-
Notifications
You must be signed in to change notification settings - Fork 293
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Negotiate individual buffer size dynamically #155
base: main
Are you sure you want to change the base?
Conversation
95d089c
to
d8e49b1
Compare
@wjliang and @arnop2 the new patch address all your comment, please review. |
d8e49b1
to
5b6813a
Compare
@andersson, could you check this proposal to see if there is any conflict to the Linux kernel implementation? |
@wjliang, @arnop2, @galak and @@andersson, here is patch for kernel side: |
How about a generic bit for the RX/TX buf size settings as I believe that virtio-net devices will also need the ability to "set" the sizes vs having random values being picked by each front end? |
5b6813a
to
64037f5
Compare
Rebase to the latest mainline, no other change. |
Move rpmsg_virtio_read_config after device status become ready. |
But from my reading virtio spec, the feature bit is specified by individual virtio device, nobody try to define the common bits before. |
6f74e9d
to
fca2f5c
Compare
@arnopo @edmooring all my PR update to the latest mainline and pass the check, feel free give me the feedback. |
c6f69a0
to
d2fcb24
Compare
Hi @xiaoxiang781216, |
This method is also based on resource table. The key difference is where we put the information:
The problem is how we pass this information from remoteproc layer to rpmsg layer in a standard and compatible way.
Since virtio has many functions to access the config space, the 2nd method is more simple and compatible.
Here is the patch set: |
65cd46a
to
a2b5b59
Compare
504482e
to
a8900e6
Compare
05c0087
to
1c5cd00
Compare
1c5cd00
to
d8e7dae
Compare
d8e7dae
to
c32c98a
Compare
What's the status of this PR? Any chances to merge it soon? If not, would it be OK to split it in two:
I understand the 2nd one must be compatible with Linux and requires more discussion. I hope the 1st one can be merged earlier. |
it seems that we come to the same conclusion (see #322 (comment)) |
380eb70
to
1039d88
Compare
1039d88
to
9ecafcf
Compare
9ecafcf
to
a56398d
Compare
456be8d
to
907f369
Compare
907f369
to
ebe691d
Compare
cce22c2
to
7f36844
Compare
42004c2
to
4ac8ac2
Compare
If slave support VIRTIO_RPMSG_F_BUFSZ(0x04) feature, master determine the buffer size from config space(first 8 bytes), otherwise the default size(512 bytes) will be used. Signed-off-by: Chao An <anchao@pinecone.net>
4ac8ac2
to
8a210fe
Compare
@wjliang and @arnop2,
512 byte buffer size isn't suitable in all case, so this patch let slave specify the best size from resource table.
To keep the compatibility, the new behavior is guarded by VIRTIO_RPMSG_F_BUFSZ.