Message ID | 1521735499-29138-1-git-send-email-yibin.gong@nxp.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Hi Robin, Several comments inside. Sascha On Fri, Mar 23, 2018 at 12:18:19AM +0800, Robin Gong wrote: > The legacy sdma driver has below limitations or drawbacks: > 1. Hardcode the max BDs number as "PAGE_SIZE / sizeof(*)", and alloc > one page size for one channel regardless of only few BDs needed > most time. But in few cases, the max PAGE_SIZE maybe not enough. > 2. One SDMA channel can't stop immediatley once channel disabled which > means SDMA interrupt may come in after this channel terminated.There > are some patches for this corner case such as commit "2746e2c389f9", > but not cover non-cyclic. > > The common virt-dma overcomes the above limitations. It can alloc bd > dynamically and free bd once this tx transfer done. No memory wasted or > maximum limititation here, only depends on how many memory can be requested > from kernel. For No.2, such issue can be workaround by checking if there > is available descript("sdmac->desc") now once the unwanted interrupt > coming. At last the common virt-dma is easier for sdma driver maintain. > > Signed-off-by: Robin Gong <yibin.gong@nxp.com> > --- > drivers/dma/Kconfig | 1 + > drivers/dma/imx-sdma.c | 395 +++++++++++++++++++++++++++++++------------------ > 2 files changed, 253 insertions(+), 143 deletions(-) > > diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig > index 27df3e2..c4ce43c 100644 > --- a/drivers/dma/Kconfig > +++ b/drivers/dma/Kconfig > @@ -247,6 +247,7 @@ config IMX_SDMA > tristate "i.MX SDMA support" > depends on ARCH_MXC > select DMA_ENGINE > + select DMA_VIRTUAL_CHANNELS > help > Support the i.MX SDMA engine. This engine is integrated into > Freescale i.MX25/31/35/51/53/6 chips. > diff --git a/drivers/dma/imx-sdma.c b/drivers/dma/imx-sdma.c > index ccd03c3..df79e73 100644 > --- a/drivers/dma/imx-sdma.c > +++ b/drivers/dma/imx-sdma.c > @@ -48,6 +48,7 @@ > #include <linux/mfd/syscon/imx6q-iomuxc-gpr.h> > > #include "dmaengine.h" > +#include "virt-dma.h" > > /* SDMA registers */ > #define SDMA_H_C0PTR 0x000 > @@ -291,10 +292,19 @@ struct sdma_context_data { > u32 scratch7; > } __attribute__ ((packed)); > > -#define NUM_BD (int)(PAGE_SIZE / sizeof(struct sdma_buffer_descriptor)) > - > struct sdma_engine; > > +struct sdma_desc { > + struct virt_dma_desc vd; > + struct list_head node; > + unsigned int num_bd; > + dma_addr_t bd_phys; > + unsigned int buf_tail; > + unsigned int buf_ptail; > + struct sdma_channel *sdmac; > + struct sdma_buffer_descriptor *bd; > +}; > + > /** > * struct sdma_channel - housekeeping for a SDMA channel > * > @@ -310,19 +320,17 @@ struct sdma_engine; > * @num_bd max NUM_BD. number of descriptors currently handling > */ > struct sdma_channel { > + struct virt_dma_chan vc; > + struct list_head pending; > struct sdma_engine *sdma; > + struct sdma_desc *desc; > unsigned int channel; > enum dma_transfer_direction direction; > enum sdma_peripheral_type peripheral_type; > unsigned int event_id0; > unsigned int event_id1; > enum dma_slave_buswidth word_size; > - unsigned int buf_tail; > - unsigned int buf_ptail; > - unsigned int num_bd; > unsigned int period_len; > - struct sdma_buffer_descriptor *bd; > - dma_addr_t bd_phys; > unsigned int pc_from_device, pc_to_device; > unsigned int device_to_device; > unsigned long flags; > @@ -330,15 +338,12 @@ struct sdma_channel { > unsigned long event_mask[2]; > unsigned long watermark_level; > u32 shp_addr, per_addr; > - struct dma_chan chan; > - spinlock_t lock; > - struct dma_async_tx_descriptor desc; > enum dma_status status; > unsigned int chn_count; > unsigned int chn_real_count; > - struct tasklet_struct tasklet; > struct imx_dma_data data; > bool enabled; Usage of this variable is removed in this patch, but not the variable itself. > + u32 bd_size_sum; This variable is never used for anything. > }; > > #define IMX_DMA_SG_LOOP BIT(0) > @@ -398,6 +403,9 @@ struct sdma_engine { > u32 spba_start_addr; > u32 spba_end_addr; > unsigned int irq; > + /* channel0 bd */ > + dma_addr_t bd0_phys; > + struct sdma_buffer_descriptor *bd0; > }; > > static struct sdma_driver_data sdma_imx31 = { > @@ -553,6 +561,8 @@ MODULE_DEVICE_TABLE(of, sdma_dt_ids); > #define SDMA_H_CONFIG_ACR BIT(4) /* indicates if AHB freq /core freq = 2 or 1 */ > #define SDMA_H_CONFIG_CSM (3) /* indicates which context switch mode is selected*/ > > +static void sdma_start_desc(struct sdma_channel *sdmac); > + > static inline u32 chnenbl_ofs(struct sdma_engine *sdma, unsigned int event) > { > u32 chnenbl0 = sdma->drvdata->chnenbl0; > @@ -597,14 +607,7 @@ static int sdma_config_ownership(struct sdma_channel *sdmac, > > static void sdma_enable_channel(struct sdma_engine *sdma, int channel) > { > - unsigned long flags; > - struct sdma_channel *sdmac = &sdma->channel[channel]; > - > writel(BIT(channel), sdma->regs + SDMA_H_START); > - > - spin_lock_irqsave(&sdmac->lock, flags); > - sdmac->enabled = true; > - spin_unlock_irqrestore(&sdmac->lock, flags); > } > > /* > @@ -632,7 +635,7 @@ static int sdma_run_channel0(struct sdma_engine *sdma) > static int sdma_load_script(struct sdma_engine *sdma, void *buf, int size, > u32 address) > { > - struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd; > + struct sdma_buffer_descriptor *bd0 = sdma->bd0; This change seems to be an orthogonal change. Please make this a separate patch. > void *buf_virt; > dma_addr_t buf_phys; > int ret; > @@ -691,23 +694,16 @@ static void sdma_event_disable(struct sdma_channel *sdmac, unsigned int event) > static void sdma_update_channel_loop(struct sdma_channel *sdmac) > { > struct sdma_buffer_descriptor *bd; > + struct sdma_desc *desc = sdmac->desc; > int error = 0; > enum dma_status old_status = sdmac->status; > - unsigned long flags; > - > - spin_lock_irqsave(&sdmac->lock, flags); > - if (!sdmac->enabled) { > - spin_unlock_irqrestore(&sdmac->lock, flags); > - return; > - } > - spin_unlock_irqrestore(&sdmac->lock, flags); > > /* > * loop mode. Iterate over descriptors, re-setup them and > * call callback function. > */ > - while (1) { > - bd = &sdmac->bd[sdmac->buf_tail]; > + while (desc) { 'desc' seems to be used as a loop counter here, but this variable is never assigned another value, so I assume it's just another way to say "skip the loop if desc is NULL". When 'desc' NULL you won't get into this function at all though, so this check for desc seems rather pointless. > + bd = &desc->bd[desc->buf_tail]; > > if (bd->mode.status & BD_DONE) > break; > @@ -726,8 +722,8 @@ static void sdma_update_channel_loop(struct sdma_channel *sdmac) > sdmac->chn_real_count = bd->mode.count; > bd->mode.status |= BD_DONE; > bd->mode.count = sdmac->period_len; > - sdmac->buf_ptail = sdmac->buf_tail; > - sdmac->buf_tail = (sdmac->buf_tail + 1) % sdmac->num_bd; > + desc->buf_ptail = desc->buf_tail; > + desc->buf_tail = (desc->buf_tail + 1) % desc->num_bd; > > /* > * The callback is called from the interrupt context in order > @@ -735,15 +731,16 @@ static void sdma_update_channel_loop(struct sdma_channel *sdmac) > * SDMA transaction status by the time the client tasklet is > * executed. > */ > - > - dmaengine_desc_get_callback_invoke(&sdmac->desc, NULL); > + spin_unlock(&sdmac->vc.lock); > + dmaengine_desc_get_callback_invoke(&desc->vd.tx, NULL); > + spin_lock(&sdmac->vc.lock); > > if (error) > sdmac->status = old_status; > } > } > > -static void mxc_sdma_handle_channel_normal(unsigned long data) > +static void mxc_sdma_handle_channel_normal(struct sdma_channel *data) > { > struct sdma_channel *sdmac = (struct sdma_channel *) data; > struct sdma_buffer_descriptor *bd; > @@ -754,8 +751,8 @@ static void mxc_sdma_handle_channel_normal(unsigned long data) > * non loop mode. Iterate over all descriptors, collect > * errors and call callback function > */ > - for (i = 0; i < sdmac->num_bd; i++) { > - bd = &sdmac->bd[i]; > + for (i = 0; i < sdmac->desc->num_bd; i++) { > + bd = &sdmac->desc->bd[i]; > > if (bd->mode.status & (BD_DONE | BD_RROR)) > error = -EIO; > @@ -766,10 +763,6 @@ static void mxc_sdma_handle_channel_normal(unsigned long data) > sdmac->status = DMA_ERROR; > else > sdmac->status = DMA_COMPLETE; > - > - dma_cookie_complete(&sdmac->desc); > - > - dmaengine_desc_get_callback_invoke(&sdmac->desc, NULL); > } > > static irqreturn_t sdma_int_handler(int irq, void *dev_id) > @@ -785,13 +778,24 @@ static irqreturn_t sdma_int_handler(int irq, void *dev_id) > while (stat) { > int channel = fls(stat) - 1; > struct sdma_channel *sdmac = &sdma->channel[channel]; > - > - if (sdmac->flags & IMX_DMA_SG_LOOP) > - sdma_update_channel_loop(sdmac); > - else > - tasklet_schedule(&sdmac->tasklet); > + struct sdma_desc *desc; > + > + spin_lock(&sdmac->vc.lock); > + desc = sdmac->desc; > + if (desc) { > + if (sdmac->flags & IMX_DMA_SG_LOOP) { > + sdma_update_channel_loop(sdmac); > + } else { > + mxc_sdma_handle_channel_normal(sdmac); > + vchan_cookie_complete(&desc->vd); > + if (!list_empty(&sdmac->pending)) > + list_del(&desc->node); What does this list_empty check protect you from? It looks like when the list really is empty then it's a bug in your internal driver logic. > + sdma_start_desc(sdmac); Whitespace damage here. > + } > + } > > __clear_bit(channel, &stat); > + spin_unlock(&sdmac->vc.lock); > } > > return IRQ_HANDLED; > @@ -897,7 +901,7 @@ static int sdma_load_context(struct sdma_channel *sdmac) > int channel = sdmac->channel; > int load_address; > struct sdma_context_data *context = sdma->context; > - struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd; > + struct sdma_buffer_descriptor *bd0 = sdma->bd0; > int ret; > unsigned long flags; > > @@ -946,7 +950,7 @@ static int sdma_load_context(struct sdma_channel *sdmac) > > static struct sdma_channel *to_sdma_chan(struct dma_chan *chan) > { > - return container_of(chan, struct sdma_channel, chan); > + return container_of(chan, struct sdma_channel, vc.chan); > } > > static