Rather than checking the DMA attribute at each callsite, just pass it
through for acpi_dma_configure() to handle directly. That can then deal
with the relatively exceptional DEV_DMA_NOT_SUPPORTED case by explicitly
installing dummy DMA ops instead of just skipping setup entirely. This
will then free up the dev->dma_ops == NULL case for some valuable
fastpath optimisations.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Reviewed-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Tested-by: NTony Luck <tony.luck@intel.com>

The dummy DMA ops are currently used by arm64 for any device which has
an invalid ACPI description and is thus barred from using DMA due to not
knowing whether is is cache-coherent or not. Factor these out into
general dma-mapping code so that they can be referenced from other
common code paths. In the process, we can prune all the optional
callbacks which just do the same thing as the default behaviour, and
fill in .map_resource for completeness.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
[hch: moved to a separate source file]
Reviewed-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NTony Luck <tony.luck@intel.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

All architectures except for sparc64 use the dma-direct code in some
form, and even for sparc64 we had the discussion of a direct mapping
mode a while ago.  In preparation for directly calling the direct
mapping code don't bother having it optionally but always build the
code in.  This is a minor hardship for some powerpc and arm configs
that don't pull it in yet (although they should in a relase ot two),
and sparc64 which currently doesn't need it at all, but it will
reduce the ifdef mess we'd otherwise need significantly.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NTony Luck <tony.luck@intel.com>

This isn't exactly a slow path routine, but it is not super critical
either, and moving it out of line will help to keep the include chain
clean for the following DMA indirection bypass work.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NTony Luck <tony.luck@intel.com>

There is no need to have all setup and coherent allocation / freeing
routines inline.  Move them out of line to keep the implemeation
nicely encapsulated and save some kernel text size.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NTony Luck <tony.luck@intel.com>

dma_get_required_mask should really be with the rest of the DMA mapping
implementation instead of in drivers/base as a lone outlier.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NTony Luck <tony.luck@intel.com>

The two functions are exactly the same, so don't bother implementing
them twice.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NTony Luck <tony.luck@intel.com>

We can just call the regular calls after adding offset the the address instead
of reimplementing them.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NJesper Dangaard Brouer <brouer@redhat.com>
Tested-by: NTony Luck <tony.luck@intel.com>

We already zero the memory after allocating it from the pool that
this function fills, and having the memset here in this form means
we can't support CMA highmem allocations.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reported-by: NRussell King - ARM Linux <linux@armlinux.org.uk>

The sparc tree already has this change for the pre-refactored code,
but pulling it into the dma-mapping tree like this should ease
the merge conflicts a bit.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NSam Ravnborg <sam@ravnborg.org>
Acked-by: NDavid Miller <davem@davemloft.net>

There are enough common defintions that a single header seems nicer.

Also drop the pointless <linux/dma-mapping.h> include.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NSam Ravnborg <sam@ravnborg.org>

It has nothing to do with the content of the pci.h header.

Suggested by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NDavid S. Miller <davem@davemloft.net>

The only thing we need to explicitly pull in is the defines for the
CPU type.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NDavid S. Miller <davem@davemloft.net>

There is no good reason to have a double indirection for the sparc32
dma ops, so remove the sparc32_dma_ops and define separate dma_map_ops
instance for the different IOMMU types.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NDavid S. Miller <davem@davemloft.net>

Factor the code to remap memory returned from the DMA coherent allocator
into two helpers that can be shared by the IOMMU and direct mapping code.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NSam Ravnborg <sam@ravnborg.org>

No need to BUG_ON() on the cache maintainance ops - they are no-ops
by default, and there is nothing in the DMA API contract that prohibits
calling them on sbus devices (even if such drivers are unlikely to
ever appear).

Similarly a dma_supported method that always returns 0 is rather
pointless.  The only thing that indicates is that no one ever calls
the method on sbus devices.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NDavid S. Miller <davem@davemloft.net>

DMA debug entries are one of those things which aren't that useful
individually - we will always want some larger quantity of them - and
which we don't really need to manage the exact number of - we only care
about having 'enough'. In that regard, the current behaviour of creating
them one-by-one leads to a lot of unwarranted function call overhead and
memory wasted on alignment padding.

Now that we don't have to worry about freeing anything via
dma_debug_resize_entries(), we can optimise the allocation behaviour by
grabbing whole pages at once, which will save considerably on the
aforementioned overheads, and probably offer a little more cache/TLB
locality benefit for traversing the lists under normal operation. This
should also give even less reason for an architecture-level override of
the preallocation size, so make the definition unconditional - if there
is still any desire to change the compile-time value for some platforms
it would be better off as a Kconfig option anyway.

Since freeing a whole page of entries at once becomes enough of a
challenge that it's not really worth complicating dma_debug_init(), we
may as well tweak the preallocation behaviour such that as long as we
manage to allocate *some* pages, we can leave debugging enabled on a
best-effort basis rather than otherwise wasting them.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NQian Cai <cai@lca.pw>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

With the only caller now gone, we can clean up this part of dma-debug's
exposed internals and make way to tweak the allocation behaviour.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NQian Cai <cai@lca.pw>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

dma-debug is now capable of adding new entries to its pool on-demand if
the initial preallocation was insufficient, so the IOMMU_LEAK logic no
longer needs to explicitly change the pool size. This does lose it the
ability to save a couple of megabytes of RAM by reducing the pool size
below its default, but it seems unlikely that that is a realistic
concern these days (or indeed that anyone is actively debugging AGP
drivers' DMA usage any more). Getting rid of dma_debug_resize_entries()
will make room for further streamlining in the dma-debug code itself.

