erofs-utils: mkfs: speed up uncompressed data handling
Currently, it only writes one uncompressed block (typically 4 KiB) and
then tries to recompress the remaining part if the entire data was
compressed and proven to be incompressible.
This is wasteful of CPU resources and slow, as such incompressible data
could be skipped entirely.
The LZMA build time for large pclusters (e.g., 512K, 1M) can be greatly
reduced if there is a significant amount of uncompressed data.
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Link: https://lore.kernel.org/r/20241220143859.643175-2-hsiangkao@linux.alibaba.com