Skip to content

Conversation

@raharper
Copy link
Contributor

@raharper raharper commented Dec 8, 2025

Newer binutils 2.45.50+ "fixes" use of --target which cli documents as setting both input and output targets. However, up until versions in 25.10 (Ubuntu Questing) this flag didn't fail like it does now where it errors out with:

objcopy: file format not recognized.

Some details in LP:#2134326 now closed as invalid after testing out use of --output-target.

@raharper raharper added the bug Something isn't working label Dec 8, 2025
Newer binutils 2.45.50+ "fixes" use of --target which cli documents
as setting both input and output targets.  However, up until versions
in 25.10 (Ubuntu Questing) this flag didn't fail like it does now
where it errors out with:

objcopy: file format not recognized.

Some details in LP:#2134326, now closed as invalid after testing out
use of --output-target.

Signed-off-by: Ryan Harper <rharper@woxford.com>
Signed-off-by: Ryan Harper <rharper@woxford.com>
@raharper raharper force-pushed the fix/objcopy-use-output-target branch from b9515a9 to 3761bb9 Compare December 8, 2025 23:41
The older focal ovmf build is not installable via apt any longer.
Further jammy and newer releases all have ovmf which has uefi-shell
disabled in secureboot which makes those releases not useful for
stubby testing.

I have found a needle in the haystack of ovmf builds somewhere in
the jammy release where we have 2M and 4M builds and secureboot is
not disabled.

I'm not sure if this build has the snakeoil key bug or not but
I'm hoping this is sufficient to get some testing that stubby itself
compiles and boots in some cases to confirm that changing the objcopy
flag creates a functioning stubby.

Signed-off-by: Ryan Harper <rharper@woxford.com>
@raharper
Copy link
Contributor Author

raharper commented Dec 9, 2025

I'm not terribly interested in debugging on the tip of stubby and the test-harness; so the results would need some time to understand and probably need to manually verify each of those failure cases. None the less, I do think it's worth merging for:

  1. the makefile fix
  2. we can now build and run on jammy or noble since I've found a working ovmf package

Alternatives to (2) would be to build our own ovmf package via edk2 builds ; but I don't want to do that either.

@hallyn
Copy link
Contributor

hallyn commented Dec 10, 2025

I'm not terribly interested in debugging on the tip of stubby and the test-harness; so the results would need some time to understand and probably need to manually verify each of those failure cases. None the less, I do think it's worth merging for:

1. the makefile fix

2. we can now build and run on jammy or noble since I've found a working ovmf package

Alternatives to (2) would be to build our own ovmf package via edk2 builds ; but I don't want to do that either.

If you're saying you think we should merge this as is, please go ahead, or I'll do it.
But as I'm not certain that's what you're saying, I'll wait until you reply :

@raharper
Copy link
Contributor Author

I'm not terribly interested in debugging on the tip of stubby and the test-harness; so the results would need some time to understand and probably need to manually verify each of those failure cases. None the less, I do think it's worth merging for:

1. the makefile fix

2. we can now build and run on jammy or noble since I've found a working ovmf package

Alternatives to (2) would be to build our own ovmf package via edk2 builds ; but I don't want to do that either.

If you're saying you think we should merge this as is, please go ahead, or I'll do it. But as I'm not certain that's what you're saying, I'll wait until you reply :

I want to merge now; Thanks!

@raharper raharper merged commit 30070a3 into puzzleos:main Dec 10, 2025
1 check failed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bug Something isn't working

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants