Inaccurate address attribution with `else if` (affects Show Condition As Inverted)

Start with the attached test.zip reproducer and inspect func1, comparing the decompiler's address attribution with the shown disassembly. Trace how the shared 00400000 address is used by Show Condition As Inverted and comments; done means the else-if is attributed to its relevant instructions and those actions target the intended condition.

Version and Platform (required):

• Binary Ninja Version: 5.1.8002-dev (9b330043)
• Edition: Commercial
• OS: macOS
• OS Version: 15.5 (24F74)
• CPU Architecture: ARM

Bug Description:

Possibly related to #7080 and #7175.

Here is the reproducer. For this report we're looking at func1 (func2 is for another report).

The initial decompilation I get is:

00400000    int64_t func1(int64_t arg1)
00400000        if (arg1 != 0)
00400010            extern3()
00400000        else if (extern1() != 0)
0040000c            extern2()
00400010            extern3()
00400010        
00400018        return extern4()

If I right-click the line containing else if (extern1() != 0) and click Show Condition As Inverted, then instead of inverting that condition, it inverts the previous if statement:

00400000    int64_t func1(int64_t arg1)
00400000        if (arg1 == 0)
00400008            if (extern1() != 0)
0040000c                extern2()
00400010                extern3()
00400000        else
00400010            extern3()
00400010        
00400018        return extern4()

This appears to be caused by the fact that the if and else if are both assigned the same address, 00400000. This doesn't really make sense. The relevant part of the disassembly is:

00400000  800000b5   cbnz    x0, 0x400010   <-- branch if arg1 != 0

00400004  13000094   bl      extern1        <-- call extern1()
00400008  600000b4   cbz     x0, 0x400014   <-- branch if return value == 0

The else if would be better attributed to 00400004-00400008.

(That said, my ideal behavior would be if the decompiler could distinguish two branches in decompiled output even if they did come from the same code address. Show Condition As Inverted is not critical functionality, but the same issue also affects comments, which never end up quite where I want them even after #2728.)