Fix integer overflow in PanasonicMakerNote::printAccelerometer

[kevinbackhouse]

Fixes: #2006

The subtraction in this code is getting flagged by OSS-Fuzz because it can overflow. It's obviously harmless because it only affects the value that gets printed, so this fix is just to make OSS-Fuzz happy.

This subtraction code is very strange though. According to the comment, it's a conversion from unsigned to signed. The comment doesn't specifically say, but it looks like the code is expecting the value to be a uint16_t. But it's weird because it's subtracting 0xffff, rather than 0x10000 which is what I would expect for two's complement arithmetic.

I have replaced the weird arithmetic with a simple static_cast, because that's what makes sense to me. Unfortunately, we do not have any test files that trigger this code.

[codecov]

Codecov Report

Merging #2007 (35f48ae) into main (fde8ed0) will increase coverage by 0.02%.
The diff coverage is 33.33%.

@@            Coverage Diff             @@
##             main    #2007      +/-   ##
==========================================
+ Coverage   61.16%   61.18%   +0.02%     
==========================================
  Files          96       96              
  Lines       19255    19252       -3     
  Branches     9862     9862              
==========================================
+ Hits        11777    11780       +3     
+ Misses       5135     5129       -6     
  Partials     2343     2343              

Impacted Files	Coverage Δ
src/panasonicmn_int.cpp	43.13% <33.33%> (+4.08%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update fde8ed0...35f48ae. Read the comment docs.

[clanmills]

Pass me the barf bag:

i = i - ((i & 0x8000) >> 15) * 0xffff;

I'm not clever enough to know what that's meant to do. I see it's repeated:

641 rmills@rmillsm1:~/gnu/github/exiv2/0.27-maintenance $ git blame src/panasonicmn_int.cpp | grep 0x8000
717d789c8 src/panasonicmn.cpp     (Andreas Huggel   2010-08-30 19:33:07 +0000 520)         TagInfo(0x8000, "MakerNoteVersion", N_("MakerNote Version"), N_("MakerNote version"), panasonicId, makerTags, undefined, -1, printExifVersion),
c41988539 src/panasonicmn.cpp     (nkbj             2014-09-08 03:57:56 +0000 697)         i = i - ((i & 0x8000) >> 15) * 0xffff;
c41988539 src/panasonicmn.cpp     (nkbj             2014-09-08 03:57:56 +0000 705)         i = i - ((i & 0x8000) >> 15) * 0xffff;
c41988539 src/panasonicmn.cpp     (nkbj             2014-09-08 03:57:56 +0000 719)         i = i - ((i & 0x8000) >> 15) * 0xffff;
642 rmills@rmillsm1:~/gnu/github/exiv2/0.27-maintenance $ 

[kmilos]

LGTM