Integrate the bsdiff code into imgdiff so we can just call it rather
than writing temp files and invoking bsdiff as a subprocess. When
computing multiple patches from the same source file, we can also save
time by doing some bsdiff preprocessing on the source file only once.
(Note this renames the existing 'bsdiff.c' -- which actually does
patching in applypatch -- to 'bspatch.c', and introduces a new
bsdiff.c linked into imgdiff.)
Make some changes needed to applypatch in order to store the recovery
image in the system partition as a binary patch relative to the boot
image:
- make applypatch use shared libraries, so it's smaller. It will
need to be on the main system so it can install the recovery
image. Make an applypatch_static binary for use in recovery
packages (still needed for updating cupcake devices to donut).
- output the results of patching to an in-memory buffer and write
that to the partition; there's no convenient /tmp for us to us.
(This should be basically a no-op in recovery, since /tmp is a
ramdisk anyway.)
Images (like boot and recovery) consist of large sections of gzipped
data interspersed with other data. To do effective binary patching of
these files, we need to apply patches to the gzipped parts in
'uncompressed space', that is, we decompress, apply a patch, then
recompress to obtain the desired output.
This change defines a new format with these patches, which is
basically a description of how the source and target files are to be
divided up into chunks and a bsdiff patch for each chunk. We add a
new host executable, "imgdiff", for generating these patches from
source and target images, and add support in applypatch for
recognizing this format and applying it on the device.
