Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

For entirely dma coherent architectures there is no requirement to ever
remap dma coherent allocation.  Move all the remap and pool code under
IS_ENABLED() checks and drop the Kconfig dependency.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Inline __iommu_dma_mmap_pfn into the main function, and use the
fact that __iommu_dma_get_pages return NULL for remapped contigous
allocations to simplify the code flow a bit.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Inline __iommu_dma_get_sgtable_page into the main function, and use the
fact that __iommu_dma_get_pages return NULL for remapped contigous
allocations to simplify the code flow a bit.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

All the logic in iommu_dma_alloc that deals with page allocation from
the CMA or page allocators can be split into a self-contained helper,
and we can than map the result of that or the atomic pool allocation
with the iommu later.  This also allows reusing __iommu_dma_free to
tear down the allocations and MMU mappings when the IOMMU mapping
fails.

Based on a patch from Robin Murphy.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Most importantly clear up the size / iosize confusion.  Also rename addr
to cpu_addr to match the surrounding code and make the intention a little
more clear.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
[hch: split from a larger patch]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Most of it can double up to serve the failure cleanup path for
iommu_dma_alloc().
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Instead of having a separate code path for the non-blocking alloc_pages
and CMA allocations paths merge them into one.  There is a slight
behavior change here in that we try the page allocator if CMA fails.
This matches what dma-direct and other iommu drivers do and will be
needed to use the dma-iommu code on architectures without DMA remapping
later on.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Always remapping CMA allocations was largely a bodge to keep the freeing
logic manageable when it was split between here and an arch wrapper. Now
that it's all together and streamlined, we can relax that limitation.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Shuffle around the self-contained atomic and non-contiguous cases to
return early and get out of the way of the CMA case that we're about to
work on next.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
[hch: slight changes to the code flow]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The freeing logic was made particularly horrible by part of it being
opaque to the arch wrapper, which led to a lot of convoluted repetition
to ensure each path did everything in the right order. Now that it's
all private, we can pick apart and consolidate the logically-distinct
steps of freeing the IOMMU mapping, the underlying pages, and the CPU
remap (if necessary) into something much more manageable.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
[various cosmetic changes to the code flow]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

We only have a single caller of this function left, so open code it there.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Move the call to dma_common_pages_remap into __iommu_dma_alloc and
rename it to iommu_dma_alloc_remap.  This creates a self-contained
helper for remapped pages allocation and mapping.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Since we duplicate the find_vm_area() logic a few times in places where
we only care aboute the pages, factor out a helper to abstract it.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
[hch: don't warn when not finding a region, as we'll rely on that later]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The remaining internal callsites don't care about having prototypes
compatible with the relevant dma_map_ops callbacks, so the extra
level of indirection just wastes space and complictaes things.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Most of the callers don't care, and the couple that do already have the
domain to hand for other reasons are in slow paths where the (trivial)
overhead of a repeated lookup will be utterly immaterial.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
[hch: dropped the hunk touching iommu_dma_get_msi_page to avoid a
 conflict with another series]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Moving this function up to its unmap counterpart helps to keep related
code together for the following changes.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

There is nothing really arm64 specific in the iommu_dma_ops
implementation, so move it to dma-iommu.c and keep a lot of symbols
self-contained.  Note the implementation does depend on the
DMA_DIRECT_REMAP infrastructure for now, so we'll have to make the
DMA_IOMMU support depend on it, but this will be relaxed soon.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

arch_dma_prep_coherent can handle physically contiguous ranges larger
than PAGE_SIZE just fine, which means we don't need a page-based
iterator.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

We now have a arch_dma_prep_coherent architecture hook that is used
for the generic DMA remap allocator, and we should use the same
interface for the dma-iommu code.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Acked-by: NCatalin Marinas <catalin.marinas@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Convert to use vm_map_pages() to map range of kernel memory to user vma.

