| Message ID | 1432483292-23109-1-git-send-email-jungseoklee85@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Sun, May 24, 2015 at 12:01 PM, Jungseok Lee <jungseoklee85@gmail.com> wrote: > A current implementation assumes thread_info address is always correctly > calculated via virt_to_page. It restricts a different approach, such as > thread_info allocation from vmalloc space. > > This patch, thus, introduces an independent function to calculate page > address from thread_info one. > > Suggested-by: Sungjinn Chung <barami97@gmail.com> > Signed-off-by: Jungseok Lee <jungseoklee85@gmail.com> > Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> > Cc: linux-arm-kernel@lists.infradead.org > --- > kernel/fork.c | 7 ++++++- > 1 file changed, 6 insertions(+), 1 deletion(-) I haven't receive a path [2/2] and haven't review whole patches. But this patch itself is OK to me. Acked-by: KOSAKI Motohiro <kosaki.motohiro@fujitsu.com>
On May 27, 2015, at 12:49 PM, KOSAKI Motohiro wrote: Hello, KOSAKI, > On Sun, May 24, 2015 at 12:01 PM, Jungseok Lee <jungseoklee85@gmail.com> wrote: >> A current implementation assumes thread_info address is always correctly >> calculated via virt_to_page. It restricts a different approach, such as >> thread_info allocation from vmalloc space. >> >> This patch, thus, introduces an independent function to calculate page >> address from thread_info one. >> >> Suggested-by: Sungjinn Chung <barami97@gmail.com> >> Signed-off-by: Jungseok Lee <jungseoklee85@gmail.com> >> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> >> Cc: linux-arm-kernel@lists.infradead.org >> --- >> kernel/fork.c | 7 ++++++- >> 1 file changed, 6 insertions(+), 1 deletion(-) > > I haven't receive a path [2/2] and haven't review whole patches. But > this patch itself is OK to me. > Acked-by: KOSAKI Motohiro <kosaki.motohiro@fujitsu.com> Thanks! I didn't add you to Cc list since [PATCH 2/2] is architecture specific. According to the feedbacks, it is needed to figure out fundamental solutions: 1) reduce stack size and 2) focus on a generic anti-fragmentation logic. Please refer to https://lkml.org/lkml/2015/5/24/121 for [PATCH 2/2]. Best Regards Jungseok Lee
diff --git a/kernel/fork.c b/kernel/fork.c index 03c1eaa..6300bbd 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -212,9 +212,14 @@ struct kmem_cache *vm_area_cachep; /* SLAB cache for mm_struct structures (tsk->mm) */ static struct kmem_cache *mm_cachep; +struct page * __weak arch_thread_info_to_page(struct thread_info *ti) +{ + return virt_to_page(ti); +} + static void account_kernel_stack(struct thread_info *ti, int account) { - struct zone *zone = page_zone(virt_to_page(ti)); + struct zone *zone = page_zone(arch_thread_info_to_page(ti)); mod_zone_page_state(zone, NR_KERNEL_STACK, account); }
A current implementation assumes thread_info address is always correctly calculated via virt_to_page. It restricts a different approach, such as thread_info allocation from vmalloc space. This patch, thus, introduces an independent function to calculate page address from thread_info one. Suggested-by: Sungjinn Chung <barami97@gmail.com> Signed-off-by: Jungseok Lee <jungseoklee85@gmail.com> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: linux-arm-kernel@lists.infradead.org --- kernel/fork.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)