Conversation

Vlastimil Babka

The mm-hotfixes pull request was merged yesterday, including my two fixes for two embarrassing bugs introduced also by myself several/many releases ago. If there's a rc8 instead of 6.8 final tomorrow, you know why...
3
0
5
@vbabka ha! Imagine doing some really embarrassing mistake that somebody fixes later that'd never happen to me *anxious cough*
0
0
1
Alright so that didn't result in rc8, so 6.8 is now final, the first kernel since 2.6.16 (March 2006) to have no choice of a slab allocator. And I need a new dead horse to beat.
5
3
15

@vbabka make THP work with PREEMPT_RT

1
0
1

@vbabka what CVE number is that?

1
0
1

@vbabka Maybe you could write some high-level docs about page allocation? We already have some, but it's a bit on the short side: https://docs.kernel.org/mm/page_allocation.html

0
0
1

@lkundrak @vbabka All Linux CNA Team members are currently busy processing earlier commits. Please hold, and your fixes will be assigned a CVE as soon as possible.

1
0
2

@ptesarik @vbabka assign a CVE number, then assign a CVE number for the CVE number

0
0
2

@vbabka obviously the logical conclusion of this work is to delete SLUB, too

1
0
1

@oleksandr @vbabka @osandov I see. Reducing the attack surface should be good for security. But can we then also remove the buddy allocator at some point in the future? The bootmem allocator looks much simpler in comparison…

0
2
1