From  issue 722979 :
a supersize diff shows 13kb even though libmonochrome.so grew by 16 kb

Maybe section breakdown would help?

readelf -S libmonochrome.so

Yep, we use that to verify that the section sizes from the linker map match what readelf -S says.

c#1 is actually wrong - the normalized APK size grew by 16 kb.

Going to leave this open for today while I address more size alerts, then I'll close this once I've verified there's nothing to do here!

Another discrepancy found in  issue 719391 .

8kb of main lib growth in relocations reported by resource_sizes, but supersize doesn't seem to catch any of it.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/ffd9b04d92c4a7052471a6a20686dbd95b53093d

commit ffd9b04d92c4a7052471a6a20686dbd95b53093d
Author: agrieve <agrieve@chromium.org>
Date: Thu Jun 08 19:34:45 2017

supersize: Show relocations in section totals

This should make it obvious when binary size growth is due to relocations.

BUG= 726151 

Review-Url: https://codereview.chromium.org/2924193004
Cr-Commit-Position: refs/heads/master@{#478055}

[modify] https://crrev.com/ffd9b04d92c4a7052471a6a20686dbd95b53093d/tools/binary_size/libsupersize/describe.py
[modify] https://crrev.com/ffd9b04d92c4a7052471a6a20686dbd95b53093d/tools/binary_size/libsupersize/testdata/Console.golden
[modify] https://crrev.com/ffd9b04d92c4a7052471a6a20686dbd95b53093d/tools/binary_size/libsupersize/testdata/Diff_Basic.golden
[modify] https://crrev.com/ffd9b04d92c4a7052471a6a20686dbd95b53093d/tools/binary_size/libsupersize/testdata/Diff_NullDiff.golden
[modify] https://crrev.com/ffd9b04d92c4a7052471a6a20686dbd95b53093d/tools/binary_size/libsupersize/testdata/FullDescription.golden