We get 2 warnings when building kernel with W=1:
drivers/dma/virt-dma.c:22:14: warning: no previous prototype for 'vchan_tx_submit' [-Wmissing-prototypes]
drivers/dma/virt-dma.c:52:5: warning: no previous prototype for 'vchan_tx_desc_free' [-Wmissing-prototypes]
In fact, these two functions are incorrectly declared in a function.
So this patch moves function declarations out of this function.
dmaengine: omap-dma: Enable burst and data pack for SG
Enable the burst and data pack modes for the scatter-gather
in order to improve the throughput of the data transfers.
The improvement has been verified with MMC HS200 mode in
the DRA72 EVM using the iozone tool to compare the read
throughput (in kB/s) with and without burst/pack for
different reclens (in kB).
Colin Ian King [Thu, 29 Sep 2016 17:45:05 +0000 (18:45 +0100)]
dmaengine: jz4780: fix resource leaks on error exit return
In two cases when jz4780_dma_setup_hwdesc fails, there is a memory
leak on the allocated desc and associated DMA pools on the error
exit return path. Fix this by free'ing the resources before
returning.
Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
lib/dma-debug.c: In function 'debug_dma_map_resource':
>> lib/dma-debug.c:1541:16: error: implicit declaration of function '__phys_to_pfn' [-Werror=implicit-function-declaration]
entry->pfn = __phys_to_pfn(addr);
^~~~~~~~~~~~~
ia64 does not provide __phys_to_pfn(), use the PHYS_PFN() alias.
Fixes: 0e74b34dfc3318bf ("dma-debug: add support for resource mappings") Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Colin Ian King [Thu, 29 Sep 2016 18:14:38 +0000 (19:14 +0100)]
dmaengine: coh901318: fix integer overflow when shifting more than 32 places
Currently U300_DMA_CHANNELS is set to 40, meaning that the shift of 1 can
be more than 32 places, which leads to a 32 bit integer overflow. Fix this
by using 1ULL instead of 1 before shifting it. Also add braces on the
for-loop to keep with coding style conventions.
Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
If edma_read_slot() gets an invalid argument, it does not set a result,
as found by "gcc -Wmaybe-uninitialized"
drivers/dma/edma.c: In function 'dma_ccerr_handler':
drivers/dma/edma.c:1499:21: error: 'p.a_b_cnt' may be used uninitialized in this function [-Werror=maybe-uninitialized]
drivers/dma/edma.c:1499:21: error: 'p.ccnt' may be used uninitialized in this function [-Werror=maybe-uninitialized]
if (p.a_b_cnt == 0 && p.ccnt == 0) {
If we change the function to return an error in this case, we can handle
the failure more gracefully and treat this the same way as a null slot
that we already catch.
In file included from include/linux/skbuff.h:34:0,
from include/linux/icmpv6.h:4,
from include/linux/ipv6.h:75,
from include/net/ipv6.h:16,
from include/linux/sunrpc/clnt.h:27,
from include/linux/nfs_fs.h:30,
from fs/lockd/clntlock.c:13:
include/linux/dma-mapping.h: In function 'dma_map_resource':
>> include/linux/dma-mapping.h:274:16: warning: unused variable 'pfn' [-Wunused-variable]
unsigned long pfn = __phys_to_pfn(phys_addr);
^~~
The pfn value is only used once in the call to pfn_valid(), remove the
variable and calculate the pfn when it's needed. Note that the kbuild
report is old and PHYS_PFN() is now used instead of __phys_to_pfn() to
calculate the pfn.
In file included from include/linux/skbuff.h:34:0,
from include/linux/tcp.h:21,
from drivers/net/ethernet/amd/xgbe/xgbe-drv.c:119:
include/linux/dma-mapping.h: In function 'dma_map_resource':
>> include/linux/dma-mapping.h:274:22: error: implicit declaration of function '__phys_to_pfn' [-Werror=implicit-function-declaration]
unsigned long pfn = __phys_to_pfn(phys_addr);
^~~~~~~~~~~~~
ia64 does not provide __phys_to_pfn(), use the PHYS_PFN() alias.
