1724629 - fix sha256-armv8 build failure on Aarch64 Big Endian

Status: UNCONFIRMED

(NSS :: Build, defect, P5)

Description

[Giulio Benetti]

User Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/92.0.4515.131 Safari/537.36

Steps to reproduce:

Build for Aarch64 Big Endian.

Actual results:

http://autobuild.buildroot.net/results/6c78efd8af401cd19635ea91f20e33cdf55691b7/build-end.log

Comment 1

[Giulio Benetti]

Also, is it correct that sha256 acceleration on armv8 is valid only for LITTLE_ENDIAN? Or is it valid for BIG_ENDIAN too?
I can test it since I don't have a target for it. Can you please clarify?

Thanks a lot

Comment 2

[Benjamin Beurdouche [:beurdouche]]

Unfortunately this is not happening on a supported architecture and we have no hardware to reproduce, we would gladly accept a patch if you can work on this. About the acceleration, from the code, it looks like it should work as well on big endian if this double definition is fixed.

Comment 3

[Giulio Benetti]

I don't even have a supported machine. Anyway another contributor in Buildroot fixed the problem 1 assuming it is compatible for both BIG and LITTLE Endian and works fine. So we should be done with this.

Best regards

Comment 4

[Benjamin Beurdouche [:beurdouche]]

Nice, thanks for pointing to this, we'll integrate that change.

Comment 5

[Giulio Benetti]

Great, look forward to see it committed.

Best regards

Comment 6

[Benjamin Beurdouche [:beurdouche]]

Attachment: Bug 1724629 - Use HW accelerated SHA2 on AArch64 Big Endian.