v9fs:Add socket support #15597

crafcat7 · 2025-01-18T05:36:24Z

Summary

In #13001, I added a new distributed file system in NuttX, which only supports VIRTIO as a data transmission method.
In this PR, I implemented a Socket-based transmission method, so that V9FS can access the content on the Host based on the Socket method, which is suitable for more scenarios.
I also added the V9FS document to explain how to use it.
In short:
- mplemented a socket-based v9fs driver
- Enhanced Net-related, now TCP supports multi-segment IOV sending
- Added V9FS Document

Impact

Enhanced NET related changes, now TCP supports multi-segment IOV mode sending.
V9FS supports socket as a transmission medium
Added V9FS Document

Testing

Build Host(s): Linux x86
Target(s): qemu
Passed local file system related tests

nuttxpr · 2025-01-18T05:45:16Z

[Experimental Bot, please feedback here]

This PR mostly meets the NuttX requirements, but could be improved by providing more specific details. Here's a breakdown:

Strengths:

Clear summary of changes: The summary explains the "why," "what," and "how" of the changes. Referencing the prior PR is also helpful.
Acknowledges impact areas: The impact section touches on several key aspects, like functionality, documentation, and networking changes.
Provides basic testing information: The testing section indicates the host and target environments.

Weaknesses & Suggestions for Improvement:

Missing Issue References: While the prior PR is mentioned, specific NuttX or NuttX Apps issues should be linked if applicable. Even if there isn't a directly related issue, creating one for tracking the feature request/bug fix is recommended.
Impact Section Lacks Detail: The impact section is too brief. Instead of just stating "Enhanced NET related changes," describe specifically how the multi-segment IOV sending impacts users, builds, etc. For example:
- Impact on user: Users can now potentially see performance improvements in network file system operations.
- Impact on build: No changes to the build process itself, but new code is added.
- Impact on hardware: No direct hardware impact, but affects network driver usage.
- Impact on security: Consider potential security implications of socket-based V9FS (e.g., authentication, data integrity). Even if there are none, explicitly state this.
- Impact on compatibility: Does this change break anything? Is it compatible with all existing network configurations?
Missing "Before/After" Testing Logs: The testing section states tests passed, but provides no actual logs. Include relevant log snippets demonstrating the change's effect (e.g., mounting the V9FS filesystem, performing file operations, demonstrating the multi-segment IOV sending). Show how the behavior differs before and after your changes.
Limited Testing Information: While QEMU is a good start, testing on actual hardware is highly recommended if possible. Specify the QEMU configuration used (e.g., qemu-system-x86_64 -M virt). More testing details like the specific tests run (if any automated test suite was used) would strengthen the PR.

In short: The PR is on the right track, but needs more detail, particularly in the Impact and Testing sections, to fully meet the NuttX requirements. Be as explicit and thorough as possible to ensure reviewers can properly assess the changes.

xiaoxiang781216 · 2025-01-18T06:56:19Z

@crafcat7 please fix the document error

crafcat7 · 2025-01-18T09:48:43Z

@crafcat7 please fix the document error

Done,now it has passed the document check.

anchao · 2025-01-18T13:29:24Z

net/tcp/tcp_recvfrom.c

+       */
+
+      if (ret < 0 ||
+          (!(flags & MSG_WAITALL) && ret < msg->msg_iov[i].iov_len))


update the nrecv if ret < iov_len

The current implementation is designed so that if an error occurs in one of the segments when receiving multiple segments of iov, the received data will be returned.
If we are not in MSG_WAITALL, then as long as there is one recv, we can directly return ret.
In the case of multiple segments of iov, when we ensure that ret >= 0, each segment of iov is received only once and returns nrev.

ret already indicates that you have received some data, why not return this value to the user?

