[prev in list] [next in list] [prev in thread] [next in thread] 

List:       busybox
Subject:    sha_ni detection  for sha hwaccel is broken
From:       Natanael Copa <ncopa () alpinelinux ! org>
Date:       2023-01-18 12:03:43
Message-ID: 20230118130343.5b4a1e00 () ncopa-desktop ! lan
[Download RAW message or body]

Hi,

We have an issue where busybox sha1sum and/or sha256sum is broken when
HWACCEL is enabled. This has been observed on github runners only so far.

I cloned busybox repo and added github runner that build only sha1sum
and sha256sum with hwaccel enabled. I added debug printf statements to
see if shaNI was enabled or not.

What happened was that sometimes shaNI is enabled, and sometimes it is
not.

You have example of this here:

https://github.com/ncopa/busybox/actions/runs/3948518291/jobs/6758600595

As you see, when sha256sum runs, shaNI is -1, but later, using the same
binary shaNI becomes 11 and ends up with illegal instruction.

Something is off, but I'm not sure what.

Disabling hwaccel makes it all go away.

I have not been able to reproduce this on my i9. It has only been
observed on github runners so far.

-nc
_______________________________________________
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
[prev in list] [next in list] [prev in thread] [next in thread]