[PATCH] btrfs-progs: doc: add more explanation on subapge limits

[PATCH] btrfs-progs: doc: add more explanation on subapge limits

[Qu Wenruo]

The current subpage support in v5.18 has several limits, the most
obvious ones are:

- Only support 64KiB page size
- No RAID56 support

The supports are already queued for v5.19.

And some minor ones:

- No inline extent write support
  Read is always supported.
  Subpage mount will always just act as "max_inline=0".

- Compression write is only for page aligned range.
  Read is always supported, no matter the alignment.

- Extra memory usage for scrub
  Patchset is hanging there for a while though.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 Documentation/Subpage.rst | 59 +++++++++++++++++++++++++++++++++++++--
 1 file changed, 56 insertions(+), 3 deletions(-)

diff --git a/Documentation/Subpage.rst b/Documentation/Subpage.rst
index 0aadf3c00b98..1f56b35f2e04 100644
--- a/Documentation/Subpage.rst
+++ b/Documentation/Subpage.rst
@@ -6,6 +6,59 @@ using a filesystem that has different size of data block size (*sectorsize*)
 and the host CPU page size. For easier implementation the support was limited
 to the exactly same size of the block and page. On x86_64 this is typically
 4KiB, but there are other architectures commonly used that make use of larger
-pages, like 64KiB on 64bit ARM or PowerPC. A filesystem created on one cannot
-be mounted on the other.  The subpage support is still work in progress in 5.18
-but the support is incrementally added with each release.
+pages, like 64KiB on 64bit ARM or PowerPC.  This means filesystems created
+with 64KiB sector size can not be mounted on system with 4KiB page size.
+
+While with subpage support, systems with 64KiB page size can create (still needs
+"-s 4k" option for mkfs.btrfs) and mount filesystems with 4KiB sectorsize,
+allowing us to push 4KiB sectorsize as default sectorsize for all platforms in the
+near-future.
+
+
+Requirements, limitations
+^^^^^^^^^^^^^^^^^^^^^^^^^
+
+The subpage support is initially added in v5.15, although it's still
+considered as experimental at the time of writing (v5.18), most features are
+already working without problems.
+
+End users can mount filesystems with 4KiB sectorsize and do their usual
+workload, while should not notice any obvious change, as long as the initial
+mount succeeded (there are cases btrfs will reject the subpage mount though).
+
+The following features has some limitations for subpage:
+
+- RAID56 support
+  This support is already queued for v5.19 cycle.
+  Any fs with RAID56 chunks will be rejected at mount time for now.
+
+- Support for page size other than 64KiB
+  The support for other page sizes (16KiB, 32KiB and more) are already queued
+  for v5.19 cycle.
+  Initially the subpage support is only for 64KiB support, but the design makes
+  it pretty easy to enable support for other page sizes.
+
+- No inline extent creation
+  This is an artificial limit, to prevent mixed inline and regular extents.
+
+  It's possible to create mixed inline and regular extents even with
+  non-subpage mount for certain corner cases, it's way easier to create such
+  mixed extents for subpage.
+
+  Thus max_inline mount option will be sliently ignored for subpage mounts,
+  and it always acts as "max_inline=0".
+
+- Compression write is limited to page aligned ranges
+  Compression write for subpage is introduced in v5.16, with the limitation
+  that only page aligned range can be compressed.
+  This limitation is due to how btrfs handles delayed allocation.
+
+- No support for v1 space cache
+  V1 space cache is considered deprecated, and we're defaulting to v2 cache
+  in btrfs-progs already.
+  The old v1 cache has quite some hard coded page size usage, and consider it
+  is already deprecated, we force v2 cache for subpage.
+
+- Slightly higher memory usage for scrub
+  This is due to how we allocate pages for scrub, and will be fixed in the coming
+  releases soon.
-- 
2.36.1

Re: [PATCH] btrfs-progs: doc: add more explanation on subapge limits

[David Sterba]

On Wed, May 18, 2022 at 01:37:50PM +0800, Qu Wenruo wrote:
> The current subpage support in v5.18 has several limits, the most
> obvious ones are:
> 
> - Only support 64KiB page size
> - No RAID56 support
> 
> The supports are already queued for v5.19.
> 
> And some minor ones:
> 
> - No inline extent write support
>   Read is always supported.
>   Subpage mount will always just act as "max_inline=0".
> 
> - Compression write is only for page aligned range.
>   Read is always supported, no matter the alignment.
> 
> - Extra memory usage for scrub
>   Patchset is hanging there for a while though.
> 
> Signed-off-by: Qu Wenruo <wqu@suse.com>

Thank you, added to devel.