On Fri, May 26, 2023 at 2:22 PM Duan, Zhenzhong
<zhenzhong.d...@intel.com> wrote:
>
>
> >-----Original Message-----
> >From: Peter Xu <pet...@redhat.com>
> >Sent: Thursday, May 25, 2023 9:54 PM
> >Subject: Re: [PATCH] intel_iommu: Optimize out some unnecessary UNMAP
> >calls
> >
> >On Thu, May 25, 2023 at 11:29:34AM +0000, Duan, Zhenzhong wrote:
> >> Hi Peter,
> >>
> >> See inline.
> >> >-----Original Message-----
> >> >From: Peter Xu <pet...@redhat.com>
> >> >Sent: Thursday, May 25, 2023 12:59 AM
> >> >Subject: Re: [PATCH] intel_iommu: Optimize out some unnecessary UNMAP
> >> >calls
> >> >
> >> >Hi, Zhenzhong,
> >> >
> >> >On Tue, May 23, 2023 at 04:07:02PM +0800, Zhenzhong Duan wrote:
> >> >> Commit 63b88968f1 ("intel-iommu: rework the page walk logic") adds
> >> >> logic to record mapped IOVA ranges so we only need to send MAP or
> >> >> UNMAP when necessary. But there are still a few corner cases of
> >> >unnecessary UNMAP.
> >> >>
> >> >> One is address space switch. During switching to iommu address
> >> >> space, all the original mappings have been dropped by VFIO memory
> >> >> listener, we don't need to unmap again in replay. The other is
> >> >> invalidation, we only need to unmap when there are recorded mapped
> >> >> IOVA ranges, presuming most of OSes allocating IOVA range
> >> >> continuously, ex. on x86, linux sets up mapping from 0xffffffff
> >downwards.
> >> >>
> >> >> Signed-off-by: Zhenzhong Duan <zhenzhong.d...@intel.com>
> >> >> ---
> >> >> Tested on x86 with a net card passed or hotpluged to kvm guest,
> >> >> ping/ssh pass.
> >> >
> >> >Since this is a performance related patch, do you have any number to
> >> >show the effect?
> >>
> >> I straced the time of UNMAP ioctl, its time is 0.000014us and we have
> >> 28 ioctl() due to the two notifiers in x86 are split into power of 2
> >> pieces.
> >>
> >> ioctl(48, VFIO_DEVICE_QUERY_GFX_PLANE or VFIO_IOMMU_UNMAP_DMA,
> >> 0x7ffffd5c42f0) = 0 <0.000014>
> >
> >Could you add some information like this into the commit message when
> >repost? E.g. UNMAP was xxx sec before, and this patch reduces it to yyy.
> Sure, will do.
>
> >
> >>
> >> >
> >> >>
> >> >> hw/i386/intel_iommu.c | 31 ++++++++++++++-----------------
> >> >> 1 file changed, 14 insertions(+), 17 deletions(-)
> >> >>
> >> >> diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c index
> >> >> 94d52f4205d2..6afd6428aaaa 100644
> >> >> --- a/hw/i386/intel_iommu.c
> >> >> +++ b/hw/i386/intel_iommu.c
> >> >> @@ -3743,6 +3743,7 @@ static void
> >> >vtd_address_space_unmap(VTDAddressSpace *as, IOMMUNotifier *n)
> >> >> hwaddr start = n->start;
> >> >> hwaddr end = n->end;
> >> >> IntelIOMMUState *s = as->iommu_state;
> >> >> + IOMMUTLBEvent event;
> >> >> DMAMap map;
> >> >>
> >> >> /*
> >> >> @@ -3762,22 +3763,25 @@ static void
> >> >vtd_address_space_unmap(VTDAddressSpace *as, IOMMUNotifier *n)
> >> >> assert(start <= end);
> >> >> size = remain = end - start + 1;
> >> >>
> >> >> + event.type = IOMMU_NOTIFIER_UNMAP;
> >> >> + event.entry.target_as = &address_space_memory;
> >> >> + event.entry.perm = IOMMU_NONE;
> >> >> + /* This field is meaningless for unmap */
> >> >> + event.entry.translated_addr = 0;
> >> >> +
> >> >> while (remain >= VTD_PAGE_SIZE) {
> >> >> - IOMMUTLBEvent event;
> >> >> uint64_t mask = dma_aligned_pow2_mask(start, end, s->aw_bits);
> >> >> uint64_t size = mask + 1;
> >> >>
> >> >> assert(size);
> >> >>
> >> >> - event.type = IOMMU_NOTIFIER_UNMAP;
> >> >> - event.entry.iova = start;
> >> >> - event.entry.addr_mask = mask;
> >> >> - event.entry.target_as = &address_space_memory;
> >> >> - event.entry.perm = IOMMU_NONE;
> >> >> - /* This field is meaningless for unmap */
> >> >> - event.entry.translated_addr = 0;
> >> >> -
> >> >> - memory_region_notify_iommu_one(n, &event);
> >> >> + map.iova = start;
> >> >> + map.size = size;
> >> >> + if (iova_tree_find(as->iova_tree, &map)) {
> >> >> + event.entry.iova = start;
> >> >> + event.entry.addr_mask = mask;
> >> >> + memory_region_notify_iommu_one(n, &event);
> >> >> + }
> >> >
> >> >This one looks fine to me, but I'm not sure how much benefit we'll
> >> >get here either as this path should be rare afaiu.