Assuming msg->msg_iovlen > 1 and MSG_WAITALL, nbytes will be returned only after all iov_lens are filled. If a recv error occurs in the middle, the received nbytes will be told to the caller, because nbytes has been added to ret in for.
If it is not MSG_WAITALL, we only need to receive it once and tell the caller directly, which is equivalent to the original method.
So I would like to ask if your idea is to fill each iov_len segment even if there is no MSG_WAITALL flag?

anchao · 2025-01-18T13:31:49Z

net/socket/recvmsg.c

@@ -88,11 +88,6 @@ ssize_t psock_recvmsg(FAR struct socket *psock, FAR struct msghdr *msg,
      return -EINVAL;
    }

-  if (msg->msg_iovlen != 1)


why not implement multi message recv in this function?

it's the work for the furture improvement, your contribution is welcome.

let us hold this commit. the protocol layer does not have multi-segment data mapping implement, so I think to resolve the multi-segment reception problem in VFS, rather than requiring similar changes for each protocol.

do you review the recent change from @yamt : #14597 #13498 #12674
it's impossible to implement the multi-segment data transfer on read(recv)/write(send) without the additional allocation.

fs/v9fs/socket_9p.c

anchao · 2025-01-18T13:33:54Z

fs/v9fs/Kconfig

@@ -21,4 +21,9 @@ config V9FS_VIRTIO_9P
 	depends on DRIVERS_VIRTIO
 	default n

+config V9FS_SOCKET_9P


Suggested change

config V9FS_SOCKET_9P

config V9FS_SOCKET

but viritio is V9FS_VIRTIO_9P

The current naming format is V9FS_[Transport NAME]_9P
If we change to V9FS_SOCKET, then VIRTIO should also be changed to V9FS_VIRTIO

Summary: 1.Add new api for socket parsing header - v9fs_parse_size 2.Add socket driver for 9pfs Signed-off-by: chenrun1 <[email protected]>

To prepare for supporting multiple iov in each protocol. Signed-off-by: Zhe Weng <[email protected]>

Summary: Implement multiple iovlen recvmsg in inet_sockif, only TCP mode support at current. Signed-off-by: chenrun1 <[email protected]>

Signed-off-by: chenrun1 <[email protected]>

github-actions bot added Area: Documentation Improvements or additions to documentation Area: Networking Effects networking subsystem Area: File System File System issues Size: L The size of the change in this PR is large labels Jan 18, 2025

xiaoxiang781216 approved these changes Jan 18, 2025

View reviewed changes

crafcat7 force-pushed the v9fs_socket branch from 353f2c0 to 8bb937c Compare January 18, 2025 08:07

crafcat7 changed the title ~~V9fs socket~~ v9fs:Add socket support Jan 18, 2025

anchao requested changes Jan 18, 2025

View reviewed changes

crafcat7 and others added 3 commits January 19, 2025 00:34

v9fs:socket driver

c24577c

Summary: 1.Add new api for socket parsing header - v9fs_parse_size 2.Add socket driver for 9pfs Signed-off-by: chenrun1 <[email protected]>

net: Split msg_iovlen check in recvmsg into each protocol

1a2b3a1

To prepare for supporting multiple iov in each protocol. Signed-off-by: Zhe Weng <[email protected]>

inet_sockif:support multiple iov recv

cb07bd8

Summary: Implement multiple iovlen recvmsg in inet_sockif, only TCP mode support at current. Signed-off-by: chenrun1 <[email protected]>

crafcat7 force-pushed the v9fs_socket branch from 8bb937c to 7a1cdc6 Compare January 18, 2025 16:34

v9fs:add Documentation

3c389a6

Signed-off-by: chenrun1 <[email protected]>

crafcat7 force-pushed the v9fs_socket branch from 7a1cdc6 to 3c389a6 Compare January 18, 2025 16:43

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

v9fs:Add socket support #15597

v9fs:Add socket support #15597

crafcat7 commented Jan 18, 2025

nuttxpr commented Jan 18, 2025

xiaoxiang781216 commented Jan 18, 2025

crafcat7 commented Jan 18, 2025

anchao Jan 18, 2025

crafcat7 Jan 18, 2025 •

edited

Loading

anchao Jan 18, 2025

crafcat7 Jan 19, 2025

anchao Jan 18, 2025

xiaoxiang781216 Jan 18, 2025

anchao Jan 18, 2025

xiaoxiang781216 Jan 18, 2025

anchao Jan 18, 2025

xiaoxiang781216 Jan 18, 2025

crafcat7 Jan 18, 2025

v9fs:Add socket support #15597

Are you sure you want to change the base?

v9fs:Add socket support #15597

Conversation

crafcat7 commented Jan 18, 2025

Summary

Impact

Testing

nuttxpr commented Jan 18, 2025

xiaoxiang781216 commented Jan 18, 2025

crafcat7 commented Jan 18, 2025

Choose a reason for hiding this comment

crafcat7 Jan 18, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

crafcat7 Jan 18, 2025 •

edited

Loading