#1034012 libzstd: Possible data corruption in zstd 1.5.x

Package:
src:libzstd
Source:
src:libzstd
Submitter:
Andrey Semashev
Date:
2025-03-05 19:36:03 UTC
Severity:
normal
#1034012#5
Date:
2023-04-06 12:02:20 UTC
From:
To:
It was discovered that zstd versions from 1.5.0 to 1.5.4, inclusively,
may exhibit data corruption on high compression levels (>= 13). This is
fixed in zstd 1.5.5 and outlined in the release notes and the related
pull request:

https://github.com/facebook/zstd/releases/tag/v1.5.5
https://github.com/facebook/zstd/pull/3517

Please consider upgrading to zstd 1.5.5 or backporting the patch with
the fix.
#1034012#10
Date:
2023-04-06 13:28:13 UTC
From:
To:
control: notfound -1 1.5.4+dfsg2-4

Hi,

Thanks for filing this bug report.

The patch is already included in libzstd 1.5.4+dfsg2-5 in unstable;
I guess I will have to ask the release managers for a freeze exception so
that this version migrates to testing.

G'luck,
Peter
#1034012#17
Date:
2024-09-25 16:26:30 UTC
From:
To:
Dear Maintainer,

I am running a clean install of Raspian OS on a Raspberry Pi 5 w/ 8 GB RAM on a 32 GB microSD w/ OMV 7.

When I run the backup plugin, it calls dd and zstd to create a compressed image of the entire drive.

Right now, the compressed image is around 800 MB, so less than 1:32 compression ratio.

zstd runs correctly only ONCE---but somehow breaks itself and/or liblz4-1, such that the backup
will not run (segmentation faults or finds corrupted libraries). When I reinstall those packages,
the backupu (zstd) will gain run correctly ONCE, but breaks itself again.

I checked the brand-new microSD with f3, and it has 0.00 errors.

I did a clean install of the entire system again, but the problem repeats itself.

I upgraded to zstd 1.5.6 testing to see if the bug had been fixed there, but the same problem occurs.

I have not tried the unstable branch.

Thank you very much!


*** Reporter, please consider answering these questions, where appropriate ***

   * What led up to the situation?
   * What exactly did you do (or not do) that was effective (or
     ineffective)?
   * What was the outcome of this action?
   * What outcome did you expect instead?

*** End of the template - remove these template lines ***
#1034012#22
Date:
2024-09-25 16:58:18 UTC
From:
To:
Dear Maintainer,

I apologize, but I updated zstd just now to the UNSTABLE branch, and it appears that the segmentation fault no longer occurs in that version (1.5.6 UNSTABLE).

Thank you!

–
Rev. Brian T. Austin, F.S.S.P.
#1034012#27
Date:
2024-09-25 18:23:37 UTC
From:
To:
Dear Maintainer,

Well … it worked once, but I rebooted and then … not a segmentation fault but the other error:

$ zstd --version
zstd: error while loading shared libraries: ead_cond_signal: cannot open shared object file: No such file or directory

After reinstalling zstd, I get:

 $ zstd --version
*** Zstandard CLI (64-bit) v1.5.6, by Yann Collet ***

And … once again … the backup plugin on OMV 7 will work … until it breaks again.

Thank you!

–
Rev. Brian T. Austin, F.S.S.P.