int sdma_disable_channel(struct dma_chan *chan) > @@ -954,15 +958,10 @@ static int sdma_disable_channel(struct dma_chan *chan) > struct sdma_channel *sdmac = to_sdma_chan(chan); > struct sdma_engine *sdma = sdmac->sdma; > int channel = sdmac->channel; > - unsigned long flags; > > writel_relaxed(BIT(channel), sdma->regs + SDMA_H_STATSTOP); > sdmac->status = DMA_ERROR; > > - spin_lock_irqsave(&sdmac->lock, flags); > - sdmac->enabled = false; > - spin_unlock_irqrestore(&sdmac->lock, flags); > - > return 0; > } > > @@ -1097,42 +1096,101 @@ static int sdma_set_channel_priority(struct sdma_channel *sdmac, > return 0; > } > > -static int sdma_request_channel(struct sdma_channel *sdmac) > +static int sdma_alloc_bd(struct sdma_desc *desc) > { > - struct sdma_engine *sdma = sdmac->sdma; > - int channel = sdmac->channel; > - int ret = -EBUSY; > + u32 bd_size = desc->num_bd * sizeof(struct sdma_buffer_descriptor); > + int ret = 0; > + unsigned long flags; > > - sdmac->bd = dma_zalloc_coherent(NULL, PAGE_SIZE, &sdmac->bd_phys, > + desc->bd = dma_zalloc_coherent(NULL, bd_size, &desc->bd_phys, > GFP_KERNEL); > - if (!sdmac->bd) { > + if (!desc->bd) { > ret = -ENOMEM; > goto out; > } > > - sdma->channel_control[channel].base_bd_ptr = sdmac->bd_phys; > - sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys; > + spin_lock_irqsave(&desc->sdmac->vc.lock, flags); > + desc->sdmac->bd_size_sum += bd_size; > + spin_unlock_irqrestore(&desc->sdmac->vc.lock, flags); > > - sdma_set_channel_priority(sdmac, MXC_SDMA_DEFAULT_PRIORITY); > - return 0; > out: > - > return ret; > } > > -static dma_cookie_t sdma_tx_submit(struct dma_async_tx_descriptor *tx) > +static void sdma_free_bd(struct sdma_desc *desc) > { > + u32 bd_size = desc->num_bd * sizeof(struct sdma_buffer_descriptor); > unsigned long flags; > - struct sdma_channel *sdmac = to_sdma_chan(tx->chan); > - dma_cookie_t cookie; > > - spin_lock_irqsave(&sdmac->lock, flags); > + if (desc->bd) { > + dma_free_coherent(NULL, bd_size, desc->bd, desc->bd_phys); > + > + spin_lock_irqsave(&desc->sdmac->vc.lock, flags); > + desc->sdmac->bd_size_sum -= bd_size; > + spin_unlock_irqrestore(&desc->sdmac->vc.lock, flags); > + } > +} > + > +static int sdma_request_channel0(struct sdma_engine *sdma) > +{ > + int ret = 0; > + > + sdma->bd0 = dma_zalloc_coherent(NULL, PAGE_SIZE, &sdma->bd0_phys, > + GFP_KERNEL); > + if (!sdma->bd0) { > + ret = -ENOMEM; > + goto out; > + } > > - cookie = dma_cookie_assign(tx); > + sdma->channel_control[0].base_bd_ptr = sdma->bd0_phys; > + sdma->channel_control[0].current_bd_ptr = sdma->bd0_phys; > > - spin_unlock_irqrestore(&sdmac->lock, flags); > + sdma_set_channel_priority(&sdma->channel[0], MXC_SDMA_DEFAULT_PRIORITY); > +out: > > - return cookie; > + return ret; > +} > + > +static struct sdma_desc *to_sdma_desc(struct dma_async_tx_descriptor *t) > +{ > + return container_of(t, struct sdma_desc, vd.tx); > +} > + > +static void sdma_desc_free(struct virt_dma_desc *vd) > +{ > + struct sdma_desc *desc = container_of(vd, struct sdma_desc, vd); > + > + if (desc) { Depending on the position of 'vd' in struct sdma_desc 'desc' will always be non-NULL, even if 'vd' is NULL. I think this test is unnecessary since this function should never be called with an invalid pointer. If it is, then the caller really deserved the resulting crash. > + sdma_free_bd(desc); > + kfree(desc); > + } > +} > + > +static int sdma_terminate_all(struct dma_chan *chan) > +{ > + struct sdma_channel *sdmac = to_sdma_chan(chan); > + unsigned long flags; > + LIST_HEAD(head); > + > + spin_lock_irqsave(&sdmac->vc.lock, flags); > + vchan_get_all_descriptors(&sdmac->vc, &head); > + while (!list_empty(&sdmac->pending)) { > + struct sdma_desc *desc = list_first_entry(&sdmac->pending, > + struct sdma_desc, node); > + > + list_del(&desc->node); > + spin_unlock_irqrestore(&sdmac->vc.lock, flags); > + sdmac->vc.desc_free(&desc->vd); > + spin_lock_irqsave(&sdmac->vc.lock, flags); > + } list_for_each_entry_safe? > + > + if (sdmac->desc) > + sdmac->desc = NULL; The test is unnecesary. > + spin_unlock_irqrestore(&sdmac->vc.lock, flags); > + vchan_dma_desc_free_list(&sdmac->vc, &head); > + sdma_disable_channel_with_delay(chan); > + > + return 0; > } > > static int sdma_alloc_chan_resources(struct dma_chan *chan) > @@ -1168,18 +1226,11 @@ static int sdma_alloc_chan_resources(struct dma_chan *chan) > if (ret) > goto disable_clk_ipg; > > - ret = sdma_request_channel(sdmac); > - if (ret) > - goto disable_clk_ahb; > - > ret = sdma_set_channel_priority(sdmac, prio); > if (ret) > goto disable_clk_ahb; > > - dma_async_tx_descriptor_init(&sdmac->desc, chan); > - sdmac->desc.tx_submit = sdma_tx_submit; > - /* txd.flags will be overwritten in prep funcs */ > - sdmac->desc.flags = DMA_CTRL_ACK; > + sdmac->bd_size_sum = 0; > > return 0; > > @@ -1195,7 +1246,7 @@ static void sdma_free_chan_resources(struct dma_chan *chan) > struct sdma_channel *sdmac = to_sdma_chan(chan); > struct sdma_engine *sdma = sdmac->sdma; > > - sdma_disable_channel(chan); > + sdma_terminate_all(chan); > > if (sdmac->event_id0) > sdma_event_disable(sdmac, sdmac->event_id0); > @@ -1207,12 +1258,43 @@ static void sdma_free_chan_resources(struct dma_chan *chan) > > sdma_set_channel_priority(sdmac, 0); > > - dma_free_coherent(NULL, PAGE_SIZE, sdmac->bd, sdmac->bd_phys); > - > clk_disable(sdma->clk_ipg); > clk_disable(sdma->clk_ahb); > } > > +static struct sdma_desc *sdma_transfer_init(struct sdma_channel *sdmac, > + enum dma_transfer_direction direction, u32 bds) > +{ > + struct sdma_desc *desc; > + > + desc = kzalloc((sizeof(*desc)), GFP_KERNEL); > + if (!desc) > + goto err_out; > + > + sdmac->status = DMA_IN_PROGRESS; > + sdmac->direction = direction; > + sdmac->flags = 0; > + sdmac->chn_count = 0; > + sdmac->chn_real_count = 0; > + > + desc->sdmac = sdmac; > + desc->num_bd = bds; > + INIT_LIST_HEAD(&desc->node); > + > + if (sdma_alloc_bd(desc)) > + goto err_desc_out; > + > + if (sdma_load_context(sdmac)) > + goto err_desc_out; > + > + return desc; > + > +err_desc_out: > + kfree(desc); > +err_out: > + return NULL; > +} > + > static struct dma_async_tx_descriptor *sdma_prep_slave_sg( > struct dma_chan *chan, struct scatterlist *sgl, > unsigned int sg_len, enum dma_transfer_direction direction, > @@ -1223,35 +1305,24 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( > int ret, i, count; > int channel = sdmac->channel; > struct scatterlist *sg; > + struct sdma_desc *desc; > > - if (sdmac->status == DMA_IN_PROGRESS) > + if (!chan) > return NULL; > - sdmac->status = DMA_IN_PROGRESS; > - > - sdmac->flags = 0; > > - sdmac->buf_tail = 0; > - sdmac->buf_ptail = 0; > - sdmac->chn_real_count = 0; > + desc = sdma_transfer_init(sdmac, direction, sg_len); > + if (!desc) > + goto err_out; > > dev_dbg(sdma->dev, "setting up %d entries for channel %d.\n", > sg_len, channel); > > - sdmac->direction = direction; > ret = sdma_load_context(sdmac); > if (ret) > goto err_out; > > - if (sg_len > NUM_BD) { > - dev_err(sdma->dev, "SDMA channel %d: maximum number of sg exceeded: %d > %d\n", > - channel, sg_len, NUM_BD); > - ret = -EINVAL; > - goto err_out; > - } > - > - sdmac->chn_count = 0; > for_each_sg(sgl, sg, sg_len, i) { > - struct sdma_buffer_descriptor *bd = &sdmac->bd[i]; > + struct sdma_buffer_descriptor *bd = &desc->bd[i]; > int param; > > bd->buffer_addr = sg->dma_address; > @@ -1262,7 +1333,7 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( > dev_err(sdma->dev, "SDMA channel %d: maximum bytes for sg entry exceeded: %d > %d\n", > channel, count, 0xffff); > ret = -EINVAL; > - goto err_out; > + goto err_bd_out; > } > > bd->mode.count = count; > @@ -1307,10 +1378,11 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( > bd->mode.status = param; > } > > - sdmac->num_bd = sg_len; > - sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys; > + return vchan_tx_prep(&sdmac->vc, &desc->vd, flags); > > - return &sdmac->desc; > +err_bd_out: > + sdma_free_bd(desc); > + kfree(desc); > err_out: > sdmac->status = DMA_ERROR; > return NULL; > @@ -1326,39 +1398,32 @@ static struct dma_async_tx_descriptor *sdma_prep_dma_cyclic( > int num_periods = buf_len / period_len; > int channel = sdmac->channel; > int ret, i = 0, buf = 0; > + struct sdma_desc *desc; > > dev_dbg(sdma->dev, "%s channel: %d\n", __func__, channel); > > - if (sdmac->status == DMA_IN_PROGRESS) > - return NULL; > - > - sdmac->status = DMA_IN_PROGRESS; > + /* Now allocate and setup the descriptor. */ > + desc = sdma_transfer_init(sdmac, direction, num_periods); > + if (!desc) > + goto err_out; > > - sdmac->buf_tail = 0; > - sdmac->buf_ptail = 0; > - sdmac->chn_real_count = 0; > + desc->buf_tail = 0; > + desc->buf_ptail = 0; > sdmac->period_len = period_len; > - > sdmac->flags |= IMX_DMA_SG_LOOP; > - sdmac->direction = direction; > + > ret = sdma_load_context(sdmac); > if (ret) > goto err_out; > > - if (num_periods > NUM_BD) { > - dev_err(sdma->dev, "SDMA channel %d: maximum number of sg exceeded: %d > %d\n", > - channel, num_periods, NUM_BD); > - goto err_out; > - } > - > if (period_len > 0xffff) { > dev_err(sdma->dev, "SDMA channel %d: maximum period size exceeded: %zu > %d\n", > channel, period_len, 0xffff); > - goto err_out; > + goto err_bd_out; > } > > while (buf < buf_len) { > - struct sdma_buffer_descriptor *bd = &sdmac->bd[i]; > + struct sdma_buffer_descriptor *bd = &desc->bd[i]; > int param; > > bd->buffer_addr = dma_addr; > @@ -1366,7 +1431,7 @@ static struct dma_async_tx_descriptor *sdma_prep_dma_cyclic( > bd->mode.count = period_len; > > if (sdmac->word_size > DMA_SLAVE_BUSWIDTH_4_BYTES) > - goto err_out; > + goto err_bd_out; > if (sdmac->word_size == DMA_SLAVE_BUSWIDTH_4_BYTES) > bd->mode.command = 0; > else > @@ -1389,10 +1454,10 @@ static struct dma_async_tx_descriptor *sdma_prep_dma_cyclic( > i++; > } > > - sdmac->num_bd = num_periods; > - sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys; > - > - return &sdmac->desc; > + return