nasm: update to 3.02.

[Rutpiv]

Testing the changes

• I tested the changes in this PR: briefly

Local build testing

• I built this PR locally for my native architecture (x86_64)
• I built this PR locally for these architectures using specific masterdirs:
  • x86_64-musl
  • i686
• I built this PR locally for these architectures (crossbuilds):
  • aarch64
  • aarch64-musl
  • armv7l
  • armv7l-musl
  • armv6l
  • armv6l-musl

[classabbyamp]

isn't this already in the google fonts package?

[Rutpiv]

You're right, I didn’t realize it was already included in the google-fonts-ttf package.
The only issue is that adding this package as a dependency significantly increases the build time before compilation starts.

How should I proceed in this case? Should I still use google-fonts-ttf as a dependency, or would it be acceptable to include a smaller subset of fonts directly in this package to keep the build faster?

In parallel, I’m currently reviewing the dependent builds for nasm on x86, which should give a good overview of what might have broken.
Below is the list of packages I’ve tested so far; checked ones are building successfully, while the others are still pending or failing:

• CPU-X
• dav1d
• dgen-sdl
• e3
• edk2-ovmf (broke after update)
• ffmpeg
• ffmpeg6
• firefox
• firefox-esr
• handbrake
• j
• kodi
• lame
• libass
• libretro-mupen64plus (broken before update)
• libx86emu (broken before update)
• limine (10.2.0)
• lrzip
• mupen64plus
• openh264
• pcsxr (broken before update)
• rav1e
• SDL2
• SDL3
• scummvm
• silc-toolkit
• srb2 (built on i686 masterdir)
• stepmania (broken before update)
• svt-av1
• syslinux
• thunderbird
• vapoursynth
• vmaf
• x264
• x265
• xpar
• xvidcore
• zsnes (built on i686 masterdir)

I’ll continue checking the remaining ones and update the results as I go.

[Rutpiv]

As an additional note, both patches 00-musl-bool-fix.patch and 01-musl-c23-test-fix.patch included in this PR have already been merged upstream into NASM’s master branch, but after the current 3.01 release.

They are still required at this point to ensure successful compilation on musl systems.

[sgn]

Merge font roboto-mono into roboto. Puts those fonts into /usr/share/fonts/roboto

[Rutpiv]

I finished merging the fonts into a single template and placed both font families under /usr/share/fonts/roboto.
For the Roboto Mono variant, I kept both the regular and wght versions to maintain compatibility.

Most dependent packages build successfully with the updated NASM, but a few issues were identified:

• edk2-ovmf: Broke after the update — previously built fine, now fails with the new NASM version.
• libretro-mupen64plus, libx86emu, pcsxr, and stepmania: Already broken before the update; unrelated to this change.

I’m currently looking into the edk2-ovmf issue to identify what caused the breakage after the update.
I found the related issue tianocore/edk2#11635, which includes a patch, but it’s currently failing to apply during the build.

How should I proceed with the packages that were already broken before this change?

[virtfunc]

I’m currently looking into the edk2-ovmf issue to identify what caused the breakage after the update. I found the related issue tianocore/edk2#11635, which includes a patch, but it’s currently failing to apply during the build.

I think its just a whitespace issue if you are copy + pasting that patch, I was able to reproduce failing to apply when using git apply without --whitespace=fix. Added a direct download link in original post (https://github.com/user-attachments/files/23197854/fix_nasm_compile.patch)

[Rutpiv]

You were right, the patch from your link applied perfectly and fixed the build. Thanks!

[Rutpiv]

Before moving this PR to Ready to Review, I made the following adjustments:

• Removed the libx86emu test suite, which was causing build issues.
• Fixed edk2-ovmf using the patch provided by VirtFunc.
• Marked the remaining problematic packages as broken, as their failures are unrelated to the NASM update.

All other dependent packages build correctly, and the update appears stable.

[Rutpiv]

Marking this PR as draft for now.
The previous edk2-ovmf release built successfully with a patch applied, but the newer version no longer builds (with nasm 3.01) on my side.
I’d prefer to keep this as draft until the regression in edk2-ovmf is properly understood and addressed.

[Rutpiv]

This PR has been updated with the following changes:

• fonts-roboto-ttf

  • Updated roboto-3-classic to 3.015.
  • Switched to the static TTF variant, matching NASM’s default font usage.

• nasm

  • Updated to 3.01.
  • Simplified the template by removing the custom doc option; documentation now follows the standard build flow.
  • Disabled tests. The previous checks (enabled by me) regenerated golden files and compared them against locally generated outputs, which does not provide meaningful validation in the packaging context.
  • Removed the vsed adjustment, no longer needed after the Roboto update.

• edk2-ovmf

  • Added upstream patch fixing the NASM 3.x compatibility issue.
  • A new upstream release including the fix is already available; the patch can be dropped when updating.

• libx86emu

  • Disabled tests. The suite relies on strict golden output comparison. A few tests differ due to minor bit-level changes (no functional impact) and updated instruction behavior producing additional padding. Updating the golden references would require further validation.

• stepmania, pcsxr, libretro-mupen64plus

  • Marked as broken.
  • Builds were already failing prior to the NASM update.

Marking as ready for review.

[Rutpiv]

• fonts-roboto-ttf
  • Master already has 3.016, so my commit is now a pure fix (license Apache-2.0 → OFL-1.1, static font path). NASM's documentation build requires the static variants.
• nasm
  • Updated to 3.02, which also resolves the performance regression reported in Very serious performance regression on huge files compared to the 2.X series netwide-assembler/nasm#253.
  • Backported the fix for CVE-2026-6067 (no upstream release with the fix yet).
  • Dropped liberation-fonts-ttf from hostmakedepends; the Roboto static fonts cover the documentation build.
• libx86emu
  • Replaced the previous disable tests commit with a proper fix. The Error 127 failures were a parallel make race (now handled with disable_parallel_check=yes), and the single remaining failure with nasm 3.x (0003_lds_real) is a prefix byte order swap in the golden file, with identical instruction semantics. Fixed with a one-line patch, so the full test suite stays active.
• stepmania, pcsxr, libretro-mupen64plus
  • Dropped the broken= commits from this PR. Their failures predate the NASM update and are unrelated to it, so the fixes landed in separate PRs to keep this one scoped to the nasm bump: stepmania: fix build with GCC 14 and binutils >= 2.41 #61436 (stepmania), pcsxr: fix build with GCC 14 and PIE crash #61437 (pcsxr), New package: libretro-mupen64plus-nx-20260520 #61438 (libretro-mupen64plus).

Marking as ready for review.

[Rutpiv]

Follow-up on edk2-ovmf: the version currently on master already includes the upstream fix for nasm 3.x, so the patch from this PR is no longer needed and the commit was dropped. Verified locally that it builds fine with nasm 3.02 (x86_64 glibc).