Date: Fri, 7 Apr 2023 13:41:47 -0400
On 4/7/23 1:24 PM, Phil Bouchard wrote:
>
>
> On 4/7/23 12:57, Tom Honermann wrote:
>>
>> Microsoft offers some useful and free tools for diagnosing such
>> problems. See GFlags and PageHeap
>> <https://learn.microsoft.com/en-us/windows-hardware/drivers/debugger/gflags-and-pageheap>
>> and the related references there for example.
>>
>> I recommend exercising more restraint with regard to the veracity of
>> the criticism you post to this list; it might just be the case that
>> you have more to learn.
>
> These tools or the Address Sanitizer don't work with CUDA so you
> require a better implementation of the STL or some run-time bounds
> checker.
They are, unsurprisingly, tied to use of heaps created by Windows. If
you are using heaps created in other ways, then, yes, you'll need to use
tools that work with those heaps. I don't see how that justifies the
criticism that you have been posting to this list.
>
> Oh wait, C++ Superset already integrates that at compile-time. Never
> mind.
Yet it didn't suffice to identify the issue in your own code. Interesting.
Tom.
>
>
> On 4/7/23 12:57, Tom Honermann wrote:
>>
>> Microsoft offers some useful and free tools for diagnosing such
>> problems. See GFlags and PageHeap
>> <https://learn.microsoft.com/en-us/windows-hardware/drivers/debugger/gflags-and-pageheap>
>> and the related references there for example.
>>
>> I recommend exercising more restraint with regard to the veracity of
>> the criticism you post to this list; it might just be the case that
>> you have more to learn.
>
> These tools or the Address Sanitizer don't work with CUDA so you
> require a better implementation of the STL or some run-time bounds
> checker.
They are, unsurprisingly, tied to use of heaps created by Windows. If
you are using heaps created in other ways, then, yes, you'll need to use
tools that work with those heaps. I don't see how that justifies the
criticism that you have been posting to this list.
>
> Oh wait, C++ Superset already integrates that at compile-time. Never
> mind.
Yet it didn't suffice to identify the issue in your own code. Interesting.
Tom.
Received on 2023-04-07 17:41:52