Skip to content

Conversation

mcimerman
Copy link
Contributor

We should try to use the maximum allowed IPC transfer buffer size. Can possibly save 4 calls.

@jxsvoboda
Copy link
Contributor

Increasing the buffer size is probably OK, although it will typically not bring huge improvement in the typical scenario as the bottleneck is between the file system and block device, with much smaller buffers.

If you want to do this, then either just continue using a literal constant, or have vfs define some constant, e.g. VFS_XFER_SIZE to make this nicely encapsulated. We don't want to leak the details of how the VFS interface works by tapping directly into IPC stuff.

@mcimerman
Copy link
Contributor Author

Increasing the buffer size is probably OK, although it will typically not bring huge improvement in the typical scenario as the bottleneck is between the file system and block device, with much smaller buffers.

I agree, but in the future if the block cache is improved and there will be buffering added, it'll help.

I will change it to literal (1024 * 64).

@mcimerman mcimerman force-pushed the mkfile-data-xfer-limit branch from 200b03c to b38f5cd Compare October 11, 2025 19:38
@mcimerman mcimerman force-pushed the mkfile-data-xfer-limit branch from b38f5cd to aa0bfe3 Compare October 11, 2025 19:40
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Development

Successfully merging this pull request may close these issues.

2 participants