From: Peter Ujfalusi <peter.ujfalusi@ti.com>
To: <vinod.koul@intel.com>, <arnd@arndb.de>, <andy.shevchenko@gmail.com>
Cc: tony@atomide.com, linux-mmc@vger.kernel.org, nsekhar@ti.com,
linux-kernel@vger.kernel.org, linux-spi@vger.kernel.org,
dmaengine@vger.kernel.org, linux-omap@vger.kernel.org,
linux-arm-kernel@lists.infradead.org
Subject: [RFC v03 03/15] dmaengine: core: Introduce new, universal API to request a channel
Date: Wed, 2 Dec 2015 15:59:13 +0200 [thread overview]
Message-ID: <1449064765-27427-4-git-send-email-peter.ujfalusi@ti.com> (raw)
In-Reply-To: <1449064765-27427-1-git-send-email-peter.ujfalusi@ti.com>
The two API function can cover most, if not all current APIs used to
request a channel. With minimal effort dmaengine drivers, platforms and
dmaengine user drivers can be converted to use the two function.
struct dma_chan *dma_request_chan_by_mask(const dma_cap_mask_t *mask);
To request any channel matching with the requested capabilities, can be
used to request channel for memcpy, memset, xor, etc where no hardware
synchronization is needed.
struct dma_chan *dma_request_chan(struct device *dev, const char *name);
To request a slave channel. The dma_request_chan() will try to find the
channel via DT, ACPI or in case if the kernel booted in non DT/ACPI mode
it will use a filter lookup table and retrieves the needed information from
the dma_filter_map provided by the DMA drivers.
This legacy mode needs changes in platform code, in dmaengine drivers and
finally the dmaengine user drivers can be converted:
For each dmaengine driver an array of DMA device, slave and the parameter
for the filter function needs to be added:
static const struct dma_filter_map da830_edma_map[] = {
{ "davinci-mcasp.0", "rx", EDMA_FILTER_PARAM(0, 0) },
{ "davinci-mcasp.0", "tx", EDMA_FILTER_PARAM(0, 1) },
{ "davinci-mcasp.1", "rx", EDMA_FILTER_PARAM(0, 2) },
{ "davinci-mcasp.1", "tx", EDMA_FILTER_PARAM(0, 3) },
{ "davinci-mcasp.2", "rx", EDMA_FILTER_PARAM(0, 4) },
{ "davinci-mcasp.2", "tx", EDMA_FILTER_PARAM(0, 5) },
{ "spi_davinci.0", "rx", EDMA_FILTER_PARAM(0, 14) },
{ "spi_davinci.0", "tx", EDMA_FILTER_PARAM(0, 15) },
{ "da830-mmc.0", "rx", EDMA_FILTER_PARAM(0, 16) },
{ "da830-mmc.0", "tx", EDMA_FILTER_PARAM(0, 17) },
{ "spi_davinci.1", "rx", EDMA_FILTER_PARAM(0, 18) },
{ "spi_davinci.1", "tx", EDMA_FILTER_PARAM(0, 19) },
};
This information is going to be needed by the dmaengine driver, so
modification to the platform_data is needed, and the driver map should be
added to the pdata of the DMA driver:
da8xx_edma0_pdata.slave_map = da830_edma_map;
da8xx_edma0_pdata.slavecnt = ARRAY_SIZE(da830_edma_map);
The DMA driver then needs to configure the needed device -> filter_fn
mapping before it registers with dma_async_device_register() :
if (info->slave_map) {
ecc->dma_slave.filter_map.map = info->slave_map;
ecc->dma_slave.filter_map.mapcnt = info->slavecnt;
ecc->dma_slave.filter_map.filter_fn = edma_filter_fn;
}
When neither DT or ACPI lookup is available the dma_request_chan() will
try to match the requester's device name with the filter_map's list of
device names, when a match found it will use the information from the
dma_filter_map to get the channel with the dma_get_channel() internal
function.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
---
Documentation/dmaengine/client.txt | 23 +++------
drivers/dma/dmaengine.c | 98 ++++++++++++++++++++++++++++++++++++++
include/linux/dmaengine.h | 27 +++++++++++
3 files changed, 131 insertions(+), 17 deletions(-)
diff --git a/Documentation/dmaengine/client.txt b/Documentation/dmaengine/client.txt
index d9f9f461102a..6c72a06eb1a5 100644
--- a/Documentation/dmaengine/client.txt
+++ b/Documentation/dmaengine/client.txt
@@ -22,25 +22,14 @@ The slave DMA usage consists of following steps:
Channel allocation is slightly different in the slave DMA context,
client drivers typically need a channel from a particular DMA
controller only and even in some cases a specific channel is desired.
- To request a channel dma_request_channel() API is used.
