Context
The 2026-05-07-evening audit shows `Expected '(' after 'asm'` is now the #1 PARSE_FAIL pattern at 10 files. After the recent parser-cluster work (#42 :: support, etc.), the next downstream blocker for the arm64/rpi/power kernel cluster is inline-asm parsing.
Affected files (10)
All under kernel/src/ in home-os:
arch/arm64/rpi/boot.homearch/arm64/rpi/devicetree.homearch/arm64/rpi/emmc.homearch/arm64/rpi/gic400.homearch/arm64/rpi/gpio.homearch/arm64/rpi/uart.homepower/battery.homepower/cpufreq_x86.homepower/thermal_x86.homedebug/bug_tracker.home
Reproduction
Need to inspect actual asm shapes in the affected files first. Likely candidates:
- Zig-style:
asm volatile ("..." : "..." (out) : "..." (in) : "memory") - Rust-style:
asm!("...", in("reg") val, out("reg") result) - Bare string:
asm "mov x0, x1"
The parser previously had @asm(...) support added in #40 (treated as opaque builtin). The asm-as-statement form in these files likely uses different syntax.
Approach
- Inspect 3-5 of the affected files to identify the actual
asmshape used - Decide: extend the parser to accept that shape, OR have the kernel codemod to a Home-supported form (e.g. wrap in
@asm(...)builtin per #40) - If parser fix: add asm-block grammar (probably as a statement, not an expression)
- If codemod: file as home-os codemod issue and close this one
Why this matters
10 modules unblocked. Predicted +2.5pp OK, pushing kernel past 80% threshold.
Reference
- 2026-05-07-evening audit: `~/Desktop/home-os-kernel-audit-2026-05-07-evening.tsv`
- Related: home #40 added
@asm(...)builtin form