lkml.org 
[lkml]   [2021]   [Nov]   [16]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [PATCH] mm/hmm.c: Allow VM_MIXEDMAP to work with hmm_range_fault
Date
On Tuesday, 16 November 2021 3:48:42 PM AEDT Andrew Morton wrote:
> On Mon, 8 Nov 2021 18:58:01 +1100 Alistair Popple <apopple@nvidia.com> wrote:
>
> > On Friday, 5 November 2021 11:25:57 PM AEDT Jason Gunthorpe wrote:
> > > On Thu, Nov 04, 2021 at 12:20:01PM +1100, Alistair Popple wrote:
> > > > hmm_range_fault() can be used instead of get_user_pages() for devices
> > > > which allow faulting however unlike get_user_pages() it will return an
> > > > error when used on a VM_MIXEDMAP range.
> > > >
> > > > To make hmm_range_fault() more closely match get_user_pages() remove
> > > > this restriction. This requires dealing with the !ARCH_HAS_PTE_SPECIAL
> > > > case in hmm_vma_handle_pte(). Rather than replicating the logic of
> > > > vm_normal_page() call it directly and do a check for the zero pfn
> > > > similar to what get_user_pages() currently does.
> > > >
> > > > Also add a test to hmm selftest to verify functionality.
> > >
> > > Please add a fixes line
> >
> > This has been the case since hmm_vma_get_pfns() was first introduced. Therefore
> > the fixes line is:
> >
> > Fixes: da4c3c735ea4 ("mm/hmm/mirror: helper to snapshot CPU page table")
> >
> > Should I send a v2 adding that tag or does Andrew normally pick these up along
> > with the Reviewed-by?
>
> I got it.
>
> It needed a bit of rework due to the newly-added !pte_devmap() change
> in hmm_vma_handle_pte(). Please check carefully?

Thanks. The rework looks good to me - changing pte_special() to
vm_insert_page() does not change the logic of the !pte_devmap() change. In
either case the value of pte_special() or vm_insert_page() is ignored when
pte_devmap()==True as intended by the !pte_devmap() change.

For good measure I rerun the HMM selftest as well.

