Re: [PATCH] handle non-constant offsets in -Wstringop-overflow (PR 77608)

[Jeff Law <law at redhat dot com>]

On 11/18/2017 09:47 AM, Martin Sebor wrote:
> On 11/18/2017 12:53 AM, Jeff Law wrote:
>> On 11/17/2017 12:36 PM, Martin Sebor wrote:
>>> The attached patch enhances -Wstringop-overflow to detect more
>>> instances of buffer overflow at compile time by handling non-
>>> constant offsets into the destination object that are known to
>>> be in some range.  The solution could be improved by handling
>>> even more cases (e.g., anti-ranges or offsets relative to
>>> pointers beyond the beginning of an object) but it's a start.
> 
> I should have made it even more clear that this change affects
> just warnings, not the runtime protection with _FORTIFY_SOURCE.
> The latter is driven by the __builtin_object_size intrinsic and
> by the compute_builtin_object_size GCC function.
> 
> This patch does not change what __builtin_object_size returns.
Understood.  But I think that the documented behavior of b_o_s directly
impacts how this routine is supposed to work.

>> 
> -Wstringop-overflow (i.e., warnings only) always uses the largest
> object size for memxxx functions (type 0) and the smallest (type
> 2) for string functions by default.  This can be changed but I
> suspect no one does because (IME) few users understand what
> the other modes mean or what good using them vs the default might
> do.  It was probably a mistake to expose the mode via the option.
Then maybe only do the deeper analysis for max mode only?  As it stands
if someone were to call it in a min mode I think they're going to be
rather surprised by the answer.

jeff