Peter Ujfalusi [Wed, 21 Sep 2016 12:41:35 +0000 (15:41 +0300)]
dmaengine: ti-dma-crossbar: enable COMPILE_TEST
To get more coverage, enable COMPILE_TEST for this driver.
When compile testing eDMA or omap-dma, select also the ti-dma-crossbar so
it is also covered by the compile testing.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Peter Ujfalusi [Wed, 21 Sep 2016 12:41:32 +0000 (15:41 +0300)]
dmaengine: ti-dma-crossbar: Fix of_device_id data parameter usage
Use pointers to static constant variables for crossbar type and for DMA
offset configuration.
Fixes compiler warnings on 64bit architectures:
drivers/dma/ti-dma-crossbar.c: In function ‘ti_dra7_xbar_probe’:
drivers/dma/ti-dma-crossbar.c:398:21: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
xbar->dma_offset = (u32)match->data;
^
drivers/dma/ti-dma-crossbar.c: In function ‘ti_dma_xbar_probe’:
drivers/dma/ti-dma-crossbar.c:431:10: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
switch ((u32)match->data) {
^
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Peter Ujfalusi [Wed, 21 Sep 2016 12:41:30 +0000 (15:41 +0300)]
dmaengine/ARM: omap-dma: Fix the DMAengine compile test on non OMAP configs
The DMAengine driver for omap-dma use three function calls from the
plat-omap legacy driver. When the DMAengine driver is built when ARCH_OMAP
is not set, the compilation will fail due to missing symbols.
Add empty inline functions to allow the DMAengine driver to be compiled
with COMPILE_TEST.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Peter Ujfalusi [Thu, 22 Sep 2016 06:31:04 +0000 (09:31 +0300)]
dmaengine: edma: Rename set_bits and remove unused clear_bits helper
The clear_bits() helper is not used by the driver so it can be removed.
powerpc architecture defines the set_bits() in
arch/powerpc/include/asm/bitops.h which results failed compile testing on
powerpc architecture:
>> drivers/dma/edma.c:415:20: error: conflicting types for 'set_bits'
static inline void set_bits(int offset, int len, unsigned long *p)
^~~~~~~~
In file included from include/linux/bitops.h:36:0,
from include/linux/kernel.h:10,
from include/linux/list.h:8,
from include/linux/kobject.h:20,
from include/linux/device.h:17,
from include/linux/dmaengine.h:20,
from drivers/dma/edma.c:16:
arch/powerpc/include/asm/bitops.h:75:14: note: previous definition of 'set_bits' was here
DEFINE_BITOP(set_bits, or, "")
^
arch/powerpc/include/asm/bitops.h:58:24: note: in definition of macro 'DEFINE_BITOP'
static __inline__ void fn(unsigned long mask, \
^~
>> drivers/dma/edma.c:421:20: error: conflicting types for 'clear_bits'
static inline void clear_bits(int offset, int len, unsigned long *p)
^~~~~~~~~~
In file included from include/linux/bitops.h:36:0,
from include/linux/kernel.h:10,
from include/linux/list.h:8,
from include/linux/kobject.h:20,
from include/linux/device.h:17,
from include/linux/dmaengine.h:20,
from drivers/dma/edma.c:16:
arch/powerpc/include/asm/bitops.h:76:14: note: previous definition of 'clear_bits' was here
DEFINE_BITOP(clear_bits, andc, "")
^
arch/powerpc/include/asm/bitops.h:58:24: note: in definition of macro 'DEFINE_BITOP'
static __inline__ void fn(unsigned long mask, \
^~
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Peter Ujfalusi [Wed, 21 Sep 2016 12:41:28 +0000 (15:41 +0300)]
dmaengine: edma: Fix of_device_id data parameter usage (legacy vs TPCC)
Use pointers to static constant variables for eDMA binding
type (legacy vs TPCC).
