pq/libfreemkv
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: correct Profile-07 comment — prime_cache(lba) reads lba-3..lba-1, not lba itself

Browse Source
This commit is contained in:
Matthew Jackson
2026-06-24 01:51:21 -07:00
parent ee0c7cebe3
commit dbc2225315
1 changed files with 9 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/passn_handler_ab.rs
+9 -8
View File
@@ -688,14 +688,15 @@ fn profile_07_medium_then_good() {
trace, trace,
); );
// GOLDEN: patch's cache-priming (`prime_cache`) issues 3 // GOLDEN: patch's cache-priming (`prime_cache`) issues 3 throwaway
// throwaway single-sector reads at lba-3..lba before each count==1 // single-sector reads at lba-3, lba-2, lba-1 (NOT lba itself) before
// recovery read. Those throwaway reads ADVANCE the per-LBA script // each count==1 recovery read. In the default REVERSE patch pass,
// step counter even though their results are discarded. So a // lba 108's prime reads lba 105 (consuming step 0 = fail) and lba 107's
// 3-step script (fail, fail, ok) gets consumed by 2 prime calls // prime reads lba 105 again (consuming step 1 = fail), so when the real
// plus 1 real read → the real read sees `Ok` and the sector is // recovery read for lba 105 occurs its script step is 2 (= ok).
// recovered. Net effect: patch fully recovers the range in one // prime_cache(105) itself reads 102, 103, 104 — it does NOT advance
// pass thanks to priming, even though the script said "fails on // lba 105's own counter. Net effect: patch fully recovers the range in
// one pass thanks to priming, even though the script said "fails on
// first two attempts." // first two attempts."
// //
// This is the documented cache-prime behavior (`disc/patch.rs` // This is the documented cache-prime behavior (`disc/patch.rs`