From: "Mathieu Dubois-Briand" <mathieu.dubois-briand@bootlin.com>
To: "Jaipaul Cheernam" <jaipaul.cheernam@est.tech>,
<openembedded-core@lists.openembedded.org>
Subject: Re: [OE-core] [PATCH 2/4] rsync: upgrade 3.5.1 -> 3.5.2
Date: Wed, 04 Nov 2026 14:52:33 +0100 [thread overview]
Message-ID: <DMMH5MBL0Y9R.19MJFCNHD82Q7@bootlin.com> (raw)
In-Reply-To: <20261104113822.45520-2-jaipaul.cheernam@est.tech>
On Wed Nov 4, 2026 at 12:38 PM CET, Jaipaul Cheernam wrote:
> NEWS: https://download.samba.org/pub/rsync/NEWS#3.5.2
>
> - Fixed interrupted `--append` and `--inplace` transfers writing the final
> 256 KiB buffer twice. An interrupted destination now remains a clean prefix
> which the next transfer can resume or repair.
> - Fixed uploads through `rrsync` when the destination file already exists.
> Existing files are now reached beneath a pinned parent directory instead of
> being passed as an unwritable `/proc/self/fd/N` destination.
> - Fixed poor matching performance on large sparse and repeated-block files
> when duplicate checksum chains exhausted the search limit. Valid basis
> blocks are no longer hidden behind duplicate entries.
> - Bounded strong-checksum work when a malicious receiver supplies weak
> checksums that collide with consecutive rolling windows
> (CVE-2026-119741).
> - Prevented the `--inplace` matcher from advancing its aligned basis index
> beyond the available checksum entries. This avoids signed overflow on large
> source files when the peer supplies a small block count.
> - Recovered from failed hard-link creation without aborting or hanging the
> generator. Remaining members fall back to a fresh hard-link group when the
> original destination inode cannot accept another link.
> - Prevented an out-of-bounds write while invalidating a sender checksum after
> a file read error when the negotiated checksum length is zero.
> - Fixed a daemon receiver stack overflow when queued deletion output reached
> `--max-alloc`. Quiet daemon transfers no longer queue deletion messages
> unless the client requested them.
> - Fixed a `--files-from` pull from a subdirectory of a daemon module, which
> read each listed file from the module root instead. It could send, and with
> `--remove-source-files` delete, a same-named file the module's filter
> excludes.
> - Fixed password files whose first line is empty. The parser no longer skips
> the empty line and uses a later password.
> - Escaped terminal control characters in filenames and diagnostics written to
> standard output, standard error and log files while preserving printable
> text in the active locale.
> - Preserved Cygwin and MSYS2 UNC roots during ownership-checked path walks.
> - Restored daemon startup from accepted named Unix sockets on standard input,
> including filesystem and Linux abstract sockets used for socket activation.
> Unnamed local socket pairs remain non-daemon channels.
> - Fixed forward-DNS hostname entries in daemon `hosts allow` and `hosts deny`
> lists for IPv6 peers. Address comparison is now binary and family-aware.
> - Preserved Unix symlink permissions on systems which require `fchmodat()` to
> change the mode of a symlink rather than its target.
> - Restored builds on systems without `O_CLOEXEC` or `O_DIRECTORY`, including
> Solaris 10.
>
> - Added `--nice` to adjust CPU and I/O scheduling priorities for local and
> remote rsync processes. Daemon clients and rrsync cannot raise server-side
> priorities.
> - Human-readable output now uses dynamic precision to show three significant
> digits and appends `i` to binary unit suffixes. `--progress` uses the same
> formatter for transfer rates and labels byte counts and rates with `B` and
> `B/s`. Example output includes `46.7M`, `48.8Mi`, `782,448B` and
> `113,295B/s`. File-list, created-directory and skipped-deletion counts also
> use thousands separators.
> - Added `--delay-symlinks` to defer creating new and changed symlinks until
> other transferred files are in place. It implies `--no-inc-recursive` and
> requires protocol 29 or newer with support on the receiving side. It cannot
> be combined with `--remove-source-files`, `--hard-links`, `--compare-dest`,
> `--copy-dest` or `--link-dest`.
>
> ------------------------------------------------------------------------------
> | 04 Nov 2026 | 3.5.2 | | 33 |
>
> Signed-off-by: Jaipaul Cheernam <jaipaul.cheernam@est.tech>
> ---
Hi Jaipaul,
https://download.samba.org looks down, so we can not fetch the sources
and put in in the mirror. I will try again later this week.
https://autobuilder.yoctoproject.org/valkyrie/#/builders/29/builds/5054
Thanks,
Mathieu
--
Mathieu Dubois-Briand, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
next prev parent reply other threads:[~2026-11-04 13:52 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-04 11:38 [PATCH 1/4] repo: upgrade 2.71 -> 2.72 Jaipaul Cheernam
2026-11-04 11:38 ` [PATCH 2/4] rsync: upgrade 3.5.1 -> 3.5.2 Jaipaul Cheernam
2026-11-04 13:52 ` Mathieu Dubois-Briand [this message]
2026-11-04 14:00 ` [OE-core] " Jaipaul Cheernam
2026-11-04 11:38 ` [PATCH 3/4] xorgproto: upgrade 2025.1 -> 2026.1 Jaipaul Cheernam
2026-11-04 11:38 ` [PATCH 4/4] ppp: upgrade 2.5.4 -> 2.5.5 Jaipaul Cheernam
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=DMMH5MBL0Y9R.19MJFCNHD82Q7@bootlin.com \
--to=mathieu.dubois-briand@bootlin.com \
--cc=jaipaul.cheernam@est.tech \
--cc=openembedded-core@lists.openembedded.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox