When reading a sized Little Endian uint from a buffer, we just need to read
N bytes from the buffer and place them at the beginning of a new 8-byte buffer,
(as if it were a guint64) and finally convert from LE to HE.
But when reading a sized Big Endian uint from a buffer, the N bytes that we read from the beginning of the buffer need to be placed at the *end* of the new 8-byte buffer before converting from BE to HE.
Same when writing sized Big Endian uints.
Created attachment 79897 [details] [review]
Patch for the issue.
Patch can be applied on top of the one for bug 64987 (issue actually found with the updated tests in that patch).
Patch pushed after testing ok in a mips BE system.