> >>
> >> Yes, I only see such UNMAP call at cold bootup/shutdown, hot plug and
> >unplug.
> >>
> >> In fact, the other purpose of this patch is to eliminate noisy error
> >> log when we work with IOMMUFD. It looks the duplicate UNMAP call will
> >> fail with IOMMUFD while always succeed with legacy container. This
> >> behavior difference lead to below error log for IOMMUFD:
A dumb question, should IOMMUFD stick the same behaviour with legacy container?
Thanks
> >>
> >> IOMMU_IOAS_UNMAP failed: No such file or directory
> >> vfio_container_dma_unmap(0x562012d6b6d0, 0x0, 0x80000000) = -2 (No
> >> such file or directory) IOMMU_IOAS_UNMAP failed: No such file or
> >> directory vfio_container_dma_unmap(0x562012d6b6d0, 0x80000000,
> >> 0x40000000) = -2 (No such file or directory)
> >
> >I see. Please also mention this in the commit log, that'll help reviewers
> >understand the goal of the patch, thanks!
> Will do.
>
> >
> >>
> >> >
> >> >>
> >> >> start += size;
> >> >> remain -= size;
> >> >> @@ -3826,13 +3830,6 @@ static void
> >> >vtd_iommu_replay(IOMMUMemoryRegion *iommu_mr, IOMMUNotifier
> >*n)
> >> >> uint8_t bus_n = pci_bus_num(vtd_as->bus);
> >> >> VTDContextEntry ce;
> >> >>
> >> >> - /*
> >> >> - * The replay can be triggered by either a invalidation or a newly
> >> >> - * created entry. No matter what, we release existing mappings
> >> >> - * (it means flushing caches for UNMAP-only registers).
> >> >> - */
> >> >> - vtd_address_space_unmap(vtd_as, n);
> >> >
> >> >IIUC this is needed to satisfy current replay() semantics:
> >> >
> >> > /**
> >> > * @replay:
> >> > *
> >> > * Called to handle memory_region_iommu_replay().
> >> > *
> >> > * The default implementation of memory_region_iommu_replay() is to
> >> > * call the IOMMU translate method for every page in the address space
> >> > * with flag == IOMMU_NONE and then call the notifier if translate
> >> > * returns a valid mapping. If this method is implemented then it
> >> > * overrides the default behaviour, and must provide the full
> >> > semantics
> >> > * of memory_region_iommu_replay(), by calling @notifier for every
> >> > * translation present in the IOMMU.
> >> Above semantics claims calling @notifier for every translation present
> >> in the IOMMU But it doesn't claim if calling @notifier for non-present
> >translation.
> >> I checked other custom replay() callback, ex. virtio_iommu_replay(),
> >> spapr_tce_replay() it looks only intel_iommu is special by calling
> >> unmap_all()
> >before rebuild mapping.
> >
> >Yes, and I'll reply below for this..
> >
> >>
> >> >
> >> >The problem is vtd_page_walk() currently by default only notifies on
> >> >page changes, so we'll notify all MAP only if we unmap all of them first.
> >> Hmm, I didn't get this point. Checked vtd_page_walk_one(), it will
> >> rebuild the mapping except the DMAMap is exactly same which it will skip.
> >See below:
> >>
> >> /* Update local IOVA mapped ranges */
> >> if (event->type == IOMMU_NOTIFIER_MAP) {
> >> if (mapped) {
> >> /* If it's exactly the same translation, skip */
> >> if (!memcmp(mapped, &target, sizeof(target))) {
> >> trace_vtd_page_walk_one_skip_map(entry->iova, entry-
> >>addr_mask,
> >> entry->translated_addr);
> >> return 0;
> >> } else {
> >> /*
> >> * Translation changed. Normally this should not
> >> * happen, but it can happen when with buggy guest
> >
> >So I haven't touched the vIOMMU code for a few years, but IIRC if we
> >replay() on an address space that has mapping already, then if without the
> >unmap_all() at the start we'll just notify nothing, because "mapped" will be
> >true for all the existing mappings, and memcmp() should return 0 too if
> >nothing changed?
> Understood, you are right. VFIO migration dirty sync needs to be notified
> even if mapping is unchanged.
>
> >
> >I think (and agree) it could be a "bug" for vtd only, mostly not affecting
> >anything at least before vfio migration.
> >
> >Do you agree, and perhaps want to fix it altogether? If so I suppose it'll
> >also
> >fix the issue below on vfio dirty sync.
> Yes, I'll write an implementation.
>
> Thanks
> Zhenzhong
