Skip to content

fix(write): report a stuck bitmap lock as EIO, not ENOSPC - #104

Merged
mattmattox merged 1 commit into
mainfrom
fix/alloc-lock-eio
Nov 13, 2026
Merged

mattmattox merged 1 commit into
mainfrom
fix/alloc-lock-eio

Conversation

@mattmattox

Copy link
Copy Markdown
Contributor

Follow-up to #66, which fixed only inode claims. Block allocation in write and fallocate still reported every bitmap or extent-tree allocation failure as ENOSPC, including a bitmap lock wait that gave up after 30s. On main, TestAllocationReportsStuckBitmapLockAsEIO gets "no space left on device" from both a write and a fallocate while another node holds the bitmap lock. ENOSPC now means only an empty bitmap or a backing device that is itself full.

Create, mkdir and copy-on-write already returned EIO for every allocation failure, so they never claimed a full device. When the device really is full, they still say EIO, not ENOSPC; that is not changed here.

🤖 Generated with Claude Code

Block allocation in write and fallocate mapped every bitmap or
extent-tree allocation error to ENOSPC, including a lock wait that gave
up and a bitmap sector that could not be read. Only an empty bitmap or
full backing storage is ENOSPC now; everything else is EIO.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@mattmattox
mattmattox merged commit 26e8517 into main Nov 13, 2026
13 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant