DMA Engine development
 help / color / mirror / Atom feed
From: Frank Li <Frank.li@oss.nxp.com>
To: "Nuno Sá" <nuno.sa@analog.com>
Cc: linux-kernel@vger.kernel.org, dmaengine@vger.kernel.org,
	linux-arm-msm@vger.kernel.org,
	linux-stm32@st-md-mailman.stormreply.com,
	linux-arm-kernel@lists.infradead.org, linux-iio@vger.kernel.org,
	linux-sound@vger.kernel.org, linux-spi@vger.kernel.org,
	"Vinod Koul" <vkoul@kernel.org>, "Frank Li" <Frank.Li@kernel.org>,
	"Lars-Peter Clausen" <lars@metafoo.de>,
	"Eugeniy Paltsev" <Eugeniy.Paltsev@synopsys.com>,
	"Amélie Delaunay" <amelie.delaunay@foss.st.com>,
	"Maxime Coquelin" <mcoquelin.stm32@gmail.com>,
	"Alexandre Torgue" <alexandre.torgue@foss.st.com>,
	"Jonathan Cameron" <jic23@kernel.org>,
	"David Lechner" <dlechner@baylibre.com>,
	"Andy Shevchenko" <andy@kernel.org>,
	"Jaroslav Kysela" <perex@perex.cz>,
	"Takashi Iwai" <tiwai@suse.com>,
	"Mark Brown" <broonie@kernel.org>
Subject: Re: [PATCH v6 1/9] dmaengine: Support bus widths of 32 bytes and above
Date: Wed, 14 Oct 2026 10:28:00 -0500	[thread overview]
Message-ID: <as-fgDViIz4oSap7@SMW015318> (raw)
In-Reply-To: <20261014-dmaengine-support-wider-dma-masks-v6-1-5b3ffdef8191@analog.com>

On Wed, Oct 14, 2026 at 03:19:11PM +0100, Nuno Sá wrote:
> The src_addr_widths and dst_addr_widths capability masks encode each
> supported width as a bit whose position equals the corresponding
> enum dma_slave_buswidth value (e.g. DMA_SLAVE_BUSWIDTH_4_BYTES sets bit
> 4). As these masks are plain u32, widths of 32 bytes and above
> (DMA_SLAVE_BUSWIDTH_32/64/128_BYTES map to bits 32, 64 and 128) cannot
> be represented at all.
>
> Introduce bitmap-based bus width capabilities that span the full enum
> range, through a new dma_buswidth_mask_t type modeled after
> dma_cap_mask_t. To allow DMA controller drivers to be converted
> incrementally, the legacy dma_device u32 fields are kept alongside the
> new masks and the core folds a legacy-only driver's u32 into the mask
> when the device is registered, so consumers only ever have to look at
> the mask.
>
> The fold is bidirectional while both representations coexist. A driver
> that only fills in the legacy u32 gets its mask derived from it, so the
> consumers already converted see it. A driver that only fills in the mask
> gets its legacy u32 derived from the mask, so the consumers not
> converted yet, which read the legacy dma_slave_caps fields, keep working.
>
> On top of that, dma_get_slave_caps() derives the legacy dma_slave_caps
> masks from the new ones when a device_caps() callback adjusted them, so
> that a converted controller narrowing its per-channel capabilities is
> still seen by the consumers not converted yet, while a driver adjusting
> the legacy masks directly keeps working.
>
> Signed-off-by: Nuno Sá <nuno.sa@analog.com>
>
> ---

Reviewed-by: Frank Li <Frank.Li@nxp.com>

>
> Andy, did not take your tag given that you're the one asking for the
> split and since we ended up not doing it, not sure if the tag is
> to be kept :)
> ---
>  drivers/dma/dmaengine.c   |  50 +++++++++++-
>  include/linux/dmaengine.h | 197 +++++++++++++++++++++++++++++++++++++++++++---
>  2 files changed, 234 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c
> index c5fef57b5adf..b47008595332 100644
> --- a/drivers/dma/dmaengine.c
> +++ b/drivers/dma/dmaengine.c
> @@ -33,6 +33,7 @@
>
>  #include <linux/acpi.h>
>  #include <linux/acpi_dma.h>
> +#include <linux/bitmap.h>
>  #include <linux/device.h>
>  #include <linux/dma-mapping.h>
>  #include <linux/dmaengine.h>
> @@ -592,6 +593,8 @@ int dma_get_slave_caps(struct dma_chan *chan, struct dma_slave_caps *caps)
>  	if (!device->directions)
>  		return -ENXIO;
>
> +	dmaengine_bus_width_copy(&caps->src_bus_widths, &device->src_bus_widths);
> +	dmaengine_bus_width_copy(&caps->dst_bus_widths, &device->dst_bus_widths);
>  	caps->src_addr_widths = device->src_addr_widths;
>  	caps->dst_addr_widths = device->dst_addr_widths;
>  	caps->directions = device->directions;
> @@ -611,9 +614,31 @@ int dma_get_slave_caps(struct dma_chan *chan, struct dma_slave_caps *caps)
>  	 * callback to override the generic capabilities with
>  	 * channel-specific ones.
>  	 */
> -	if (device->device_caps)
> +	if (device->device_caps) {
>  		device->device_caps(chan, caps);
>
> +		/*
> +		 * A driver already converted to the bus width interface
> +		 * adjusts the masks, so derive the legacy capabilities from
> +		 * them for the consumers not converted yet. Drivers not
> +		 * converted adjust the legacy capabilities directly, in which
> +		 * case there is nothing to do.
> +		 *
> +		 * Goes away with the legacy dma_slave_caps fields.
> +		 */
> +		if (!bitmap_equal(caps->src_bus_widths.bits,
> +				  device->src_bus_widths.bits,
> +				  DMA_SLAVE_BUSWIDTH_MAX))
> +			caps->src_addr_widths = bitmap_read(caps->src_bus_widths.bits,
> +							    0, 32);
> +
> +		if (!bitmap_equal(caps->dst_bus_widths.bits,
> +				  device->dst_bus_widths.bits,
> +				  DMA_SLAVE_BUSWIDTH_MAX))
> +			caps->dst_addr_widths = bitmap_read(caps->dst_bus_widths.bits,
> +							    0, 32);
> +	}
> +
>  	return 0;
>  }
>  EXPORT_SYMBOL_GPL(dma_get_slave_caps);
> @@ -1173,6 +1198,27 @@ void dma_async_device_channel_unregister(struct dma_device *device,
>  }
>  EXPORT_SYMBOL_GPL(dma_async_device_channel_unregister);
>
> +/*
> + * DMA controller drivers not converted to the bus width helpers only fill in
> + * the legacy u32 masks, which cannot hold widths of 32 bytes and above. Fold
> + * them into the mask so that consumers only ever have to look at the mask.
> + *
> + * Goes away with the legacy dma_device fields.
> + */
> +static void dma_device_fold_legacy_bus_widths(struct dma_device *device)
> +{
> +	if (bitmap_empty(device->src_bus_widths.bits, DMA_SLAVE_BUSWIDTH_MAX))
> +		bitmap_from_arr32(device->src_bus_widths.bits, &device->src_addr_widths, 32);
> +	/* consumers not converted yet still read the legacy caps */
> +	else if (!device->src_addr_widths)
> +		device->src_addr_widths = bitmap_read(device->src_bus_widths.bits, 0, 32);
> +
> +	if (bitmap_empty(device->dst_bus_widths.bits, DMA_SLAVE_BUSWIDTH_MAX))
> +		bitmap_from_arr32(device->dst_bus_widths.bits, &device->dst_addr_widths, 32);
> +	else if (!device->dst_addr_widths)
> +		device->dst_addr_widths = bitmap_read(device->dst_bus_widths.bits, 0, 32);
> +}
> +
>  /**
>   * dma_async_device_register - registers DMA devices found
>   * @device:	pointer to &struct dma_device
> @@ -1234,6 +1280,8 @@ int dma_async_device_register(struct dma_device *device)
>  		dev_dbg(device->dev,
>  			 "WARN: Device release is not defined so it is not safe to unbind this driver while in use\n");
>
> +	dma_device_fold_legacy_bus_widths(device);
> +
>  	kref_init(&device->ref);
>
>  	/* note: this only matters in the
> diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
> index 71cde09ff240..b190bfb42df3 100644
> --- a/include/linux/dmaengine.h
> +++ b/include/linux/dmaengine.h
> @@ -427,8 +427,168 @@ enum dma_slave_buswidth {
>  	DMA_SLAVE_BUSWIDTH_32_BYTES = 32,
>  	DMA_SLAVE_BUSWIDTH_64_BYTES = 64,
>  	DMA_SLAVE_BUSWIDTH_128_BYTES = 128,
> +	DMA_SLAVE_BUSWIDTH_MAX
>  };
>
> +/**
> + * typedef dma_buswidth_mask_t - bus width capabilities bitmap modeled after
> + * dma_cap_mask_t.
> + *
> + * Each supported bus width is represented by the bit whose position equals the
> + * corresponding enum dma_slave_buswidth value, e.g. a device supporting a bus
> + * width of 4 bytes has bit 4 set.
> + */
> +typedef struct {
> +	DECLARE_BITMAP(bits, DMA_SLAVE_BUSWIDTH_MAX);
> +} dma_buswidth_mask_t;
> +
> +/**
> + * dmaengine_bus_width_valid - test if a bus width is a valid one
> + * @width: bus width to validate
> + *
> + * Return: true if @width is a valid &enum dma_slave_buswidth, false otherwise.
> + */
> +static inline bool dmaengine_bus_width_valid(enum dma_slave_buswidth width)
> +{
> +	switch (width) {
> +	case DMA_SLAVE_BUSWIDTH_UNDEFINED:
> +	case DMA_SLAVE_BUSWIDTH_1_BYTE:
> +	case DMA_SLAVE_BUSWIDTH_2_BYTES:
> +	case DMA_SLAVE_BUSWIDTH_3_BYTES:
> +	case DMA_SLAVE_BUSWIDTH_4_BYTES:
> +	case DMA_SLAVE_BUSWIDTH_8_BYTES:
> +	case DMA_SLAVE_BUSWIDTH_16_BYTES:
> +	case DMA_SLAVE_BUSWIDTH_32_BYTES:
> +	case DMA_SLAVE_BUSWIDTH_64_BYTES:
> +	case DMA_SLAVE_BUSWIDTH_128_BYTES:
> +		return true;
> +	default:
> +		return false;
> +	}
> +}
> +
> +/**
> + * dmaengine_bus_width_set_many - set the supported bus widths
> + * @mask: bus width mask
> + * @widths: array of supported bus widths
> + * @n_widths: number of entries in @widths
> + *
> + * Like __set_bit(), the widths are added to @mask and the bits already set in it
> + * are left untouched, so @mask must be initialized by the caller.
> + *
> + * Return: 0 on success, -EINVAL if @widths contains an invalid bus width. Note
> + * that the bus widths validated before the failing one are still set.
> + */
> +static inline int dmaengine_bus_width_set_many(dma_buswidth_mask_t *mask,
> +					       const enum dma_slave_buswidth *widths,
> +					       unsigned int n_widths)
> +{
> +	for (unsigned int i = 0; i < n_widths; i++) {
> +		if (!dmaengine_bus_width_valid(widths[i]))
> +			return -EINVAL;
> +
> +		__set_bit(widths[i], mask->bits);
> +	}
> +
> +	return 0;
> +}
> +
> +/**
> + * dmaengine_bus_width_set - set a single supported bus width
> + * @mask: bus width mask
> + * @width: supported bus width
> + *
> + * Like __set_bit(), @width is added to @mask and the bits already set in it are
> + * left untouched, so @mask must be initialized by the caller.
> + *
> + * Return: 0 on success, -EINVAL if @width is invalid.
> + */
> +static inline int dmaengine_bus_width_set(dma_buswidth_mask_t *mask,
> +					  enum dma_slave_buswidth width)
> +{
> +	return dmaengine_bus_width_set_many(mask, &width, 1);
> +}
> +
> +/**
> + * dmaengine_bus_width_clear - remove a bus width from a bus width mask
> + * @mask: bus width mask
> + * @width: bus width to clear
> + *
> + * Return: 0 on success, -EINVAL if @width is invalid.
