You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
cmd/compile: inline marker targets not reachable after assembly on arm #40796
The linux-arm buildbot is failing the test added in CL 247697. This is the same fundamental issue as seen in #40689 (ppc64x) and #40473 (s390x), and should be a similar fix.
The arm assembler is removing a Prog from the function while processing it. That same Prog is then being used to generate inlining data after assembly. Since the Prog was never assembled its PC is incorrect and so the generated inlining data is broken. This can cause problems with tracebacks, such as the infinite loop seen in #40473.
This bug can cause tracebacks to hang. There is no workaround. The fix for this issue will need to be backported before the test added in CL 247697 can be backported.
Activity
mundaym commentedon Aug 14, 2020
/cc @randall77
mundaym commentedon Aug 14, 2020
@gopherbot Please open backport issues.
This bug can cause tracebacks to hang. There is no workaround. The fix for this issue will need to be backported before the test added in CL 247697 can be backported.
gopherbot commentedon Aug 14, 2020
Backport issue(s) opened: #40797 (for 1.14), #40798 (for 1.15).
Remember to create the cherry-pick CL(s) as soon as the patch is submitted to master, according to https://golang.org/wiki/MinorReleases.
gopherbot commentedon Aug 15, 2020
Change https://golang.org/cl/248684 mentions this issue:
cmd/internal/obj: stop removing NOPs from instruction streamcmd/internal/obj: stop removing NOPs from instruction stream