Skip to contentSkip to conversation

dma engine drivers for 2.6.25?

DISCUSSION Kumar Gala 3 participants 6 messages Feb 14, 2008 – Feb 25, 2008

Subsystems (2)
lore
Share & tools
lore

Kumar Gala

author
#0 lore raw

Dan,

What's going on with the dma engine drivers for 2.6.25? We had a Freescale dma driver from Zhang Wei queued up but seems to have been lost.

- k

#1 lore raw
Quoted 5 lines · Kumar Gala

Dan,

What's going on with the dma engine drivers for 2.6.25? We had a Freescale dma driver from Zhang Wei queued up but seems to have been lost.

I pulled it into my tree and am holding it until Zhang has an opportunity to address the pending review comments [1]. I also did not feel comfortable pushing it to Linus without a PPC maintainer's Acked-by/Reviewed-by.

I have attached the version I am carrying.

--
Dan

[1] http://marc.info/?l=linux-kernel&m=119998939916613&w=2

dmaengine: add driver for Freescale MPC85xx DMA controller

From: Zhang Wei <[email protected]>

The driver implements DMA engine API for Freescale MPC85xx DMA controller, which could be used by devices in the silicon. The driver supports the Basic mode of Freescale MPC85xx DMA controller. The MPC85xx processors supported include MPC8540/60, MPC8555, MPC8548, MPC8641 and so on.

The MPC83xx(MPC8349, MPC8360) are also supported.

[[email protected]: build fix]
[[email protected]: merge mm fixes, rebase on async_tx-2.6.25]
Cc Shannon Nelson <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Kumar Gala <[email protected]>

---

 MAINTAINERS          |    7 
 drivers/dma/Kconfig  |   19 +
 drivers/dma/Makefile |    1 
 drivers/dma/fsldma.c | 1060 ++++++++++++++++++++++++++++++++++++++++++++++++++
 drivers/dma/fsldma.h |  189 +++++++++
 5 files changed, 1275 insertions(+), 1 deletions(-)
 create mode 100644 drivers/dma/fsldma.c
 create mode 100644 drivers/dma/fsldma.h
MAINTAINERS +7 -0
index 6680ec4..9ba1630 100644
@@ -1584,6 +1584,13 @@ L:	[email protected] (subscribers-only)
 W:	http://linux-fbdev.sourceforge.net/
 S:	Maintained
 
+FREESCALE DMA DRIVER
+P;	Zhang Wei
+M:	[email protected]
+L:	[email protected]
+L:	[email protected]
+S:	Maintained
+
 FREESCALE SOC FS_ENET DRIVER
 P:	Pantelis Antoniou
 M:	[email protected]
drivers/dma/Kconfig +18 -1
index a703def..27340a7 100644
@@ -4,7 +4,7 @@
 
 menuconfig DMADEVICES
 	bool "DMA Engine support"
-	depends on (PCI && X86) || ARCH_IOP32X || ARCH_IOP33X || ARCH_IOP13XX
+	depends on (PCI && X86) || ARCH_IOP32X || ARCH_IOP33X || ARCH_IOP13XX || PPC
 	depends on !HIGHMEM64G
 	help
 	  DMA engines can do asynchronous data transfers without
@@ -37,6 +37,23 @@ config INTEL_IOP_ADMA
 	help
 	  Enable support for the Intel(R) IOP Series RAID engines.
 
+config FSL_DMA
+	bool "Freescale MPC85xx/MPC83xx DMA support"
+	depends on PPC
+	select DMA_ENGINE
+	---help---
+	  Enable support for the Freescale DMA engine. Now, it support
+	  MPC8560/40, MPC8555, MPC8548 and MPC8641 processors.
+	  The MPC8349, MPC8360 is also supported.
+
+config FSL_DMA_SELFTEST
+	bool "Enable the self test for each DMA channel"
+	depends on FSL_DMA
+	default y
+	---help---
+	  Enable the self test for each DMA channel. A self test will be
+	  performed after the channel probed to ensure the DMA works well.
+
 config DMA_ENGINE
 	bool
 
drivers/dma/Makefile +1 -0
index b152cd8..c8036d9 100644
@@ -3,3 +3,4 @@ obj-$(CONFIG_NET_DMA) += iovlock.o
 obj-$(CONFIG_INTEL_IOATDMA) += ioatdma.o
 ioatdma-objs := ioat.o ioat_dma.o ioat_dca.o
 obj-$(CONFIG_INTEL_IOP_ADMA) += iop-adma.o
+obj-$(CONFIG_FSL_DMA) += fsldma.o
drivers/dma/fsldma.c new file +1060 -0

1,060 lines not on this page · raw message

drivers/dma/fsldma.h new file +189 -0

189 lines not on this page · raw message

author
#2 lore raw
Quoted 7 lines · Dan Williams
Quoted 5 lines · Kumar Gala

Dan,

What's going on with the dma engine drivers for 2.6.25? We had a Freescale dma driver from Zhang Wei queued up but seems to have been lost.

Quoted 6 lines

I pulled it into my tree and am holding it until Zhang has an opportunity to address the pending review comments [1]. I also did not feel comfortable pushing it to Linus without a PPC maintainer's Acked-by/Reviewed-by.

I have attached the version I am carrying.