> + */
> +static inline int dmaengine_bus_width_clear(dma_buswidth_mask_t *mask,
> +					    enum dma_slave_buswidth width)
> +{
> +	if (!dmaengine_bus_width_valid(width))
> +		return -EINVAL;
> +
> +	__clear_bit(width, mask->bits);
> +
> +	return 0;
> +}
> +
> +/**
> + * dmaengine_bus_width_test - test if a bus width is part of a bus width mask
> + * @mask: bus width mask
> + * @width: bus width to test
> + *
> + * Return: true if @width is set in @mask, false otherwise.
> + */
> +static inline bool dmaengine_bus_width_test(const dma_buswidth_mask_t *mask,
> +					    enum dma_slave_buswidth width)
> +{
> +	if (!dmaengine_bus_width_valid(width))
> +		return false;
> +
> +	return test_bit(width, mask->bits);
> +}
> +
> +/**
> + * dmaengine_bus_width_min - get the smallest bus width of a bus width mask
> + * @mask: bus width mask
> + *
> + * Return: the smallest bus width set in @mask, or
> + * %DMA_SLAVE_BUSWIDTH_UNDEFINED if @mask is empty.
> + */
> +static inline enum dma_slave_buswidth
> +dmaengine_bus_width_min(const dma_buswidth_mask_t *mask)
> +{
> +	enum dma_slave_buswidth width;
> +
> +	width = find_first_bit(mask->bits, DMA_SLAVE_BUSWIDTH_MAX);
> +	if (width == DMA_SLAVE_BUSWIDTH_MAX)
> +		return DMA_SLAVE_BUSWIDTH_UNDEFINED;
> +
> +	return width;
> +}
> +
> +/**
> + * dmaengine_bus_width_copy - copy a bus width mask
> + * @dst: bus width mask to copy to
> + * @src: bus width mask to copy from
> + */
> +static inline void dmaengine_bus_width_copy(dma_buswidth_mask_t *dst,
> +					    const dma_buswidth_mask_t *src)
> +{
> +	bitmap_copy(dst->bits, src->bits, DMA_SLAVE_BUSWIDTH_MAX);
> +}
> +
> +/**
> + * dmaengine_bus_width_and - intersect two bus width masks
> + * @dst: bus width mask to store the result in
> + * @src1: first bus width mask
> + * @src2: second bus width mask
> + *
> + * Return: true if @dst has at least one bus width set, false otherwise.
> + */
> +static inline bool dmaengine_bus_width_and(dma_buswidth_mask_t *dst,
> +					   const dma_buswidth_mask_t *src1,
> +					   const dma_buswidth_mask_t *src2)
> +{
> +	return bitmap_and(dst->bits, src1->bits, src2->bits,
> +			  DMA_SLAVE_BUSWIDTH_MAX);
> +}
> +
>  /**
>   * struct dma_slave_config - dma slave channel runtime config
>   * @direction: whether the data shall go in or out on this slave
> @@ -523,10 +683,11 @@ enum dma_residue_granularity {
>
>  /**
>   * struct dma_slave_caps - expose capabilities of a slave channel only
> - * @src_addr_widths: bit mask of src addr widths the channel supports.
> - *	Width is specified in bytes, e.g. for a channel supporting
> - *	a width of 4 the mask should have BIT(4) set.
> - * @dst_addr_widths: bit mask of dst addr widths the channel supports
> + * @src_bus_widths: mask of source bus widths the channel supports.
> + * @src_addr_widths: legacy bit mask of source bus widths the channel supports.
> + * @dst_bus_widths: mask of destination bus widths the channel supports.
> + * @dst_addr_widths: legacy bit mask of destination bus widths the channel
> + *	supports.
>   * @directions: bit mask of slave directions the channel supports.
>   *	Since the enum dma_transfer_direction is not defined as bit flag for
>   *	each type, the dma controller should set BIT(<TYPE>) and same
> @@ -545,8 +706,14 @@ enum dma_residue_granularity {
>   * resubmitted multiple times
>   */
>  struct dma_slave_caps {
> -	u32 src_addr_widths;
> -	u32 dst_addr_widths;
> +	struct {
> +		dma_buswidth_mask_t src_bus_widths;
> +		u32 src_addr_widths;
> +	};
> +	struct {
> +		dma_buswidth_mask_t dst_bus_widths;
> +		u32 dst_addr_widths;
> +	};
>  	u32 directions;
>  	u32 min_burst;
>  	u32 max_burst;
> @@ -839,10 +1006,10 @@ struct dma_filter {
>   * @dev: struct device reference for dma mapping api
>   * @owner: owner module (automatically set based on the provided dev)
>   * @chan_ida: unique channel ID
> - * @src_addr_widths: bit mask of src addr widths the device supports
> - *	Width is specified in bytes, e.g. for a device supporting
> - *	a width of 4 the mask should have BIT(4) set.