vchan_tx_prep(&sdmac->vc, &desc->vd, flags); > +err_bd_out: > + sdma_free_bd(desc); > + kfree(desc); > err_out: > sdmac->status = DMA_ERROR; > return NULL; > @@ -1432,26 +1497,74 @@ static enum dma_status sdma_tx_status(struct dma_chan *chan, > { > struct sdma_channel *sdmac = to_sdma_chan(chan); > u32 residue; > + struct virt_dma_desc *vd; > + struct sdma_desc *desc; > + enum dma_status ret; > + unsigned long flags; > > - if (sdmac->flags & IMX_DMA_SG_LOOP) > - residue = (sdmac->num_bd - sdmac->buf_ptail) * > + ret = dma_cookie_status(chan, cookie, txstate); > + if (ret == DMA_COMPLETE && txstate) { > + residue = sdmac->chn_count - sdmac->chn_real_count; > + return ret; > + } > + > + spin_lock_irqsave(&sdmac->vc.lock, flags); > + vd = vchan_find_desc(&sdmac->vc, cookie); > + desc = to_sdma_desc(&vd->tx); You should use 'vd' only after you have made sure it is valid (though I see it causes no harm in this case, but let's be nice to the readers of this code) > + if (vd) { > + if (sdmac->flags & IMX_DMA_SG_LOOP) > + residue = (desc->num_bd - desc->buf_ptail) * > sdmac->period_len - sdmac->chn_real_count; > - else > + else > + residue = sdmac->chn_count - sdmac->chn_real_count; > + } else if (sdmac->desc && sdmac->desc->vd.tx.cookie == cookie) { > residue = sdmac->chn_count - sdmac->chn_real_count; > + } else { > + residue = 0; > + } > + ret = sdmac->status; > + spin_unlock_irqrestore(&sdmac->vc.lock, flags); > > dma_set_tx_state(txstate, chan->completed_cookie, chan->cookie, > residue); > > - return sdmac->status; > + return ret; > +} > + > +static void sdma_start_desc(struct sdma_channel *sdmac) > +{ > + struct virt_dma_desc *vd = vchan_next_desc(&sdmac->vc); > + struct sdma_desc *desc; > + struct sdma_engine *sdma = sdmac->sdma; > + int channel = sdmac->channel; > + > + if (!vd) { > + sdmac->desc = NULL; > + return; > + } > + sdmac->desc = desc = to_sdma_desc(&vd->tx); > + /* > + * Do not delete the node in desc_issued list in cyclic mode, otherwise > + * the desc alloced will never be freed in vchan_dma_desc_free_list > + */ > + if (!(sdmac->flags & IMX_DMA_SG_LOOP)) { > + list_add_tail(&sdmac->desc->node, &sdmac->pending); > + list_del(&vd->node); > + } > + sdma->channel_control[channel].base_bd_ptr = desc->bd_phys; > + sdma->channel_control[channel].current_bd_ptr = desc->bd_phys; > + sdma_enable_channel(sdma, sdmac->channel); > } > > static void sdma_issue_pending(struct dma_chan *chan) > { > struct sdma_channel *sdmac = to_sdma_chan(chan); > - struct sdma_engine *sdma = sdmac->sdma; > + unsigned long flags; > > - if (sdmac->status == DMA_IN_PROGRESS) > - sdma_enable_channel(sdma, sdmac->channel); > + spin_lock_irqsave(&sdmac->vc.lock, flags); > + if (vchan_issue_pending(&sdmac->vc) && !sdmac->desc) > + sdma_start_desc(sdmac); > + spin_unlock_irqrestore(&sdmac->vc.lock, flags); > } > > #define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1 34 > @@ -1657,7 +1770,7 @@ static int sdma_init(struct sdma_engine *sdma) > for (i = 0; i < MAX_DMA_CHANNELS; i++) > writel_relaxed(0, sdma->regs + SDMA_CHNPRI_0 + i * 4); > > - ret = sdma_request_channel(&sdma->channel[0]); > + ret = sdma_request_channel0(sdma); > if (ret) > goto err_dma_alloc; > > @@ -1819,22 +1932,17 @@ static int sdma_probe(struct platform_device *pdev) > struct sdma_channel *sdmac = &sdma->channel[i]; > > sdmac->sdma = sdma; > - spin_lock_init(&sdmac->lock); > - > - sdmac->chan.device = &sdma->dma_device; > - dma_cookie_init(&sdmac->chan); > sdmac->channel = i; > - > - tasklet_init(&sdmac->tasklet, mxc_sdma_handle_channel_normal, > - (unsigned long) sdmac); > + sdmac->status = DMA_IN_PROGRESS; > + sdmac->vc.desc_free = sdma_desc_free; > + INIT_LIST_HEAD(&sdmac->pending); > /* > * Add the channel to the DMAC list. Do not add channel 0 though > * because we need it internally in the SDMA driver. This also means > * that channel 0 in dmaengine counting matches sdma channel 1. > */ > if (i) > - list_add_tail(&sdmac->chan.device_node, > - &sdma->dma_device.channels); > + vchan_init(&sdmac->vc, &sdma->dma_device); > } > > ret = sdma_init(sdma); > @@ -1879,7 +1987,7 @@ static int sdma_probe(struct platform_device *pdev) > sdma->dma_device.device_prep_slave_sg = sdma_prep_slave_sg; > sdma->dma_device.device_prep_dma_cyclic = sdma_prep_dma_cyclic; > sdma->dma_device.device_config = sdma_config; > - sdma->dma_device.device_terminate_all = sdma_disable_channel_with_delay; > + sdma->dma_device.device_terminate_all = sdma_terminate_all; > sdma->dma_device.src_addr_widths = SDMA_DMA_BUSWIDTHS; > sdma->dma_device.dst_addr_widths = SDMA_DMA_BUSWIDTHS; > sdma->dma_device.directions = SDMA_DMA_DIRECTIONS; > @@ -1939,7 +2047,8 @@ static int sdma_remove(struct platform_device *pdev) > for (i = 0; i < MAX_DMA_CHANNELS; i++) { > struct sdma_channel *sdmac = &sdma->channel[i]; > > - tasklet_kill(&sdmac->tasklet); > + tasklet_kill(&sdmac->vc.task); > + sdma_free_chan_resources(&sdmac->vc.chan); > } > > platform_set_drvdata(pdev, NULL); > -- > 2.7.4 > > > _______________________________________________ > linux-arm-kernel mailing list > linux-arm-kernel@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel >
On 二, 2018-05-22 at 12:09 +0200, Sascha Hauer wrote: > Hi Robin, > > Several comments inside. > > Sascha > > On Fri, Mar 23, 2018 at 12:18:19AM +0800, Robin Gong wrote: > > > > The legacy sdma driver has below limitations or drawbacks: > > 1. Hardcode the max BDs number as "PAGE_SIZE / sizeof(*)", and > > alloc > > one page size for one channel regardless of only few BDs > > needed > > most time. But in few cases, the max PAGE_SIZE maybe not > > enough. > > 2. One SDMA channel can't stop immediatley once channel disabled > > which > > means SDMA interrupt may come in after this channel > > terminated.There > > are some patches for this corner case such as commit > > "2746e2c389f9", > > but not cover non-cyclic. > > > > The common virt-dma overcomes the above limitations. It can alloc > > bd > > dynamically and free bd once this tx transfer done. No memory > > wasted or > > maximum limititation here, only depends on how many memory can be > > requested > > from kernel. For No.2, such issue can be workaround by checking if > > there > > is available descript("sdmac->desc") now once the unwanted > > interrupt > > coming. At last the common virt-dma is easier for sdma driver > > maintain. > > > > Signed-off-by: Robin Gong <yibin.gong@nxp.com> > > --- > > drivers/dma/Kconfig | 1 + > > drivers/dma/imx-sdma.c | 395 +++++++++++++++++++++++++++++++---- > > -------------- > > 2 files changed, 253 insertions(+), 143 deletions(-) > > > > diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig > > index 27df3e2..c4ce43c 100644 > > --- a/drivers/dma/Kconfig > > +++ b/drivers/dma/Kconfig > > @@ -247,6 +247,7 @@ config IMX_SDMA > > tristate "i.MX SDMA support" > > depends on ARCH_MXC > > select DMA_ENGINE > > + select DMA_VIRTUAL_CHANNELS > > help > > Support the i.MX SDMA engine. This engine is integrated > > into > > Freescale i.MX25/31/35/51/53/6 chips. > > diff --git a/drivers/dma/imx-sdma.c b/drivers/dma/imx-sdma.c > > index ccd03c3..df79e73 100644 > > --- a/drivers/dma/imx-sdma.c > > +++ b/drivers/dma/imx-sdma.c > > @@ -48,6 +48,7 @@ > > #include <linux/mfd/syscon/imx6q-iomuxc-gpr.h> > > > > #include "dmaengine.h" > > +#include "virt-dma.h" > > > > /* SDMA registers */ > > #define SDMA_H_C0PTR 0x000 > > @@ -291,10 +292,19 @@ struct sdma_context_data { > > u32 scratch7; > > } __attribute__ ((packed)); > > > > -#define NUM_BD (int)(PAGE_SIZE / sizeof(struct > > sdma_buffer_descriptor)) > > - > > struct sdma_engine; > > > > +struct sdma_desc { > > + struct virt_dma_desc vd; > > + struct list_head node; > > + unsigned int num_bd; > > + dma_addr_t bd_phys; > > + unsigned int buf_tail; > > + unsigned int buf_ptail; > > + struct sdma_channel *sdmac; > > + struct sdma_buffer_descriptor *bd; > > +}; > > + > > /** > > * struct sdma_channel - housekeeping for a SDMA channel > > * > > @@ -310,19 +320,17 @@ struct sdma_engine; > > * @num_bd max NUM_BD. number of descriptors > > currently handling > > */ > > struct sdma_channel { > > + struct virt_dma_chan vc; > > + struct list_head pending; > > struct sdma_engine *sdma; > > + struct sdma_desc *desc; > > unsigned int channel; > > enum dma_transfer_direction direction; > > enum sdma_peripheral_type peripheral_type; > > unsigned int event_id0; > > unsigned int event_id1; > > enum dma_slave_buswidth word_size; > > - unsigned int buf_tail; > > - unsigned int buf_ptail; > > - unsigned int num_bd; > > unsigned int period_len; > > - struct sdma_buffer_descriptor *bd; > > - dma_addr_t bd_phys; > > unsigned int pc_from_device, > > pc_to_device; > > unsigned int device_to_device; > > unsigned long flags; > > @@ -330,15 +338,12 @@ struct sdma_channel { > > unsigned long event_mask[2]; > > unsigned long watermark_level; > > u32 shp_addr, per_addr; > > - struct dma_chan chan; > > - spinlock_t lock; > > - struct dma_async_tx_descriptor desc; > > enum dma_status status; > > unsigned int chn_count; > > unsigned int chn_real_count; > > - struct tasklet_struct tasklet; > > struct imx_dma_data data; > > bool enabled; > Usage of this variable is removed in this patch, but not the variable > itself. Yes, will remove the usless 'enabled' in v2. > > > > > + u32 bd_size_sum; > This variable is never used for anything. Yes, it's not for significative use but debug to see how many current bds used. > > > > > }; > > > > #define IMX_DMA_SG_LOOP BIT(0) > > @@ -398,6 +403,9 @@ struct sdma_engine { > > u32 spba_start_addr; > > u32 spba_end_addr; > > unsigned int irq; > > + /* channel0 bd */ > > + dma_addr_t bd0_phys; > > + struct sdma_buffer_descriptor *bd0; > > }; > > > > static struct sdma_driver_data sdma_imx31 = { > > @@ -553,6 +561,8 @@ MODULE_DEVICE_TABLE(of, sdma_dt_ids); > > #define SDMA_H_CONFIG_ACR BIT(4) /* indicates if AHB freq > > /core freq = 2 or 1 */ > > #define SDMA_H_CONFIG_CSM (3) /* indicates