Re: [PATCH v4 0/1] Use HMM for ODP v4
From: Jason Gunthorpe
Date: Wed May 22 2019 - 15:25:09 EST
On Wed, May 22, 2019 at 01:48:52PM -0400, Jerome Glisse wrote:
> > > +long ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp,
> > > + struct hmm_range *range)
> > > {
> > > + struct device *device = umem_odp->umem.context->device->dma_device;
> > > + struct ib_ucontext_per_mm *per_mm = umem_odp->per_mm;
> > > struct ib_umem *umem = &umem_odp->umem;
> > > - struct task_struct *owning_process = NULL;
> > > - struct mm_struct *owning_mm = umem_odp->umem.owning_mm;
> > > - struct page **local_page_list = NULL;
> > > - u64 page_mask, off;
> > > - int j, k, ret = 0, start_idx, npages = 0, page_shift;
> > > - unsigned int flags = 0;
> > > - phys_addr_t p = 0;
> > > -
> > > - if (access_mask == 0)
> > > + struct mm_struct *mm = per_mm->mm;
> > > + unsigned long idx, npages;
> > > + long ret;
> > > +
> > > + if (mm == NULL)
> > > + return -ENOENT;
> > > +
> > > + /* Only drivers with invalidate support can use this function. */
> > > + if (!umem->context->invalidate_range)
> > > return -EINVAL;
> > >
> > > - if (user_virt < ib_umem_start(umem) ||
> > > - user_virt + bcnt > ib_umem_end(umem))
> > > - return -EFAULT;
> > > + /* Sanity checks. */
> > > + if (range->default_flags == 0)
> > > + return -EINVAL;
> > >
> > > - local_page_list = (struct page **)__get_free_page(GFP_KERNEL);
> > > - if (!local_page_list)
> > > - return -ENOMEM;
> > > + if (range->start < ib_umem_start(umem) ||
> > > + range->end > ib_umem_end(umem))
> > > + return -EINVAL;
> > >
> > > - page_shift = umem->page_shift;
> > > - page_mask = ~(BIT(page_shift) - 1);
> > > - off = user_virt & (~page_mask);
> > > - user_virt = user_virt & page_mask;
> > > - bcnt += off; /* Charge for the first page offset as well. */
> > > + idx = (range->start - ib_umem_start(umem)) >> umem->page_shift;
> >
> > Is this math OK? What is supposed to happen if the range->start is not
> > page aligned to the internal page size?
>
> range->start is align on 1 << page_shift boundary within pagefault_mr
> thus the above math is ok. We can add a BUG_ON() and comments if you
> want.
OK
> > > + range->pfns = &umem_odp->pfns[idx];
> > > + range->pfn_shift = ODP_FLAGS_BITS;
> > > + range->values = odp_hmm_values;
> > > + range->flags = odp_hmm_flags;
> > >
> > > /*
> > > - * owning_process is allowed to be NULL, this means somehow the mm is
> > > - * existing beyond the lifetime of the originating process.. Presumably
> > > - * mmget_not_zero will fail in this case.
> > > + * If mm is dying just bail out early without trying to take mmap_sem.
> > > + * Note that this might race with mm destruction but that is fine the
> > > + * is properly refcounted so are all HMM structure.
> > > */
> > > - owning_process = get_pid_task(umem_odp->per_mm->tgid, PIDTYPE_PID);
> > > - if (!owning_process || !mmget_not_zero(owning_mm)) {
> >
> > But we are not in a HMM context here, and per_mm is not a HMM
> > structure.
> >
> > So why is mm suddenly guarenteed valid? It was a bug report that
> > triggered the race the mmget_not_zero is fixing, so I need a better
> > explanation why it is now safe. From what I see the hmm_range_fault
> > is doing stuff like find_vma without an active mmget??
>
> So the mm struct can not go away as long as we hold a reference on
> the hmm struct and we hold a reference on it through both hmm_mirror
> and hmm_range struct. So struct mm can not go away and thus it is
> safe to try to take its mmap_sem.
This was always true here, though, so long as the umem_odp exists the
the mm has a grab on it. But a grab is not a get..
The point here was the old code needed an mmget() in order to do
get_user_pages_remote()
If hmm does not need an external mmget() then fine, we delete this
stuff and rely on hmm.
But I don't think that is true as we have:
CPU 0 CPU1
mmput()
__mmput()
exit_mmap()
down_read(&mm->mmap_sem);
hmm_range_dma_map(range, device,..
ret = hmm_range_fault(range, block);
if (hmm->mm == NULL || hmm->dead)
mmu_notifier_release()
hmm->dead = true
vma = find_vma(hmm->mm, start);
.. rb traversal .. while (vma) remove_vma()
*goes boom*
I think this is violating the basic constraint of the mm by acting on
a mm's VMA's without holding a mmget() to prevent concurrent
destruction.
In other words, mmput() destruction does not respect the mmap_sem - so
holding the mmap sem alone is not enough locking.
The unlucked hmm->dead simply can't save this. Frankly every time I
look a struct with 'dead' in it, I find races like this.
Thus we should put the mmget_notzero back in.
I saw some other funky looking stuff in hmm as well..
> Hence it is safe to take mmap_sem and it is safe to call in hmm, if
> mm have been kill it will return EFAULT and this will propagate to
> RDMA.
> As per_mm i removed the per_mm->mm = NULL from release so that it is
> always safe to use that field even in face of racing mm "killing".
Yes, that certainly wasn't good.
> > > - * An array of the pages included in the on-demand paging umem.
> > > - * Indices of pages that are currently not mapped into the device will
> > > - * contain NULL.
> > > + * An array of the pages included in the on-demand paging umem. Indices
> > > + * of pages that are currently not mapped into the device will contain
> > > + * 0.
> > > */
> > > - struct page **page_list;
> > > + uint64_t *pfns;
> >
> > Are these actually pfns, or are they mangled with some shift? (what is range->pfn_shift?)
>
> They are not pfns they have flags (hence range->pfn_shift) at the
> bottoms i just do not have a better name for this.
I think you need to have a better name then
Jason