What issues are still open. I was under the belief that Zhang had resolved all the issues.

- k

#3 lore raw
Quoted 16 lines · Kumar Gala
Quoted 13 lines · Dan Williams
Quoted 5 lines · Kumar Gala

Dan,

What's going on with the dma engine drivers for 2.6.25? We had a Freescale dma driver from Zhang Wei queued up but seems to have been lost.

I pulled it into my tree and am holding it until Zhang has an opportunity to address the pending review comments [1]. I also did not feel comfortable pushing it to Linus without a PPC maintainer's Acked-by/Reviewed-by.

Quoted 5 lines
Quoted 1 line

I have attached the version I am carrying.

What issues are still open. I was under the belief that Zhang had resolved all the issues.

The high priority review item is that the driver performs operation completion callbacks in hardirq context. Clients of the API assume that callbacks will happen in softirq context. Of lesser concern is the use of GFP_ATOMIC in fsl_dma_alloc_descriptor. Other drivers preallocate a small pool of descriptors.

Regards,
Dan

#4 lore raw
Quoted 23 lines · Dan Williams
Quoted 16 lines · Kumar Gala
Quoted 13 lines · Dan Williams
Quoted 5 lines · Kumar Gala

Dan,

What's going on with the dma engine drivers for 2.6.25? We had a Freescale dma driver from Zhang Wei queued up but seems to have been lost.

I pulled it into my tree and am holding it until Zhang has an opportunity to address the pending review comments [1]. I also did not feel comfortable pushing it to Linus without a PPC maintainer's Acked-by/Reviewed-by.

Quoted 5 lines
Quoted 1 line

I have attached the version I am carrying.

What issues are still open. I was under the belief that Zhang had resolved all the issues.

Quoted 6 lines

The high priority review item is that the driver performs operation completion callbacks in hardirq context. Clients of the API assume that callbacks will happen in softirq context. Of lesser concern is the use of GFP_ATOMIC in fsl_dma_alloc_descriptor. Other drivers preallocate a small pool of descriptors.

Have not received a response, so let's try this the other way. I took a closer look and found that my concern should be addressed by the following one-liner:

drivers/dma/fsldma.c +0 -0
index 902e852..cc9a681 100644
@@ -685,7 +685,6 @@ static irqreturn_t fsl_dma_chan_do_interrupt(int

irq, void *data)

                                "nlndar 0x%016llx\n", (u64)get_cdar(fsl_chan),
                                (u64)get_ndar(fsl_chan));
                stat &= ~FSL_DMA_SR_EOSI;
-               fsl_chan_ld_cleanup(fsl_chan);
        }

        /* If it current transfer is the end-of-transfer,

With your ack I'll push the driver plus this fixlet for the current kernel.

Regards,
Dan

#5 lore raw

Hi,

Sorry for missing lots of emails in my long vacation. :)

Quoted 64 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Dan Williams
Sent: Sunday, February 24, 2008 10:06 AM
To: Kumar Gala
Cc: LKML Kernel; Zhang Wei
Subject: Re: dma engine drivers for 2.6.25?

Quoted 1 line · Dan Williams

On Thu, Feb 14, 2008 at 8:44 PM, Kumar Gala

Quoted 8 lines · <[email protected]> wrote:
Quoted 8 lines · Dan Williams
Quoted 6 lines
Quoted 3 lines · Kumar Gala

Dan,

What's going on with the dma engine drivers for

2.6.25? We had a

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Freescale dma driver from Zhang Wei queued up but

seems to have been

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 1 line

lost.

I pulled it into my tree and am holding it until Zhang has an opportunity to address the pending review comments

[1]. I also did

Quoted 1 line
Quoted 1 line
Quoted 1 line

not feel comfortable pushing it to Linus without a

PPC maintainer's

Quoted 5 lines
Quoted 5 lines
Quoted 3 lines

Acked-by/Reviewed-by.

I have attached the version I am carrying.

What issues are still open. I was under the belief

that Zhang had

Quoted 6 lines
Quoted 2 lines

resolved all the issues.

The high priority review item is that the driver performs operation completion callbacks in hardirq context. Clients of the API assume that callbacks will happen in softirq context. Of lesser

concern is

Quoted 3 lines

the use of GFP_ATOMIC in fsl_dma_alloc_descriptor. Other drivers preallocate a small pool of descriptors.

Have not received a response, so let's try this the other way. I took a closer look and found that my concern should be addressed by the following one-liner:

drivers/dma/fsldma.c +0 -1
index 902e852..cc9a681 100644
@@ -685,7 +685,6 @@ static irqreturn_t fsl_dma_chan_do_interrupt(int
 irq, void *data)
                                "nlndar 0x%016llx\n", 
 (u64)get_cdar(fsl_chan),
                                (u64)get_ndar(fsl_chan));
                stat &= ~FSL_DMA_SR_EOSI;
-               fsl_chan_ld_cleanup(fsl_chan);
        }
Quoted 4 lines
        /* If it current transfer is the end-of-transfer,

With your ack I'll push the driver plus this fixlet for the current kernel.

I agree with you to remove the fsl_chan_ld_cleanup() function calling in Interrupt action, which is already called in tasklet. I forgot to remove it. Thanks!

Cheers!
Wei.