which > > context switch mode is selected*/ > > > > +static void sdma_start_desc(struct sdma_channel *sdmac); > > + > > static inline u32 chnenbl_ofs(struct sdma_engine *sdma, unsigned > > int event) > > { > > u32 chnenbl0 = sdma->drvdata->chnenbl0; > > @@ -597,14 +607,7 @@ static int sdma_config_ownership(struct > > sdma_channel *sdmac, > > > > static void sdma_enable_channel(struct sdma_engine *sdma, int > > channel) > > { > > - unsigned long flags; > > - struct sdma_channel *sdmac = &sdma->channel[channel]; > > - > > writel(BIT(channel), sdma->regs + SDMA_H_START); > > - > > - spin_lock_irqsave(&sdmac->lock, flags); > > - sdmac->enabled = true; > > - spin_unlock_irqrestore(&sdmac->lock, flags); > > } > > > > /* > > @@ -632,7 +635,7 @@ static int sdma_run_channel0(struct sdma_engine > > *sdma) > > static int sdma_load_script(struct sdma_engine *sdma, void *buf, > > int size, > > u32 address) > > { > > - struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd; > > + struct sdma_buffer_descriptor *bd0 = sdma->bd0; > This change seems to be an orthogonal change. Please make this a > separate patch. It's something related with virtual dma support, because in virtual dma framework, all bds should be allocated dynamically if they used. but bd0 is a specail case since it's must and basic for load sdma firmware and context for other channels. So here alloc 'bd0' for other channels. > > > > > void *buf_virt; > > dma_addr_t buf_phys; > > int ret; > > @@ -691,23 +694,16 @@ static void sdma_event_disable(struct > > sdma_channel *sdmac, unsigned int event) > > static void sdma_update_channel_loop(struct sdma_channel *sdmac) > > { > > struct sdma_buffer_descriptor *bd; > > + struct sdma_desc *desc = sdmac->desc; > > int error = 0; > > enum dma_status old_status = sdmac->status; > > - unsigned long flags; > > - > > - spin_lock_irqsave(&sdmac->lock, flags); > > - if (!sdmac->enabled) { > > - spin_unlock_irqrestore(&sdmac->lock, flags); > > - return; > > - } > > - spin_unlock_irqrestore(&sdmac->lock, flags); > > > > /* > > * loop mode. Iterate over descriptors, re-setup them and > > * call callback function. > > */ > > - while (1) { > > - bd = &sdmac->bd[sdmac->buf_tail]; > > + while (desc) { > 'desc' seems to be used as a loop counter here, but this variable is > never assigned another value, so I assume it's just another way to > say > "skip the loop if desc is NULL". When 'desc' NULL you won't get into > this function at all though, so this check for desc seems rather > pointless. Good catch, should check 'sdmac->desc' here instead of 'desc' since in the following 'sdmac->desc' may be set to NULL by sdma_terminate_all during folllowing spin_unlock and spin_lock narrow window. Will improve it in V2. > > > > > + bd = &desc->bd[desc->buf_tail]; > > > > if (bd->mode.status & BD_DONE) > > break; > > @@ -726,8 +722,8 @@ static void sdma_update_channel_loop(struct > > sdma_channel *sdmac) > > sdmac->chn_real_count = bd->mode.count; > > bd->mode.status |= BD_DONE; > > bd->mode.count = sdmac->period_len; > > - sdmac->buf_ptail = sdmac->buf_tail; > > - sdmac->buf_tail = (sdmac->buf_tail + 1) % sdmac- > > >num_bd; > > + desc->buf_ptail = desc->buf_tail; > > + desc->buf_tail = (desc->buf_tail + 1) % desc- > > >num_bd; > > > > /* > > * The callback is called from the interrupt > > context in order > > @@ -735,15 +731,16 @@ static void sdma_update_channel_loop(struct > > sdma_channel *sdmac) > > * SDMA transaction status by the time the client > > tasklet is > > * executed. > > */ > > - > > - dmaengine_desc_get_callback_invoke(&sdmac->desc, > > NULL); > > + spin_unlock(&sdmac->vc.lock); > > + dmaengine_desc_get_callback_invoke(&desc->vd.tx, > > NULL); > > + spin_lock(&sdmac->vc.lock); > > > > if (error) > > sdmac->status = old_status; > > } > > } > > > > -static void mxc_sdma_handle_channel_normal(unsigned long data) > > +static void mxc_sdma_handle_channel_normal(struct sdma_channel > > *data) > > { > > struct sdma_channel *sdmac = (struct sdma_channel *) data; > > struct sdma_buffer_descriptor *bd; > > @@ -754,8 +751,8 @@ static void > > mxc_sdma_handle_channel_normal(unsigned long data) > > * non loop mode. Iterate over all descriptors, collect > > * errors and call callback function > > */ > > - for (i = 0; i < sdmac->num_bd; i++) { > > - bd = &sdmac->bd[i]; > > + for (i = 0; i < sdmac->desc->num_bd; i++) { > > + bd = &sdmac->desc->bd[i]; > > > > if (bd->mode.status & (BD_DONE | BD_RROR)) > > error = -EIO; > > @@ -766,10 +763,6 @@ static void > > mxc_sdma_handle_channel_normal(unsigned long data) > > sdmac->status = DMA_ERROR; > > else > > sdmac->status = DMA_COMPLETE; > > - > > - dma_cookie_complete(&sdmac->desc); > > - > > - dmaengine_desc_get_callback_invoke(&sdmac->desc, NULL); > > } > > > > static irqreturn_t sdma_int_handler(int irq, void *dev_id) > > @@ -785,13 +778,24 @@ static irqreturn_t sdma_int_handler(int irq, > > void *dev_id) > > while (stat) { > > int channel = fls(stat) - 1; > > struct sdma_channel *sdmac = &sdma- > > >channel[channel]; > > - > > - if (sdmac->flags & IMX_DMA_SG_LOOP) > > - sdma_update_channel_loop(sdmac); > > - else > > - tasklet_schedule(&sdmac->tasklet); > > + struct sdma_desc *desc; > > + > > + spin_lock(&sdmac->vc.lock); > > + desc = sdmac->desc; > > + if (desc) { > > + if (sdmac->flags & IMX_DMA_SG_LOOP) { > > + sdma_update_channel_loop(sdmac); > > + } else { > > + mxc_sdma_handle_channel_normal(sdm > > ac); > > + vchan_cookie_complete(&desc->vd); > > + if (!list_empty(&sdmac->pending)) > > + list_del(&desc->node); > What does this list_empty check protect you from? It looks like when > the > list really is empty then it's a bug in your internal driver logic. Yes, no need here check local sdmac->pending since I directly start setup next desc flowing in isr instead of local tasklet and virt_dma framework will handle all lists such as desc_issued/desc_completed etc. Will remove sdmac->pending in V2. > > > > > + sdma_start_desc(sdmac); > Whitespace damage here. Will fix in V2. > > > > > + } > > + } > > > > __clear_bit(channel, &stat); > > + spin_unlock(&sdmac->vc.lock); > > } > > > > return IRQ_HANDLED; > > @@ -897,7 +901,7 @@ static int sdma_load_context(struct > > sdma_channel *sdmac) > > int channel = sdmac->channel; > > int load_address; > > struct sdma_context_data *context = sdma->context; > > - struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd; > > + struct sdma_buffer_descriptor *bd0 = sdma->bd0; > > int ret; > > unsigned long flags; > > > > @@ -946,7 +950,7 @@ static int sdma_load_context(struct > > sdma_channel *sdmac) > > > > static struct sdma_channel *to_sdma_chan(struct dma_chan *chan) > > { > > - return container_of(chan, struct sdma_channel, chan); > > + return container_of(chan, struct sdma_channel, vc.chan); > > } > > > > static int sdma_disable_channel(struct dma_chan *chan) > > @@ -954,15 +958,10 @@ static int sdma_disable_channel(struct > > dma_chan *chan) > > struct sdma_channel *sdmac = to_sdma_chan(chan); > > struct sdma_engine *sdma = sdmac->sdma; > > int channel = sdmac->channel; > > - unsigned long flags; > > > > writel_relaxed(BIT(channel), sdma->regs + > > SDMA_H_STATSTOP); > > sdmac->status = DMA_ERROR; > > > > - spin_lock_irqsave(&sdmac->lock, flags); > > - sdmac->enabled = false; > > - spin_unlock_irqrestore(&sdmac->lock, flags); > > - > > return 0; > > } > > > > @@ -1097,42 +1096,101 @@ static int > > sdma_set_channel_priority(struct sdma_channel *sdmac, > > return 0; > > } > > > > -static int sdma_request_channel(struct sdma_channel *sdmac) > > +static int sdma_alloc_bd(struct sdma_desc *desc) > > { > > - struct sdma_engine *sdma = sdmac->sdma; > > - int channel = sdmac->channel; > > - int ret = -EBUSY; > > + u32 bd_size = desc->num_bd * sizeof(struct > > sdma_buffer_descriptor); > > + int ret = 0; > > + unsigned long flags; > > > > - sdmac->bd = dma_zalloc_coherent(NULL, PAGE_SIZE, &sdmac- > > >bd_phys, > > + desc->bd = dma_zalloc_coherent(NULL, bd_size, &desc- > > >bd_phys, > > GFP_KERNEL); > > - if (!sdmac->bd) { > > + if (!desc->bd) { > > ret = -ENOMEM; > > goto out; > > } > > > > - sdma->channel_control[channel].base_bd_ptr = sdmac- > > >bd_phys; > > - sdma->channel_control[channel].current_bd_ptr = sdmac- > > >bd_phys; > > + spin_lock_irqsave(&desc->sdmac->vc.lock, flags); > > + desc->sdmac->bd_size_sum += bd_size; > > + spin_unlock_irqrestore(&desc->sdmac->vc.lock, flags); > > > > - sdma_set_channel_priority(sdmac, > > MXC_SDMA_DEFAULT_PRIORITY); > > - return 0; > > out: > > - > > return ret; > > } > > > > -static dma_cookie_t sdma_tx_submit(struct dma_async_tx_descriptor > > *tx) > > +static void sdma_free_bd(struct sdma_desc *desc) > > { > > + u32 bd_size = desc->num_bd * sizeof(struct > > sdma_buffer_descriptor); > > unsigned long flags; > > - struct sdma_channel *sdmac = to_sdma_chan(tx->chan); > > - dma_cookie_t cookie; > > > > - spin_lock_irqsave(&sdmac->lock, flags); > > + if (desc->bd) { > > + dma_free_coherent(NULL, bd_size, desc->bd, desc- > > >bd_phys); > > + > > + spin_lock_irqsave(&desc->sdmac->vc.lock, flags); > > + desc->sdmac->bd_size_sum -= bd_size; > > + spin_unlock_irqrestore(&desc->sdmac->vc.lock, > > flags); > > + } > > +} > > + > > +static int sdma_request_channel0(struct sdma_engine *sdma) > > +{ > > + int ret = 0; > > + > > + sdma->bd0 = dma_zalloc_coherent(NULL, PAGE_SIZE, &sdma- > > >bd0_phys, > > + GFP_KERNEL); > > + if (!sdma->bd0) { > > + ret = -ENOMEM; > > + goto out; > > + } > > > > - cookie = dma_cookie_assign(tx); > > + sdma->channel_control[0].base_bd_ptr = sdma->bd0_phys; > > + sdma->channel_control[0].current_bd_ptr = sdma->bd0_phys; > > > > - spin_unlock_irqrestore(&sdmac->lock, flags); > > + sdma_set_channel_priority(&sdma->channel[0], > > MXC_SDMA_DEFAULT_PRIORITY); > > +out: > > > > - return cookie; > > + return ret; > > +} > > + > > +static struct sdma_desc *to_sdma_desc(struct > > dma_async_tx_descriptor *t) > > +{ > > + return container_of(t, struct