Implement the new domain_alloc and domain_free call-backs
and remove the old domain_init/destroy ones.
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The OMAP IOMMU driver unconditionally executes code and registers a
struct iommu_ops with the platform bus irrespective of whether it runs
on an OMAP SoC or not. This causes problems in multi-platform kernels
where drivers for other SoCs will no longer be able to register their
own struct iommu_ops or even try to use a struct iommu_ops for an IOMMU
that obviously isn't there.

The smallest fix I could think of is to check for the existence of any
OMAP IOMMU devices in the device tree and skip initialization otherwise.

This fixes a problem on Tegra20 where the DRM driver will try to use the
obviously non-existent OMAP IOMMU.
Reported-by: NNicolas Chauvet <kwizart@gmail.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Suman Anna <s-anna@ti.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NThierry Reding <treding@nvidia.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Fixes this compile warning:

drivers/iommu/omap-iommu.c: In function 'omap_iommu_map':
drivers/iommu/omap-iommu.c:1139:2: warning: format '%x' expects argument of type 'unsigned int', but argument 5 has type 'phys_addr_t' [-Wformat=]
  dev_dbg(dev, "mapping da 0x%lx to pa 0x%x size 0x%x\n", da, pa, bytes);
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Use the wrapper function for retrieving the platform data instead of
accessing dev->platform_data directly.
Signed-off-by: NKiran Padwal <kiran.padwal@smartplayin.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Mapping and unmapping are more often than not in the critical path.
map_sg allows IOMMU driver implementations to optimize the process
of mapping buffers into the IOMMU page tables.

Instead of mapping a buffer one page at a time and requiring potentially
expensive TLB operations for each page, this function allows the driver
to map all pages in one go and defer TLB maintenance until after all
pages have been mapped.

Additionally, the mapping operation would be faster in general since
clients does not have to keep calling map API over and over again for
each physically contiguous chunk of memory that needs to be mapped to a
virtually contiguous region.
Signed-off-by: NOlav Haugan <ohaugan@codeaurora.org>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The .domain field in omap_iommu struct is set properly when the
OMAP IOMMU device is attached to, but is never reset properly
on detach. Reset this properly so that the OMAP IOMMU debugfs
logic can depend on this field before allowing the debugfs
operations.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The following functions were exported previously for usage by
the OMAP IOMMU debug module:
	omap_iommu_dump_ctx()
	omap_dump_tlb_entries()
	omap_iopgtable_store_entry()

These functions need not be exported anymore as the OMAP IOMMU
debugfs code is integrated with the OMAP IOMMU driver, and
there won't be external users for these functions. So, remove
the EXPORT_SYMBOL_GPL on these. The omap_iopgtable_store_entry()
is also made internal only, after making the 'pagetable' debugfs
entry read-only.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The exported functions omap_foreach_iommu_device() and
omap_iotlb_cr_to_e() have been deleted, as they are no
longer needed.

The function omap_foreach_iommu_device() is not required
after the consolidation of the OMAP IOMMU debug module,
and the function omap_iotlb_cr_to_e() is not required
after making the debugfs entry 'pagetable' read-only.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The debugfs support for OMAP IOMMU is currently implemented
as a module, warranting certain OMAP-specific IOMMU API to
be exported. The OMAP IOMMU, when enabled, can only be built-in
into the kernel, so integrate the OMAP IOMMU debug module
into the OMAP IOMMU driver. This helps in eliminating the
need to export most of the current OMAP IOMMU API.

The following are the main changes:
- The debugfs directory and entry creation logic is reversed,
  the calls are invoked by the OMAP IOMMU driver now.
- The current iffy circular logic of adding IOMMU archdata
  to the IOMMU devices itself to get a pointer to the omap_iommu
  object in the debugfs support code is replaced by directly
  using the omap_iommu structure while creating the debugfs
  entries.
- The debugfs root directory is renamed from the generic name
  "iommu" to a specific name "omap_iommu".
- Unneeded headers have also been cleaned up while at this.
- There will no longer be a omap-iommu-debug.ko module after
  this patch.
- The OMAP_IOMMU_DEBUG Kconfig option is converted to boolean
  only, the OMAP IOMMU debugfs support is built alongside the
  OMAP IOMMU driver only when this option is enabled.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The OMAP IOMMU driver was originally designed as modules, and split
into a core module and a thin arch-specific module through the OMAP
arch-specific struct iommu_functions, to scale for both OMAP1 and
OMAP2+ IOMMU variants. The driver can only be built for OMAP2+
platforms currently, and also can only be built-in after the
adaptation to generic IOMMU API. The OMAP1 variant was never added
and will most probably be never added (the code for the only potential
user, its parent, DSP processor has already been cleaned up). So,
consolidate the OMAP2 specific omap-iommu2 module into the core OMAP
IOMMU driver - this eliminates the arch-specific ops structure and
simplifies the driver into a single module that only implements the
generic IOMMU API's iommu_ops.

