[DeviceSanitizer] bugfix for buffer logger type error

[yingcong-wu]

Buffer impl use char* pointer for Ptr and HostPtr, but that will confuse the logger and make it consider those as string and try to print them as string, which will result in SEGV.

[pbalcer]

Buffer impl use char* pointer for Ptr and HostPtr, but that will confuse the logger and make it consider those as string and try to print them as string, which will result in SEGV.

we should probably be using uint8_t instead of char if something is not a c-style string.

[yingcong-wu]

Buffer impl use char* pointer for Ptr and HostPtr, but that will confuse the logger and make it consider those as string and try to print them as string, which will result in SEGV.

we should probably be using uint8_t instead of char if something is not a c-style string.

That is a good idea. I will update this PR with a TODO.

[yingcong-wu]

Hi @intel/llvm-gatekeepers , the CI failure is not related and still fails for multiple rerun. I have created #19033 to track it. Could you please help merge this patch? Thanks.

[uditagarwal97]

Hi @intel/llvm-gatekeepers , the CI failure is not related and still fails for multiple rerun. I have created #19033 to track it. Could you please help merge this patch? Thanks.

Yes, the CUDA failure is unrelated, I see the same failure on other PRs as well. Thanks for creating a ticket.