[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20231003121720.3139967-1-kory.maincent@bootlin.com>
Date: Tue, 3 Oct 2023 14:17:20 +0200
From: Köry Maincent <kory.maincent@...tlin.com>
To: Cai Huoqing <cai.huoqing@...ux.dev>,
Manivannan Sadhasivam <mani@...nel.org>,
Serge Semin <fancer.lancer@...il.com>,
Vinod Koul <vkoul@...nel.org>,
Gustavo Pimentel <Gustavo.Pimentel@...opsys.com>,
dmaengine@...r.kernel.org, linux-kernel@...r.kernel.org
Cc: Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
Gustavo Pimentel <gustavo.pimentel@...opsys.com>,
Herve Codina <herve.codina@...tlin.com>,
Kory Maincent <kory.maincent@...tlin.com>
Subject: [PATCH v2 5/5] dmaengine: dw-edma: eDMA: Add sync read before starting the DMA transfer in remote setup
From: Kory Maincent <kory.maincent@...tlin.com>
The Linked list element and pointer are not stored in the same memory as
the eDMA controller register. If the doorbell register is toggled before
the full write of the linked list a race condition error can appears.
In remote setup we can only use a readl to the memory to assured the full
write has occurred.
Fixes: 7e4b8a4fbe2c ("dmaengine: Add Synopsys eDMA IP version 0 support")
Signed-off-by: Kory Maincent <kory.maincent@...tlin.com>
---
drivers/dma/dw-edma/dw-edma-v0-core.c | 17 +++++++++++++++++
1 file changed, 17 insertions(+)
diff --git a/drivers/dma/dw-edma/dw-edma-v0-core.c b/drivers/dma/dw-edma/dw-edma-v0-core.c
index b38786f0ad79..6245b720fbfe 100644
--- a/drivers/dma/dw-edma/dw-edma-v0-core.c
+++ b/drivers/dma/dw-edma/dw-edma-v0-core.c
@@ -346,6 +346,20 @@ static void dw_edma_v0_core_write_chunk(struct dw_edma_chunk *chunk)
dw_edma_v0_write_ll_link(chunk, i, control, chunk->ll_region.paddr);
}
+static void dw_edma_v0_sync_ll_data(struct dw_edma_chunk *chunk)
+{
+ /*
+ * In case of remote eDMA engine setup, the DW PCIe RP/EP internals
+ * configuration registers and Application memory are normally accessed
+ * over different buses. Ensure LL-data reaches the memory before the
+ * doorbell register is toggled by issuing the dummy-read from the remote
+ * LL memory in a hope that the posted MRd TLP will return only after the
+ * last MWr TLP is completed
+ */
+ if (!(chunk->chan->dw->chip->flags & DW_EDMA_CHIP_LOCAL))
+ readl(chunk->ll_region.vaddr.io);
+}
+
static void dw_edma_v0_core_start(struct dw_edma_chunk *chunk, bool first)
{
struct dw_edma_chan *chan = chunk->chan;
@@ -412,6 +426,9 @@ static void dw_edma_v0_core_start(struct dw_edma_chunk *chunk, bool first)
SET_CH_32(dw, chan->dir, chan->id, llp.msb,
upper_32_bits(chunk->ll_region.paddr));
}
+
+ dw_edma_v0_sync_ll_data(chunk);
+
/* Doorbell */
SET_RW_32(dw, chan->dir, doorbell,
FIELD_PREP(EDMA_V0_DOORBELL_CH_MASK, chan->id));
--
2.25.1
Powered by blists - more mailing lists