That's a bit surprising (particularly the IDA case), and probably an indication that the ARM code/data separator in these disassemblers needs some work... because my experience with x86 IDA is that it is not going to think some address is the start of a function and disassemble it unless it actually finds a control flow instruction like call/jump leading to it, and it starts at the entry point when it does code/data separation so it shouldn't falsely identify data as code (indirect jumps, however, do tend to make it stumble, so code that doesn't get disassembled is the more common case.)

Another possibility could be blind reliance on debug information that's causing it to think that symbol is a function; strip the binary and try again.