runtime: missing deferreturn on linux/ppc64le [1.14 backport]

[gopherbot]

@ianlancetaylor requested issue #39049 to be considered for backport to the next 1.14 minor release.

@gopherbot please open a backport to 1.14

[laboger]

I can work on the patch for this unless someone else has started.

[laboger]

Oh I think @cherrymui has to do it.

[laboger]

Is it possible that this is a bug due to the new linker? Is there still a way to build with the old linker in Go 1.14 to test it out?

[cherrymui]

@laboger Yeah, I can do the backport.

No, I don't think it is the new linker. The original issue (#39049) was reported for Go 1.14.

[gopherbot]

Change https://golang.org/cl/240917 mentions this issue: [release-branch.go1.14] cmd/link: detect trampoline of deferreturn call

[laboger]

Thanks Cherry!

[dmitshur]

@ianlancetaylor Can you please include a short rationale about why the backport might be needed? (Per MinorReleases.) Thanks.

[laboger]

RedHat has moved their build compiler to Go 1.14.4 and is seeing the problem reported in issue #39049. The fix for that issue was merged upstream but not backported to the Go 1.14 branch.

[ianlancetaylor]

@dmitshur Sorry for missing the rationale in the backport request. It is the preceding comment on the main issue: #39049 (comment)

This is a critical bug for kubernetes project, can someone help us cherry-picking this to 1.14 release?

[mkumatag]

@dmitshur Sorry for missing the rationale in the backport request. It is the preceding comment on the main issue: #39049 (comment)

This is a critical bug for kubernetes project, can someone help us cherry-picking this to 1.14 release?

yes, @ianlancetaylor is right, kubernetes community recently adopted golang 1.14 version and hit with this issue for ppc64le platform, need to be fixed asap. -Thanks.

[mkumatag]

@ianlancetaylor @dmitshur I'm wondering when this will get merged into the 1.14 branch and any tentative timeline for the 1.14.5 release so that I can plan a golang bump for Kubernetes project accordingly!

[dmitshur]

Approving as it is a serious problem without a workaround. It is a 1.14-only regression, so a fix for 1.13 isn't needed.