Removing the call reveals quite a lot of cruft which has been useless
for nearly a decade since commit 19c1a6f5 ("x86 gart: reimplement
IOMMU_LEAK feature by using DMA_API_DEBUG"), including the entire
'iommu=leak' parameter, which controlled nothing except whether
dma_debug_resize_entries() was called or not.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Acked-by: NThomas Gleixner <tglx@linutronix.de>
Tested-by: NQian Cai <cai@lca.pw>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

Now that we can dynamically allocate DMA debug entries to cope with
drivers maintaining excessively large numbers of live mappings, a driver
which *does* actually have a bug leaking mappings (and is not unloaded)
will no longer trigger the "DMA-API: debugging out of memory - disabling"
message until it gets to actual kernel OOM conditions, which means it
could go unnoticed for a while. To that end, let's inform the user each
time the pool has grown to a multiple of its initial size, which should
make it apparent that they either have a leak or might want to increase
the preallocation size.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NQian Cai <cai@lca.pw>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

Certain drivers such as large multi-queue network adapters can use pools
of mapped DMA buffers larger than the default dma_debug_entry pool of
65536 entries, with the result that merely probing such a device can
cause DMA debug to disable itself during boot unless explicitly given an
appropriate "dma_debug_entries=..." option.

Developers trying to debug some other driver on such a system may not be
immediately aware of this, and at worst it can hide bugs if they fail to
realise that dma-debug has already disabled itself unexpectedly by the
time their code of interest gets to run. Even once they do realise, it
can be a bit of a pain to emprirically determine a suitable number of
preallocated entries to configure, short of massively over-allocating.

There's really no need for such a static limit, though, since we can
quite easily expand the pool at runtime in those rare cases that the
preallocated entries are insufficient, which is arguably the least
surprising and most useful behaviour. To that end, refactor the
prealloc_memory() logic a little bit to generalise it for runtime
reallocations as well.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NQian Cai <cai@lca.pw>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

Use pr_fmt() to generate the "DMA-API: " prefix consistently. This
results in it being added to a couple of pr_*() messages which were
missing it before, and for the err_printk() calls moves it to the actual
start of the message instead of somewhere in the middle.
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NQian Cai <cai@lca.pw>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

Expose nr_total_entries in debugfs, so that {num,min}_free_entries
become even more meaningful to users interested in current/maximum
utilisation. This becomes even more relevant once nr_total_entries
may change at runtime beyond just the existing AMD GART debug code.
Suggested-by: NJohn Garry <john.garry@huawei.com>
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Tested-by: NQian Cai <cai@lca.pw>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

These days architectures are mostly out of the business of dealing with
struct scatterlist at all, unless they have architecture specific iommu
drivers.  Replace the ARCH_HAS_SG_CHAIN symbol with a ARCH_NO_SG_CHAIN
one only enabled for architectures with horrible legacy iommu drivers
like alpha and parisc, and conditionally for arm which wants to keep it
disable for legacy platforms.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NPalmer Dabbelt <palmer@sifive.com>

Currently dma_mapping_error returns a boolean as int, with 1 meaning
error.  This is rather unusual and many callers have to convert it to
errno value.  The callers are highly inconsistent with error codes
ranging from -ENOMEM over -EIO, -EINVAL and -EFAULT ranging to -EAGAIN.
Return -ENOMEM which seems to be what the largest number of callers
convert it to, and which also matches the typical error case where
we are out of resources.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NRussell King <rmk+kernel@armlinux.org.uk>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>

No users left except for vmd which just forwards it.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>

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>

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>

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>

Pass the page + offset to the low-level __iommu_map_single helper
(which gets renamed to fit the new calling conventions) as both
callers have the page at hand.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>

Return DMA_MAPPING_ERROR instead of 0 on a dma mapping failure and let
the core dma-mapping code handle the rest.

Note that the existing code used AMD_IOMMU_MAPPING_ERROR to check from
a 0 return from the IOVA allocator, which is replaced with an explicit
0 as in the implementation and other users of that interface.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>

Return DMA_MAPPING_ERROR instead of the magic bad_dma_addr on a dma
mapping failure and let the core dma-mapping code handle the rest.

Remove the magic EMERGENCY_PAGES that the bad_dma_addr gets redirected to.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>

Return DMA_MAPPING_ERROR instead of the magic bad_dma_addr on a dma
mapping failure and let the core dma-mapping code handle the rest.

Remove the magic EMERGENCY_PAGES that the bad_dma_addr gets redirected to.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>

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>

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>

Remove the odd sba_{un,}map_single_attrs wrappers, check errors
everywhere.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>

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>

Just return DMA_MAPPING_ERROR from __dummy_map_page 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>

The SBA iommu code already returns (~(dma_addr_t)0x0) on mapping
failures, so we can switch over to returning DMA_MAPPING_ERROR 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>

The CCIO iommu code already returns (~(dma_addr_t)0x0) on mapping
failures, so we can switch over to returning DMA_MAPPING_ERROR 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>