On Wed, Feb 05, 2025 at 04:10:15PM -0800, Dan Williams wrote:

Vivek Goyal wrote:

quoted

On Fri, Jan 10, 2025 at 05:00:29PM +1100, Alistair Popple wrote:

quoted

FS DAX requires file systems to call into the DAX layout prior to unlinking
inodes to ensure there is no ongoing DMA or other remote access to the
direct mapped page. The fuse file system implements
fuse_dax_break_layouts() to do this which includes a comment indicating
that passing dmap_end == 0 leads to unmapping of the whole file.

However this is not true - passing dmap_end == 0 will not unmap anything
before dmap_start, and further more dax_layout_busy_page_range() will not
scan any of the range to see if there maybe ongoing DMA access to the
range. Fix this by passing -1 for dmap_end to fuse_dax_break_layouts()
which will invalidate the entire file range to
dax_layout_busy_page_range().

Hi Alistair,

Thanks for fixing DAX related issues for virtiofs. I am wondering how are
you testing DAX with virtiofs. AFAIK, we don't have DAX support in Rust
virtiofsd. C version of virtiofsd used to have out of the tree patches
for DAX. But C version got deprecated long time ago.

Do you have another implementation of virtiofsd somewhere else which
supports DAX and allows for testing DAX related changes?

I have personally never seen a virtiofs-dax test. It sounds like you are
saying we can deprecate that support if there are no longer any users.
Or, do you expect that C-virtiofsd is alive in the ecosystem?

Ashai Lina responded that they need and test DAX using libkrun.

C version of virtiofsd is now gone. We are actively working and testing
Rust version of virtiofsd. We have not been able to add DAX support to
it yet for various reasons. 

Biggest unsolved problem with viritofsd DAX mode is guest process should
get a SIGBUS if it tries to access a file beyond the file. This can happen
if file has been truncated on the host (while it is still mapped inside
the guest). 

I had tried to summarize the problem in this presentation in the section
"KVM Page fault error handling".

https://kvm-forum.qemu.org/2020/KVMForum2020_APF.pdf

This is a tricky problem to handle. Once this gets handled, it becomes
safer to use DAX with virtiofs. Otherwise you can't share the filesystem
with other guests in DAX mode and use cases are limited.

And then there are challenges at QEMU level. virtiofsd needs additional
vhost-user commands to implement DAX and these never went upstream in
QEMU. I hope these challenges are sorted at some point of time.

I think virtiofs DAX is a very cool piece of technology. I would not like
to deprecate it. It has its own problems and challenges and once we
are able to solve these, it might see wider usage/adoption.

Thanks
Vivek