https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119737
--- Comment #4 from Thomas Schwinge <tschwinge at gcc dot gnu.org> ---
Andrew P., thanks for your analysis -- that indeed looks like a plausible
explanation.
(In reply to myself from comment #0)
> I've not yet looked into creating a non-offloading, GCN target test case.
No need to create one; the problem is already visible for (only) one test case
for GCN target testing:
FAIL: g++.dg/init/call1.C -std=gnu++17 (internal compiler error: Aborted
signal terminated program as)
FAIL: g++.dg/init/call1.C -std=gnu++17 (test for excess errors)
UNRESOLVED: g++.dg/init/call1.C -std=gnu++17 compilation failed to produce
executable
UNSUPPORTED: g++.dg/init/call1.C -std=gnu++98: exception handling not
supported
FAIL: g++.dg/init/call1.C -std=gnu++26 (internal compiler error: Aborted
signal terminated program as)
FAIL: g++.dg/init/call1.C -std=gnu++26 (test for excess errors)
UNRESOLVED: g++.dg/init/call1.C -std=gnu++26 compilation failed to produce
executable
..., which likewise FAILs due to 'LLVM ERROR: Size expression must be
absolute.', where we've got:
[...]
.text
.align 256
.type main,@function
main:
[...]
s_endpgm
.L15:
s_getpc_b64 s[12:13]
s_add_u32 s12, s12, abort@rel32@lo+4
s_addc_u32 s13, s13, abort@rel32@hi+4
s_swappc_b64 s[18:19], s[12:13]
.LFE17:
.section .fake_gcc_except_table
[...]
.section .text.startup
.size main, .-main
[...]
So, here it's '.text' vs. '.text.startup'.