sdma_desc, vd.tx); > > +} > > + > > +static void sdma_desc_free(struct virt_dma_desc *vd) > > +{ > > + struct sdma_desc *desc = container_of(vd, struct > > sdma_desc, vd); > > + > > + if (desc) { > Depending on the position of 'vd' in struct sdma_desc 'desc' will > always > be non-NULL, even if 'vd' is NULL. > > I think this test is unnecessary since this function should never be > called with an invalid pointer. If it is, then the caller really > deserved the resulting crash. Yes, will remove it. > > > > > + sdma_free_bd(desc); > > + kfree(desc); > > + } > > +} > > + > > +static int sdma_terminate_all(struct dma_chan *chan) > > +{ > > + struct sdma_channel *sdmac = to_sdma_chan(chan); > > + unsigned long flags; > > + LIST_HEAD(head); > > + > > + spin_lock_irqsave(&sdmac->vc.lock, flags); > > + vchan_get_all_descriptors(&sdmac->vc, &head); > > + while (!list_empty(&sdmac->pending)) { > > + struct sdma_desc *desc = list_first_entry(&sdmac- > > >pending, > > + struct sdma_desc, node); > > + > > + list_del(&desc->node); > > + spin_unlock_irqrestore(&sdmac->vc.lock, flags); > > + sdmac->vc.desc_free(&desc->vd); > > + spin_lock_irqsave(&sdmac->vc.lock, flags); > > + } > list_for_each_entry_safe? Will remove here all while(sdmac->pending) checking.No need here. > > > > > + > > + if (sdmac->desc) > > + sdmac->desc = NULL; > The test is unnecesary. This 'NULL' is meaningful in case that dma done interrupt come after terminate as you know sdma will actually stop after current transfer done. > > > > + spin_unlock_irqrestore(&sdmac->vc.lock, flags); > > + vchan_dma_desc_free_list(&sdmac->vc, &head); > > + sdma_disable_channel_with_delay(chan); > > + > > + return 0; > > } > > > > static int sdma_alloc_chan_resources(struct dma_chan *chan) > > @@ -1168,18 +1226,11 @@ static int sdma_alloc_chan_resources(struct > > dma_chan *chan) > > if (ret) > > goto disable_clk_ipg; > > > > - ret = sdma_request_channel(sdmac); > > - if (ret) > > - goto disable_clk_ahb; > > - > > ret = sdma_set_channel_priority(sdmac, prio); > > if (ret) > > goto disable_clk_ahb; > > > > - dma_async_tx_descriptor_init(&sdmac->desc, chan); > > - sdmac->desc.tx_submit = sdma_tx_submit; > > - /* txd.flags will be overwritten in prep funcs */ > > - sdmac->desc.flags = DMA_CTRL_ACK; > > + sdmac->bd_size_sum = 0; > > > > return 0; > > > > @@ -1195,7 +1246,7 @@ static void sdma_free_chan_resources(struct > > dma_chan *chan) > > struct sdma_channel *sdmac = to_sdma_chan(chan); > > struct sdma_engine *sdma = sdmac->sdma; > > > > - sdma_disable_channel(chan); > > + sdma_terminate_all(chan); > > > > if (sdmac->event_id0) > > sdma_event_disable(sdmac, sdmac->event_id0); > > @@ -1207,12 +1258,43 @@ static void sdma_free_chan_resources(struct > > dma_chan *chan) > > > > sdma_set_channel_priority(sdmac, 0); > > > > - dma_free_coherent(NULL, PAGE_SIZE, sdmac->bd, sdmac- > > >bd_phys); > > - > > clk_disable(sdma->clk_ipg); > > clk_disable(sdma->clk_ahb); > > } > > > > +static struct sdma_desc *sdma_transfer_init(struct sdma_channel > > *sdmac, > > + enum dma_transfer_direction > > direction, u32 bds) > > +{ > > + struct sdma_desc *desc; > > + > > + desc = kzalloc((sizeof(*desc)), GFP_KERNEL); > > + if (!desc) > > + goto err_out; > > + > > + sdmac->status = DMA_IN_PROGRESS; > > + sdmac->direction = direction; > > + sdmac->flags = 0; > > + sdmac->chn_count = 0; > > + sdmac->chn_real_count = 0; > > + > > + desc->sdmac = sdmac; > > + desc->num_bd = bds; > > + INIT_LIST_HEAD(&desc->node); > > + > > + if (sdma_alloc_bd(desc)) > > + goto err_desc_out; > > + > > + if (sdma_load_context(sdmac)) > > + goto err_desc_out; > > + > > + return desc; > > + > > +err_desc_out: > > + kfree(desc); > > +err_out: > > + return NULL; > > +} > > + > > static struct dma_async_tx_descriptor *sdma_prep_slave_sg( > > struct dma_chan *chan, struct scatterlist *sgl, > > unsigned int sg_len, enum dma_transfer_direction > > direction, > > @@ -1223,35 +1305,24 @@ static struct dma_async_tx_descriptor > > *sdma_prep_slave_sg( > > int ret, i, count; > > int channel = sdmac->channel; > > struct scatterlist *sg; > > + struct sdma_desc *desc; > > > > - if (sdmac->status == DMA_IN_PROGRESS) > > + if (!chan) > > return NULL; > > - sdmac->status = DMA_IN_PROGRESS; > > - > > - sdmac->flags = 0; > > > > - sdmac->buf_tail = 0; > > - sdmac->buf_ptail = 0; > > - sdmac->chn_real_count = 0; > > + desc = sdma_transfer_init(sdmac, direction, sg_len); > > + if (!desc) > > + goto err_out; > > > > dev_dbg(sdma->dev, "setting up %d entries for channel > > %d.\n", > > sg_len, channel); > > > > - sdmac->direction = direction; > > ret = sdma_load_context(sdmac); > > if (ret) > > goto err_out; > > > > - if (sg_len > NUM_BD) { > > - dev_err(sdma->dev, "SDMA channel %d: maximum > > number of sg exceeded: %d > %d\n", > > - channel, sg_len, NUM_BD); > > - ret = -EINVAL; > > - goto err_out; > > - } > > - > > - sdmac->chn_count = 0; > > for_each_sg(sgl, sg, sg_len, i) { > > - struct sdma_buffer_descriptor *bd = &sdmac->bd[i]; > > + struct sdma_buffer_descriptor *bd = &desc->bd[i]; > > int param; > > > > bd->buffer_addr = sg->dma_address; > > @@ -1262,7 +1333,7 @@ static struct dma_async_tx_descriptor > > *sdma_prep_slave_sg( > > dev_err(sdma->dev, "SDMA channel %d: > > maximum bytes for sg entry exceeded: %d > %d\n", > > channel, count, 0xffff); > > ret = -EINVAL; > > - goto err_out; > > + goto err_bd_out; > > } > > > > bd->mode.count = count; > > @@ -1307,10 +1378,11 @@ static struct dma_async_tx_descriptor > > *sdma_prep_slave_sg( > > bd->mode.status = param; > > } > > > > - sdmac->num_bd = sg_len; > > - sdma->channel_control[channel].current_bd_ptr = sdmac- > > >bd_phys; > > + return vchan_tx_prep(&sdmac->vc, &desc->vd, flags); > > > > - return &sdmac->desc; > > +err_bd_out: > > + sdma_free_bd(desc); > > + kfree(desc); > > err_out: > > sdmac->status = DMA_ERROR; > > return NULL; > > @@ -1326,39 +1398,32 @@ static struct dma_async_tx_descriptor > > *sdma_prep_dma_cyclic( > > int num_periods = buf_len / period_len; > > int channel = sdmac->channel; > > int ret, i = 0, buf = 0; > > + struct sdma_desc *desc; > > > > dev_dbg(sdma->dev, "%s channel: %d\n", __func__, channel); > > > > - if (sdmac->status == DMA_IN_PROGRESS) > > - return NULL; > > - > > - sdmac->status = DMA_IN_PROGRESS; > > + /* Now allocate and setup the descriptor. */ > > + desc = sdma_transfer_init(sdmac, direction, num_periods); > > + if (!desc) > > + goto err_out; > > > > - sdmac->buf_tail = 0; > > - sdmac->buf_ptail = 0; > > - sdmac->chn_real_count = 0; > > + desc->buf_tail = 0; > > + desc->buf_ptail = 0; > > sdmac->period_len = period_len; > > - > > sdmac->flags |= IMX_DMA_SG_LOOP; > > - sdmac->direction = direction; > > + > > ret = sdma_load_context(sdmac); > > if (ret) > > goto err_out; > > > > - if (num_periods > NUM_BD) { > > - dev_err(sdma->dev, "SDMA channel %d: maximum > > number of sg exceeded: %d > %d\n", > > - channel, num_periods, NUM_BD); > > - goto err_out; > > - } > > - > > if (period_len > 0xffff) { > > dev_err(sdma->dev, "SDMA channel %d: maximum > > period size exceeded: %zu > %d\n", > > channel, period_len, 0xffff); > > - goto err_out; > > + goto err_bd_out; > > } > > > > while (buf < buf_len) { > > - struct sdma_buffer_descriptor *bd = &sdmac->bd[i]; > > + struct sdma_buffer_descriptor *bd = &desc->bd[i]; > > int param; > > > > bd->buffer_addr = dma_addr; > > @@ -1366,7 +1431,7 @@ static struct dma_async_tx_descriptor > > *sdma_prep_dma_cyclic( > > bd->mode.count = period_len; > > > > if (sdmac->word_size > DMA_SLAVE_BUSWIDTH_4_BYTES) > > - goto err_out; > > + goto err_bd_out; > > if (sdmac->word_size == > > DMA_SLAVE_BUSWIDTH_4_BYTES) > > bd->mode.command = 0; > > else > > @@ -1389,10 +1454,10 @@ static struct dma_async_tx_descriptor > > *sdma_prep_dma_cyclic( > > i++; > > } > > > > - sdmac->num_bd = num_periods; > > - sdma->channel_control[channel].current_bd_ptr = sdmac- > > >bd_phys; > > - > > - return &sdmac->desc; > > + return vchan_tx_prep(&sdmac->vc, &desc->vd, flags); > > +err_bd_out: > > + sdma_free_bd(desc); > > + kfree(desc); > > err_out: > > sdmac->status = DMA_ERROR; > > return NULL; > > @@ -1432,26 +1497,74 @@ static enum dma_status > > sdma_tx_status(struct dma_chan *chan, > > { > > struct sdma_channel *sdmac = to_sdma_chan(chan); > > u32 residue; > > + struct virt_dma_desc *vd; > > + struct sdma_desc *desc; > > + enum dma_status ret; > > + unsigned long flags; > > > > - if (sdmac->flags & IMX_DMA_SG_LOOP) > > - residue = (sdmac->num_bd - sdmac->buf_ptail) * > > + ret = dma_cookie_status(chan, cookie, txstate); > > + if (ret == DMA_COMPLETE && txstate) { > > + residue = sdmac->chn_count - sdmac- > > >chn_real_count; > > + return ret; > > + } > > + > > + spin_lock_irqsave(&sdmac->vc.lock, flags); > > + vd = vchan_find_desc(&sdmac->vc, cookie); > > + desc = to_sdma_desc(&vd->tx); > You should use 'vd' only after you have made sure it is valid (though > I > see it causes no harm in this case, but let's be nice to the readers > of > this code) Ok, will move desc = to_sdma_desc(&vd->tx) into the below if(vd)... > > > > > + if (vd) { > > + if (sdmac->flags & IMX_DMA_SG_LOOP) > > + residue = (desc->num_bd - desc->buf_ptail) > > * > > sdmac->period_len - sdmac- > > >chn_real_count; > > - else > > + else > > + residue = sdmac->chn_count - sdmac- > > >chn_real_count; > > + } else if (sdmac->desc && sdmac->desc->vd.tx.cookie == > > cookie) { > > residue = sdmac->chn_count - sdmac- > > >chn_real_count; > > + } else { > > + residue = 0; > > + } > > + ret = sdmac->status; > > + spin_unlock_irqrestore(&sdmac->vc.lock, flags); > > > > dma_set_tx_state(txstate, chan->completed_cookie, chan- > > >cookie, > > residue); > > > > - return sdmac->status; > > + return ret; > > +} > > + > > +static void sdma_start_desc(struct sdma_channel *sdmac) > > +{ > > + struct virt_dma_desc *vd = vchan_next_desc(&sdmac->vc); > > + struct sdma_desc *desc; > > + struct sdma_engine *sdma = sdmac->sdma; > > + int channel = sdmac->channel; > > + > > + if (!vd) { > > + sdmac->desc = NULL; > > + return; > > + } > > + sdmac->desc = desc = to_sdma_desc(&vd->tx); > > + /* > > + * Do not delete the node in desc_issued list in cyclic > > mode, otherwise > > + * the desc alloced will never be freed in > > vchan_dma_desc_free_list > > + */ > > + if (!