From: Hannes Laimer <h.laimer@proxmox.com>
To: "Proxmox Backup Server development discussion"
<pbs-devel@lists.proxmox.com>,
"Fabian Grünbichler" <f.gruenbichler@proxmox.com>
Subject: Re: [pbs-devel] [PATCH proxmox-backup] tape: media_catalog: replace deprecated flock() with Flock
Date: Thu, 20 Nov 2025 11:11:25 +0100 [thread overview]
Message-ID: <bb17abdf-f830-45ae-9a68-c69673621f17@proxmox.com> (raw)
In-Reply-To: <1763022246.tus7z1wzjz.astroid@yuna.none>
not sure why, but my brain skipped this mail...
anyway, thanks for taking a look! comments inline
On 11/13/25 09:39, Fabian Grünbichler wrote:
> On November 12, 2025 10:18 am, Hannes Laimer wrote:
>> nix 0.28.0 deprecated fcntl::flock() in favor of fcntl::Flock.
>> In commit() lock a cloned fd to avoid moving self.file
>>
>> No functional change intended. Fixes deprecation warnings.
>>
>> Signed-off-by: Hannes Laimer <h.laimer@proxmox.com>
>> ---
>> src/tape/media_catalog.rs | 27 ++++++++++++++++++---------
>> 1 file changed, 18 insertions(+), 9 deletions(-)
>>
>> diff --git a/src/tape/media_catalog.rs b/src/tape/media_catalog.rs
>> index 63329a65..7307ca0f 100644
>> --- a/src/tape/media_catalog.rs
>> +++ b/src/tape/media_catalog.rs
>> @@ -9,6 +9,7 @@ use endian_trait::Endian;
>>
>> use proxmox_sys::fs::read_subdir;
>>
>> +use nix::fcntl;
>> use proxmox_io::{ReadExt, WriteExt};
>> use proxmox_sys::fs::{create_path, fchown, CreateOptions};
>> use proxmox_uuid::Uuid;
>> @@ -194,7 +195,7 @@ impl MediaCatalog {
>> let me = proxmox_lang::try_block!({
>> Self::create_basedir(base_path)?;
>>
>> - let mut file = std::fs::OpenOptions::new()
>> + let file = std::fs::OpenOptions::new()
>> .read(true)
>> .write(write)
>> .create(create)
>> @@ -219,9 +220,12 @@ impl MediaCatalog {
>> };
>>
>> // Note: lock file, to get a consistent view with load_catalog
>> - nix::fcntl::flock(file.as_raw_fd(), nix::fcntl::FlockArg::LockExclusive)?;
>> - let result = me.load_catalog(&mut file, media_id.media_set_label.as_ref());
>> - nix::fcntl::flock(file.as_raw_fd(), nix::fcntl::FlockArg::Unlock)?;
>> + let mut locked = fcntl::Flock::lock(file, nix::fcntl::FlockArg::LockExclusive)
>> + .map_err(|(_, e)| format_err!("flock failed - {}", e))?;
>
> this part here is fine
>
>> + let result = me.load_catalog(&mut locked, media_id.media_set_label.as_ref());
>> + let file = locked
>> + .unlock()
>> + .map_err(|(_, e)| format_err!("flock unlock failed - {}", e))?;
>
> but this is not - if unlocking fails, the returned Error tuple has the
> Flock instance as first member. and if you drop an Flock, it will try to
> unlock it again, and panic if that fails (which seems likely if the
> first unlock attempt failed, e.g. because the FD got closed for some
> reason, or the underlying FS went away, or ..).
>
good catch! I think a
`std::mem::forget(_guard);`
in the mapper should fix that. As for the migration in proxmox-sys...
> if we switch to Flock here, we should also migrate our helpers in
> proxmox_sys over, and ideally, make them usable here as well ;)
>
... I'm not sure that that would work? I don't think we could do timeout.
we could add a separate/new `safe-locking-thing` that uses the new Flock
with RAII though... but I'm not sure we want two
>>
>> let (found_magic_number, _) = result?;
>>
>> @@ -371,14 +375,19 @@ impl MediaCatalog {
>> Some(ref mut file) => {
>> let pending = &self.pending;
>> // Note: lock file, to get a consistent view with load_catalog
>> - nix::fcntl::flock(file.as_raw_fd(), nix::fcntl::FlockArg::LockExclusive)?;
>> + let file_clone = file.try_clone()?;
>> + let mut locked =
>> + fcntl::Flock::lock(file_clone, nix::fcntl::FlockArg::LockExclusive)
>> + .map_err(|(_, e)| format_err!("flock failed - {}", e))?;
>> let result: Result<(), Error> = proxmox_lang::try_block!({
>> - file.write_all(pending)?;
>> - file.flush()?;
>> - file.sync_data()?;
>> + locked.write_all(pending)?;
>> + locked.flush()?;
>> + locked.sync_data()?;
>> Ok(())
>> });
>> - nix::fcntl::flock(file.as_raw_fd(), nix::fcntl::FlockArg::Unlock)?;
>> + let _ = locked
>> + .unlock()
>> + .map_err(|(_, e)| format_err!("flock unlock failed - {}", e))?;
>
> same here
>
>>
>> result?;
>> }
>> --
>> 2.47.3
>>
>>
>>
>> _______________________________________________
>> pbs-devel mailing list
>> pbs-devel@lists.proxmox.com
>> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pbs-devel
>>
>>
>>
>
>
> _______________________________________________
> pbs-devel mailing list
> pbs-devel@lists.proxmox.com
> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pbs-devel
>
>
_______________________________________________
pbs-devel mailing list
pbs-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pbs-devel
next prev parent reply other threads:[~2025-11-20 10:11 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-11-12 9:18 Hannes Laimer
2025-11-13 8:39 ` Fabian Grünbichler
2025-11-20 10:11 ` Hannes Laimer [this message]
2025-11-20 10:34 ` Fabian Grünbichler
2025-11-20 8:12 Hannes Laimer
2025-11-20 9:32 ` Fabian Grünbichler
2025-11-20 9:35 ` Hannes Laimer
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=bb17abdf-f830-45ae-9a68-c69673621f17@proxmox.com \
--to=h.laimer@proxmox.com \
--cc=f.gruenbichler@proxmox.com \
--cc=pbs-devel@lists.proxmox.com \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.