Re: Racy manipulation of task_struct->flags in cgroups code causes hard to reproduce kernel panics
From: Kees Cook <hidden>
Date: 2014-09-20 17:15:52
Also in:
lkml
On Sat, Sep 20, 2014 at 7:30 AM, Peter Zijlstra [off-list ref] wrote:
On Sat, Sep 20, 2014 at 01:55:54PM +0800, Zefan Li wrote:quoted
We should make the updating of this flag atomic.quoted
/* Per-process atomic flags. */ #define PFA_NO_NEW_PRIVS 0x00000001 /* May not gain new privileges. */ +#define PFA_SPREAD_PAGE 0x00000002 /* Spread page cache over cpuset */ +#define PFA_SPREAD_SLAB 0x00000004 /* Spread some slab caches over cpuset */Ooh, I was not ware we had those.. /me checks where that came from. Hmm weird, while I did get that patch it had a seccomp prefix when landing in my inbox so I ignored it. However the commit has a sched prefix (which I would not have ignored). Dubious things happened here.
The series went through a lot of revisions, so it probably gained the sched prefix later in its life. Is there anything that needs changing about how this has been implemented? -Kees -- Kees Cook Chrome OS Security