Fixes the following warning when compiling the driver for 64bit
architectures (x86_64 for example):
drivers/dma/edma.c:2185:16: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
if (match && (u32)match->data == EDMA_BINDING_TPCC)
^
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
dmaengine: s3c24xx: Add dma_slave_map for s3c2440 devices
This patch updates the s3c24xx dma driver to be able to pass a
dma_slave_map array via the platform data. This is needed to
be able to use the new, simpler dmaengine API [1].
I used the virtual DMA channels as a parameter for the dma_filter
function. By doing that, I could reuse the existing filter function in
drivers/dma/s3c24xx-dma.c.
I have tested this on my mini2440 board with the audio driver.
According to my observations, dma_request_slave_channel in the
function dmaengine_pcm_new in the file
sound/soc/soc-generic-dmaengine-pcm.c now returns a valid DMA channel
whereas before no DMA channel was returned at that point.
Entries for DMACH_XD0, DMACH_XD1 and DMACH_TIMER are missing because I
don't realy know which driver to use for these.
On some systems, pointer can be large than unsigned int, triggering warning
pointer-to-int-cast on conversion.
drivers/dma/s3c24xx-dma.c: In function 's3c24xx_dma_filter':
drivers/dma/s3c24xx-dma.c:1421:24: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
dmaengine: s3c24xx: use correct print specifiers for size_t
This driver warns:
drivers/dma/s3c24xx-dma.c: In function 's3c24xx_dma_prep_memcpy':
drivers/dma/s3c24xx-dma.c:826:2: warning: format '%d' expects argument of type 'int', but argument 4 has type 'size_t' [-Wformat=]
drivers/dma/s3c24xx-dma.c:830:3: warning: format '%d' expects argument of type 'int', but argument 3 has type 'size_t' [-Wformat=]
We should use %zu to print 'size_t' values.
Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
dmaengine: mmp_tdma: use correct print specifiers for size_t
This driver warns:
drivers/dma/mmp_tdma.c: In function 'mmp_tdma_prep_dma_cyclic':
drivers/dma/mmp_tdma.c:437:5: warning: format '%d' expects argument of type 'int', but argument 3 has type 'size_t' [-Wformat=]
dmaengine: ep93xx: use correct print specifiers for size_t
This driver warns:
drivers/dma/ep93xx_dma.c: In function 'ep93xx_dma_prep_slave_sg':
drivers/dma/ep93xx_dma.c:1054:6: warning: format '%d' expects argument of type 'int', but argument 3 has type 'size_t' [-Wformat=]
drivers/dma/ep93xx_dma.c: In function 'ep93xx_dma_prep_dma_cyclic':
drivers/dma/ep93xx_dma.c:1129:5: warning: format '%d' expects argument of type 'int', but argument 3 has type 'size_t' [-Wformat=]
dmaengine: sa11x0: use correct print specifiers for dma_addr_t
This driver when compiled on 64 bits gave warnings:
drivers/dma/sa11x0-dma.c:466:27: warning: format '%x' expects argument of type 'unsigned int', but argument 4 has type 'dma_addr_t' [-Wformat=]
drivers/dma/sa11x0-dma.c:554:31: warning: format '%x' expects argument of type 'unsigned int', but argument 5 has type 'dma_addr_t' [-Wformat=]
drivers/dma/sa11x0-dma.c:696:34: warning: format '%x' expects argument of type 'unsigned int', but argument 5 has type 'dma_addr_t' [-Wformat=]
We should use %pad to print 'dma_addr_t' values.
