Recently we got a strange bug report that, one 32bit systems like armv6
or non-64bit x86, certain large btrfs can't be mounted.

It turns out that, since page->index is just unsigned long, and on 32bit
systemts, that can just be 32bit.

And when filesystems is utilizing any page offset over 4T, page->index
get truncated, causing various problems.

This is especially a big problem for btrfs, as btrfs uses its internal
address space, which is from 0 to U64_MAX, but still sometimes relies on
page->index, just like most filesystems.

If a metadata is at or beyond 4T boundary (which is not rare, even with
small btrfs, as btrfs can related its chunks to much higher bytenr than
device boundary), then page->index will be truncated and may even
conflicts with existing pages.

I'm wonder if this is a known problem, and if so is there any plan to fix?
If not a known one, does it mean we have to make page->index u64 to fix
it? (this is definitely not going to be easy)


Reply via email to