Re: [PATCH rdma-rc 3/3] IB/core: Fix ODP with IB_ACCESS_HUGETLB handling

From: Yishai Hadas
Date: Wed Jan 08 2020 - 10:24:24 EST


On 1/8/2020 2:56 PM, Geert Uytterhoeven wrote:
ÂÂÂÂHi Leon,

On Thu, 19 Dec 2019, Leon Romanovsky wrote:
From: Yishai Hadas <yishaih@xxxxxxxxxxxx>

As VMAs for a given range might not be available as part of the
registration phase in ODP, IB_ACCESS_HUGETLB/page_shift must be checked
as part of the page fault flow.

If the application didn't mmap the backed memory with huge pages or
released part of that hugepage area, an error will be set as part of the
page fault flow once be detected.

Fixes: 0008b84ea9af ("IB/umem: Add support to huge ODP")
Signed-off-by: Yishai Hadas <yishaih@xxxxxxxxxxxx>
Reviewed-by: Artemy Kovalyov <artemyko@xxxxxxxxxxxx>
Reviewed-by: Aviad Yehezkel <aviadye@xxxxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>

Thanks for your patch!

--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -241,22 +241,10 @@ struct ib_umem_odp *ib_umem_odp_get(struct ib_udata *udata, unsigned long addr,
ÂÂÂÂumem_odp->umem.owning_mm = mm = current->mm;
ÂÂÂÂumem_odp->notifier.ops = ops;

-ÂÂÂ umem_odp->page_shift = PAGE_SHIFT;
-ÂÂÂ if (access & IB_ACCESS_HUGETLB) {
-ÂÂÂÂÂÂÂ struct vm_area_struct *vma;
-ÂÂÂÂÂÂÂ struct hstate *h;
-
-ÂÂÂÂÂÂÂ down_read(&mm->mmap_sem);
-ÂÂÂÂÂÂÂ vma = find_vma(mm, ib_umem_start(umem_odp));
-ÂÂÂÂÂÂÂ if (!vma || !is_vm_hugetlb_page(vma)) {
-ÂÂÂÂÂÂÂÂÂÂÂ up_read(&mm->mmap_sem);
-ÂÂÂÂÂÂÂÂÂÂÂ ret = -EINVAL;
-ÂÂÂÂÂÂÂÂÂÂÂ goto err_free;
-ÂÂÂÂÂÂÂ }
-ÂÂÂÂÂÂÂ h = hstate_vma(vma);
-ÂÂÂÂÂÂÂ umem_odp->page_shift = huge_page_shift(h);
-ÂÂÂÂÂÂÂ up_read(&mm->mmap_sem);
-ÂÂÂ }
+ÂÂÂ if (access & IB_ACCESS_HUGETLB)
+ÂÂÂÂÂÂÂ umem_odp->page_shift = HPAGE_SHIFT;
+ÂÂÂ else
+ÂÂÂÂÂÂÂ umem_odp->page_shift = PAGE_SHIFT;

ÂÂÂÂumem_odp->tgid = get_task_pid(current->group_leader, PIDTYPE_PID);
ÂÂÂÂret = ib_init_umem_odp(umem_odp, ops);

noreply@xxxxxxxxxxxxxx reports for linux-next/m68k-allmodconfig/m68k:

ÂÂÂ drivers/infiniband/core/umem_odp.c:245:26: error: 'HPAGE_SHIFT' undeclared (first use in this function); did you mean 'PAGE_SHIFT'?

Should this depend on some HUGETLBFS option?


Thanks for pointing on,
We would expect to use #ifdef CONFIG_HUGETLB_PAGE as done in below kernel code [1] that also used HPAGE_SHIFT.

I'll send some patch to 'for-next' to handle it.

[1] https://elixir.bootlin.com/linux/v5.3-rc7/source/drivers/misc/sgi-gru/grufault.c#L183