Link: http://lkml.kernel.org/r/80c3d220fc6ada73a88ce43ca049afb55a889258.1552921225.git.jrdr.linux@gmail.comSigned-off-by: NSouptick Joarder <jrdr.linux@gmail.com>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Heiko Stuebner <heiko@sntech.de>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Cc: Pawel Osciak <pawel@osciak.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Sandy Huang <hjc@rock-chips.com>
Cc: Stefan Richter <stefanr@s5r6.in-berlin.de>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Thierry Reding <treding@nvidia.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The dma_ranges list field of PCI host bridge structure has resource entries
in sorted order representing address ranges allowed for DMA transfers.

Process the list and reserve IOVA addresses that are not present in its
resource entries (ie DMA memory holes) to prevent allocating IOVA addresses
that cannot be accessed by PCI devices.
Based-on-a-patch-by: NOza Pawandeep <oza.oza@broadcom.com>
Signed-off-by: NSrinath Mannam <srinath.mannam@broadcom.com>
Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Reviewed-by: NOza Pawandeep <poza@codeaurora.org>
Acked-by: NRobin Murphy <robin.murphy@arm.com>

A recent change split iommu_dma_map_msi_msg() in two new functions. The
function was still implemented to avoid modifying all the callers at
once.

Now that all the callers have been reworked, iommu_dma_map_msi_msg() can
be removed.
Signed-off-by: NJulien Grall <julien.grall@arm.com>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Reviewed-by: NEric Auger <eric.auger@redhat.com>
Acked-by: NJoerg Roedel <jroedel@suse.de>
Signed-off-by: NMarc Zyngier <marc.zyngier@arm.com>

On RT, iommu_dma_map_msi_msg() may be called from non-preemptible
context. This will lead to a splat with CONFIG_DEBUG_ATOMIC_SLEEP as
the function is using spin_lock (they can sleep on RT).

iommu_dma_map_msi_msg() is used to map the MSI page in the IOMMU PT
and update the MSI message with the IOVA.

Only the part to lookup for the MSI page requires to be called in
preemptible context. As the MSI page cannot change over the lifecycle
of the MSI interrupt, the lookup can be cached and re-used later on.

iomma_dma_map_msi_msg() is now split in two functions:
    - iommu_dma_prepare_msi(): This function will prepare the mapping
    in the IOMMU and store the cookie in the structure msi_desc. This
    function should be called in preemptible context.
    - iommu_dma_compose_msi_msg(): This function will update the MSI
    message with the IOVA when the device is behind an IOMMU.
Signed-off-by: NJulien Grall <julien.grall@arm.com>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Reviewed-by: NEric Auger <eric.auger@redhat.com>
Acked-by: NJoerg Roedel <jroedel@suse.de>
Signed-off-by: NMarc Zyngier <marc.zyngier@arm.com>

end_pfn is never used after commit <aa3ac946> ('iommu/iova: Make dma
32bit pfn implicit'), cleanup it.

Cc: Joerg Roedel <jroedel@suse.de>
Cc: Robin Murphy <robin.murphy@arm.com>
Cc: Zhen Lei <thunder.leizhen@huawei.com>
Signed-off-by: NShaokun Zhang <zhangshaokun@hisilicon.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Use the new helpers dev_iommu_fwspec_get()/set() to access
the dev->iommu_fwspec pointer. This makes it easier to move
that pointer later into another struct.
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Change function __iommu_dma_alloc_pages() to allocate pages for DMA from
respective device NUMA node. The ternary operator which would be for
alloc_pages_node() is tidied along with this.

The motivation for this change is to have a policy for page allocation
consistent with direct DMA mapping, which attempts to allocate pages local
to the device, as mentioned in [1].

In addition, for certain workloads it has been observed a marginal
performance improvement. The patch caused an observation of 0.9% average
throughput improvement for running tcrypt with HiSilicon crypto engine.

We also include a modification to use kvzalloc() for kzalloc()/vzalloc()
combination.

