News: 0001576373

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

Linux 6.18 Will Further Complicate Non-GPL Out-Of-Tree File-Systems

([Linux Storage] 5 Hours Ago write_cache_pages Removed)


Out-of-tree file-system drivers not licensed/compatible with the GPL will have a new obstacle to deal with come time for [1]Linux 6.18 later this year.

A patch queued up this week in advance of the Linux 6.18 merge window opening removes write_cache_pages. In turn this will cause issues for non-GPL out-of-tree file-systems for writing dirty data from the page cache. After the NTFS3 and Bcachefs in-tree users of the iterator were moved off of it, for Linux 6.18 the "write_cache_pages" will be removed that is depended upon by out-of-tree, non-GPL file-systems.

[2]This patch from Christoph Hellwig is what's now in linux-next ahead of Linux 6.18 and kills off the write_cache_pages now that there are no longer any in-tree users.

The out-of-tree OpenZFS file-system is among the [3]users of write_cache_pages.

The Phoronix reader who pointed me to this pending conflict with out-of-tree file-system drivers commented in a message:

"With the removal of both file_operation.writepage and write_cache_pages() it is no longer possible for out-of-tree non-GPL filesystems to write dirty pages out of the page cache.

It might be possible to use non-RCU interfaces to walk the internal xarray to search for dirty pages to write but that would be a major layer violation. Not to mention performance would be severely impacted."

We'll see how the situation plays out in the coming weeks/months with the stable Linux 6.18 kernel due out in December.



[1] https://www.phoronix.com/search/Linux+6.18

[2] https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=9c5518f1bacf98b20c3ad0fa5873b4da92122ced

[3] https://github.com/openzfs/zfs/blob/d64711c20205b3ffa6ce1e0495d55207d6bbd80b/module/os/linux/zfs/zpl_file.c#L490



Alexmitter

sophisticles

nyanbinary

ahrs

OneTimeShot

rb777

touchdown

cb88

dibal

Once there was this conductor see, who had a bass problem. You see, during
a portion of Beethovan's Ninth Symphony in which there are no bass violin
parts, one of the bassists always passed a bottle of scotch around. So,
to remind himself that the basses usually required an extra cue towards the
end of the symphony, the conductor would fasten a piece of string around the
page of the score before the bass cue. As the basses grew more and more
inebriated, two of them fell asleep. The conductor grew quite nervous (he
was very concerned about the pitch) because it was the bottom of the ninth;
the score was tied and the basses were loaded with two out.