Re: [RISCV] redundant return-only BB not eliminated
Project / Subsystem
gcc / gcc
Date
2026-09-08
Proposer
Robert Guthrie <forkbombidable@gmail.com>
Source type
public_inbox
Consensus
Proposed
Sentiment
—/10
Technical tradeoffs
- • Reliance on `prev_bb`/`next_bb` vs. predecessor/successor lists for CFG manipulation.
All attributes
- project
- gcc
- subsystem
- gcc
- patch_id
- —
- discussion_id
- ap9pjXUOhYGVixBO@bertguix
- source_type
- public_inbox
- title
- Re: [RISCV] redundant return-only BB not eliminated
- headline
- Redundant return-only BB not eliminated
- tldr
- A patch to optimize CFG cleanup by removing redundant return-only basic blocks is discussed, with suggestions on implementation details.
- proposer
- Robert Guthrie <forkbombidable@gmail.com>
- consensus
- Proposed
- outcome
- proposed
- sentiment_score
- —
- technical_tradeoffs
-
- • Reliance on `prev_bb`/`next_bb` vs. predecessor/successor lists for CFG manipulation.
- series_id
- —
- series_role
- standalone
- series_parts
- []
- tags
-
- • optimization
- • cfg
- bugzilla_url
- —
- date
- 2026-09-08T00:00:00.000Z
Re: [RISCV] redundant return-only BB not eliminated
Robert Guthrie proposes a patch to eliminate redundant consecutive return-only basic blocks in the CFG cleanup pass. He initially struggled with debugging but found the solution straightforward after resolving misunderstandings. Another participant (implied by the quote) suggests a different approach, focusing on predecessor and successor lists rather than prev_bb/next_bb for block ordering.