Re: [PATCH v2 1/2] tracing: ring-buffer: Have the ring buffer code do the vmap of physical memory

[Steven Rostedt]

On Mon, 31 Mar 2025 18:30:33 -0700
Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:

> On Mon, 31 Mar 2025 at 18:01, Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> >
> > Note, I believe that Linus brought up the issue that because this physical
> > memory is not currently part of the memory allocator (it's not aware of it
> > yet), that the getting struct page or a "pfn" for it may not be reliable.  
> 
> 'pfn' is always reliable.
> 
> The pfn ('page frame number') is literally just the physical address
> expressed in 'page units', ie just shifted down by the page shift.

Ah, for some reason I thought the pfn mapped directly to struct pages.

> 
> So pfn and phys_addr_t are interchangeable when it comes to mapping
> pages. The pfn is in fact often the preferred form, because on 32-bit
> architectures a pfn is 32-bit, but a phys_addr_t is often 64-bit and
> generates extra code.
> 
> I think 'pfn' was introduced as a name ong long ago because it was
> what the alpha architecture used in the VM documentation. It probably
> predates that too, but it's where I got it from, iirc.
> 

It is old, as I remember using it when I first started Linux kernel
development back in 1998. But my memory of it was it was also used as an
index into a struct page array. Which is why I was thinking it was somewhat
interchangeable with struct page. But that was a long time ago when I was
an embedded developer, but I've only been using struct page for my needs in
the last couple of decades.

-- Steve