See original thread at: https://lists.gnu.org/archive/html/bug-patch/2011-08/msg00004.html

Short summary - I encountered a case of an ambiguous patch generated with diff -p and containing 3 lines of context (the patch needed 6 lines before it became unambiguous), and where the patch file did not list the right line compared to the version of the file I was patching.

When passed through patch, the offset chosen was the minimal offset (the closest location that matched the context), even though this was NOT the function labeled in the @@ hunk header.  When passed through 'git am', the offset chosen was further away from the line number in the patch, but happened to match the function name in the @@ header.

Patch should be taught how to use any and all information in the @@ header as part of its disambiguation algorithm when determining an offset to apply to a context patch.