On Sunday 21 October 2007 15:10, Eric W. Biederman wrote:
Yes it does. It is exactly breaking the coherency between block
device and filesystem metadata coherency that Andrew cared about.
Whether or not that matters, that is a much bigger conceptual
change than simply using slightly more (reclaimable) memory in
some situations that my patch does.
If you want to try convincing people to break that coherency,
fine, but it has to be done consistently and everywhere rather than
for a special case in rd.c.
No, it doesn't. A real block device driver does have its own
buffer cache as it's backing store. It doesn't know about
readpage or writepage or set_page_dirty or buffers or pagecache.
What's the considered rewrite here? The rewrite I posted is the
only one so far that's come up that I would consider [worthy],
while these patches are just more of the same wrongness.
It fixes the bug in question, but not because it makes any
fundamental improvement to the conceptual ickyness of rd.c. I
don't know what you mean about attempting to reconcile the
incompatible [stuff]?
Using radix_tree_insert instead of add_to_page_cache is hardly
complicated. If anything it is simpler because it isn't actually
confusing the issues and it is much better fit with real block
device drivers, which is actually the thing that is familiar to
block device driver maintainers.
My opinion is that it is a much bigger behavioural change because
it results in incoherent buffer cache / blockdev cache, and it
results in code which is still ugly and conceptually wrong.
-
