[v2] dmaengine: ti: k3-udma: Workaround errata i2234

Commit Message

Jayesh Choudhary March 23, 2023, 12:01 p.m. UTC

From: Vignesh Raghavendra <[e-mail]>

Per [1], UDMA TR15 transactions may hang if ICNT0 is less than 64B
Work around is to set EOL flag is to 1 for ICNT0.

Since, there is no performance penalty / side effects of setting EOL
flag event ICNTO > 64B, just set the flag for all UDMAP TR15
descriptors.

[1] https://www.ti.com/lit/er/sprz455a/sprz455a.pdf
Errata doc for J721E DRA829/TDA4VM Processors Silicon Revision 1.1/1.0
(Rev. A)

Signed-off-by: Vignesh Raghavendra <[e-mail]>
[[e-mail]: minor cleanups]
Signed-off-by: Jayesh Choudhary <[e-mail]>
---

Changelog v1->v2:
- Correct the commit subject
- Update functions that uses only tr_type15_t, and not tr_type_1_t
  as PDMAs and CSI uses tr_type_1_t (to address Peter's comment in v1)

v1 patch link:
<https://lore.kernel.org/all/[e-mail]/>

 drivers/dma/ti/k3-udma.c     | 20 +++++++++++---------
 include/linux/dma/ti-cppi5.h |  1 +
 2 files changed, 12 insertions(+), 9 deletions(-)

Comments

Vinod Koul March 31, 2023, 12:19 p.m. UTC | #1

On 23-03-23, 17:31, Jayesh Choudhary wrote:
> From: Vignesh Raghavendra <[e-mail]>
> 
> Per [1], UDMA TR15 transactions may hang if ICNT0 is less than 64B
> Work around is to set EOL flag is to 1 for ICNT0.
> 
> Since, there is no performance penalty / side effects of setting EOL
> flag event ICNTO > 64B, just set the flag for all UDMAP TR15
> descriptors.

Applied, thanks

Patch

diff --git a/drivers/dma/ti/k3-udma.c b/drivers/dma/ti/k3-udma.c
index 7e23a6fdef95..e33a7fec2c0f 100644
--- a/drivers/dma/ti/k3-udma.c
+++ b/drivers/dma/ti/k3-udma.c
@@ -2964,6 +2964,7 @@  udma_prep_slave_sg_triggered_tr(struct udma_chan *uc, struct scatterlist *sgl,
 	struct scatterlist *sgent;
 	struct cppi5_tr_type15_t *tr_req = NULL;
 	enum dma_slave_buswidth dev_width;
+	u32 csf = CPPI5_TR_CSF_SUPR_EVT;
 	u16 tr_cnt0, tr_cnt1;
 	dma_addr_t dev_addr;
 	struct udma_desc *d;
@@ -3034,6 +3035,7 @@  udma_prep_slave_sg_triggered_tr(struct udma_chan *uc, struct scatterlist *sgl,
 
 	if (uc->ud->match_data->type == DMA_TYPE_UDMA) {
 		asel = 0;
+		csf |= CPPI5_TR_CSF_EOL_ICNT0;
 	} else {
 		asel = (u64)uc->config.asel << K3_ADDRESS_ASEL_SHIFT;
 		dev_addr |= asel;
@@ -3057,7 +3059,7 @@  udma_prep_slave_sg_triggered_tr(struct udma_chan *uc, struct scatterlist *sgl,
 
 		cppi5_tr_init(&tr_req[tr_idx].flags, CPPI5_TR_TYPE15, false,
 			      true, CPPI5_TR_EVENT_SIZE_COMPLETION, 0);
-		cppi5_tr_csf_set(&tr_req[tr_idx].flags, CPPI5_TR_CSF_SUPR_EVT);
+		cppi5_tr_csf_set(&tr_req[tr_idx].flags, csf);
 		cppi5_tr_set_trigger(&tr_req[tr_idx].flags,
 				     uc->config.tr_trigger_type,
 				     CPPI5_TR_TRIGGER_TYPE_ICNT2_DEC, 0, 0);
@@ -3103,8 +3105,7 @@  udma_prep_slave_sg_triggered_tr(struct udma_chan *uc, struct scatterlist *sgl,
 			cppi5_tr_init(&tr_req[tr_idx].flags, CPPI5_TR_TYPE15,
 				      false, true,
 				      CPPI5_TR_EVENT_SIZE_COMPLETION, 0);
-			cppi5_tr_csf_set(&tr_req[tr_idx].flags,
-					 CPPI5_TR_CSF_SUPR_EVT);
+			cppi5_tr_csf_set(&tr_req[tr_idx].flags, csf);
 			cppi5_tr_set_trigger(&tr_req[tr_idx].flags,
 					     uc->config.tr_trigger_type,
 					     CPPI5_TR_TRIGGER_TYPE_ICNT2_DEC,
@@ -3148,8 +3149,7 @@  udma_prep_slave_sg_triggered_tr(struct udma_chan *uc, struct scatterlist *sgl,
 		d->residue += sg_len;
 	}
 