> - * @dst_addr_widths: bit mask of dst addr widths the device supports
> + * @src_bus_widths: mask of source bus widths the device supports.
> + * @src_addr_widths: legacy bit mask of source bus widths the device supports.
> + * @dst_bus_widths: mask of destination bus widths the device supports.
> + * @dst_addr_widths: legacy bit mask of destination bus widths the device supports.
>   * @directions: bit mask of slave directions the device supports.
>   *	Since the enum dma_transfer_direction is not defined as bit flag for
>   *	each type, the dma controller should set BIT(<TYPE>) and same
> @@ -924,8 +1091,14 @@ struct dma_device {
>  	struct module *owner;
>  	struct ida chan_ida;
>
> -	u32 src_addr_widths;
> -	u32 dst_addr_widths;
> +	struct {
> +		dma_buswidth_mask_t src_bus_widths;
> +		u32 src_addr_widths;
> +	};
> +	struct {
> +		dma_buswidth_mask_t dst_bus_widths;
> +		u32 dst_addr_widths;
> +	};
>  	u32 directions;
>  	u32 min_burst;
>  	u32 max_burst;
>
> --
> 2.56.0
>

  reply	other threads:[~2026-10-14 15:28 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-14 14:19 [PATCH v6 0/9] dmaengine: Support bus widths of 32 bytes and above Nuno Sá
2026-10-14 14:19 ` [PATCH v6 1/9] " Nuno Sá
2026-10-14 15:28   ` Frank Li [this message]
2026-10-14 14:19 ` [PATCH v6 2/9] dmaengine: dma-axi-dmac: Use bus width capability helpers Nuno Sá
2026-10-14 14:33   ` sashiko-bot
2026-10-14 14:19 ` [PATCH v6 3/9] dmaengine: dw-axi-dmac: " Nuno Sá
2026-10-14 14:19 ` [PATCH v6 4/9] dmaengine: qcom: gpi: " Nuno Sá
2026-10-14 14:19 ` [PATCH v6 5/9] dmaengine: stm32-dma3: " Nuno Sá
2026-10-14 14:19 ` [PATCH v6 6/9] iio: buffer-dmaengine: Use dma_slave_caps bus width accessors Nuno Sá
2026-10-14 14:19 ` [PATCH v6 7/9] ALSA: pcm_dmaengine: Use dma_slave_caps bus width helpers Nuno Sá
2026-10-14 14:19 ` [PATCH v6 8/9] spi: dw: " Nuno Sá
2026-10-14 14:19 ` [PATCH v6 9/9] dmaengine: Drop legacy bus width fields from dma_slave_caps Nuno Sá

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=as-fgDViIz4oSap7@SMW015318 \
    --to=frank.li@oss.nxp.com \
    --cc=Eugeniy.Paltsev@synopsys.com \
    --cc=Frank.Li@kernel.org \
    --cc=alexandre.torgue@foss.st.com \
    --cc=amelie.delaunay@foss.st.com \
    --cc=andy@kernel.org \
    --cc=broonie@kernel.org \
    --cc=dlechner@baylibre.com \
    --cc=dmaengine@vger.kernel.org \
    --cc=jic23@kernel.org \
    --cc=lars@metafoo.de \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-arm-msm@vger.kernel.org \
    --cc=linux-iio@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-sound@vger.kernel.org \
    --cc=linux-spi@vger.kernel.org \
    --cc=linux-stm32@st-md-mailman.stormreply.com \
    --cc=mcoquelin.stm32@gmail.com \
    --cc=nuno.sa@analog.com \
    --cc=perex@perex.cz \
    --cc=tiwai@suse.com \
    --cc=vkoul@kernel.org \
    /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