Re: [PATCH linux-evl 0/2] Add RISC-V support to EVL

On Wed, 2026-07-08 at 17:32 +0200, Jan Kiszka wrote:
> On 08.07.26 11:51, Florian Bezdeka wrote:
> > On Wed, 2026-07-08 at 10:55 +0200, Philippe Gerum wrote:
> > > Jan Kiszka <[email protected]> writes:
> > > 
> > > > On 08.07.26 08:26, Tobias Schaffner wrote:
> > > > > 
> > > > > 
> > > > > On 7/8/26 08:07, Jan Kiszka wrote:
> > > > > > On 08.07.26 07:17, Tobias Schaffner wrote:
> > > > > > > On 7/8/26 00:30, Florian Bezdeka wrote:
> > > > > > > > On Fri, 2025-10-10 at 12:10 +0200, Tobias Schaffner wrote:
> > > > > > > > > Hi all,
> > > > > > > > > 
> > > > > > > > > this series introduces an initial EVL port for RISC-V.
> > > > > > > > > 
> > > > > > > > > While some tests are still failing, I’d like to share this early
> > > > > > > > > version with the community to gather feedback, reviews, and hopefully
> > > > > > > > > some assistance in refining the port.
> > > > > > > > > 
> > > > > > > > > The patches have been tested in combination with the Dovetail RISC-V
> > > > > > > > > port on top of v6.12-evl1-rebase.
> > > > > > > > > 
> > > > > > > > > Looking forward to your feedback and suggestions.
> > > > > > > > > 
> > > > > > > > 
> > > > > > > > I tried to apply this series today, and run into the following issues
> > > > > > > > reported by my tooling.
> > > > > > > > 
> > > > > > > > Patch 1 is fine, patch 2 has some checkpatch findings.
> > > > > > > > 
> > > > > > > > A couple of debatable warnings like:
> > > > > > > > 
> > > > > > > > WARNING: Misplaced SPDX-License-Identifier tag - use line 1 instead
> > > > > > > > #58: FILE: arch/riscv/include/uapi/asm/evl/fptest.h:2:
> > > > > > > > + * SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note
> > > > > > > > 
> > > > > > > > WARNING: unnecessary whitespace before a quoted newline
> > > > > > > > #95: FILE: arch/riscv/include/uapi/asm/evl/fptest.h:34:
> > > > > > > > +
> > > > > > > > 
> > > > > > > > In addition to a serious one:
> > > > > > > > 
> > > > > > > > ERROR: Missing Signed-off-by: line by nominal patch author 'shannmu
> > > > > > > > <[email protected]>'
> > > > > > > 
> > > > > > > I did not get any responses from this email anymore. I can try to reach
> > > > > > > out to him one last time, but what do we do about it if he will not
> > > > > > > reply?
> > > > > > > 
> > > > > > 
> > > > > > If I read that patch correctly, it's majority is taken from the
> > > > > > corresponding Xenomai 3 logic. How much is unique contribution, how much
> > > > > > was your work, Tobias?
> > > > > 
> > > > > I added the evl_set_fpregs/evl_check_fpregs content which is a
> > > > > translation from what Gilles Chanteperdrix
> > > > > <[email protected]> did for the x86 and arm architectures
> > > > > with some isa enablement bits added.
> > > > > 
> > > > > This should be the initial version by shanmu where I added this on top:
> > > > > https://github.com/shannmu/linux/blob/xenomai4/wip/dovetail-riscv/
> > > > > evl_port/arch/riscv/include/asm/evl/fptest.h
> > > > > 
> > > > 
> > > > At least for this patch, I think it would be fair to claim authorship
> > > > for the final one and only refer to shanmu as provider of a baseline.
> > > > 
> > > 
> > > Agreed.  This is a trivial adaptation of the pre-existing
> > > arch/arm*/include/asm/evl/fptest.h, which is in and of itself
> > > boilerplate code.
> > > 
> > 
> > We had a short chat about this topic in the Xenomai community call
> > today. To avoid further delay of the dovetail 7.1 release we decided
> > that we will move out the risc-v series again.
> > 
> 
> As you already removed wip/flo/v7.1-evl-rebase+riscv (which broke
> xenomai-images CI, unfortunately), where will be the new transitional
> branch in the meantime?
> 

Oh, those temporary branches shouldn't be referenced in public CI
things. If necessary we can bring it back.