On Fri, Dec 18, 2020 at 5:58 AM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
[..]
> 5) The DAX case which you made "work" with dev_access_enable() and
> dev_access_disable(), i.e. with yet another lazy approach of
> avoiding to change a handful of usage sites.
>
> The use cases are strictly context local which means the global
> magic is not used at all. Why does it exist in the first place?
>
> Aside of that this global thing would never work at all because the
> refcounting is per thread and not global.
>
> So that DAX use case is just a matter of:
>
> grant/revoke_access(DEV_PKS_KEY, READ/WRITE)
>
> which is effective for the current execution context and really
> wants to be a distinct READ/WRITE protection and not the magic
> global thing which just has on/off. All usage sites know whether
> they want to read or write.
I was tracking and nodding until this point. Yes, kill the global /
kmap() support, but if grant/revoke_access is not integrated behind
kmap_{local,atomic}() then it's not a "handful" of sites that need to
be instrumented it's 100s. Are you suggesting that "relaxed" mode
enforcement is a way to distribute the work of teaching driver writers
that they need to incorporate explicit grant/revoke-read/write in
addition to kmap? The entire reason PTE_DEVMAP exists was to allow
get_user_pages() for PMEM and not require every downstream-GUP code
path to specifically consider whether it was talking to PMEM or RAM
pages, and certainly not whether they were reading or writing to it.
