guix-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bug#42023] [PATCH] Retry deduplication on ENOENT


From: Ludovic Courtès
Subject: [bug#42023] [PATCH] Retry deduplication on ENOENT
Date: Wed, 16 Sep 2020 22:37:05 +0200
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/27.1 (gnu/linux)

Hi!

Caleb Ristvedt <caleb.ristvedt@cune.org> skribis:

[...]

>> There’s an issue with this patch.  I gave it a spin (offloading a few
>> builds) and it got stuck in a infinite loop:
>>
>> stat("/gnu/store/.links/0k63r0n3681r2gqd00blq4z5xd7cw1knv0x049p99f0pq31brhrk",
>>  0x7ffe43898cd0) = -1 ENOENT (Dosiero aŭ dosierujo ne ekzistas)
>> link("/gnu/store/83jy739bn644w3pnwgb5kwjig0kzs92f-libwps-0.4.12/lib/libwps-0.4.so",
>>  "/gnu/store/.links/0k63r0n3681r2gqd00blq4z5xd7cw1knv0x049p99f0pq31brhrk") = 
>> -1 EEXIST (Dosiero jam ekzistas)
>> stat("/gnu/store/.links/0k63r0n3681r2gqd00blq4z5xd7cw1knv0x049p99f0pq31brhrk",
>>  0x7ffe43898cd0) = -1 ENOENT (Dosiero aŭ dosierujo ne ekzistas)
>> link("/gnu/store/83jy739bn644w3pnwgb5kwjig0kzs92f-libwps-0.4.12/lib/libwps-0.4.so",
>>  "/gnu/store/.links/0k63r0n3681r2gqd00blq4z5xd7cw1knv0x049p99f0pq31brhrk") = 
>> -1 EEXIST (Dosiero jam ekzistas)
>> stat("/gnu/store/.links/0k63r0n3681r2gqd00blq4z5xd7cw1knv0x049p99f0pq31brhrk",
>>  0x7ffe43898cd0) = -1 ENOENT (Dosiero aŭ dosierujo ne ekzistas)
>> link("/gnu/store/83jy739bn644w3pnwgb5kwjig0kzs92f-libwps-0.4.12/lib/libwps-0.4.so",
>>  "/gnu/store/.links/0k63r0n3681r2gqd00blq4z5xd7cw1knv0x049p99f0pq31brhrk") = 
>> -1 EEXIST (Dosiero jam ekzistas)
>>
>
> I believe I can explain this. In 'deduplicate' we currently treat
> anything that isn't a directory as a hardlinkable thing. This includes
> symlinks (although it's implementation-defined whether symlinks can be
> hardlinked to - we use CAN_LINK_SYMLINK to test this in
> nix/libstore/optimise-store.cc). This means that at present we
> unconditionally attempt to deduplicate symlinks (which happens to work
> on linux). However, 'file-exists?' uses stat, not lstat, to check for
> file existence. Thus, if there is a dangling symlink, 'file-exists?'
> will return #f when passed it, but of course attempting to call link()
> to create it will fail with EEXIST. Attached is a modified patch that
> tests for file existence with lstat instead. I expect that will fix the
> problem.

Ah ha!

> We should probably still add a test in 'deduplicate' for whether
> symlinks can be hardlinked to.

If GNU/Linux and GNU/Hurd support it, it’s unnecessary.

> Tangent: I was curious why libwps-0.4.so would be a dangling symlink,
> and it turns out that it's actually a relative symlink, so when
> accessing it via /gnu/store/...-libwps-0.4.12/lib/libwps-0.4.so it isn't
> dangling, but when accessing it via /gnu/store/.links/0k63r... it is.

I see, good catch!

> From 12f5848e79b0ede95babebea240264b32e39812c Mon Sep 17 00:00:00 2001
> From: Caleb Ristvedt <caleb.ristvedt@cune.org>
> Date: Sat, 8 Aug 2020 11:25:57 -0500
> Subject: [PATCH] deduplication: retry on ENOENT.
>
> It's possible for the garbage collector to remove the "canonical" link after
> it's been detected as existing by 'deduplicate'.  This would cause an ENOENT
> error when replace-with-link attempts to create the temporary link.  This
> changes it so that it will properly handle that by retrying.
>
> * guix/store/deduplication.scm (replace-with-link): renamed to
>   canonicalize-with-link, now also handles the case where the target link
>   doesn't exist yet, and retries on ENOENT.  Also modified to support
>   canonicalizing symbolic links, though it is the caller's responsibility to
>   ensure that the system supports hardlinking to a symbolic link (on Linux it
>   does).
>   (deduplicate): modified to use canonicalize-with-link.

[...]

> +            (lambda args
> +              (let ((errno (system-error-errno args)))
> +                (cond
> +                 ((= errno ENOENT)
> +                  ;; either SWAP-DIRECTORY has missing directory
> +                  ;; components or TARGET was deleted - this is a
> +                  ;; fundamental ambiguity to the errno produced by
> +                  ;; link()
> +                  (if (file-exists? swap-directory)
> +                      ;; we must assume link failed because target doesn't
> +                      ;; exist, so create it.

Nitpick: Please capitalize sentences, add a period at the end, and write
“'link'” instead of “link()” or “link” for clarity.

Otherwise LGTM.

I think we’ll have to stress-test it through offloading to catch any
remaining issues.

Thank you!

Ludo’.





reply via email to

[Prev in Thread] Current Thread [Next in Thread]