Re: [PATCH][gdb] fix unsigned overflow in charset.c

[John Baldwin <jhb@FreeBSD.org>]

On 10/9/18 10:40 AM, Paul Koning wrote:
> 
> 
>> On Oct 9, 2018, at 1:31 PM, Pedro Alves <palves@redhat.com> wrote:
>>
>> On 10/09/2018 06:19 PM, Paul Koning wrote:
>>> This fixed an overflow in pointer arithmetic that crashes GDB on Mac OS.
>>
>> _unsigned_ overflow?  That isn't undefined.  Do we really want to trap
>> those?  I don't think GCC's version does that.
>>
>> From: 
>>  https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html#silencing-unsigned-integer-overflow
>> seems like there's a way to disable it.
>>
>> Thanks,
>> Pedro Alves
> 
> You're right, it was an LLVM build.  I know unsigned overflow is well defined with integers; is that true for pointers?
> 
> Given that GDB triggers this issue, should the GDB build do that LLVM workaround if LLVM is used to build it?  
> 
> But it seems simpler to use the proposed patch; clearly the intent is to back up a pointer by -(space_left) and doing that operation on a signed type seems like a logical thing to do, it makes the intended meaning clear.

I also ran into the same failure using LLVM's ubsan on FreeBSD but in a different
use of obstack_blank_fast().  If we wanted to fix this, I wonder if we'd instead
want to fix it centrally in obstack_blank_fast (e.g. by using a ptrdiff_t cast)
rather than fixing various consumers of the API.  That would be a change to
libiberty though, not just gdb.

-- 
John Baldwin