+ To request a channel dma_request_chan() API is used.
Interface:
- struct dma_chan *dma_request_channel(dma_cap_mask_t mask,
- dma_filter_fn filter_fn,
- void *filter_param);
- where dma_filter_fn is defined as:
- typedef bool (*dma_filter_fn)(struct dma_chan *chan, void *filter_param);
-
- The 'filter_fn' parameter is optional, but highly recommended for
- slave and cyclic channels as they typically need to obtain a specific
- DMA channel.
-
- When the optional 'filter_fn' parameter is NULL, dma_request_channel()
- simply returns the first channel that satisfies the capability mask.
-
- Otherwise, the 'filter_fn' routine will be called once for each free
- channel which has a capability in 'mask'. 'filter_fn' is expected to
- return 'true' when the desired DMA channel is found.
+ struct dma_chan *dma_request_chan(struct device *dev, const char *name);
+
+ Which will find and return the 'name' DMA channel associated with the 'dev'
+ device. The association is done via DT, ACPI or board file based
+ dma_map_filter matching table.
A channel allocated via this interface is exclusive to the caller,
until dma_release_channel() is called.
diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c
index ea9d66982d40..b9dc1512c4aa 100644
--- a/drivers/dma/dmaengine.c
+++ b/drivers/dma/dmaengine.c
@@ -43,6 +43,7 @@
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+#include <linux/platform_device.h>
#include <linux/dma-mapping.h>
#include <linux/init.h>
#include <linux/module.h>
@@ -712,6 +713,103 @@ struct dma_chan *dma_request_slave_channel(struct device *dev,
}
EXPORT_SYMBOL_GPL(dma_request_slave_channel);
+const static struct dma_filter_map *dma_filter_match(struct dma_device *device,
+ const char *name,
+ struct device *dev)
+{
+ const struct dma_filter_map *map_found = NULL;
+ int i;
+
+ if (!device->filter_map.mapcnt)
+ return NULL;
+
+ for (i = 0; i < device->filter_map.mapcnt; i++) {
+ const struct dma_filter_map *map = &device->filter_map.map[i];
+
+ if (!strcmp(map->devname, dev_name(dev)) &&
+ !strcmp(map->slave, name)) {
+ map_found = map;
+ break;
+ }
+ }
+
+ return map_found;
+}
+
+/**
+ * dma_request_chan - try to allocate an exclusive slave channel
+ * @dev: pointer to client device structure
+ * @name: slave channel name
+ *
+ * Returns pointer to appropriate DMA channel on success or an error pointer.
+ */
+struct dma_chan *dma_request_chan(struct device *dev, const char *name)
+{
+ struct dma_device *device, *_d;
+ struct dma_chan *chan = NULL;
+
+ /* If device-tree is present get slave info from here */
+ if (dev->of_node)
+ chan = of_dma_request_slave_channel(dev->of_node, name);
+
+ /* If device was enumerated by ACPI get slave info from here */
+ if (has_acpi_companion(dev) && !chan)
+ chan = acpi_dma_request_slave_chan_by_name(dev, name);
+
+ if (chan) {
+ /* Valid channel found */
+ if (!IS_ERR(chan))
+ return chan;
+
+ pr_warn("%s: %s DMA request failed, falling back to legacy\n",
+ __func__, dev->of_node ? "OF" : "ACPI");
+ }
+
+ /* Try to find the channel via the DMA filter map(s) */
+ mutex_lock(&dma_list_mutex);
+ list_for_each_entry_safe(device, _d, &dma_device_list, global_node) {
+ dma_cap_mask_t mask;
+ const struct dma_filter_map *map = dma_filter_match(device,
+ name, dev);
+
+ if (!map)
+ continue;
+
+ dma_cap_zero(mask);
+ dma_cap_set(DMA_SLAVE, mask);
+
+ chan = find_candidate(device, &mask,
+ device->filter_map.filter_fn, map->param);
+ if (!IS_ERR(chan))
+ break;
+ }
+ mutex_unlock(&dma_list_mutex);
+
+ return chan ? chan : ERR_PTR(-EPROBE_DEFER);
+}
+EXPORT_SYMBOL_GPL(dma_request_chan);
+
+/**
+ * dma_request_chan_by_mask - allocate a channel satisfying certain capabilities
+ * @mask: capabilities that the channel must satisfy
+ *
+ * Returns pointer to appropriate DMA channel on success or an error pointer.