(sdmac->flags & IMX_DMA_SG_LOOP)) { > > + list_add_tail(&sdmac->desc->node, &sdmac- > > >pending); > > + list_del(&vd->node); > > + } > > + sdma->channel_control[channel].base_bd_ptr = desc- > > >bd_phys; > > + sdma->channel_control[channel].current_bd_ptr = desc- > > >bd_phys; > > + sdma_enable_channel(sdma, sdmac->channel); > > } > > > > static void sdma_issue_pending(struct dma_chan *chan) > > { > > struct sdma_channel *sdmac = to_sdma_chan(chan); > > - struct sdma_engine *sdma = sdmac->sdma; > > + unsigned long flags; > > > > - if (sdmac->status == DMA_IN_PROGRESS) > > - sdma_enable_channel(sdma, sdmac->channel); > > + spin_lock_irqsave(&sdmac->vc.lock, flags); > > + if (vchan_issue_pending(&sdmac->vc) && !sdmac->desc) > > + sdma_start_desc(sdmac); > > + spin_unlock_irqrestore(&sdmac->vc.lock, flags); > > } > > > > #define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1 34 > > @@ -1657,7 +1770,7 @@ static int sdma_init(struct sdma_engine > > *sdma) > > for (i = 0; i < MAX_DMA_CHANNELS; i++) > > writel_relaxed(0, sdma->regs + SDMA_CHNPRI_0 + i * > > 4); > > > > - ret = sdma_request_channel(&sdma->channel[0]); > > + ret = sdma_request_channel0(sdma); > > if (ret) > > goto err_dma_alloc; > > > > @@ -1819,22 +1932,17 @@ static int sdma_probe(struct > > platform_device *pdev) > > struct sdma_channel *sdmac = &sdma->channel[i]; > > > > sdmac->sdma = sdma; > > - spin_lock_init(&sdmac->lock); > > - > > - sdmac->chan.device = &sdma->dma_device; > > - dma_cookie_init(&sdmac->chan); > > sdmac->channel = i; > > - > > - tasklet_init(&sdmac->tasklet, > > mxc_sdma_handle_channel_normal, > > - (unsigned long) sdmac); > > + sdmac->status = DMA_IN_PROGRESS; > > + sdmac->vc.desc_free = sdma_desc_free; > > + INIT_LIST_HEAD(&sdmac->pending); > > /* > > * Add the channel to the DMAC list. Do not add > > channel 0 though > > * because we need it internally in the SDMA > > driver. This also means > > * that channel 0 in dmaengine counting matches > > sdma channel 1. > > */ > > if (i) > > - list_add_tail(&sdmac->chan.device_node, > > - &sdma- > > >dma_device.channels); > > + vchan_init(&sdmac->vc, &sdma->dma_device); > > } > > > > ret = sdma_init(sdma); > > @@ -1879,7 +1987,7 @@ static int sdma_probe(struct platform_device > > *pdev) > > sdma->dma_device.device_prep_slave_sg = > > sdma_prep_slave_sg; > > sdma->dma_device.device_prep_dma_cyclic = > > sdma_prep_dma_cyclic; > > sdma->dma_device.device_config = sdma_config; > > - sdma->dma_device.device_terminate_all = > > sdma_disable_channel_with_delay; > > + sdma->dma_device.device_terminate_all = > > sdma_terminate_all; > > sdma->dma_device.src_addr_widths = SDMA_DMA_BUSWIDTHS; > > sdma->dma_device.dst_addr_widths = SDMA_DMA_BUSWIDTHS; > > sdma->dma_device.directions = SDMA_DMA_DIRECTIONS; > > @@ -1939,7 +2047,8 @@ static int sdma_remove(struct platform_device > > *pdev) > > for (i = 0; i < MAX_DMA_CHANNELS; i++) { > > struct sdma_channel *sdmac = &sdma->channel[i]; > > > > - tasklet_kill(&sdmac->tasklet); > > + tasklet_kill(&sdmac->vc.task); > > + sdma_free_chan_resources(&sdmac->vc.chan); > > } > > > > platform_set_drvdata(pdev, NULL); > > -- > > 2.7.4 > > > > > > _______________________________________________ > > linux-arm-kernel mailing list > > linux-arm-kernel@lists.infradead.org > > https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fl > > ists.infradead.org%2Fmailman%2Flistinfo%2Flinux-arm- > > kernel&data=02%7C01%7Cyibin.gong%40nxp.com%7C3323f6aae75e45a3155f08 > > d5bfcc314f%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63662580608 > > 2347660&sdata=5eDirTg4boJfw0zu320d9GZTeeDwnfCPfHFY8HXt1nI%3D&reserv > > ed=0 > >
On Wed, May 23, 2018 at 10:26:23AM +0000, Robin Gong wrote: > > > > > > > > + u32 bd_size_sum; > > This variable is never used for anything. > Yes, it's not for significative use but debug to see how many current > bds used. I am not convinced this is useful. The variable could easily be added back by someone who debugs this driver. The code has to be changed anyway to make use of this variable. > > > @@ -632,7 +635,7 @@ static int sdma_run_channel0(struct sdma_engine > > > *sdma) > > > static int sdma_load_script(struct sdma_engine *sdma, void *buf, > > > int size, > > > u32 address) > > > { > > > - struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd; > > > + struct sdma_buffer_descriptor *bd0 = sdma->bd0; > > This change seems to be an orthogonal change. Please make this a > > separate patch. > It's something related with virtual dma support, because in virtual > dma framework, all bds should be allocated dynamically if they used. > but bd0 is a specail case since it's must and basic for load sdma > firmware and context for other channels. So here alloc 'bd0' for other > channels. Well, it's somewhat related to virtual dma support, but that's not my point. My point is that this patch is quite big and thus hard to review. If we find ways to make it smaller and to split it up in multiple patches then we should do so, because it makes it easier to review and in case you break something here we raise the chance that a "git bisect" lands on a smaller patch which is easier to understand. Please try and make that a separate change. I haven't really looked into it and it may not be possible due to reasons I haven't seen, but please at least give it a try. > > > > > > > > + > > > + if (sdmac->desc) > > > + sdmac->desc = NULL; > > The test is unnecesary. > This 'NULL' is meaningful in case that dma done interrupt come after > terminate as you know sdma will actually stop after current transfer > done. The setting of the variable to NULL is ok, but the test is useless. if (sdmac->desc) sdmac->desc = NULL; is equivalent to: sdmac->desc = NULL; Sascha
On 23-05-18, 12:56, s.hauer@pengutronix.de wrote: > Well, it's somewhat related to virtual dma support, but that's not my > point. My point is that this patch is quite big and thus hard to review. > If we find ways to make it smaller and to split it up in multiple > patches then we should do so, because it makes it easier to review and > in case you break something here we raise the chance that a "git bisect" > lands on a smaller patch which is easier to understand. > > Please try and make that a separate change. I haven't really looked into > it and it may not be possible due to reasons I haven't seen, but please > at least give it a try. That is something would help me as well. I have reviewed the patch and am not sure I fully understand the changes, so breaking up stuff would definitely help in the review..
Okay, I'll try to split it. On 三, 2018-05-23 at 19:04 +0530, Vinod wrote: > On 23-05-18, 12:56, s.hauer@pengutronix.de wrote: > > > > > Well, it's somewhat related to virtual dma support, but that's not > > my > > point. My point is that this patch is quite big and thus hard to > > review. > > If we find ways to make it smaller and to split it up in multiple > > patches then we should do so, because it makes it easier to review > > and > > in case you break something here we raise the chance that a "git > > bisect" > > lands on a smaller patch which is easier to understand. > > > > Please try and make that a separate change. I haven't really looked > > into > > it and it may not be possible due to reasons I haven't seen, but > > please > > at least give it a try. > That is something would help me as well. I have reviewed the patch > and am not > sure I fully understand the changes, so breaking up stuff would > definitely help > in the review.. >
diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig index 27df3e2..c4ce43c 100644 --- a/drivers/dma/Kconfig +++ b/drivers/dma/Kconfig @@ -247,6 +247,7 @@ config IMX_SDMA tristate "i.MX SDMA support" depends on ARCH_MXC select DMA_ENGINE + select DMA_VIRTUAL_CHANNELS help Support the i.MX SDMA engine. This engine is integrated into Freescale i.MX25/31/35/51/53/6 chips. diff --git a/drivers/dma/imx-sdma.c b/drivers/dma/imx-sdma.c index ccd03c3..df79e73 100644 --- a/drivers/dma/imx-sdma.c +++ b/drivers/dma/imx-sdma.c @@ -48,6 +48,7 @@ #include <linux/mfd/syscon/imx6q-iomuxc-gpr.h> #include "dmaengine.h" +#include "virt-dma.h" /* SDMA registers */ #define SDMA_H_C0PTR 0x000 @@ -291,10 +292,19 @@ struct sdma_context_data { u32 scratch7; } __attribute__ ((packed)); -#define NUM_BD (int)(PAGE_SIZE / sizeof(struct sdma_buffer_descriptor)) - struct sdma_engine; +struct sdma_desc { + struct virt_dma_desc vd; + struct list_head node; + unsigned int num_bd; + dma_addr_t bd_phys; + unsigned int buf_tail; + unsigned int buf_ptail; + struct sdma_channel *sdmac; + struct sdma_buffer_descriptor *bd; +}; + /** * struct sdma_channel - housekeeping for a SDMA channel * @@ -310,19 +320,17 @@ struct sdma_engine; * @num_bd max NUM_BD. number of descriptors currently handling */ struct sdma_channel { + struct virt_dma_chan vc; + struct list_head pending; struct sdma_engine *sdma; + struct sdma_desc *desc; unsigned int channel; enum dma_transfer_direction direction; enum sdma_peripheral_type peripheral_type; unsigned int event_id0; unsigned int event_id1; enum dma_slave_buswidth word_size; - unsigned int buf_tail; - unsigned int buf_ptail; - unsigned int num_bd; unsigned int period_len; - struct sdma_buffer_descriptor *bd; - dma_addr_t bd_phys; unsigned int pc_from_device, pc_to_device; unsigned int device_to_device; unsigned long flags; @@ -330,15 +338,12 @@ struct sdma_channel { unsigned long event_mask[2]; unsigned long watermark_level; u32 shp_addr, per_addr; - struct dma_chan chan; - spinlock_t lock; - struct dma_async_tx_descriptor desc; enum dma_status status; unsigned int