Closed CV-Bowen closed 7 months ago
hello @CV-Bowen ,
Could you details what was the issue you are facing that needs to move the virtio_create_virtqueues to the transport layer? that would help to understand your work
@arnopo Hi, we want to support different transport layer (remoteproc, mmio) for virtio drivers. After this PR, we can implement the mmio tranport layer base on OpenAmp. I think this patch is also helpful for virtio-mmio framework in openamp/virtio-exp branch to support remoteproc transport layer.
@arnopo Hi, we want to support different transport layer (remoteproc, mmio) for virtio drivers. After this PR, we can implement the mmio tranport layer base on OpenAmp. I think this patch is also helpful for virtio-mmio framework in openamp/virtio-exp branch to support remoteproc transport layer.
I understand the objective., what is not clear to me as a first review is why you need to move virtio_create_virtqueues
this should be independent from the transport layer, no?
@arnopo Hi, we want to support different transport layer (remoteproc, mmio) for virtio drivers. After this PR, we can implement the mmio tranport layer base on OpenAmp. I think this patch is also helpful for virtio-mmio framework in openamp/virtio-exp branch to support remoteproc transport layer.
I understand the objective., what is not clear to me as a first review is why you need to move
virtio_create_virtqueues
this should be independent from the transport layer, no?
I found only rpmsg_virtio_create_virtqueues
call virtio_create_virtqueues
in OpenAmp, so I delete virtio_create_virtqueues
after moving it's implementation to rproc_virtio_create_virtqueues
. I do not consider that some other codes use it, I will restore this api and call it directly in rproc_virtio_create_virtqueues()
.
@arnopo @edmooring Commit 1: virtio: follow virtio 1.2 spec, add more virtio status and device id Seems commit 1 can be an independent PR. I will upload it later, and this PR will be only for Commit 2.
@CV-Bowen seems that you remove the Commit 2: virtio: decoupling the transport layer and virtio device layer
Please notice also https://github.com/OpenAMP/open-amp/pull/494, we should find a common solution...
@arnopo Sorry, I has fixed this and I will look #494.
@CV-Bowen seems that you remove the Commit 2: virtio: decoupling the transport layer and virtio device layer
Please notice also #494, we should find a common solution...
@CV-Bowen 's change extract and generalize from #494, after this change get merged:
@CV-Bowen : Please, could you push your work on lib/include/openamp/virtio.h in a separate pull request as requested in (https://github.com/OpenAMP/open-amp/issues/390#issuecomment-1602605560) That well help to move forward on the topic? Thanks in advance.
@arnopo Sorry, i'm busy with other things, and the new PR has been created ( https://github.com/OpenAMP/open-amp/pull/495). But a mistake is this PR is also updated, i will restore this PR.
@CV-Bowen Zephyr Build fails because the sample application has defined same functions(https://elixir.bootlin.com/zephyr/latest/source/samples/subsys/ipc/openamp/src/main.c#L66). Just ignore the issue this has to be fixed in Zephyr ( I will send a PR)
Thank you for your prompt update!
LGTM after addressing the 2 comments and fix "deivce" typo in commit
Just see that you create the https://github.com/OpenAMP/open-amp/pull/495 to create the dispatch functions. So please take into account my remarks to apply them on https://github.com/OpenAMP/open-amp/pull/495.
@arnopo Done in #495
I will continue this PR to implement the remoteproc transport ops: create_virtqueue and delete_virtqueues.
All the comments should have been addressed. Could you take a look again? @arnopo
@arnopo, @edmooring If you are okay, we can merge the change.
@CV-Bowen, Please, could you rebase it to solve merge conflict, that i test it and merge it
@arnopo Rebased.
tests Ok on my side @edmooring , @tnmysh , could you make a short test on Xilinx platform for complementary test on remoteproc ?
This pull request has been marked as a stale pull request because it has been open (more than) 45 days with no activity.
@arnopo Rebase to the last to fix the conflict. Could we merge this PR? There isn’t much time until release in April.
Sorry @CV-Bowen, it go out of my radar,
@edmooring , @tnmysh , as requested could you confirm that it does not create regression on your platforms
remoteproc_virtio: optimize the remoteproc virtio transport layer
rpmsg_init_vdev_with_config()
also