forked from OSchip/llvm-project
a1ffe2fa49
The DataSecEentries is defined as an unordered_map since
order does not really matter.
std::unordered_map<std::string, std::unique_ptr<BTFKindDataSec>>
DataSecEntries;
This seems causing the test static-var-derived-type.ll flaky
as two sections ".bss" and ".readonly" have undeterministic
ordering when performing map iterating, which decides the
output assembly code sequence of BTF_KIND_DATASEC entries.
Fix the test to have only one data section to remove
flakiness.
Signed-off-by: Yonghong Song <yhs@fb.com>
llvm-svn: 356731
|
||
|---|---|---|
| .. | ||
| BTF | ||
| 32-bit-subreg-alu.ll | ||
| 32-bit-subreg-cond-select.ll | ||
| 32-bit-subreg-load-store.ll | ||
| 32-bit-subreg-peephole.ll | ||
| alu8.ll | ||
| atomics.ll | ||
| basictest.ll | ||
| byval.ll | ||
| cc_args.ll | ||
| cc_args_be.ll | ||
| cc_ret.ll | ||
| cmp.ll | ||
| dwarfdump.ll | ||
| elf-symbol-information.ll | ||
| ex1.ll | ||
| fi_ri.ll | ||
| inline_asm.ll | ||
| intrinsics.ll | ||
| lit.local.cfg | ||
| load.ll | ||
| loops.ll | ||
| many_args1.ll | ||
| many_args2.ll | ||
| mem_offset.ll | ||
| mem_offset_be.ll | ||
| memcpy-expand-in-order.ll | ||
| objdump_atomics.ll | ||
| objdump_cond_op.ll | ||
| objdump_cond_op_2.ll | ||
| objdump_imm_hex.ll | ||
| objdump_intrinsics.ll | ||
| objdump_static_var.ll | ||
| objdump_trivial.ll | ||
| reloc-btf.ll | ||
| reloc.ll | ||
| remove_truncate_1.ll | ||
| remove_truncate_2.ll | ||
| remove_truncate_3.ll | ||
| remove_truncate_4.ll | ||
| remove_truncate_5.ll | ||
| rodata_1.ll | ||
| rodata_2.ll | ||
| rodata_3.ll | ||
| rodata_4.ll | ||
| sanity.ll | ||
| sdiv_error.ll | ||
| select_ri.ll | ||
| setcc.ll | ||
| shifts.ll | ||
| sockex2.ll | ||
| struct_ret1.ll | ||
| struct_ret2.ll | ||
| undef.ll | ||
| vararg1.ll | ||
| warn-call.ll | ||
| warn-stack.ll | ||
| xadd.ll | ||
| xadd_legal.ll | ||