erofs-utils: use cmpsgn(x, y) for standardized large value comparsion

Previously, roundup(bb->buffers.off % EROFS_BLKSIZ, alignsize)
+ incr + extrasize is an unsigned 64bit value and sgn(x) didn't
work properly. Fix it.

Link: https://lore.kernel.org/r/20191015155025.13215-1-hsiangkao@aol.com
Fixes: b0ca535297b6 ("erofs-utils: support 64-bit internal buffer cache")
Reviewed-and-tested-by: Li Guifu <blucerlee@gmail.com>
Signed-off-by: Gao Xiang <hsiangkao@aol.com>
2 files changed