diff options
| author | Jakob Unterwurzacher | 2026-11-11 21:22:00 +0100 |
|---|---|---|
| committer | Jakob Unterwurzacher | 2026-11-11 21:22:00 +0100 |
| commit | 851188164274de96cd581c3da1b5e847a42e2948 (patch) | |
| tree | c1d33d2b5086e9f5bfad8abddb5c6261068ff029 /internal | |
| parent | 0363a138f8125604d75d9ca84be4092482b622df (diff) | |
fusefrontend: -sharedstorage: always check for file holes on write
Write() skips writePadHole() when a write directly follows the previous
one (isConsecutiveWrite), because such a write cannot create a hole.
That decision is based on process-local state. With -sharedstorage,
another mount may have shrunk the file in between, so the old partial
last block was never padded and ended up as a short block in the middle
of the file, making it unreadable (EIO).
Disable the optimization with -sharedstorage.
Add TestConsecutiveWriteAfterRemoteTruncate as a regression test.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Diffstat (limited to 'internal')
| -rw-r--r-- | internal/fusefrontend/file.go | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/internal/fusefrontend/file.go b/internal/fusefrontend/file.go index fb97b9a..feb40b5 100644 --- a/internal/fusefrontend/file.go +++ b/internal/fusefrontend/file.go @@ -431,7 +431,9 @@ func (f *File) Write(ctx context.Context, data []byte, off int64) (uint32, sysca // If the write creates a file hole, we have to zero-pad the last block. // But if the write directly follows an earlier write, it cannot create a // hole, and we can save one Stat() call. - if !f.isConsecutiveWrite(off) { + // With -sharedstorage, another mount may have truncated the file since our + // last write, so we have to check every time. + if f.rootNode.args.SharedStorage || !f.isConsecutiveWrite(off) { errno := f.writePadHole(off) if errno != 0 { return 0, errno |
