[PATCH v2 0/3] ARM: NEON based fast(er) AES in CBC/CTR/XTS modes

On Fri, Oct 04, 2013 at 08:41:35PM +0200, Ard Biesheuvel wrote:

On 4 October 2013 20:34, Nicolas Pitre [off-list ref] wrote:

quoted

On Fri, 4 Oct 2013, Will Deacon wrote:

[...]

quoted

quoted

Why do you consider it unsuitable to ship the perl script with the kernel?
Perl 5 is already documented as a build dependency in Documentation/Changes

Do you have an example of something that does require perl to build the
kernel on ARM?  I was under the impression that people try to avoid it
as much as possible in general.

I'm personally sitting on the fence between effectively adding a new
kernel build dependencies or carrying the output of the perl script.
But if the kernel build does already require perl in practice then this
might tip the balance.

I like Russell's suggestion the most, in fact. In this case, the build
time requirement for Perl effectively gets suspended until you start
making modifications to the perl script, and the relation between the
.S and the .pl files is made explicit by the make rule.

Should I put the cmd_perl rule in scripts/Makefile.build ? Or can I
just keep it under arch/arm/crypto ?

Just running through the Makefiles, it seems we have a fair amount
of stuff already using perl in various ways.  So I wouldn't worry too
much about where it's placed.  It's probably something that should
eventually end up in scripts/ at _some_ point.