[1] https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg1692998.htmlSigned-off-by: NGanapatrao Kulkarni <ganapatrao.kulkarni@cavium.com>
[JPG: Added kvzalloc(), drop pages ** being device local, remove ternary operator, update message]
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Return DMA_MAPPING_ERROR instead of 0 on a dma mapping failure and let
the core dma-mapping code handle the rest.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>

With the flush queue infrastructure already abstracted into IOVA
domains, hooking it up in iommu-dma is pretty simple. Since there is a
degree of dependency on the IOMMU driver knowing what to do to play
along, we key the whole thing off a domain attribute which will be set
on default DMA ops domains to request non-strict invalidation. That way,
drivers can indicate the appropriate support by acknowledging the
attribute, and we can easily fall back to strict invalidation otherwise.

The flush queue callback needs a handle on the iommu_domain which owns
our cookie, so we have to add a pointer back to that, but neatly, that's
also sufficient to indicate whether we're using a flush queue or not,
and thus which way to release IOVAs. The only slight subtlety is
switching __iommu_dma_unmap() from calling iommu_unmap() to explicit
iommu_unmap_fast()/iommu_tlb_sync() so that we can elide the sync
entirely in non-strict mode.
Signed-off-by: NZhen Lei <thunder.leizhen@huawei.com>
[rm: convert to domain attribute, tweak comments and commit message]
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NWill Deacon <will.deacon@arm.com>

Most parts of iommu-dma already assume they are operating on a default
domain set up by iommu_dma_init_domain(), and can be converted straight
over to avoid the refcounting bottleneck. MSI page mappings may be in
an unmanaged domain with an explicit MSI-only cookie, so retain the
non-specific lookup, but that's OK since they're far from a contended
fast path either way.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NWill Deacon <will.deacon@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Take the new bus limit into account (when present) for IOVA allocations,
to accommodate those SoCs which integrate off-the-shelf IP blocks with
narrower interconnects such that the link between a device output and an
IOMMU input can truncate DMA addresses to even fewer bits than the
native size of either block's interface would imply.

Eventually it might make sense for the DMA core to apply this constraint
up-front in dma_set_mask() and friends, but for now this seems like the
least risky approach.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Acked-by: NArd Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: NJoerg Roedel <jroedel@suse.de>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

This pretty much reverts commit 273df963 ("iommu/dma: Make PCI
window reservation generic")  by moving the PCI window region
reservation back into the dma specific path so that these regions
doesn't get exposed via the IOMMU API interface. With this change,
the vfio interface will report only iommu specific reserved regions
to the user space.

Cc: Joerg Roedel <joro@8bytes.org>
Signed-off-by: NShameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Fixes: 273df963 ('iommu/dma: Make PCI window reservation generic')
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Modified iommu_dma_get_resv_regions() to include GICv3 ITS
region on ACPI based ARM platfiorms which may require HW MSI
reservations.
Signed-off-by: NShameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Since IOVA allocation failure is not unusual case we need to flush
CPUs' rcache in hope we will succeed in next round.

However, it is useful to decide whether we need rcache flush step because
of two reasons:
- Not scalability. On large system with ~100 CPUs iterating and flushing
  rcache for each CPU becomes serious bottleneck so we may want to defer it.
- free_cpu_cached_iovas() does not care about max PFN we are interested in.
  Thus we may flush our rcaches and still get no new IOVA like in the
  commonly used scenario:

    if (dma_limit > DMA_BIT_MASK(32) && dev_is_pci(dev))
        iova = alloc_iova_fast(iovad, iova_len, DMA_BIT_MASK(32) >> shift);

    if (!iova)
        iova = alloc_iova_fast(iovad, iova_len, dma_limit >> shift);

   1. First alloc_iova_fast() call is limited to DMA_BIT_MASK(32) to get
      PCI devices a SAC address
   2. alloc_iova() fails due to full 32-bit space
   3. rcaches contain PFNs out of 32-bit space so free_cpu_cached_iovas()
      throws entries away for nothing and alloc_iova() fails again
   4. Next alloc_iova_fast() call cannot take advantage of rcache since we
      have just defeated caches. In this case we pick the slowest option
      to proceed.

