On 2020/12/18 上午12:02, Josef Bacik wrote:
On 12/10/20 1:38 AM, Qu Wenruo wrote:
For subpage case, grab_extent_buffer_from_page() can't really get an
extent buffer just from btrfs_subpage.
Although we have btrfs_subpage::tree_block_bitmap, which can be used to
grab the bytenr of an existing extent buffer, and can then go radix tree
search to grab that existing eb.
However we are still doing radix tree insert check in
alloc_extent_buffer(), thus we don't really need to do the extra hassle,
just let alloc_extent_buffer() to handle existing eb in radix tree.
So for grab_extent_buffer_from_page(), just always return NULL for
subpage case.
This is fundamentally flawed. The extent buffer radix tree look up is
done _after_ the pages are init'ed. This is why there's that
complicated dance of checking for existing extent buffers attached to to
a page, because we can race at the initialization stage and attach an EB
to a page before it's in the radix tree. What you'll end up doing here
is overwriting your existing subpage stuff anytime there's a race, and
it'll end very badly. Thanks,
We have page lock preventing two eb getting the same page.
And btrfs_attach_subpage() won't overwrite the existing page::private,
thus it's safe.
Thanks,
Qu
Josef
