On Sat, Feb 25, 2023 at 7:27 PM Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> Will bisect further, but adding Liam to the participants because it's
> now narrowed down to his changes.
Ok, it's commit 2286a6914c77 ("mm: change mprotect_fixup to vma iterator")
It was entirely consistent, and bisected right to that all the way
from my current git tip.
Without that commit, google-chrome works fine.
With that commit, I get "Aww snap" and a
traps: ThreadPoolForeg[4337] trap invalid opcode ip:55d5542363ee
sp:7fa5e04f1f80 error:0 in chrome[55d5537d3000+a14c000]
message in the kernel dumps (and core dump noise in journalctl).
The commit before is fine.
Sadly, it doesn't revert cleanly on my current top-of-tree (or even
_remotely_ cleanly_ because of all the other vma changes), so I can't
test just reverting that on the current state.
Also, it's not like I can debug google-chrome very much. It presumably
does complex vma's and unusual mprotect() stuff to trigger this, when
nothing else seems to care.
Liam?
Linus
---
2286a6914c776ec34cd97e4573b1466d055cb9de is the first bad commit
commit 2286a6914c776ec34cd97e4573b1466d055cb9de
Author: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx>
Date: Fri Jan 20 11:26:18 2023 -0500
mm: change mprotect_fixup to vma iterator
Use the vma iterator so that the iterator can be invalidated or updated to
avoid each caller doing so.
Link: https://lkml.kernel.org/r/20230120162650.984577-18-Liam.Howlett@xxxxxxxxxx
Signed-off-by: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
fs/exec.c | 5 ++++-
include/linux/mm.h | 6 +++---
mm/mprotect.c | 47 ++++++++++++++++++++++-------------------------
3 files changed, 29 insertions(+), 29 deletions(-)