+ */
+struct dma_chan *dma_request_chan_by_mask(const dma_cap_mask_t *mask)
+{
+ struct dma_chan *chan;
+
+ if (!mask)
+ return ERR_PTR(-ENODEV);
+
+ chan = __dma_request_channel(mask, NULL, NULL);
+ if (!chan)
+ chan = ERR_PTR(-ENODEV);
+
+ return chan;
+}
+EXPORT_SYMBOL_GPL(dma_request_chan_by_mask);
+
void dma_release_channel(struct dma_chan *chan)
{
mutex_lock(&dma_list_mutex);
diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
index a2b7c2071cf4..49d48e69c179 100644
--- a/include/linux/dmaengine.h
+++ b/include/linux/dmaengine.h
@@ -606,6 +606,18 @@ enum dmaengine_alignment {
DMAENGINE_ALIGN_64_BYTES = 6,
};
+struct dma_filter_map {
+ char *devname;
+ char *slave;
+ void *param;
+};
+
+struct dma_filter {
+ dma_filter_fn filter_fn;
+ int mapcnt;
+ const struct dma_filter_map *map;
+};
+
/**
* struct dma_device - info on the entity supplying DMA services
* @chancnt: how many DMA channels are supported
@@ -669,6 +681,7 @@ struct dma_device {
unsigned int privatecnt;
struct list_head channels;
struct list_head global_node;
+ struct dma_filter filter_map;
dma_cap_mask_t cap_mask;
unsigned short max_xor;
unsigned short max_pq;
@@ -1235,6 +1248,10 @@ struct dma_chan *__dma_request_channel(const dma_cap_mask_t *mask,
struct dma_chan *dma_request_slave_channel_reason(struct device *dev,
const char *name);
struct dma_chan *dma_request_slave_channel(struct device *dev, const char *name);
+
+struct dma_chan *dma_request_chan(struct device *dev, const char *name);
+struct dma_chan *dma_request_chan_by_mask(const dma_cap_mask_t *mask);
+
void dma_release_channel(struct dma_chan *chan);
int dma_get_slave_caps(struct dma_chan *chan, struct dma_slave_caps *caps);
#else
@@ -1268,6 +1285,16 @@ static inline struct dma_chan *dma_request_slave_channel(struct device *dev,
{
return NULL;
}
+static inline struct dma_chan *dma_request_chan(struct device *dev,
+ const char *name)
+{
+ return ERR_PTR(-ENODEV);
+}
+static inline struct dma_chan *dma_request_chan_by_mask(
+ const dma_cap_mask_t *mask)
+{
+ return ERR_PTR(-ENODEV);
+}
static inline void dma_release_channel(struct dma_chan *chan)
{
}
--
2.6.3
next prev parent reply other threads:[~2015-12-02 13:59 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-12-02 13:59 [RFC v03 00/15] dmaengine: New 'universal' API for requesting channel Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 01/15] dmaengine: core: Skip mask matching when it is not provided to private_candidate Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 02/15] dmaengine: core: Move and merge the code paths using private_candidate Peter Ujfalusi
2015-12-02 13:59 ` Peter Ujfalusi [this message]
[not found] ` <1449064765-27427-4-git-send-email-peter.ujfalusi-l0cyMroinI0@public.gmane.org>
2015-12-02 14:35 ` [RFC v03 03/15] dmaengine: core: Introduce new, universal API to request a channel Andy Shevchenko
2015-12-03 10:34 ` Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 04/15] dmaengine: edma: Add support for DMA filter mapping to slave devices Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 05/15] ARM: davinci: devices-da8xx: Add dma_filter_map to edma Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 06/15] ARM: davinci: dm355: " Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 07/15] ARM: davinci: dm365: " Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 08/15] ARM: davinci: dm644x: " Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 09/15] ARM: davinci: dm646x: " Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 10/15] mmc: davinci_mmc: Use dma_request_chan() to requesting DMA channel Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 11/15] spi: davinci: " Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 12/15] ARM: davinci: devices-da8xx: Remove DMA resources for MMC and SPI Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 13/15] ARM: davinci: devices: Remove DMA resources for MMC Peter Ujfalusi
[not found] ` <1449064765-27427-1-git-send-email-peter.ujfalusi-l0cyMroinI0@public.gmane.org>
2015-12-02 13:59 ` [RFC v03 14/15] ARM: davinci: dm355: Remove DMA resources for SPI Peter Ujfalusi
2015-12-02 13:59 ` [RFC v03 15/15] ARM: davinci: dm365: " Peter Ujfalusi
2015-12-02 14:38 ` [RFC v03 00/15] dmaengine: New 'universal' API for requesting channel Andy Shevchenko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1449064765-27427-4-git-send-email-peter.ujfalusi@ti.com \
--to=peter.ujfalusi@ti.com \
--cc=andy.shevchenko@gmail.com \
--cc=arnd@arndb.de \
--cc=dmaengine@vger.kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mmc@vger.kernel.org \
--cc=linux-omap@vger.kernel.org \
--cc=linux-spi@vger.kernel.org \
--cc=nsekhar@ti.com \
--cc=tony@atomide.com \
--cc=vinod.koul@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).