> From: Alistair Popple <apopple@nvidia.com>
> Subject: mm/hmm.c: Allow VM_MIXEDMAP to work with hmm_range_fault
>
> hmm_range_fault() can be used instead of get_user_pages() for devices
> which allow faulting however unlike get_user_pages() it will return an
> error when used on a VM_MIXEDMAP range.
>
> To make hmm_range_fault() more closely match get_user_pages() remove
> this restriction. This requires dealing with the !ARCH_HAS_PTE_SPECIAL
> case in hmm_vma_handle_pte(). Rather than replicating the logic of
> vm_normal_page() call it directly and do a check for the zero pfn
> similar to what get_user_pages() currently does.
>
> Also add a test to hmm selftest to verify functionality.
>
> Link: https://lkml.kernel.org/r/20211104012001.2555676-1-apopple@nvidia.com
> Fixes: da4c3c735ea4 ("mm/hmm/mirror: helper to snapshot CPU page table")
> Signed-off-by: Alistair Popple <apopple@nvidia.com>
> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
> Cc: Jerome Glisse <jglisse@redhat.com>
> Cc: John Hubbard <jhubbard@nvidia.com>
> Cc: Zi Yan <ziy@nvidia.com>
> Cc: Ralph Campbell <rcampbell@nvidia.com>
> Cc: Felix Kuehling <Felix.Kuehling@amd.com>
> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
> ---
>
> lib/test_hmm.c | 24 +++++++++++++
> mm/hmm.c | 5 +-
> tools/testing/selftests/vm/hmm-tests.c | 42 +++++++++++++++++++++++
> 3 files changed, 69 insertions(+), 2 deletions(-)
>
> --- a/lib/test_hmm.c~mm-hmmc-allow-vm_mixedmap-to-work-with-hmm_range_fault
> +++ a/lib/test_hmm.c
> @@ -1086,9 +1086,33 @@ static long dmirror_fops_unlocked_ioctl(
> return 0;
> }
>
> +static int dmirror_fops_mmap(struct file *file, struct vm_area_struct *vma)
> +{
> + unsigned long addr;
> +
> + for (addr = vma->vm_start; addr < vma->vm_end; addr += PAGE_SIZE) {
> + struct page *page;
> + int ret;
> +
> + page = alloc_page(GFP_KERNEL | __GFP_ZERO);
> + if (!page)
> + return -ENOMEM;
> +
> + ret = vm_insert_page(vma, addr, page);
> + if (ret) {
> + __free_page(page);
> + return ret;
> + }
> + put_page(page);
> + }
> +
> + return 0;
> +}
> +
> static const struct file_operations dmirror_fops = {
> .open = dmirror_fops_open,
> .release = dmirror_fops_release,
> + .mmap = dmirror_fops_mmap,
> .unlocked_ioctl = dmirror_fops_unlocked_ioctl,
> .llseek = default_llseek,
> .owner = THIS_MODULE,
> --- a/mm/hmm.c~mm-hmmc-allow-vm_mixedmap-to-work-with-hmm_range_fault
> +++ a/mm/hmm.c
> @@ -300,7 +300,8 @@ static int hmm_vma_handle_pte(struct mm_
> * Since each architecture defines a struct page for the zero page, just
> * fall through and treat it like a normal page.
> */
> - if (pte_special(pte) && !pte_devmap(pte) &&
> + if (!vm_normal_page(walk->vma, addr, pte) &&
> + !pte_devmap(pte) &&
> !is_zero_pfn(pte_pfn(pte))) {
> if (hmm_pte_need_fault(hmm_vma_walk, pfn_req_flags, 0)) {
> pte_unmap(ptep);
> @@ -518,7 +519,7 @@ static int hmm_vma_walk_test(unsigned lo
> struct hmm_range *range = hmm_vma_walk->range;
> struct vm_area_struct *vma = walk->vma;
>
> - if (!(vma->vm_flags & (VM_IO | VM_PFNMAP | VM_MIXEDMAP)) &&
> + if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)) &&
> vma->vm_flags & VM_READ)
> return 0;
>
> --- a/tools/testing/selftests/vm/hmm-tests.c~mm-hmmc-allow-vm_mixedmap-to-work-with-hmm_range_fault
> +++ a/tools/testing/selftests/vm/hmm-tests.c
> @@ -1251,6 +1251,48 @@ TEST_F(hmm, anon_teardown)
> /*
> * Test memory snapshot without faulting in pages accessed by the device.
> */
> +TEST_F(hmm, mixedmap)
> +{
> + struct hmm_buffer *buffer;
> + unsigned long npages;
> + unsigned long size;
> + unsigned char *m;
> + int ret;
> +
> + npages = 1;
> + size = npages << self->page_shift;
> +
> + buffer = malloc(sizeof(*buffer));
> + ASSERT_NE(buffer, NULL);
> +
> + buffer->fd = -1;
> + buffer->size = size;
> + buffer->mirror = malloc(npages);
> + ASSERT_NE(buffer->mirror, NULL);
> +
> +
> + /* Reserve a range of addresses. */
> + buffer->ptr = mmap(NULL, size,
> + PROT_READ | PROT_WRITE,
> + MAP_PRIVATE,
> + self->fd, 0);
> + ASSERT_NE(buffer->ptr, MAP_FAILED);
> +
> + /* Simulate a device snapshotting CPU pagetables. */
> + ret = hmm_dmirror_cmd(self->fd, HMM_DMIRROR_SNAPSHOT, buffer, npages);
> + ASSERT_EQ(ret, 0);
> + ASSERT_EQ(buffer->cpages, npages);
> +
> + /* Check what the device saw. */
> + m = buffer->mirror;
> + ASSERT_EQ(m[0], HMM_DMIRROR_PROT_READ);
> +
> + hmm_buffer_free(buffer);
> +}
> +
> +/*
> + * Test memory snapshot without faulting in pages accessed by the device.
> + */
> TEST_F(hmm2, snapshot)
> {
> struct hmm_buffer *buffer;
> _
>
>
>




\
 
 \ /
  Last update: 2021-11-16 08:16    [W:0.190 / U:0.124 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site