chn_count; unsigned int chn_real_count; - struct tasklet_struct tasklet; struct imx_dma_data data; bool enabled; + u32 bd_size_sum; }; #define IMX_DMA_SG_LOOP BIT(0) @@ -398,6 +403,9 @@ struct sdma_engine { u32 spba_start_addr; u32 spba_end_addr; unsigned int irq; + /* channel0 bd */ + dma_addr_t bd0_phys; + struct sdma_buffer_descriptor *bd0; }; static struct sdma_driver_data sdma_imx31 = { @@ -553,6 +561,8 @@ MODULE_DEVICE_TABLE(of, sdma_dt_ids); #define SDMA_H_CONFIG_ACR BIT(4) /* indicates if AHB freq /core freq = 2 or 1 */ #define SDMA_H_CONFIG_CSM (3) /* indicates which context switch mode is selected*/ +static void sdma_start_desc(struct sdma_channel *sdmac); + static inline u32 chnenbl_ofs(struct sdma_engine *sdma, unsigned int event) { u32 chnenbl0 = sdma->drvdata->chnenbl0; @@ -597,14 +607,7 @@ static int sdma_config_ownership(struct sdma_channel *sdmac, static void sdma_enable_channel(struct sdma_engine *sdma, int channel) { - unsigned long flags; - struct sdma_channel *sdmac = &sdma->channel[channel]; - writel(BIT(channel), sdma->regs + SDMA_H_START); - - spin_lock_irqsave(&sdmac->lock, flags); - sdmac->enabled = true; - spin_unlock_irqrestore(&sdmac->lock, flags); } /* @@ -632,7 +635,7 @@ static int sdma_run_channel0(struct sdma_engine *sdma) static int sdma_load_script(struct sdma_engine *sdma, void *buf, int size, u32 address) { - struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd; + struct sdma_buffer_descriptor *bd0 = sdma->bd0; void *buf_virt; dma_addr_t buf_phys; int ret; @@ -691,23 +694,16 @@ static void sdma_event_disable(struct sdma_channel *sdmac, unsigned int event) static void sdma_update_channel_loop(struct sdma_channel *sdmac) { struct sdma_buffer_descriptor *bd; + struct sdma_desc *desc = sdmac->desc; int error = 0; enum dma_status old_status = sdmac->status; - unsigned long flags; - - spin_lock_irqsave(&sdmac->lock, flags); - if (!sdmac->enabled) { - spin_unlock_irqrestore(&sdmac->lock, flags); - return; - } - spin_unlock_irqrestore(&sdmac->lock, flags); /* * loop mode. Iterate over descriptors, re-setup them and * call callback function. */ - while (1) { - bd = &sdmac->bd[sdmac->buf_tail]; + while (desc) { + bd = &desc->bd[desc->buf_tail]; if (bd->mode.status & BD_DONE) break; @@ -726,8 +722,8 @@ static void sdma_update_channel_loop(struct sdma_channel *sdmac) sdmac->chn_real_count = bd->mode.count; bd->mode.status |= BD_DONE; bd->mode.count = sdmac->period_len; - sdmac->buf_ptail = sdmac->buf_tail; - sdmac->buf_tail = (sdmac->buf_tail + 1) % sdmac->num_bd; + desc->buf_ptail = desc->buf_tail; + desc->buf_tail = (desc->buf_tail + 1) % desc->num_bd; /* * The callback is called from the interrupt context in order @@ -735,15 +731,16 @@ static void sdma_update_channel_loop(struct sdma_channel *sdmac) * SDMA transaction status by the time the client tasklet is * executed. */ - - dmaengine_desc_get_callback_invoke(&sdmac->desc, NULL); + spin_unlock(&sdmac->vc.lock); + dmaengine_desc_get_callback_invoke(&desc->vd.tx, NULL); + spin_lock(&sdmac->vc.lock); if (error) sdmac->status = old_status; } } -static void mxc_sdma_handle_channel_normal(unsigned long data) +static void mxc_sdma_handle_channel_normal(struct sdma_channel *data) { struct sdma_channel *sdmac = (struct sdma_channel *) data; struct sdma_buffer_descriptor *bd; @@ -754,8 +751,8 @@ static void mxc_sdma_handle_channel_normal(unsigned long data) * non loop mode. Iterate over all descriptors, collect * errors and call callback function */ - for (i = 0; i < sdmac->num_bd; i++) { - bd = &sdmac->bd[i]; + for (i = 0; i < sdmac->desc->num_bd; i++) { + bd = &sdmac->desc->bd[i]; if (bd->mode.status & (BD_DONE | BD_RROR)) error = -EIO; @@ -766,10 +763,6 @@ static void mxc_sdma_handle_channel_normal(unsigned long data) sdmac->status = DMA_ERROR; else sdmac->status = DMA_COMPLETE; - - dma_cookie_complete(&sdmac->desc); - - dmaengine_desc_get_callback_invoke(&sdmac->desc, NULL); } static irqreturn_t sdma_int_handler(int irq, void *dev_id) @@ -785,13 +778,24 @@ static irqreturn_t sdma_int_handler(int irq, void *dev_id) while (stat) { int channel = fls(stat) - 1; struct sdma_channel *sdmac = &sdma->channel[channel]; - - if (sdmac->flags & IMX_DMA_SG_LOOP) - sdma_update_channel_loop(sdmac); - else - tasklet_schedule(&sdmac->tasklet); + struct sdma_desc *desc; + + spin_lock(&sdmac->vc.lock); + desc = sdmac->desc; + if (desc) { + if (sdmac->flags & IMX_DMA_SG_LOOP) { + sdma_update_channel_loop(sdmac); + } else { + mxc_sdma_handle_channel_normal(sdmac); + vchan_cookie_complete(&desc->vd); + if (!list_empty(&sdmac->pending)) + list_del(&desc->node); + sdma_start_desc(sdmac); + } + } __clear_bit(channel, &stat); + spin_unlock(&sdmac->vc.lock); } return IRQ_HANDLED; @@ -897,7 +901,7 @@ static int sdma_load_context(struct sdma_channel *sdmac) int channel = sdmac->channel; int load_address; struct sdma_context_data *context = sdma->context; - struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd; + struct sdma_buffer_descriptor *bd0 = sdma->bd0; int ret; unsigned long flags; @@ -946,7 +950,7 @@ static int sdma_load_context(struct sdma_channel *sdmac) static struct sdma_channel *to_sdma_chan(struct dma_chan *chan) { - return container_of(chan, struct sdma_channel, chan); + return container_of(chan, struct sdma_channel, vc.chan); } static int sdma_disable_channel(struct dma_chan *chan) @@ -954,15 +958,10 @@ static int sdma_disable_channel(struct dma_chan *chan) struct sdma_channel *sdmac = to_sdma_chan(chan); struct sdma_engine *sdma = sdmac->sdma; int channel = sdmac->channel; - unsigned long flags; writel_relaxed(BIT(channel), sdma->regs + SDMA_H_STATSTOP); sdmac->status = DMA_ERROR; - spin_lock_irqsave(&sdmac->lock, flags); - sdmac->enabled = false; - spin_unlock_irqrestore(&sdmac->lock, flags); - return 0; } @@ -1097,42 +1096,101 @@ static int sdma_set_channel_priority(struct sdma_channel *sdmac, return 0; } -static int sdma_request_channel(struct sdma_channel *sdmac) +static int sdma_alloc_bd(struct sdma_desc *desc) { - struct sdma_engine *sdma = sdmac->sdma; - int channel = sdmac->channel; - int ret = -EBUSY; + u32 bd_size = desc->num_bd * sizeof(struct sdma_buffer_descriptor); + int ret = 0; + unsigned long flags; - sdmac->bd = dma_zalloc_coherent(NULL, PAGE_SIZE, &sdmac->bd_phys, + desc->bd = dma_zalloc_coherent(NULL, bd_size, &desc->bd_phys, GFP_KERNEL); - if (!sdmac->bd) { + if (!desc->bd) { ret = -ENOMEM; goto out; } - sdma->channel_control[channel].base_bd_ptr = sdmac->bd_phys; - sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys; + spin_lock_irqsave(&desc->sdmac->vc.lock, flags); + desc->sdmac->bd_size_sum += bd_size; + spin_unlock_irqrestore(&desc->sdmac->vc.lock, flags); - sdma_set_channel_priority(sdmac, MXC_SDMA_DEFAULT_PRIORITY); - return 0; out: - return ret; } -static dma_cookie_t sdma_tx_submit(struct dma_async_tx_descriptor *tx) +static void sdma_free_bd(struct sdma_desc *desc) { + u32 bd_size = desc->num_bd * sizeof(struct sdma_buffer_descriptor); unsigned long flags; - struct sdma_channel *sdmac = to_sdma_chan(tx->chan); - dma_cookie_t cookie; - spin_lock_irqsave(&sdmac->lock, flags); + if (desc->bd) { + dma_free_coherent(NULL, bd_size, desc->bd, desc->bd_phys); + + spin_lock_irqsave(&desc->sdmac->vc.lock, flags); + desc->sdmac->bd_size_sum -= bd_size; + spin_unlock_irqrestore(&desc->sdmac->vc.lock, flags); + } +} + +static int sdma_request_channel0(struct sdma_engine *sdma) +{ + int ret = 0; + + sdma->bd0 = dma_zalloc_coherent(NULL, PAGE_SIZE, &sdma->bd0_phys, + GFP_KERNEL); + if (!sdma->bd0) { + ret = -ENOMEM; + goto out; + } - cookie = dma_cookie_assign(tx); + sdma->channel_control[0].base_bd_ptr = sdma->bd0_phys; + sdma->channel_control[0].current_bd_ptr = sdma->bd0_phys; - spin_unlock_irqrestore(&sdmac->lock, flags); + sdma_set_channel_priority(&sdma->channel[0], MXC_SDMA_DEFAULT_PRIORITY); +out: - return cookie; + return ret; +} + +static struct sdma_desc *to_sdma_desc(struct dma_async_tx_descriptor *t) +{ + return container_of(t, struct sdma_desc, vd.tx); +} + +static void sdma_desc_free(struct virt_dma_desc *vd) +{ + struct sdma_desc *desc = container_of(vd, struct sdma_desc, vd); + + if (desc) { + sdma_free_bd(desc); + kfree(desc); + } +} + +static int sdma_terminate_all(struct dma_chan *chan) +{ + struct sdma_channel *sdmac = to_sdma_chan(chan); + unsigned long flags; + LIST_HEAD(head); + + spin_lock_irqsave(&sdmac->vc.lock, flags); + vchan_get_all_descriptors(&sdmac->vc, &head); + while (!list_empty(&sdmac->pending)) { + struct sdma_desc *desc = list_first_entry(&sdmac->pending, + struct sdma_desc, node); + + list_del(&desc->node); + spin_unlock_irqrestore(&sdmac->vc.lock, flags); + sdmac->vc.desc_free(&desc->vd); + spin_lock_irqsave(&sdmac->vc.lock, flags); + } + + if (sdmac->desc) + sdmac->desc = NULL; + spin_unlock_irqrestore(&sdmac->vc.lock, flags); + vchan_dma_desc_free_list(&sdmac->vc, &head); + sdma_disable_channel_with_delay(chan); + + return 0; } static int sdma_alloc_chan_resources(struct dma_chan *chan) @@ -1168,18 +1226,11 @@ static int sdma_alloc_chan_resources(struct dma_chan *chan) if (ret) goto disable_clk_ipg; - ret = sdma_request_channel(sdmac); - if (ret) - goto disable_clk_ahb; - ret = sdma_set_channel_priority(sdmac, prio); if (ret) goto disable_clk_ahb; - dma_async_tx_descriptor_init(&sdmac->desc, chan); - sdmac->desc.tx_submit = sdma_tx_submit; - /* txd.flags will be overwritten in prep funcs */ - sdmac->desc.flags = DMA_CTRL_ACK; + sdmac->bd_size_sum = 0; return 0; @@ -1195,7 +1246,7 @@ static void sdma_free_chan_resources(struct dma_chan *chan) struct sdma_channel *sdmac = to_sdma_chan(chan); struct sdma_engine *sdma = sdmac->sdma; - sdma_disable_channel(chan); + sdma_terminate_all(chan); if (sdmac->event_id0) sdma_event_disable(sdmac, sdmac->event_id0); @@ -1207,12 +1258,43 @@ static void sdma_free_chan_resources(struct dma_chan *chan) sdma_set_channel_priority(sdmac, 0); - dma_free_coherent(NULL, PAGE_SIZE, sdmac->bd, sdmac->bd_phys); - clk_disable(sdma->clk_ipg); clk_disable(sdma->clk_ahb); } +static struct sdma_desc *sdma_transfer_init(struct sdma_channel *sdmac, + enum dma_transfer_direction direction, u32 bds) +{ + struct sdma_desc *desc; + + desc = kzalloc((sizeof(*desc)), GFP_KERNEL); + if (!desc) + goto err_out; + + sdmac->status = DMA_IN_PROGRESS; + sdmac->direction = direction; + sdmac->flags = 0; + sdmac->chn_count = 0; + sdmac->chn_real_count = 0; + + desc->sdmac = sdmac; + desc->num_bd = bds; + INIT_LIST_HEAD(&desc->node); + + if (sdma_alloc_bd(desc)) + goto err_desc_out; + + if (sdma_load_context(sdmac)) + goto err_desc_out; + + return desc; + +err_desc_out: + kfree(desc); +err_out: + return NULL; +} + static struct dma_async_tx_descriptor *sdma_prep_slave_sg( struct dma_chan *chan, struct scatterlist *sgl, unsigned int sg_len, enum dma_transfer_direction direction, @@ -1223,35 +1305,24 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( int ret, i, count; int channel = sdmac->channel; struct scatterlist *sg; + struct sdma_desc *desc; - if (sdmac->status == DMA_IN_PROGRESS) + if (!chan) return NULL; - sdmac->status = DMA_IN_PROGRESS; - - sdmac->flags = 0; - sdmac->buf_tail = 0; - sdmac->buf_ptail = 0; - sdmac->chn_real_count = 0; + desc = sdma_transfer_init(sdmac, direction, sg_len); + if (!desc) + goto err_out; dev_dbg(sdma->dev, "setting up %d entries for channel %d.