On 2021/9/24 22:43, Marciniszyn, Mike wrote:

quoted

On 9/22/21 23:45, Leon Romanovsky wrote:

quoted

Isn't kptr_restrict sysctl is for that?

Hi Leon,

After I sent my email I discovered the following commit: 5ead723a20e0
("lib/vsprintf: no_hash_pointers prints all addresses as unhashed"; v5.12).
I think that commit does what we need?

Thanks Bart,

I agree.

Jason, as to traces, I suspect we need to do the same %p thing there for existing code and any new work.

For situations for debugging in the wild, a command line arg can show the actual value.   I'm ok with that.

Mike

Can this patch which changes %llx to %p in infiniband hfi1 to avoid 
kernel pointer release be applied?


Thanks.


Guo