On Fri, Jun 13, 2014 at 12:23:50PM +0800, kbuild test robot wrote:
> tree: git://git.cmpxchg.org/linux-mmotm.git master
> head: a621774e0e7bbd9e8a024230af4704cc489bd40e
> commit: 5e2db44974d6ea928e257ac99031b5c5bf2ddaa4 [80/178] memcg: separate mem_cgroup_move_charge_pte_range()
> config: make ARCH=i386 allmodconfig
>
> All warnings:
>
> mm/memcontrol.c: In function 'mem_cgroup_count_precharge_pmd':
> mm/memcontrol.c:6674:3: error: 'skip' undeclared (first use in this function)
> skip->control = PTWALK_DOWN;
> ^
My apology, I did like below in "mm/pagewalk: replace mm_walk->skip with
more general mm_walk->control"
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index e98be7500549..46994e55e778 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -6670,9 +6670,8 @@ static int mem_cgroup_count_precharge_pmd(pmd_t *pmd,
if (get_mctgt_type_thp(vma, addr, *pmd, NULL) == MC_TARGET_PAGE)
mc.precharge += HPAGE_PMD_NR;
spin_unlock(ptl);
- /* don't call mem_cgroup_count_precharge_pte() */
- walk->skip = 1;
- }
+ } else
+ skip->control = PTWALK_DOWN;
return 0;
}
And it's fixed in "mm/pagewalk: move pmd_trans_huge_lock() from callbacks
to common code." So this was a wrong intermediate state in my series.
Ver.2 (I posted yesterday) should not have this problem.
Thanks,
Naoya Horiguchi
> mm/memcontrol.c:6674:3: note: each undeclared identifier is reported only once for each function it appears in
> mm/memcontrol.c: In function 'mem_cgroup_move_charge_pte':
> >> mm/memcontrol.c:6847:184: warning: value computed is not used [-Wunused-value]
> pte_offset_map(walk->pmd, addr & PMD_MASK);
> ^
> mm/memcontrol.c: In function 'mem_cgroup_move_charge_pmd':
> mm/memcontrol.c:6905:37: error: 'ptl' undeclared (first use in this function)
> if (pmd_trans_huge_lock(pmd, vma, &ptl) == 1) {
> ^
>
> vim +6847 mm/memcontrol.c
>
> 6668
> 6669 if (pmd_trans_huge_lock(pmd, vma, &ptl) == 1) {
> 6670 if (get_mctgt_type_thp(vma, addr, *pmd, NULL) == MC_TARGET_PAGE)
> 6671 mc.precharge += HPAGE_PMD_NR;
> 6672 spin_unlock(ptl);
> 6673 } else
> > 6674 skip->control = PTWALK_DOWN;
> 6675 return 0;
> 6676 }
> 6677
> 6678 static unsigned long mem_cgroup_count_precharge(struct mm_struct *mm)
> 6679 {
> 6680 unsigned long precharge;
> 6681 struct vm_area_struct *vma;
> 6682
> 6683 struct mm_walk mem_cgroup_count_precharge_walk = {
> 6684 .pmd_entry = mem_cgroup_count_precharge_pmd,
> 6685 .pte_entry = mem_cgroup_count_precharge_pte,
> 6686 .mm = mm,
> 6687 };
> 6688 down_read(&mm->mmap_sem);
> 6689 for (vma = mm->mmap; vma; vma = vma->vm_next)
> 6690 walk_page_vma(vma, &mem_cgroup_count_precharge_walk);
> 6691 up_read(&mm->mmap_sem);
> 6692
> 6693 precharge = mc.precharge;
> 6694 mc.precharge = 0;
> 6695
> 6696 return precharge;
> 6697 }
> 6698
> 6699 static int mem_cgroup_precharge_mc(struct mm_struct *mm)
> 6700 {
> 6701 unsigned long precharge = mem_cgroup_count_precharge(mm);
> 6702
> 6703 VM_BUG_ON(mc.moving_task);
> 6704 mc.moving_task = current;
> 6705 return mem_cgroup_do_precharge(precharge);
> 6706 }
> 6707
> 6708 /* cancels all extra charges on mc.from and mc.to, and wakes up all waiters. */
> 6709 static void __mem_cgroup_clear_mc(void)
> 6710 {
> 6711 struct mem_cgroup *from = mc.from;
> 6712 struct mem_cgroup *to = mc.to;
> 6713 int i;
> 6714
> 6715 /* we must uncharge all the leftover precharges from mc.to */
> 6716 if (mc.precharge) {
> 6717 __mem_cgroup_cancel_charge(mc.to, mc.precharge);
> 6718 mc.precharge = 0;
> 6719 }
> 6720 /*
> 6721 * we didn't uncharge from mc.from at mem_cgroup_move_account(), so
> 6722 * we must uncharge here.
> 6723 */
> 6724 if (mc.moved_charge) {
> 6725 __mem_cgroup_cancel_charge(mc.from, mc.moved_charge);
> 6726 mc.moved_charge = 0;
> 6727 }
> 6728 /* we must fixup refcnts and charges */
> 6729 if (mc.moved_swap) {
> 6730 /* uncharge swap account from the old cgroup */
> 6731 if (!mem_cgroup_is_root(mc.from))
> 6732 res_counter_uncharge(&mc.from->memsw,
> 6733 PAGE_SIZE * mc.moved_swap);
> 6734
> 6735 for (i = 0; i < mc.moved_swap; i++)
> 6736 css_put(&mc.from->css);
> 6737
> 6738 if (!mem_cgroup_is_root(mc.to)) {
> 6739 /*
> 6740 * we charged both to->res and to->memsw, so we should
> 6741 * uncharge to->res.
