| Date | Thu, 21 Aug 2014 21:00:20 +0200 | From | Oleg Nesterov <> | Subject | [PATCH 0/4] mempolicy: fix show_numa_map() vs exec() + do_set_mempolicy() race |
| |
On 08/20, Oleg Nesterov wrote: > > I belive that 9e7814404b77c3e8 "hold task->mempolicy while numa_maps scans" > was wrong and it doesn't fix the problem.
See 2/4.
> I'll try to send the fix tomorrow, could you review the initial preparations? > To me, these changes make sense as cleanups in any case, but I don't really > understand this code and this was only compile tested.
Untested again because I do not know how, please review ;) On top of "mempolicy: get_task_policy() cleanups/preparations" series.
I'll send the final cleanups later.
Oleg.
fs/proc/task_mmu.c | 33 ++++++----------------- include/linux/mempolicy.h | 5 ++- mm/mempolicy.c | 64 +++++++++++++++++++------------------------- 3 files changed, 40 insertions(+), 62 deletions(-)
|