Acked-by: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
dmaengine: coh901318: use NULL for pointer initialization
Sparse complains:
drivers/dma/coh901318_lli.c:78:31: warning: Using plain integer as NULL pointer
drivers/dma/coh901318_lli.c:91:39: warning: Using plain integer as NULL pointer
Use NULL for pointer initialization rather than plain integer
On some systems, pointer can be large than unsigned int, triggering warning
pointer-to-int-cast on conversion.
drivers/dma/coh901318.c: In function 'coh901318_filter_id':
drivers/dma/coh901318.c:1769:23: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
unsigned int ch_nr = (unsigned int) chan_id;
Also, converting an iomem pointer for print leads to warn on some system
which can be avoided with %p specfier
drivers/dma/coh901318.c: In function 'coh901318_probe':
drivers/dma/coh901318.c:2748:3: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
(u32) base->virtbase);
dmaengine: coh901318: use correct print specifiers
This driver when compiled on 64 bits gave warnings:
drivers/dma/coh901318.c: In function 'coh901318_list_print':
warning: format '%x' expects argument of type 'unsigned int', but argument 7 has type 'dma_addr_t' [-Wformat=]
warning: format '%x' expects argument of type 'unsigned int', but argument 8 has type 'dma_addr_t' [-Wformat=]
warning: format '%x' expects argument of type 'unsigned int', but argument 9 has type 'dma_addr_t' [-Wformat=]
drivers/dma/coh901318.c: In function 'coh901318_prep_memcpy':
warning: format '%x' expects argument of type 'unsigned int', but argument 6 has type 'dma_addr_t' [-Wformat=]
warning: format '%x' expects argument of type 'unsigned int', but argument 7 has type 'dma_addr_t' [-Wformat=]
warning: format '%d' expects argument of type 'int', but argument 8 has type 'size_t' [-Wformat=]
We should use %pad to print 'dma_addr_t' values and %zu to print size_t
values
Niklas Söderlund [Wed, 10 Aug 2016 11:22:18 +0000 (13:22 +0200)]
dmaengine: rcar-dmac: group slave configuration
Group slave address and transfer size in own structs for source and
destination. This is in preparation for hooking up the dma-mapping API
to the slave addresses.
Niklas Söderlund [Wed, 10 Aug 2016 11:22:17 +0000 (13:22 +0200)]
arm: dma-mapping: add {map,unmap}_resource for iommu ops
Add methods to map/unmap device resources addresses for dma_map_ops that
are IOMMU aware. This is needed to map a device MMIO register from a
physical address.
Niklas Söderlund [Wed, 10 Aug 2016 11:22:15 +0000 (13:22 +0200)]
dma-debug: add support for resource mappings
A MMIO mapped resource can not be represented by a struct page so a new
debug type is needed to handle this. This patch add such type and
functionality to add/remove entries and how to translate them to a
physical address.
Niklas Söderlund [Wed, 10 Aug 2016 11:22:14 +0000 (13:22 +0200)]
dma-mapping: add {map,unmap}_resource to dma_map_ops
Add methods to handle mapping of device resources from a physical
address. This is needed for example to be able to map MMIO FIFO
registers to a IOMMU.
dmaengine: dmatest: Restore "memcpy" as default mode
Commit 0d4cb44da6ca0e8 ("dmaengine: dmatest: Add support for
scatter-gather DMA mode") changes default "dmatest" behavior by
changing default mode from "memcpy" to "scatter-gather".
Now "memcpy" gets back as default mode.
Andy Shevchenko [Tue, 23 Aug 2016 13:09:40 +0000 (16:09 +0300)]
dmaengine: hsu: refactor hsu_dma_do_irq() to return int
Since we have nice macro IRQ_RETVAL() we would use it to convert a flag of
handled interrupt from int to irqreturn_t.
The rationale of doing this is:
a) hence we implicitly mark hsu_dma_do_irq() as an auxiliary function that
can't be used as interrupt handler directly, and
b) to be in align with serial driver which is using serial8250_handle_irq()
that returns plain int by design.
Stefan Roese [Thu, 15 Sep 2016 05:37:31 +0000 (07:37 +0200)]
dmaengine: mv_xor: Add support for IO (PCIe) src/dst areas
To enable the access to a specific area, the MVEBU XOR controllers needs
to have this area enabled / mapped via an address window. Right now,
only the DRAM memory area is enabled via such memory windows. So
using this driver to DMA to / from a e.g. PCIe memory region is
currently not supported.
