Re: [PATCH] RISC-V: Fix fmv.x testcase for non-default --with-arch
Project / Subsystem
binutils / risc-v
Date
2026-09-23
Proposer
Kito Cheng <kito.cheng@gmail.com>
Source type
public_inbox
Consensus
Proposed
Sentiment
—/10
Technical tradeoffs
- • Hardcodes a specific architecture flag in a testcase, reducing its flexibility but increasing its reliability.
- • Ensures consistent test results across different build configurations.
All attributes
- project
- binutils
- subsystem
- risc-v
- patch_id
- —
- discussion_id
- CA+yXCZBg+HbM+TODOVsYfuV0_o8g841p38ePt39NLVowm0AbZA@mail.gmail.com
- source_type
- public_inbox
- title
- Re: [PATCH] RISC-V: Fix fmv.x testcase for non-default --with-arch
- headline
- RISC-V: Fix fmv.x testcase for non-default --with-arch
- tldr
- RISC-V testcase for fmv.x instruction fixed to explicitly use -march=rv32i.
- proposer
- Kito Cheng <kito.cheng@gmail.com>
- consensus
- Proposed
- outcome
- committed
- sentiment_score
- —
- technical_tradeoffs
-
- • Hardcodes a specific architecture flag in a testcase, reducing its flexibility but increasing its reliability.
- • Ensures consistent test results across different build configurations.
- series_id
- —
- series_role
- standalone
- series_parts
- []
- tags
-
- • risc-v
- • gas
- • testsuite
- • regression
- bugzilla_url
- —
- date
- 2026-09-23T00:00:00.000Z
Re: [PATCH] RISC-V: Fix fmv.x testcase for non-default --with-arch
A RISC-V testcase for the fmv.x instruction has been fixed by explicitly setting the -march=rv32i flag. Previously, the testcase did not specify -march, causing it to inherit the configured default architecture. This led to failures when using --with-arch configurations that lacked the necessary extensions, such as rv32e or rv32i_zfinx. The fix ensures the testcase correctly checks the aliasing of fmv.x.s/fmv.s.x to fmv.x.w/fmv.w.x regardless of the base ISA.