[clang][AIX] Fix Overly Strict LTO Option Checking against `data-sections` when `mxcoff-roptr` is in Effect

[qiongsiwu1]

Summary

The LTO -mxcoff-roptr check against data sections is overly strict and it ignores the fact that data sections is on by default on AIX, causing valid LTO compilation to fail when -fdata-sections is not explicitly specified.

This patch revises the check so that an error is reported only if data sections is explicitly turned off for LTO.

[qiongsiwu1]

Gentle ping for review. Thanks so much!

[MaskRay]

clang/lib/Driver/ToolChains/CommonArgs.cpp

749–750

Consider the hasFlagNoClaim variant. OPT_fdata_sections is mainly claimed elsewhere.

[MaskRay]

The LTO mxcoff-roptr check against data

-mxcoff-roptr. Consider not dropping the leading -.

[qiongsiwu1]

Address code review. Thanks @MaskRay for the feedback!

[qiongsiwu1]

Ping for review. Thank you!

[qiongsiwu1]

Revising the logic to pass -data-sections= to libLTO. We now pass -data-sections=1 to libLTO by default unless it is explicitly turned off.

Old behaviour:

1. Pass -data-sections=1 if the target requires uses of separate sections, or the user explicitly sets -fdata-sections.
2. Pass -data-sections=0 if the user explicitly specifies -fno-data-sections.
3. Pass nothing otherwise.

New behaviour:

1. Pass -data-section=1 if nothing is specified, or if the target requires uses of separate sections, or the user explicitly sets -fdata-sections. (Combining cases 1 and 3 above.)
2. Pass -data-sections=0 if the user explicitly specifies -fno-data-sections.

I think the new behaviour is safe because libLTO defaults to use data sections when nothing is specified https://github.com/llvm/llvm-project/blob/829ed96b779c10cb023d7e3dbcfad22413979ec4/llvm/lib/LTO/LTOCodeGenerator.cpp#L426.

@hubert.reinterpretcast @MaskRay could you let me know what you think? Thanks so much!

[qiongsiwu1]

Revert to the DataSections logic. It is better not to change that in this patch.

[hubert.reinterpretcast]

LGTM with minor comment.

clang/lib/Driver/ToolChains/CommonArgs.cpp

730–734

Coding standards suggest that braces be used consistently for each part of an if/else chain.

[qiongsiwu1]

Fixing inconsistent braces. Thanks for the feedback @hubert.reinterpretcast !