This patch now adds support for such PCIe / IO regions by checking
if the src / dst address is located in an IO memory area in contrast
to being located in DRAM. This is done by using the newly introduced
MBus function mvebu_mbus_get_io_win_info(). If the src / dst address
is located in such an IO area, a new address window is created in
the XOR DMA controller. Enabling the controller to access this area.
Signed-off-by: Stefan Roese <sr@denx.de> Cc: Gregory CLEMENT <gregory.clement@free-electrons.com> Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Cc: Marcin Wojtas <mw@semihalf.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Andrew Lunn <andrew@lunn.ch> Cc: Vinod Koul <vinod.koul@intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Stefan Roese [Thu, 15 Sep 2016 05:37:30 +0000 (07:37 +0200)]
bus: mvebu-mbus: Provide stub function for mvebu_mbus_get_io_win_info()
This patch provides a stub function for mvebu_mbus_get_io_win_info(),
which will be used for all non-Orion (ARM32 MVEBU) platforms for
compile test coverage.
On such platforms this function will return an error so that drivers
might detect a potential problem.
Signed-off-by: Stefan Roese <sr@denx.de> Acked-by: Gregory CLEMENT <gregory.clement@free-electrons.com> Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Cc: Marcin Wojtas <mw@semihalf.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Andrew Lunn <andrew@lunn.ch> Cc: Vinod Koul <vinod.koul@intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Tony Lindgren [Tue, 13 Sep 2016 17:22:43 +0000 (10:22 -0700)]
dmaengine: cppi41: Ignore EINPROGRESS for PM runtime
We can occasionally get -EINPROGRESS for pm_runtime_get. In that case
we can just continue as we're queueing transfers anyways when
pm_runtime_active is not set.
Hsin-Yu Chao [Tue, 23 Aug 2016 09:16:55 +0000 (17:16 +0800)]
dmaengine: pl330: Acquire dmac's spinlock in pl330_tx_status
There is a racing when accessing dmac thread in pl330_tx_status that
the pl330_update is handling active request at the same time and
changing the status of descriptors. This could cause an invalid
transferred count from BUSY descriptor added up to the residual number.
Fix the bug by using the dmac's spinlock in pl330_tx_status to protect
thread resources from changing.
Note that the nested order of holding dmac's and dma_chan's spinlock is
consistent with the rest of the driver: dma_chan first and then dmac,
so it is safe from deadlock scenario.
Stephen Barber [Fri, 19 Aug 2016 00:59:59 +0000 (17:59 -0700)]
dmaengine: pl330: fix residual for non-running BUSY descriptors
Only one descriptor in the work list should be running at
any given time, but it's possible to have an enqueued BUSY
descriptor that has not yet transferred any data, or for
a BUSY descriptor to linger briefly before transitioning
to DONE. These cases should be handled to keep residual
calculations consistent even with the non-running BUSY
descriptors in the work list.
Signed-off-by: Stephen Barber <smbarber@chromium.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
dmaengine: dmatest: Apply copy_align to DMA_SG as well
The DMA_SG is still a type of memory copy operation that should
conform the hardware restriction. So this patch just applies the
copy_align to DMA_SG as well.
dmaengine: cppi41: mark PM functions as __maybe_unused
When CONFIG_PM_SLEEP is disabled, we get a build error in
the cppi41 dmaengine driver, since the runtime-pm functions
are hidden within the wrong #ifdef:
drivers/dma/cppi41.c:1158:21: error: 'cppi41_runtime_suspend' undeclared here (not in a function)
This removes the #ifdef and instead uses __maybe_unused
annotations that cannot have this problem.
dmaengine: k3dma: use correct format string for debug output
The newly added k3_dma_prep_dma_cyclic function has some debug output
that uses incorrect typecasts, some of which cause a warning like:
drivers/dma/k3dma.c: In function 'k3_dma_prep_dma_cyclic':
drivers/dma/k3dma.c:589:671: error: cast to pointer from integer of different size [-Werror=int-to-pointer-cast]
In general, we have to print 'dma_addr_t' values using special
'%pad' format to get the correct behavior on kernels that have
a 64-bit dma_addr_t type but 32-bit pointers.
