Details
Details
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Comment Actions
WIll take a look in more detail next week. IIUC so far, we have a quirk in isUndefWeak because we don't, currently, demote lazy symbols that weren't resolved, due to a problem that we now think is resolved. At a glance it sounds reasonable to me, just need to understand what the original problem was and how D79300 fixes it.
Comment Actions
LGTM. Following this through I can't think of a test case that we can write as if all is well it should error out before we get to the problem case.
lld/ELF/SyntheticSections.cpp | ||
---|---|---|
3165 | IIUC we could write assert(!s.sym->isLazy()); here. Not sure whether it would be worth doing it in the long term as it will look somewhat arbitrary. Maybe worth a test run with it in just to see if it hits anything unexpected. |
IIUC we could write assert(!s.sym->isLazy()); here. Not sure whether it would be worth doing it in the long term as it will look somewhat arbitrary. Maybe worth a test run with it in just to see if it hits anything unexpected.