Issues and questions regarding PowerPC 32-bits e500v2 tuning

[Elinor Montmasson <elinor.montmasson@savoirfairelinux.com>]

Hello,

My company is currently interested in building a Yocto distribution for
a PowerPC 32-bit e500v2 platform using the associated tuning available
in OE-Core.
I have a few questions related to the e500v2 tuning in OE-Core.

I understand that PowerPC e500v2 was considered part of a specific architecture
category, PowerPC SPE, which isn't supported anymore in GCC as SPE support
was dropped in GCC 9.1.
In the e500v2 core, the SPE was used as FPU, meaning code for this architecture
can now only be compiled with soft fpu if using active GCC releases.
Do you confirm this analysis?

Building a core-image-minimal with the ppce500v2 tuning in OE-Core doesn't work
out-of-the-box on Yocto Wrynose.
Recipes building with meson (e.g. xorgproto) fail because the do_configure task
runs a "meson-qemuwrapper" binary which fails with SIGILL.
This looks like a floating point configuration problem as adding the tuning
feature "fpu-soft" solves this issue.
This would confirm the lack of SPE support and the need to specify soft fpu
usage or any usage of floating point will result in an execution error.

* Was any testing done for the e500v2 tuning or the e500 tuning?
	* The e500 shares the same SPE specificity.
* Should the e500v2 tuning configuration be considered not up-to-date?
* Is there any plan in updating and testing the e500 and e500v2 tuning
  configurations?
	* Especially, tune-ppce500.inc still enables the "spe" feature,
	  but should encounter the same GCC compatibility problem as the e500v2.

Best regards,
Elinor Montmasson