On 2020/9/8 上午8:56, Tom Rini wrote:
On Tue, Sep 08, 2020 at 08:26:27AM +0800, Qu Wenruo wrote:
On 2020/9/8 上午6:35, Tom Rini wrote:
On Wed, Jun 24, 2020 at 06:03:11PM +0200, Marek Behún wrote:
From: Qu Wenruo wqu@suse.com
This version of btrfs_file_read() has the following new features:
- Tries all mirrors
- More handling on unaligned size
- Better compressed extent handling The old implementation doesn't handle compressed extent with offset properly: we need to read out the whole compressed extent, then decompress the whole extent, and only then copy the requested part.
Signed-off-by: Qu Wenruo wqu@suse.com Reviewed-by: Marek Behún marek.behun@nic.cz
Note that this introduces a warning with LLVM-10: fs/btrfs/btrfs.c:246:6: warning: variable 'real_size' is used uninitialized whenever 'if' condition is false [-Wsometimes-uninitialized] if (!len) { ^~~~ fs/btrfs/btrfs.c:255:12: note: uninitialized use occurs here if (len > real_size - offset) ^~~~~~~~~ fs/btrfs/btrfs.c:246:2: note: remove the 'if' if its condition is always true if (!len) { ^~~~~~~~~~ fs/btrfs/btrfs.c:228:18: note: initialize the variable 'real_size' to silence this warning loff_t real_size; ^ = 0 1 warning generated.
and I have silenced as suggested. I'm not 100% happy with that, but leave fixing it here and in upstream btrfs-progs to the btrfs-experts.
My bad. The warning is correct, and since the code is U-boot specific, it doesn't affect kernel (using page) nor btrfs-progs (not really do file read with offset).
The fix is a little complex.
In fact we need to always call btrfs_size() to grab the real_size, and then modify @len using real_size, either using real_size directly, or do some basic calculation.
Ah, thanks. I'll fold in your changes.
BTW, I didn't see the btrfs rebuild work merged upstream. Is this planned or you just grab from some specific branch?
Yes, I'm testing them for -next right now.
That's awesome!
Thanks for your effort, Qu