Similarly, printing size_t values should be done using the %z
modifier to get the correct behavior on 64-bit kernels.
A workaround for a warning introduced a use of the NO_IRQ
macro that should have been gone for a long time.
It is clear from the code that the value cannot actually
be used, but apparently there was a configuration at
some point that caused a warning, so instead of just
reverting that patch, this rearranges the code in a way that
the warning cannot reappear.
The mxs_chan->chan_irq variable is guaranteed to never be NO_IRQ,
as it gets assigned the result of platform_get_irq() that returns
either a valid positive interrupt number, or a negative failure
code that leads to the channel not being used.
This removes the redundant check, eliminating one more instance
of NO_IRQ.
Sinan Kaya [Thu, 1 Sep 2016 14:02:55 +0000 (10:02 -0400)]
dmaengine: dmatest: exclude compare and fill time during perf report
Dmatest is currently including compare and fill time into the
calculated performance numbers. This does not reflect the HW
capability and the results vary based on the CPU speed instead of
the HW speed.
Signed-off-by: Sinan Kaya <okaya@codeaurora.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Sinan Kaya [Wed, 31 Aug 2016 15:10:29 +0000 (11:10 -0400)]
dmaengine: qcom_hidma: add error reporting for tx_status
The HIDMA driver is capable of error detection. However, the error was
not being passed back to the client when tx_status API is called.
Changing the error handling behavior to follow this oder.
1. dmaengine asserts error interrupt
2. Driver receives and mark's the txn as error
3. Driver completes the txn and intimates the client. No further
submissions. Drop the locks before calling callback, as subsequent
processing by client maybe in callback thread.
4. Client invokes status and you can return error
5. On error, client calls terminate_all. You can reset channel, free all
descriptors in the active, pending and completed lists
6. Client prepares new txn and so on.
As part of this work, got rid of the reset in the interrupt handler when
an error happens and the HW is put into disabled state. The only way to
recover is for the client to terminate the channel.
Signed-off-by: Sinan Kaya <okaya@codeaurora.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Sinan Kaya [Wed, 31 Aug 2016 15:10:28 +0000 (11:10 -0400)]
dmaengine: qcom_hidma: report transfer errors with new interface
Pass the DMA errors to the client by passing a result argument. The HW only
supports a generic error when something goes wrong. That's why, using
DMA_TRANS_ABORTED all the time.
Signed-off-by: Sinan Kaya <okaya@codeaurora.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Sinan Kaya [Wed, 31 Aug 2016 15:10:27 +0000 (11:10 -0400)]
dmaengine: qcom_hidma: release the descriptor before the callback
There is a race condition between data transfer callback and descriptor
free code. The callback routine may decide to clear the resources even
though the descriptor has not yet been freed.
Instead of calling the callback first and then releasing the memory,
this code is changing the order to return the descriptor back to the
free pool and then call the user provided callback.
Signed-off-by: Sinan Kaya <okaya@codeaurora.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Tony Lindgren [Wed, 31 Aug 2016 14:19:59 +0000 (07:19 -0700)]
dmaengine: cppi41: Add basic PM runtime support
Let's keep the device enabled between cppi41_dma_issue_pending()
and dmaengine_desc_get_callback_invoke() and rely on the PM runtime
autoidle timeout elsewhere.
As the PM runtime is for whole device, not for each channel,
we need to queue pending transfers if the device is PM runtime
suspended. Then we start the pending transfers in PM runtime
resume.
Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
John Stultz [Mon, 29 Aug 2016 17:30:53 +0000 (10:30 -0700)]
Kconfig: Allow k3dma driver to be selected for more then HISI3xx platforms
This allows the k3dma driver to be selected on HiKey via the ARCH_HISI
dependency.
Cc: Zhangfei Gao <zhangfei.gao@linaro.org> Cc: Jingoo Han <jg1.han@samsung.com> Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Vinod Koul <vinod.koul@intel.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Andy Green <andy@warmcat.com> Acked-by: Zhangfei Gao <zhangfei.gao@linaro.org> Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Andy Green [Mon, 29 Aug 2016 17:30:52 +0000 (10:30 -0700)]
k3dma: Add cyclic mode for audio
Currently the k3dma driver doesn't offer the cyclic mode
necessary for handling audio.
