Diff Detail
Diff Detail
Event Timeline
Comment Actions
Yes, but I'm not entirely sure that's a bad thing. We're certainly missing test coverage for this.
If we have a null pointer in addrspace(0), cast to a non-0 address space pointer and accessed, can we assume it would trap? My interpretation of the LangRef is ptr addrspace(0) null cannot be addrspacecasted to produce a dereferenceable result so this would be OK (so long as it's not a volatile access)
Comment Actions
Should restrict this to just addrspacecast. Plus there's some other bitcast handling code to trim
Comment Actions
Apparently I got confused and closed the wrong revision, this is really f3059e223dc137ee4111f6e3bb45e7613d6ff7d5