Re: [PATCH 3/4] mm/secretmem: zeroize secret pages before kdump

["Jan Sebastian Götte" <contact@jaseg.de>]

On 8/11/26 17:14, David Hildenbrand (Arm) wrote:
> On 8/11/26 17:05, Jan Sebastian Götte wrote:
>> On 8/11/26 16:53, David Hildenbrand (Arm) wrote:
>>> On 7/31/26 18:27, Jan Sebastian Götte wrote:
>>>> Register a CRASH_ZEROIZE notifier that wipes secretmem folios. As a
>>>> result, when CONFIG_CRASH_ZEROIZE is set, secretmem areas will be
>>>> cleared before the kdump kernel is kexec'ed.
>>>
>>> Are you actually using secretmem in your use case? I heard some rumors that
>>> secretmem isn't used all that much in practice :)
>>
>> I'm aware nobody is really using secretmem. I decided to build this on top of
>> secretmem because it's an existing API that actually kind of fits the use case.
>> While I could have added a new API just for this (marking userspace memory as
>> wipe-on-crash), making it part of secretmem instead gives some additional
>> defense-in-depth protection for both users who only need wiping, and for users
>> who only want secretmem's additional protection against kernel memory read
>> primitives so I think everyone wins here.
> 
> What I was thinking when I looked at patch #3+#4: 50% of the users you add are
> irrelevant in practice.
> 
> Or are you saying that you are going to start using secretmem for your use case?

I'm in fact going to start using secretmem.