Classified as: {THALES GROUP LIMITED DISTRIBUTION}
Thanks for your quick answer Joel.
Sorry for not using the “git send-email” command, my company blocks smtp ports
on dev machines.
Patch 0001:
I corrected the comment.
The modification in riscv/abi.yml is indeed not needed, it has been removed.
The copyright in bspcv32a6.yml is corrected.
Patch 0002:
I did not check if the stdout-path is correctly handled in
other architectures.
I used the strlen function because it was the one used
previously in fdt_path_offset.
Patch 0003:
Indeed, my modification will break things.
I observe that the interrupts are enabled here :
cpukit/score/cpu/riscv/riscv-context-switch.S#L190
which depends on RTEMS_SMP flag. My bsp does not have multi-processor but I
would like multi-tasking and ISR enabled.
I think something is wrong here, multi-tasking and SMP should be independent.
De : Joel Sherrill
Envoyé : mardi 1 août 2023 20:54
À : EYSSARTIER Kevin
Cc : devel@rtems.org
Objet : Re: CVA6 RISC-V support
Thanks for the submission. It's a bit hard to review with the patches as
attachments rather than having been sent via git send-email but here goes.
Hopefully someone with more RISC-V knowledge can comment also.
0001 - long comment appears to be one line. Please break into multiple lines.
0001 - riscv/abi.yml is this just adding another set of compiler options? Hard
to tell.
0001 - bspcv32a6.yml - Did this file end up with a copyright?
0002 - same issue with long comment. Blank line between paragraphs.
0002 - Is this a common issue which applies to other BSPs?
0002 - Is there a maximum length of the strings so strnlen() can be used
instead of strlen()?
0002 - Call strlen/strnlen once on stdout_patch and assign to a variable of
type size_t
0003 - This looks like it will break every other architecture since they will
not have had restart renamed to start.
Can you explain what is really not working so we can find a solution that is
strictly within the RISC-V? There is
also a context method which is invoked in the context of the thread once it
begins execution. Maybe this will
work. See threadhandler.c for the call to
_Context_Initialization_at_thread_begin.
I personally am ok for 01 and 02. 03 will require discussion.
--joel
On Tue, Aug 1, 2023 at 10:53 AM EYSSARTIER Kevin
mailto:kevin.eyssart...@thalesgroup.com>>
wrote:
Classified as: {THALES GROUP LIMITED DISTRIBUTION}
Hello,
I am working on the port of RTEMS 6 on our RISC-V processor, the CORE-V CVA6
processor
GitHub - ThalesGroup/cva6: The CORE-V CVA6 is an Application class 6-stage
RISC-V CPU capable of booting Linux<https://github.com/ThalesGroup/cva6>
The port is working and I think it is a good idea to publish my work on the
official repository.
During the port I also found two bugs in the RISC-V generic BSP that I
corrected.
I hope it satisfy the quality standards of RTEMS.
Best regards,
---
Kevin EYSSARTIER, Research Engineer
Thales Research & Technology France – High Performance Computing Lab
Campus Polytechnique - 1, avenue Augustin Fresnel - 91767 Palaiseau cedex
Phone : +33 (0)1 69 41 55 14 Internal : 341 55 14
{THALES GROUP LIMITED DISTRIBUTION}
___
devel mailing list
devel@rtems.org<mailto:devel@rtems.org>
http://lists.rtems.org/mailman/listinfo/devel
{THALES GROUP LIMITED DISTRIBUTION}
0001-Adding-core-v-cv32a6-support.patch
Description: 0001-Adding-core-v-cv32a6-support.patch
0002-bsps-riscv-Handle-in-chosen-stdout-path.patch
Description: 0002-bsps-riscv-Handle-in-chosen-stdout-path.patch
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel
