This reverts commit 35d9f510b6.
Quoth Jiri Slaby:
"It fixes mmap when IOMMU is used on x86 only, but breaks architectures
like ARM or PPC where virt_to_phys(dma_alloc_coherent) doesn't work.
We need there dma_mmap_coherent or similar (the trickery what
snd_pcm_default_mmap does but in some saner way). But this cannot be
done at this phase."
Requested-by: Jiri Slaby <jslaby@suse.cz>
Cc: Russell King - ARM Linux <linux@arm.linux.org.uk>
Cc: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mem->dma_handle is a dma address obtained by dma_alloc_coherent which
needn't be a physical address in presence of IOMMU, as
a hardware IOMMU can (and most likely) will return a bus address where
physical != bus address.
So ensure we are remapping (remap_pfn_range) the right page in
__videobuf_mmap_mapper by using virt_to_phys(mem->vaddr) and not
mem->dma_handle.
While at it, use PFN_DOWN instead of explicit shift.
[mchehab@redhat.com: Fix compilation breakage due to the lack of a comma]
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add an ext_lock argument to the videobuf init functions. This allows
drivers to pass the vdev->lock pointer (or any other externally held lock)
to videobuf. For now all drivers just pass NULL.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Currently videobuf uses the vb_lock mutex to lock its data structures.
But this locking will (optionally) move into the v4l2 core, which means
that in that case vb_lock shouldn't be used since the external lock is already
held.
Prepare for this by adding a pointer to such an external mutex and
don't lock if that pointer is set.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
When a driver that uses videobuf-dma-contig is used with the USERPTR
memory access method a kernel oops might happen: a NULL address may be
passed to dma_free_coherent(). This happens when an application calls
REQBUFS and then exits without queuing any buffers. This patch fixes
that bug.
Signed-off-by: Pawel Osciak <p.osciak@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The fields are assigned but never used, remove them.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
These functions allocate videobuf_buffer structures only. Renaming in order
to prevent confusion with functions allocating actual video buffer memory.
Rename the functions in videobuf-core.h videobuf-dma-sg.c as well.
Signed-off-by: Pawel Osciak <p.osciak@samsung.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
mmap_mapper should operate on a buffer, not on a complete queue. So let
the videobuf-core find the correct buffer instead of duplicating that
code in each mmap_mapper implementation.
The dma-sg implementation has backwards compatibility code for handling
the V4L1_COMPAT layer. This code is now under the v4L1_COMPAT config option.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The video_copy_to_user and copy_stream ops are almost identical for all
videobuf memtype variants. All that is needed is to use the new vaddr
op and these functions can be moved into the core, ensuring we have just
one single implementation instead of three.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Rename the .vmalloc op in struct videobuf_qtype_ops to .vaddr. This op returns
the virtual kernel address of a buffer. vaddr is a lot less confusing than
vmalloc since this callback does do any allocations.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
videobuf_alloc() returned a void *. Change to struct videobuf_buffer *
to get better type checking.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Remove the mmap_free callback from struct videobuf_qtype_ops.
All implementations of this callback do the same trivial check: return
-EBUSY if any buffer is mmapped. That can also be tested in the
videobuf core.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
If a USERPTR address that is not aligned to page boundary is passed to the
videobuf_dma_contig_user_get() function, it saves a page aligned address to
the dma_handle. This is not correct. This issue is observed when using USERPTR
IO machism for buffer exchange.
Updates from last version:-
Adding offset for size calculation as per comment from Magnus Damm. This
ensures the last page is also included for checking if memory is
contiguous.
Signed-off-by: Muralidharan Karicheri <m-karicheri2@ti.com>
Acked-by: Magnus Damm <damm@opensource.se>
Signed-off-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The videobuf_queue_ops function vector is not declared constant, but
there's no need for the videobuf layer to ever change it. Make it const
so that videobuf users can make their operations const without warnings.
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
* mark struct vm_area_struct::vm_ops as const
* mark vm_ops in AGP code
But leave TTM code alone, something is fishy there with global vm_ops
being used.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since videobuf-dma-contig is designed to handle physically contiguous
memory, this patch modifies the videobuf-dma-contig code to only accept a
user space pointer to physically contiguous memory. For now only
VM_PFNMAP vmas are supported, so forget hotplug.
On SuperH Mobile we use this with our sh_mobile_ceu_camera driver together
with various multimedia accelerator blocks that are exported to user space
using UIO. The UIO kernel code exports physically contiguous memory to
user space and lets the user space application mmap() this memory and pass
a pointer using the USERPTR interface for V4L2 zero copy operation.
With this approach we support zero copy capture, hardware scaling and
various forms of hardware encoding and decoding.
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Acked-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Remove the videobuf-dma-contig sync operation. Sync is only needed
for noncoherent buffers, and since videobuf-dma-contig is built on
coherent memory allocators the memory is by definition always in sync.
Reported-by: Matthieu CASTET <matthieu.castet@parrot.com>
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Reviewed-by: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This patch fixes a free-without-alloc bug for V4L2_MEMORY_USERPTR
video buffers.
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This is V3 of the physically contiguous videobuf queues patch.
Useful for hardware such as the SuperH Mobile CEU which doesn't
support scatter gatter bus mastering.
Since it may be difficult to allocate large chunks of physically
contiguous memory after some uptime due to fragmentation, this code
allocates memory using dma_alloc_coherent(). Architectures supporting
dma_declare_coherent_memory() can easily avoid fragmentation issues
by using dma_declare_coherent_memory() to force dma_alloc_coherent()
to allocate from a certain pre-allocated memory area.
Changes since V2
- use dma_handle for physical address
- use "scatter gather" instead of "scatter gatter"
Changes since V1:
- use dev_err() instead of pr_err()
- remember size in struct videobuf_dma_contig_memory
- keep struct videobuf_dma_contig_memory in .c file
- let videobuf_to_dma_contig() return dma_addr_t
- implement __videobuf_sync()
- return statements, white space and other minor fixes
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Federico Vaga