The following are the main changes:
- omap-iommu2 module is completely eliminated, with the common
  definitions moved to the internal omap-iommu.h, and the ops
  implementations moved into omap-iommu.c
- OMAP arch-specific struct iommu_functions is also eliminated,
  with the ops implementations directly absorbed into the calling
  functions
- iotlb_alloc_cr() is no longer inlined and defined only when
  PREFETCH_IOTLB is defined
- iotlb_dump_cr() is similarly defined only when CONFIG_OMAP_IOMMU_DEBUG
  is defined
- Elimination of the OMAP IOMMU exported functions to register the
  arch ops, omap_install_iommu_arch() & omap_uninstall_iommu_arch()
- Any stale comments about OMAP1 are also cleaned up
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The function omap_iommu_arch_version() is not used anymore,
and is not required either, so remove it. The .version field
in struct iommu_functions that this function uses is also
removed, as it is not really an ops to retrieve a version and
there won't be any usage for this field either.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The refcount field in omap_iommu object is primarily used to check
if an IOMMU device has already been enabled, but this is already
implicit in the omap_iommu_attach_dev() which ensures that only
a single device can attach to an IOMMU. This field is redundant,
and so has been cleaned up.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Make of_device_id array const, because all OF functions handle it as const.
Signed-off-by: NKiran Padwal <kiran.padwal@smartplayin.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The owner field is never set. Remove it.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

A device is tied to an iommu through its archdata field. The archdata
is allocated on the fly for DT-based devices automatically through the
.add_device iommu ops. The current logic incorrectly assigned the name
of the IOMMU user device, instead of the name of the IOMMU device as
required by the attach logic. Fix this issue so that DT-based devices
can attach successfully to an IOMMU domain.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Any device requiring to be attached to an iommu_domain must have
valid archdata containing the necessary iommu information, which
is SoC-specific. Add a check in the omap_iommu_attach_dev to make
sure that the device has valid archdata before accessing
different SoC-specific fields of the archdata. This prevents a
NULL pointer dereference on any misconfigured devices.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The OMAP3 ISP driver was the only user of the OMAP IOVMM API. Now that
is has been ported to the DMA API, remove the unused virtual memory
manager.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

This structure is read-only data and should never be modified.
Signed-off-by: NThierry Reding <treding@nvidia.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

The prot flags passed to the IOMMU map handler are defined in
include/linux/iommu.h as IOMMU_(READ|WRITE|CACHE|EXEC). However, the
driver expects to receive MMU_RAM_* OMAP-specific flags. This causes
IOMMU flags being interpreted as page sizes, leading to failures.

Hardcode the OMAP mapping parameters to little-endian, 8-bits and
non-mixed page attributes. Furthermore, as the OMAP IOMMU doesn't
support read-only or write-only mappings, ignore the prot value.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: NSakari Ailus <sakari.ailus@iki.fi>
Acked-by: NSuman Anna <s-anna@ti.com>

The IOMMU core breaks out mappings into pages already, there's no
need to support mapping multiple pages in one go.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: NSakari Ailus <sakari.ailus@iki.fi>

The flush_iotlb_page() function prints a debug message when no
corresponding page was found in the TLB. That condition is incorrectly
checked and always resolves to true, given that the for_each_iotlb_cr()
loop is never interrupted and always reaches obj->nr_tlb_entries.