This patch reworks flushed_rcache local flag to be additional function
argument instead and control rcache flush step. Also, it updates all users
to do the flush as the last chance.
Signed-off-by: NTomasz Nowicki <Tomasz.Nowicki@caviumnetworks.com>
Reviewed-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NNate Watterson <nwatters@codeaurora.org>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Now that the cached node optimisation can apply to all allocations, the
couple of users which were playing tricks with dma_32bit_pfn in order to
benefit from it can stop doing so. Conversely, there is also no need for
all the other users to explicitly calculate a 'real' 32-bit PFN, when
init_iova_domain() can happily do that itself from the page granularity.

CC: Thierry Reding <thierry.reding@gmail.com>
CC: Jonathan Hunter <jonathanh@nvidia.com>
CC: David Airlie <airlied@linux.ie>
CC: Sudeep Dutt <sudeep.dutt@intel.com>
CC: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: NZhen Lei <thunder.leizhen@huawei.com>
Tested-by: NArd Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: NZhen Lei <thunder.leizhen@huawei.com>
Tested-by: NNate Watterson <nwatters@codeaurora.org>
[rm: use iova_shift(), rewrote commit message]
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

DMA_ERROR_CODE is not a public API and will go away soon.  dma dma-iommu
driver already implements a proper ->mapping_error method, so it's only
using the value internally.  Add a new local define using the value
that arm64 which is the only current user of dma-iommu.
Signed-off-by: NChristoph Hellwig <hch@lst.de>

When walking the rbtree, the fact that iovad->start_pfn and limit_pfn
are both inclusive limits creates an ambiguity once limit_pfn reaches
the bottom of the address space and they overlap. Commit 5016bdb7
("iommu/iova: Fix underflow bug in __alloc_and_insert_iova_range") fixed
the worst side-effect of this, that of underflow wraparound leading to
bogus allocations, but the remaining fallout is that any attempt to
allocate start_pfn itself erroneously fails.

The cleanest way to resolve the ambiguity is to simply make limit_pfn an
exclusive limit when inside the guts of the rbtree. Since we're working
with PFNs, representing one past the top of the address space is always
possible without fear of overflow, and elsewhere it just makes life a
little more straightforward.
Reported-by: NAaron Sierra <asierra@xes-inc.com>
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

When __iommu_dma_map() and iommu_dma_free_iova() are called from
iommu_dma_get_msi_page(), various iova_*() helpers are still invoked in
the process, whcih is unwise since they access a different member of the
union (the iova_domain) from that which was last written, and there's no
guarantee that sensible values will result anyway.

CLean up the code paths that are valid for an MSI cookie to ensure we
only do iova_domain-specific things when we're actually dealing with one.

Fixes: a44e6657 ("iommu/dma: Clean up MSI IOVA allocation")
Reported-by: NNate Watterson <nwatters@codeaurora.org>
Tested-by: NShanker Donthineni <shankerd@codeaurora.org>
Tested-by: NBharat Bhushan <bharat.bhushan@nxp.com>
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NEric Auger <eric.auger@redhat.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

With IOVA allocation suitably tidied up, we are finally free to opt in
to the per-CPU caching mechanism. The caching alone can provide a modest
improvement over walking the rbtree for weedier systems (iperf3 shows
~10% more ethernet throughput on an ARM Juno r1 constrained to a single
650MHz Cortex-A53), but the real gain will be in sidestepping the rbtree
lock contention which larger ARM-based systems with lots of parallel I/O
are starting to feel the pain of.
Reviewed-by: NNate Watterson <nwatters@codeaurora.org>
Tested-by: NNate Watterson <nwatters@codeaurora.org>
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Now that allocation is suitably abstracted, our private alloc/free
helpers can drive the trivial MSI cookie allocator directly as well,
which lets us clean up its exposed guts from iommu_dma_map_msi_msg() and
simplify things quite a bit.
Reviewed-by: NNate Watterson <nwatters@codeaurora.org>
Tested-by: NNate Watterson <nwatters@codeaurora.org>
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>