-	cppi5_tr_csf_set(&tr_req[tr_idx - 1].flags,
-			 CPPI5_TR_CSF_SUPR_EVT | CPPI5_TR_CSF_EOP);
+	cppi5_tr_csf_set(&tr_req[tr_idx - 1].flags, csf | CPPI5_TR_CSF_EOP);
 
 	return d;
 }
@@ -3678,6 +3678,7 @@  udma_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 	int num_tr;
 	size_t tr_size = sizeof(struct cppi5_tr_type15_t);
 	u16 tr0_cnt0, tr0_cnt1, tr1_cnt0;
+	u32 csf = CPPI5_TR_CSF_SUPR_EVT;
 
 	if (uc->config.dir != DMA_MEM_TO_MEM) {
 		dev_err(chan->device->dev,
@@ -3708,13 +3709,15 @@  udma_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 	if (uc->ud->match_data->type != DMA_TYPE_UDMA) {
 		src |= (u64)uc->ud->asel << K3_ADDRESS_ASEL_SHIFT;
 		dest |= (u64)uc->ud->asel << K3_ADDRESS_ASEL_SHIFT;
+	} else {
+		csf |= CPPI5_TR_CSF_EOL_ICNT0;
 	}
 
 	tr_req = d->hwdesc[0].tr_req_base;
 
 	cppi5_tr_init(&tr_req[0].flags, CPPI5_TR_TYPE15, false, true,
 		      CPPI5_TR_EVENT_SIZE_COMPLETION, 0);
-	cppi5_tr_csf_set(&tr_req[0].flags, CPPI5_TR_CSF_SUPR_EVT);
+	cppi5_tr_csf_set(&tr_req[0].flags, csf);
 
 	tr_req[0].addr = src;
 	tr_req[0].icnt0 = tr0_cnt0;
@@ -3733,7 +3736,7 @@  udma_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 	if (num_tr == 2) {
 		cppi5_tr_init(&tr_req[1].flags, CPPI5_TR_TYPE15, false, true,
 			      CPPI5_TR_EVENT_SIZE_COMPLETION, 0);
-		cppi5_tr_csf_set(&tr_req[1].flags, CPPI5_TR_CSF_SUPR_EVT);
+		cppi5_tr_csf_set(&tr_req[1].flags, csf);
 
 		tr_req[1].addr = src + tr0_cnt1 * tr0_cnt0;
 		tr_req[1].icnt0 = tr1_cnt0;
@@ -3748,8 +3751,7 @@  udma_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 		tr_req[1].dicnt3 = 1;
 	}
 
-	cppi5_tr_csf_set(&tr_req[num_tr - 1].flags,
-			 CPPI5_TR_CSF_SUPR_EVT | CPPI5_TR_CSF_EOP);
+	cppi5_tr_csf_set(&tr_req[num_tr - 1].flags, csf | CPPI5_TR_CSF_EOP);
 
 	if (uc->config.metadata_size)
 		d->vd.tx.metadata_ops = &metadata_ops;
diff --git a/include/linux/dma/ti-cppi5.h b/include/linux/dma/ti-cppi5.h
index efa2f0309f00..c53c0f6e3b1a 100644
--- a/include/linux/dma/ti-cppi5.h
+++ b/include/linux/dma/ti-cppi5.h
@@ -616,6 +616,7 @@  static inline void *cppi5_hdesc_get_swdata(struct cppi5_host_desc_t *desc)
 #define   CPPI5_TR_CSF_SUPR_EVT			BIT(2)
 #define   CPPI5_TR_CSF_EOL_ADV_SHIFT		(4U)
 #define   CPPI5_TR_CSF_EOL_ADV_MASK		GENMASK(6, 4)
+#define   CPPI5_TR_CSF_EOL_ICNT0		BIT(4)
 #define   CPPI5_TR_CSF_EOP			BIT(7)
 
 /**