[boaks]

No description provided.

[obgm]

In general, these changes address the OOB read issue. There are two pitfalls regarding the break condition for the for loops that decrease i by 2 due to the now unsigned data type.

Coding style: This PR mixes uint16_t (the standardized type from stdint.h and uint16 (the local typedef for compatibility with platforms that do not have the standard C library). For consistency, I suggest using uint16 throughout the code.

[boaks]

I suggest using uint16 throughout the code.

typedef unsigned char uint16[2];

   uint16 i = 0;
   i++;

if we use the uint16, arithmetic will fail. Those types are only usable for sizeof().

I will add the checks for odd lengths.

[obgm]

typedef unsigned char uint16[2];

if we use the uint16, arithmetic will fail. Those types are only usable for sizeof().

Hm, a quick check shows 2 as a result for sizeof(i) as well as sizeof(uint16) with this typedef.

[boaks]

I suggest using uint16 throughout the code.

if uint16 is used for i, the compiler reports errors. There I use uint16_t.

[mrdeep1]

Yes, uint16 is an array of 2 bytes and so cannot be a single number.

[obgm]

Very valid point.