Re: [RFC][PATCH] x86/mm: Sync all vmalloc mappings before text_poke()

[Peter Zijlstra]

On Wed, Apr 29, 2020 at 10:32:16AM -0400, Steven Rostedt wrote:
> On Wed, 29 Apr 2020 16:10:04 +0200
> Joerg Roedel <jroedel@xxxxxxx> wrote:
> 
> > On Wed, Apr 29, 2020 at 10:07:31AM -0400, Steven Rostedt wrote:
> > > Talking with Mathieu about this on IRC, he pointed out that my code does
> > > have a vzalloc() that is called:
> > > 
> > > in trace_pid_write()
> > > 
> > > 	pid_list->pids = vzalloc((pid_list->pid_max + 7) >> 3);
> > > 
> > > This is done when -P1,2 is on the trace-cmd command line.  
> > 
> > Yeah, I was guessing something like this, init_mm has a mapping which
> > poking_mm has not. I currently try to reproduce this on one of my
> > machines.
> 
> Note, in use_temporary_mm(poking_mm), poking_mm only contains the page that

poking_mm should contain the full kernel map, the userspace part is 2
pages in a random location to alias two kernel pages.

So a vmalloc fault _should_ work.

> needs to be updated. But will get added to per_cpu(cpu_tlbstate.loaded_mm),
> which appears to be the one that's not updated.
> 
> -- Steve