Given that we can't have two TLB entries for the same VA, break from the
loop when a match is found.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NSuman Anna <s-anna@ti.com>

The to_iommu definition is used only locally to the omap-iommu.c
source file, and it has nothing to do with the page attributes
defined in omap-iopgtable.h. So, move the definition out of
omap-iopgtable.h header file.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>

The current OMAP IOMMU ops for .domain_has_cap is a stub,
and the iommu core already returns a value of 0 if the
domain doesn't have a .domain_has_cap ops plugged in. So,
clean up this stub function.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>

The iotlb_entry field values are used directly in omap2_alloc_cr,
a function used in preparing the MMU_CAM and MMU_RAM registers.
The iotlb_entry.valid value is being set incorrectly to 1 at the
moment, and this would result in overriding the PAGESIZE bit field
of the MMU_CAM register if prefetching of the entries were to be
supported.

The bug has not caused any MMU faults due to incorrect size
programming so far as the prefetching is disabled by default. Fix
this by using the correct init value for the iotlb_entry.valid
field.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>

The iopte_free() function should check for NULL because
kmem_cache_free() will panic on NULL argument.
Signed-off-by: NZhouyi Zhou <yizhouzhou@ict.ac.cn>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

The OMAP IOMMU driver locates the IOMMU associated to a device using the
IOMMU name stored in the device archdata iommu field. That field is
expected to be populated by platform code and is left unset for DT-based
devices. This results in a crash when the IOMMU driver attaches a domain
to a device.

Fix this by allocating the archdata iommu structure when devices are
added and freeing when they are removed. Devices without an OF node, and
devices without an iommus property in their OF node are ignored. The
iommu name is initialized from the IOMMU device node name.

This should be simplified when removing non-DT support completely from
the IOMMU users as the IOMMU name won't be needed anymore, and the
IOMMU device pointer could then be stored in the archdata iommu field
directly.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
[s-anna@ti.com: updated to use device name instead of OF name]
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

The remoteproc MMUs in OMAP4+ SoCs have some additional debug
registers that can give out the PC value in addition to the
MMU fault address. The PC value can be extracted properly only
on the DSP cores, and is not available on the ARM processors
within the IPU sub-systems. Instead, the MMUs have been enhanced
to throw a bus-error response back to the IPU processors.

This functionality is programmable through the MMU_GP_REG register.
The cores are simply stalled if the MMU_GP_REG.BUS_ERR_BACK_EN bit
is not set. When set, a bus-error exception is raised allowing the
processor to handle it as a bus fault and provide additional debug
information. This feature is turned on by default by the driver on
iommus supporting it.
Signed-off-by: NSubramaniam Chanderashekarapuram <subramaniam.ca@ti.com>
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

As OMAP2+ is moving to a full DT boot for all SoC families, commit
7ce93f31 "ARM: OMAP2+: Fix more missing data for omap3.dtsi file"
adds basic DT bits for OMAP3. But the driver is not yet converted,
so this will not work and driver will not be probed. Convert it!

The legacy boot mode is still supported until OMAP3 is converted
to DT-boot only.
Signed-off-by: NFlorian Vaussard <florian.vaussard@epfl.ch>
[s-anna@ti.com: dev_name adaptation and improved error checking]
Signed-off-by: NSuman Anna <s-anna@ti.com>
[tony@atomide.com: Ack for arch/arm/*omap* parts]
Acked-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

When booting with a devicetree, no platform data is provided.
Do not prematurely exit iommu_enable() and iommu_disable() in
such a case.

Note: As OMAP do not yet has a proper reset controller driver,
IOMMUs requiring a reset signal should use pdata-quirks as a
transitional solution.
Signed-off-by: NFlorian Vaussard <florian.vaussard@epfl.ch>
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

There are couple of issues with the error return paths in
omap_iommu_attach():
1. omap_iommu_attach() returns NULL or ERR_PTR in case of error,
   but omap_iommu_attach_dev() only checks for IS_ERR. Thus a NULL
   return value (in case driver_find_device fails) will cause the
   kernel to panic when omap_iommu_attach_dev() dereferences the
   pointer.
2. A try_module_get() failure returns a valid success value as
   returned from iommu_enable().

