update CHANGES

This commit is contained in:
Thomas Waldmann 2026-05-15 21:32:24 +02:00
parent ec366d5b5f
commit 074c0c0e18
No known key found for this signature in database
GPG key ID: 243ACFA951F78E01

View file

@ -423,13 +423,22 @@ above.
New features:
- create/import-tar --quick-stats: faster than --stats by omitting "All archives"
and repository chunks statistics, #9579
and repository chunk statistics, #9579
- prune: show total vs matching archives in output, #9262
- Minimal implementation of "related repositories", #9645
This feature allows multiple repositories to share deduplication-relevant secrets (id_key and chunk_seed) while maintaining secure, independent encryption keys.
- borg key export-related-secrets <REPO> <SPATH>
- borg init --import-related-secrets <SPATH> <REPO>
Fixes:
- hashindex: fix new checks for big endian archs, #9521
- slashdot hack: fix exclusion of source directory metadata, #9534
Note: Many of the fixed issues below here relate to rather rare or theoretical
Note: Many of the fixed issues listed below relate to rather rare or theoretical
issues and were found by automated code checking.
- LRUCache: resolve KeyError and memory leaks, #9587
@ -439,7 +448,7 @@ Note: Many of the fixed issues below here relate to rather rare or theoretical
or if the archived file had all-zero replacement chunks or inconsistent size.
- Chunker fixes, #9586:
- Better check the return value of fd.read(n) and reject if it returns more
- Strictly check the return value of fd.read(n) and reject if it returns more
bytes than requested.
- Avoid giving len <= 0 to posix_fadvise(), which could drop the rest of the
file from the cache.
@ -448,7 +457,7 @@ Note: Many of the fixed issues below here relate to rather rare or theoretical
- Check for malloc/calloc failures.
- Hashindex fixes, #9575:
- Make it possible to lookup in compacted hashtables.
- Make it possible to look up in compacted hashtables.
- Avoid buckets_length integer overflow on 32-bit systems via huge num_buckets.
- Deal safely with empty index: we must use num_buckets = 1 to avoid division
by zero and sanity check in hashindex_read.
@ -456,7 +465,7 @@ Note: Many of the fixed issues below here relate to rather rare or theoretical
- Reinitialize upper/lower limit and min_empty after compact.
- Fix size_idx / fit_size / grow_size / shrink_size (mind array bounds).
- Deal with growing when already at max capacity.
- hashindex_resize: replace num_entries assertion, rather return error.
- hashindex_resize: replace num_entries assertion, return an error instead.
- Correctly free memory when header validation fails.
- BaseIndex.clear: always stay in valid state.
Do not free the old index before we successfully have allocated a new one.
@ -465,6 +474,20 @@ Other changes:
- use F_FULLFSYNC on macOS for SyncFile data durability, #9383
- mount: improve error msg when uid/gid cannot be resolved, #9574
- docs:
- impact of the slashdot hack on pattern matching, #9647
- pull-backup.rst minor fixes
- sshfs + chroot does not support different CPU architectures, #6878
- document max_segment_size adjustment, #7858
- add section about rolling back a transaction, #9270
- clarify storage quota run-time settings, #3948
- add FAQ entry about scalability, #4742
- add FAQ entry for full repository filesystem, #9573
- add FAQ entry for bad backups and deduplication, #4744
- add FAQ entry for SSH connection timeouts, #5629
- improve macOS Keychain instructions, #5156
- add DoS warning for none encryption mode, #6715
- document error handling in borg create, #4912
Version 1.4.4 (2026-03-19)