sec, "i53") }}},
-          ctime: {{{ makeGetValue('statBuf', C_STRUCTS.stat.st_ctim.tv_sec, "i53") }}},
-          ino: {{{ makeGetValue('statBuf', C_STRUCTS.stat.st_ino, "u53") }}}
+          ctime: {{{ makeGetValue('statBuf', C_STRUCTS.stat.st_ctim.tv_sec, "i53") }}}


sbc100

Thanks for working on this!

sbc100 · 2026-05-21T16:57:29Z

+  "a.out.nodebug.wasm": 102186,
+  "a.out.nodebug.wasm.gz": 39572,
+  "total": 120848,
+  "total_gz": 47263,


The endless creep of progress :)

This reverts commit 7378ee6.

This is an automatic change generated by tools/maint/rebaseline_tests.py. The following (1) test expectation files were updated by running the tests with `--rebaseline`: ``` codesize/test_codesize_hello_dylink_all.json: 855738 => 855809 [+71 bytes / +0.01%] Average change: +0.01% (+0.01% - +0.01%) ```

kleisauke added 14 commits May 5, 2026 11:25

musl: revert Emscripten-specific changes

8afc084

This is an automatic change generated with: ``` ./system/lib/update_musl.py ~/Downloads/musl-1.2.5 ```

musl: update arch ignore list

aabc8b2

In preparation for the next commit.

Update musl to v1.2.6

fa6d187

This is an automatic change generated with: ``` ./system/lib/update_musl.py ~/Downloads/musl-1.2.6 ```

musl: reapply Emscripten-specific changes

a67d130

This reverts commit 8afc084.

Revise Emscripten-specific changes

82cc2d3

Revise Emscripten-specific include changes

2bd5f45

Revise arch/emscripten/bits changes

85d99f0

Deduplicate arch/emscripten/bits directory

e019d33

Regenerate struct info files

b97ab49

Done via: ``` ./tools/gen_struct_info.py ./tools/gen_struct_info.py --wasm64 ```

Update README.md and ChangeLog.md

0da9cf9

Re-order to match stat struct

7378ee6

Update system_libs.py after musl v1.2.6

6d50b1c

See: https://git.musl-libc.org/cgit/musl/commit/?id=ef7d0ae21240eac9fc1e8088112bfb0fac507578

Fix other.test_standalone_whole_archive failure

3d7b8f3

Details: ``` wasm-ld: error: --shared-memory is disallowed by exit.o because it was not compiled with 'atomics' or 'bulk-memory' features. ```

Merge branch 'main' into musl-1.2.6

1ef60da

kleisauke mentioned this pull request May 6, 2026

Update to v1.2.6 emscripten-core/musl#3

Merged

sbc100 reviewed May 6, 2026

View reviewed changes

kleisauke added 2 commits May 6, 2026 16:46

Revert "Revise Emscripten-specific include changes"

1f05268

This reverts commit 2bd5f45.

Revert "Revise Emscripten-specific changes"

a499fc2

This partially reverts commit 82cc2d3.

sbc100 reviewed May 6, 2026

View reviewed changes

Comment thread system/lib/libc/musl/src/internal/syscall.h Outdated

kleisauke added a commit to emscripten-core/musl that referenced this pull request May 6, 2026

Update to emscripten-core/emscripten#26860

3714aa1

This change is the result of running ./system/lib/push_musl_changes.py

Remove generic <bits/stdarg.h> header

9ac8462

In-line with: https://git.musl-libc.org/cgit/musl/commit?id=ffaaa6d230512f3a7f3d040b943517728f3dc3cf

kleisauke added a commit to emscripten-core/musl that referenced this pull request May 6, 2026

Update to emscripten-core/emscripten#26860

b427eeb

This change was generated by running ./system/lib/push_musl_changes.py.

kleisauke added a commit to emscripten-core/musl that referenced this pull request May 6, 2026

Update to emscripten-core/emscripten#26860

7147854

This change was generated by running ./system/lib/push_musl_changes.py within the Emscripten repo.

kleisauke added 2 commits May 20, 2026 16:44

Merge branch 'main' into musl-1.2.6

48c2c80

Remove outdated comment

a62785a

kleisauke added a commit to emscripten-core/musl that referenced this pull request May 20, 2026

Update to emscripten-core/emscripten#26860

a132375

This change was generated by running ./system/lib/push_musl_changes.py within the Emscripten repo.

kleisauke added 2 commits May 20, 2026 19:22

Revert "Fix other.test_standalone_whole_archive failure"

abfd52c

This reverts commit 3d7b8f3.

Fix other.test_standalone_whole_archive failure

0f0d7c0

Alternative to commit 3d7b8f3.

kleisauke added a commit to emscripten-core/musl that referenced this pull request May 20, 2026

Update to emscripten-core/emscripten#26860

0403aee

This change was generated by running ./system/lib/push_musl_changes.py within the Emscripten repo.

sbc100 reviewed May 20, 2026

View reviewed changes

kleisauke added 4 commits May 21, 2026 10:47

Merge branch 'main' into musl-1.2.6

4d7f929

Make update_musl.py more precise. NFC

4289e41

In preparation for the next commit.

Remove arch/generic/bits/stdint.h

ae38f6f

It's no longer needed.

kleisauke force-pushed the musl-1.2.6 branch from 8cf09a4 to 518eaad Compare May 21, 2026 11:01

sbc100 reviewed May 21, 2026

View reviewed changes

sbc100 approved these changes May 21, 2026

View reviewed changes

kleisauke mentioned this pull request May 21, 2026

Update to Emscripten 5.0.8 emscripten-core/musl#7

Open

kleisauke added 4 commits May 21, 2026 19:02

Revert "Re-order to match stat struct"

d7aeb05

This reverts commit 7378ee6.

Merge branch 'main' into musl-1.2.6

b93b9e1

Revert stat struct field reordering

29fc8dd

sbc100 approved these changes May 21, 2026

View reviewed changes

sbc100 merged commit 4bf5fb7 into emscripten-core:main May 21, 2026
30 checks passed

Conversation

kleisauke commented May 5, 2026

Uh oh!

sbc100 commented May 5, 2026

Uh oh!

kleisauke commented May 5, 2026

Uh oh!

sbc100 commented May 5, 2026

Uh oh!

kleisauke commented May 5, 2026

Uh oh!

kleisauke commented May 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

sbc100 commented May 6, 2026

Uh oh!

kleisauke commented May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kleisauke commented May 20, 2026

Uh oh!

sbc100 commented May 20, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kleisauke May 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sbc100 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

kleisauke commented May 6, 2026 •

edited

Loading

kleisauke commented May 20, 2026 •

edited

Loading

kleisauke May 21, 2026 •

edited

Loading