Re: [POC][RFC][PATCH 0/2] pstore/mm/x86: Add wildcard memmap to map pstore consistently

From
Steven Rostedt <rostedt@goodmis.org>
Date
2024-05-09 20:33:12
Also in
linux-hardening, linux-mm, lkml
Reply to
Mike Rapoport, 2024-05-09
On Thu, 9 May 2024 23:24:20 +0300
Mike Rapoport [off-list ref] wrote:
On Thu, May 09, 2024 at 01:31:22PM -0400, Steven Rostedt wrote:
quoted
On Thu, 9 May 2024 00:00:23 -0400
Steven Rostedt [off-list ref] wrote:
  
quoted
I tried this approach and it unfortunately picks a different physical
location every time :-(

So it is either adding to e820 tables or we create a new way to
allocate memory at early boot up.
  
Hmm, now I'm testing it more and it always seems to end up in the same
location. I'm not sure why it failed the first three times I tried it :-/  
If the kernel and the command line were the same, they shouldn't have. 
e820 translates to memblock the same way every boot and the early
allocations also do not change from boot to boot.

Could it be that three runs that failed had some differences in the kernel
parameters?
I wonder if KASLR caused it or not. But when I first added it to
another machine, it failed to get the same address on the second boot,
but was fine after that.

Could be just something with my setup. I'm going to backport this to
5.15 and test this on a Chromebook and see what happens there (as
that's the motivation behind this work).

-- Steve
Thread (34 messages) 34 messages, 7 authors, last reply 2024-05-09
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help