Both the above issues have been fixed up to return the proper
ERR_PTR.
Signed-off-by: NFlorian Vaussard <florian.vaussard@epfl.ch>
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

Use the various devm_ interfaces to simplify the cleanup in
probe and remove functions.
Signed-off-by: NFlorian Vaussard <florian.vaussard@epfl.ch>
Signed-off-by: NSuman Anna <s-anna@ti.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

The driver core clears the driver data to NULL after device_release or
on probe failure.  Thus, it is not needed to manually clear the device
driver data to NULL.
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Cc: David Brown <davidb@codeaurora.org>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Suman Anna <s-anna@ti.com>
Acked-by: NLibo Chen <libo.chen@huawei.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch fixes the checkpatch warnings in omap iommu
code, most of them are related to broken strings.
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

Fixed the following printk format warnings for dma_addr_t
for OMAP IOMMU.

drivers/iommu/omap-iommu.c: In function 'omap_iommu_iova_to_phys':
drivers/iommu/omap-iommu.c:1238:4: warning: format '%lx' expects type 'long unsigned int', but argument 4 has type 'dma_addr_t'
drivers/iommu/omap-iommu.c:1245:4: warning: format '%lx' expects type 'long unsigned int', but argument 4 has type 'dma_addr_t'
Signed-off-by: NSuman Anna <s-anna@ti.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

This is required in case of PAMU, as it can support a window size of up
to 64G (even on 32bit).
Signed-off-by: NVarun Sethi <Varun.Sethi@freescale.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

The pointer obj is dereferenced in line 146 and 149 respectively, so it is not
necessary to check null again in line 149 and 175. And I have checked that all
the callers of these two functions guarantee the parameter obj passed is not
null.
Signed-off-by: NCong Ding <dinggnu@gmail.com>
Acked-by: NOhad Ben-Cohen <ohad@wizery.com>
Acked-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

CONFIG_HOTPLUG is going away as an option.  As a result, the __dev*
markings need to be removed.

This change removes the use of __devinit, __devexit_p, __devinitdata,
and __devexit from these drivers.

Based on patches originally written by Bill Pemberton, but redone by me
in order to handle some of the coding style issues better, by hand.

Cc: Bill Pemberton <wfp5p@virginia.edu>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Ohad Ben-Cohen <ohad@wizery.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Omar Ramirez Luna <omar.luna@linaro.org>
Cc: Mauro Carvalho Chehab <mchehab@redhat.com>
Cc: Hiroshi Doyu <hdoyu@nvidia.com>
Cc: Stephen Warren <swarren@wwwdotorg.org>
Cc: Bharat Nihalani <bnihalani@nvidia.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Use runtime PM functionality interfaced with hwmod enable/idle
functions, to replace direct clock operations and sysconfig
handling.

Due to reset sequence, pm_runtime_[get|put]_sync must be used, to
avoid possible operations with the module under reset. Because of
this and given that the driver uses spin_locks to protect their
critical sections, we must use pm_runtime_irq_safe in order for the
runtime ops to be happy, otherwise might_sleep_if checks in runtime
framework will complain.

The remaining pm_runtime out of iommu_enable and iommu_disable
corresponds to paths that can be accessed through debugfs, some of
them doesn't work if the module is not enabled first, but in future
if the mmu is idled withouth freeing, these are needed to debug.
Signed-off-by: NOmar Ramirez Luna <omar.luna@linaro.org>
Tested-by: NOhad Ben-Cohen <ohad@wizery.com>
Acked-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

Use hwmod data and device attributes to build and register an
omap device for iommu driver.

 - Update the naming convention in isp module.
 - Remove unneeded check for number of resources, as this is now
   handled by omap_device and prevents driver from loading.
 - Now unused, remove platform device and resource data, handling
   of sysconfig register for softreset purposes, use default
   latency structure.
 - Use hwmod API for reset handling.
Signed-off-by: NOmar Ramirez Luna <omar.luna@linaro.org>
Tested-by: NOhad Ben-Cohen <ohad@wizery.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

The purpose of the mmu is to handle the memory accesses requested by
its users. Typically, the mmu is bundled with the processing unit in
a single IP block, which makes them to share the same clock to be
functional.

Currently, iommu code assumes that its user will be indirectly
clocking it, but being a separate mmu driver, it should handle
its own clocks, so as long as the mmu is requested it will be
powered ON and once detached it will be powered OFF.

The remaining clock handling out of iommu_enable and iommu_disable
corresponds to paths that can be accessed through debugfs, some of
them doesn't work if the module is not enabled first, but in future
if the mmu is idled withouth freeing, these are needed to debug.
Signed-off-by: NOmar Ramirez Luna <omar.luna@linaro.org>
Tested-by: NOhad Ben-Cohen <ohad@wizery.com>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>