> 6742 */
> 6743 res_counter_uncharge(&mc.to->res,
> 6744 PAGE_SIZE * mc.moved_swap);
> 6745 }
> 6746 /* we've already done css_get(mc.to) */
> 6747 mc.moved_swap = 0;
> 6748 }
> 6749 memcg_oom_recover(from);
> 6750 memcg_oom_recover(to);
> 6751 wake_up_all(&mc.waitq);
> 6752 }
> 6753
> 6754 static void mem_cgroup_clear_mc(void)
> 6755 {
> 6756 struct mem_cgroup *from = mc.from;
> 6757
> 6758 /*
> 6759 * we must clear moving_task before waking up waiters at the end of
> 6760 * task migration.
> 6761 */
> 6762 mc.moving_task = NULL;
> 6763 __mem_cgroup_clear_mc();
> 6764 spin_lock(&mc.lock);
> 6765 mc.from = NULL;
> 6766 mc.to = NULL;
> 6767 spin_unlock(&mc.lock);
> 6768 mem_cgroup_end_move(from);
> 6769 }
> 6770
> 6771 static int mem_cgroup_can_attach(struct cgroup_subsys_state *css,
> 6772 struct cgroup_taskset *tset)
> 6773 {
> 6774 struct task_struct *p = cgroup_taskset_first(tset);
> 6775 int ret = 0;
> 6776 struct mem_cgroup *memcg = mem_cgroup_from_css(css);
> 6777 unsigned long move_charge_at_immigrate;
> 6778
> 6779 /*
> 6780 * We are now commited to this value whatever it is. Changes in this
> 6781 * tunable will only affect upcoming migrations, not the current one.
> 6782 * So we need to save it, and keep it going.
> 6783 */
> 6784 move_charge_at_immigrate = memcg->move_charge_at_immigrate;
> 6785 if (move_charge_at_immigrate) {
> 6786 struct mm_struct *mm;
> 6787 struct mem_cgroup *from = mem_cgroup_from_task(p);
> 6788
> 6789 VM_BUG_ON(from == memcg);
> 6790
> 6791 mm = get_task_mm(p);
> 6792 if (!mm)
> 6793 return 0;
> 6794 /* We move charges only when we move a owner of the mm */
> 6795 if (mm->owner == p) {
> 6796 VM_BUG_ON(mc.from);
> 6797 VM_BUG_ON(mc.to);
> 6798 VM_BUG_ON(mc.precharge);
> 6799 VM_BUG_ON(mc.moved_charge);
> 6800 VM_BUG_ON(mc.moved_swap);
> 6801 mem_cgroup_start_move(from);
> 6802 spin_lock(&mc.lock);
> 6803 mc.from = from;
> 6804 mc.to = memcg;
> 6805 mc.immigrate_flags = move_charge_at_immigrate;
> 6806 spin_unlock(&mc.lock);
> 6807 /* We set mc.moving_task later */
> 6808
> 6809 ret = mem_cgroup_precharge_mc(mm);
> 6810 if (ret)
> 6811 mem_cgroup_clear_mc();
> 6812 }
> 6813 mmput(mm);
> 6814 }
> 6815 return ret;
> 6816 }
> 6817
> 6818 static void mem_cgroup_cancel_attach(struct cgroup_subsys_state *css,
> 6819 struct cgroup_taskset *tset)
> 6820 {
> 6821 mem_cgroup_clear_mc();
> 6822 }
> 6823
> 6824 static int mem_cgroup_move_charge_pte(pte_t *pte,
> 6825 unsigned long addr, unsigned long end,
> 6826 struct mm_walk *walk)
> 6827 {
> 6828 int ret = 0;
> 6829 struct vm_area_struct *vma = walk->vma;
> 6830 union mc_target target;
> 6831 struct page *page;
> 6832 struct page_cgroup *pc;
> 6833 swp_entry_t ent;
> 6834
> 6835 retry:
> 6836 if (!mc.precharge) {
> 6837 pte_t *orig_pte = pte - ((addr & (PMD_SIZE - 1)) >> PAGE_SHIFT);
> 6838 pte_unmap_unlock(orig_pte, walk->ptl);
> 6839 cond_resched();
> 6840 /*
> 6841 * We have consumed all precharges we got in can_attach().
> 6842 * We try charge one by one, but don't do any additional
> 6843 * charges to mc.to if we have failed in charge once in attach()
> 6844 * phase.
> 6845 */
> 6846 ret = mem_cgroup_do_precharge(1);
> > 6847 pte_offset_map(walk->pmd, addr & PMD_MASK);
> 6848 spin_lock(walk->ptl);
> 6849 if (!ret)
> 6850 goto retry;
>
> ---
> 0-DAY kernel build testing backend Open Source Technology Center
> http://lists.01.org/mailman/listinfo/kbuild Intel Corporation
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@xxxxxxxxx. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>