This patch adds it.
Cc: Zhangfei Gao <zhangfei.gao@linaro.org> Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Vinod Koul <vinod.koul@intel.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Andy Green <andy@warmcat.com> Acked-by: Zhangfei Gao <zhangfei.gao@linaro.org> Signed-off-by: Andy Green <andy.green@linaro.org>
[jstultz: Forward ported to mainline, removed a few
bits of logic that didn't seem to have much effect] Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
John Stultz [Mon, 29 Aug 2016 17:30:51 +0000 (10:30 -0700)]
k3dma: Fix memory handling in preparation for cyclic mode
With cyclic mode, the shared virt-dma logic doesn't actually
manage the descriptor state, nor the calling of the descriptor
free callback. This results in leaking a desc structure every
time we start an audio transfer.
Thus we must manage it ourselves. The k3dma driver already keeps
track of the active and finished descriptors via ds_run and ds_done
pointers, so cleanup how we handle those two values, so when we
tear down everything in terminate_all, call free_desc on the ds_run
and ds_done pointers if they are not null.
NOTE: HiKey doesn't use the non-cyclic dma modes, so I'm not been
able to test those modes. But with this patch we no longer leak
the desc structures.
Cc: Zhangfei Gao <zhangfei.gao@linaro.org> Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Vinod Koul <vinod.koul@intel.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Andy Green <andy@warmcat.com> Acked-by: Zhangfei Gao <zhangfei.gao@linaro.org> Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
John Stultz [Mon, 29 Aug 2016 17:30:50 +0000 (10:30 -0700)]
k3dma: Fix occasional DMA ERR issue by using proper dma api
After lots of debugging on an occasional DMA ERR issue, I realized
that the desc structures which we point the dma hardware are being
allocated out of regular memory. This means when we fill the desc
structures, that data doesn't always get flushed out to memory by
the time we start the dma transfer, resulting in the dma engine getting
some null values, resulting in a DMA ERR on the first irq.
Thus, this patch adopts mechanism similar to the zx296702_dma of
allocating the desc structures from a dma pool, so the memory caching
rules are properly set to avoid this issue.
Cc: Zhangfei Gao <zhangfei.gao@linaro.org> Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Vinod Koul <vinod.koul@intel.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Andy Green <andy@warmcat.com> Acked-by: Zhangfei Gao <zhangfei.gao@linaro.org> Signed-off-by: John Stutlz <john.stultz@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Andy Green [Mon, 29 Aug 2016 17:30:49 +0000 (10:30 -0700)]
k3dma: Fix "nobody cared" message seen on any error
As it was before, as soon as the DMAC IP felt there was an error
he would return IRQ_NONE since no actual transfer had completed.
After spinning on that for 100K interrupts, Linux yanks the IRQ with
a "nobody cared" error.
This patch lets it handle the interrupt and keep the IRQ alive.
Cc: Zhangfei Gao <zhangfei.gao@linaro.org> Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Vinod Koul <vinod.koul@intel.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Andy Green <andy@warmcat.com> Acked-by: Zhangfei Gao <zhangfei.gao@linaro.org> Signed-off-by: Andy Green <andy.green@linaro.org>
[jstultz: Forward ported to mainline] Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Andy Green [Mon, 29 Aug 2016 17:30:48 +0000 (10:30 -0700)]
k3dma: Fix dma err offsets
The offsets for ERR1 and ERR2 are wrong actually.
That's why you can never clear an error.
Cc: Zhangfei Gao <zhangfei.gao@linaro.org> Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Vinod Koul <vinod.koul@intel.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Andy Green <andy@warmcat.com> Acked-by: Zhangfei Gao <zhangfei.gao@linaro.org> Signed-off-by: Andy Green <andy.green@linaro.org>
[jstultz: Forward ported to mainline] Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>