\n", sg_len, channel); - sdmac->direction = direction; ret = sdma_load_context(sdmac); if (ret) goto err_out; - if (sg_len > NUM_BD) { - dev_err(sdma->dev, "SDMA channel %d: maximum number of sg exceeded: %d > %d\n", - channel, sg_len, NUM_BD); - ret = -EINVAL; - goto err_out; - } - - sdmac->chn_count = 0; for_each_sg(sgl, sg, sg_len, i) { - struct sdma_buffer_descriptor *bd = &sdmac->bd[i]; + struct sdma_buffer_descriptor *bd = &desc->bd[i]; int param; bd->buffer_addr = sg->dma_address; @@ -1262,7 +1333,7 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( dev_err(sdma->dev, "SDMA channel %d: maximum bytes for sg entry exceeded: %d > %d\n", channel, count, 0xffff); ret = -EINVAL; - goto err_out; + goto err_bd_out; } bd->mode.count = count; @@ -1307,10 +1378,11 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( bd->mode.status = param; } - sdmac->num_bd = sg_len; - sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys; + return vchan_tx_prep(&sdmac->vc, &desc->vd, flags); - return &sdmac->desc; +err_bd_out: + sdma_free_bd(desc); + kfree(desc); err_out: sdmac->status = DMA_ERROR; return NULL; @@ -1326,39 +1398,32 @@ static struct dma_async_tx_descriptor *sdma_prep_dma_cyclic( int num_periods = buf_len / period_len; int channel = sdmac->channel; int ret, i = 0, buf = 0; + struct sdma_desc *desc; dev_dbg(sdma->dev, "%s channel: %d\n", __func__, channel); - if (sdmac->status == DMA_IN_PROGRESS) - return NULL; - - sdmac->status = DMA_IN_PROGRESS; + /* Now allocate and setup the descriptor. */ + desc = sdma_transfer_init(sdmac, direction, num_periods); + if (!desc) + goto err_out; - sdmac->buf_tail = 0; - sdmac->buf_ptail = 0; - sdmac->chn_real_count = 0; + desc->buf_tail = 0; + desc->buf_ptail = 0; sdmac->period_len = period_len; - sdmac->flags |= IMX_DMA_SG_LOOP; - sdmac->direction = direction; + ret = sdma_load_context(sdmac); if (ret) goto err_out; - if (num_periods > NUM_BD) { - dev_err(sdma->dev, "SDMA channel %d: maximum number of sg exceeded: %d > %d\n", - channel, num_periods, NUM_BD); - goto err_out; - } - if (period_len > 0xffff) { dev_err(sdma->dev, "SDMA channel %d: maximum period size exceeded: %zu > %d\n", channel, period_len, 0xffff); - goto err_out; + goto err_bd_out; } while (buf < buf_len) { - struct sdma_buffer_descriptor *bd = &sdmac->bd[i]; + struct sdma_buffer_descriptor *bd = &desc->bd[i]; int param; bd->buffer_addr = dma_addr; @@ -1366,7 +1431,7 @@ static struct dma_async_tx_descriptor *sdma_prep_dma_cyclic( bd->mode.count = period_len; if (sdmac->word_size > DMA_SLAVE_BUSWIDTH_4_BYTES) - goto err_out; + goto err_bd_out; if (sdmac->word_size == DMA_SLAVE_BUSWIDTH_4_BYTES) bd->mode.command = 0; else @@ -1389,10 +1454,10 @@ static struct dma_async_tx_descriptor *sdma_prep_dma_cyclic( i++; } - sdmac->num_bd = num_periods; - sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys; - - return &sdmac->desc; + return vchan_tx_prep(&sdmac->vc, &desc->vd, flags); +err_bd_out: + sdma_free_bd(desc); + kfree(desc); err_out: sdmac->status = DMA_ERROR; return NULL; @@ -1432,26 +1497,74 @@ static enum dma_status sdma_tx_status(struct dma_chan *chan, { struct sdma_channel *sdmac = to_sdma_chan(chan); u32 residue; + struct virt_dma_desc *vd; + struct sdma_desc *desc; + enum dma_status ret; + unsigned long flags; - if (sdmac->flags & IMX_DMA_SG_LOOP) - residue = (sdmac->num_bd - sdmac->buf_ptail) * + ret = dma_cookie_status(chan, cookie, txstate); + if (ret == DMA_COMPLETE && txstate) { + residue = sdmac->chn_count - sdmac->chn_real_count; + return ret; + } + + spin_lock_irqsave(&sdmac->vc.lock, flags); + vd = vchan_find_desc(&sdmac->vc, cookie); + desc = to_sdma_desc(&vd->tx); + if (vd) { + if (sdmac->flags & IMX_DMA_SG_LOOP) + residue = (desc->num_bd - desc->buf_ptail) * sdmac->period_len - sdmac->chn_real_count; - else + else + residue = sdmac->chn_count - sdmac->chn_real_count; + } else if (sdmac->desc && sdmac->desc->vd.tx.cookie == cookie) { residue = sdmac->chn_count - sdmac->chn_real_count; + } else { + residue = 0; + } + ret = sdmac->status; + spin_unlock_irqrestore(&sdmac->vc.lock, flags); dma_set_tx_state(txstate, chan->completed_cookie, chan->cookie, residue); - return sdmac->status; + return ret; +} + +static void sdma_start_desc(struct sdma_channel *sdmac) +{ + struct virt_dma_desc *vd = vchan_next_desc(&sdmac->vc); + struct sdma_desc *desc; + struct sdma_engine *sdma = sdmac->sdma; + int channel = sdmac->channel; + + if (!vd) { + sdmac->desc = NULL; + return; + } + sdmac->desc = desc = to_sdma_desc(&vd->tx); + /* + * Do not delete the node in desc_issued list in cyclic mode, otherwise + * the desc alloced will never be freed in vchan_dma_desc_free_list + */ + if (!(sdmac->flags & IMX_DMA_SG_LOOP)) { + list_add_tail(&sdmac->desc->node, &sdmac->pending); + list_del(&vd->node); + } + sdma->channel_control[channel].base_bd_ptr = desc->bd_phys; + sdma->channel_control[channel].current_bd_ptr = desc->bd_phys; + sdma_enable_channel(sdma, sdmac->channel); } static void sdma_issue_pending(struct dma_chan *chan) { struct sdma_channel *sdmac = to_sdma_chan(chan); - struct sdma_engine *sdma = sdmac->sdma; + unsigned long flags; - if (sdmac->status == DMA_IN_PROGRESS) - sdma_enable_channel(sdma, sdmac->channel); + spin_lock_irqsave(&sdmac->vc.lock, flags); + if (vchan_issue_pending(&sdmac->vc) && !sdmac->desc) + sdma_start_desc(sdmac); + spin_unlock_irqrestore(&sdmac->vc.lock, flags); } #define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1 34 @@ -1657,7 +1770,7 @@ static int sdma_init(struct sdma_engine *sdma) for (i = 0; i < MAX_DMA_CHANNELS; i++) writel_relaxed(0, sdma->regs + SDMA_CHNPRI_0 + i * 4); - ret = sdma_request_channel(&sdma->channel[0]); + ret = sdma_request_channel0(sdma); if (ret) goto err_dma_alloc; @@ -1819,22 +1932,17 @@ static int sdma_probe(struct platform_device *pdev) struct sdma_channel *sdmac = &sdma->channel[i]; sdmac->sdma = sdma; - spin_lock_init(&sdmac->lock); - - sdmac->chan.device = &sdma->dma_device; - dma_cookie_init(&sdmac->chan); sdmac->channel = i; - - tasklet_init(&sdmac->tasklet, mxc_sdma_handle_channel_normal, - (unsigned long) sdmac); + sdmac->status = DMA_IN_PROGRESS; + sdmac->vc.desc_free = sdma_desc_free; + INIT_LIST_HEAD(&sdmac->pending); /* * Add the channel to the DMAC list. Do not add channel 0 though * because we need it internally in the SDMA driver. This also means * that channel 0 in dmaengine counting matches sdma channel 1. */ if (i) - list_add_tail(&sdmac->chan.device_node, - &sdma->dma_device.channels); + vchan_init(&sdmac->vc, &sdma->dma_device); } ret = sdma_init(sdma); @@ -1879,7 +1987,7 @@ static int sdma_probe(struct platform_device *pdev) sdma->dma_device.device_prep_slave_sg = sdma_prep_slave_sg; sdma->dma_device.device_prep_dma_cyclic = sdma_prep_dma_cyclic; sdma->dma_device.device_config = sdma_config; - sdma->dma_device.device_terminate_all = sdma_disable_channel_with_delay; + sdma->dma_device.device_terminate_all = sdma_terminate_all; sdma->dma_device.src_addr_widths = SDMA_DMA_BUSWIDTHS; sdma->dma_device.dst_addr_widths = SDMA_DMA_BUSWIDTHS; sdma->dma_device.directions = SDMA_DMA_DIRECTIONS; @@ -1939,7 +2047,8 @@ static int sdma_remove(struct platform_device *pdev) for (i = 0; i < MAX_DMA_CHANNELS; i++) { struct sdma_channel *sdmac = &sdma->channel[i]; - tasklet_kill(&sdmac->tasklet); + tasklet_kill(&sdmac->vc.task); + sdma_free_chan_resources(&sdmac->vc.chan); } platform_set_drvdata(pdev, NULL);
The legacy sdma driver has below limitations or drawbacks: 1. Hardcode the max BDs number as "PAGE_SIZE / sizeof(*)", and alloc one page size for one channel regardless of only few BDs needed most time. But in few cases, the max PAGE_SIZE maybe not enough. 2. One SDMA channel can't stop immediatley once channel disabled which means SDMA interrupt may come in after this channel terminated.There are some patches for this corner case such as commit "2746e2c389f9", but not cover non-cyclic. The common virt-dma overcomes the above limitations. It can alloc bd dynamically and free bd once this tx transfer done. No memory wasted or maximum limititation here, only depends on how many memory can be requested from kernel. For No.2, such issue can be workaround by checking if there is available descript("sdmac->desc") now once the unwanted interrupt coming. At last the common virt-dma is easier for sdma driver maintain. Signed-off-by: Robin Gong <yibin.gong@nxp.com> --- drivers/dma/Kconfig | 1 + drivers/dma/imx-sdma.c | 395 +++++++++++++++++++++++++++++++------------------ 2 files changed, 253 insertions(+), 143 deletions(-)