* [PATCH 2/2] dw_dmac: max_mem_width limits value for SRC/DST_TR_WID register
@ 2012-08-26 20:41 Hein Tibosch
[not found] ` <CAKohpontcFpMiju1LwO3fL43Oud0jAixYf8ZNN+6-hTj35ownA@mail.gmail.com>
2012-08-31 23:49 ` Andrew Morton
0 siblings, 2 replies; 4+ messages in thread
From: Hein Tibosch @ 2012-08-26 20:41 UTC (permalink / raw)
To: viresh kumar
Cc: spear-devel, Linux Kernel Mailing List, ludovic.desroches,
Havard Skinnemoen, Nicolas Ferre, egtvedt, Andrew Morton,
Arnd Bergmann
The dw_dmac driver was earlier adapted to do 64-bit transfers
on the memory side (https://lkml.org/lkml/2012/1/18/52)
This works on ARM platforms but for AVR32 (AP700x) the maximum
allowed transfer size is 32-bits.
This patch allows the arch to set a new slave property
max_mem_width to limit the size.
Allowable values for dw_dma_slave::max_mem_width are:
0 : leave it up to dw_dmac (64 bits)
1 : 16-bits
2 : 32-bits
Signed-off-by: Hein Tibosch <hein_tibosch@yahoo.es>
---
drivers/dma/dw_dmac.c | 8 ++++++++
include/linux/dw_dmac.h | 3 +++
2 files changed, 11 insertions(+), 0 deletions(-)
diff --git a/drivers/dma/dw_dmac.c b/drivers/dma/dw_dmac.c
index 7212961..a4bdf1d 100644
--- a/drivers/dma/dw_dmac.c
+++ b/drivers/dma/dw_dmac.c
@@ -754,6 +754,10 @@ dwc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl,
mem_width = 1;
else
mem_width = 0;
+ /* Some controllers don't support 64-bits mem access */
+ if (dws->max_mem_width &&
+ mem_width > dws->max_mem_width)
+ mem_width = dws->max_mem_width;
slave_sg_todev_fill_desc:
desc = dwc_desc_get(dwc);
@@ -821,6 +825,10 @@ slave_sg_todev_fill_desc:
mem_width = 1;
else
mem_width = 0;
+ /* Some controllers don't support 64-bits mem access */
+ if (dws->max_mem_width &&
+ mem_width > dws->max_mem_width)
+ mem_width = dws->max_mem_width;
slave_sg_fromdev_fill_desc:
desc = dwc_desc_get(dwc);
diff --git a/include/linux/dw_dmac.h b/include/linux/dw_dmac.h
index 2412e02..60b01c3 100644
--- a/include/linux/dw_dmac.h
+++ b/include/linux/dw_dmac.h
@@ -51,6 +51,8 @@ enum dw_dma_msize {
* @cfg_lo: Platform-specific initializer for the CFG_LO register
* @src_master: src master for transfers on allocated channel.
* @dst_master: dest master for transfers on allocated channel.
+ * @max_mem_width: max value for SRC/DST_TR_WIDTH register
+ * default 0 means: leave it up to driver
*/
struct dw_dma_slave {
struct device *dma_dev;
@@ -58,6 +60,7 @@ struct dw_dma_slave {
u32 cfg_lo;
u8 src_master;
u8 dst_master;
+ u8 max_mem_width;
};
/* Platform-configurable bits in CFG_HI */
--
1.7.8.0
^ permalink raw reply related [flat|nested] 4+ messages in thread[parent not found: <CAKohpontcFpMiju1LwO3fL43Oud0jAixYf8ZNN+6-hTj35ownA@mail.gmail.com>]
* Re: [PATCH 2/2] dw_dmac: max_mem_width limits value for SRC/DST_TR_WID register [not found] ` <CAKohpontcFpMiju1LwO3fL43Oud0jAixYf8ZNN+6-hTj35ownA@mail.gmail.com> @ 2012-08-27 7:48 ` Hein Tibosch 2012-08-27 8:26 ` Viresh Kumar 0 siblings, 1 reply; 4+ messages in thread From: Hein Tibosch @ 2012-08-27 7:48 UTC (permalink / raw) To: Viresh Kumar Cc: spear-devel, Linux Kernel Mailing List, ludovic.desroches, Havard Skinnemoen, Nicolas Ferre, egtvedt, Andrew Morton, Arnd Bergmann, vinod.koul, dan.j.williams On 8/27/2012 11:47 AM, Viresh Kumar wrote: > On 27 August 2012 02:11, Hein Tibosch <hein_tibosch@yahoo.es <mailto:hein_tibosch@yahoo.es>> wrote: > > The dw_dmac driver was earlier adapted to do 64-bit transfers > on the memory side (https://lkml.org/lkml/2012/1/18/52) > This works on ARM platforms but for AVR32 (AP700x) the maximum > allowed transfer size is 32-bits. > This patch allows the arch to set a new slave property > max_mem_width to limit the size. > Allowable values for dw_dma_slave::max_mem_width are: > > 0 : leave it up to dw_dmac (64 bits) > 1 : 16-bits > 2 : 32-bits > > > Either this should be in increasing order or decreasing. That will make it > more sensible, isn't it? > > So it should be: > 0-64bit > 1-32bit > 2-16bit > > Also, i am not sure if we should actually support 16 bit here at all. Maybe nobody > will have 16 bit requirement. So can remove it for now. If in future it is required > then can accept that. > Viresh, If you're OK with the version below, I'll send both dw_dmac patches again as v2 Now cc Vinod Koul and Dan Williams The dw_dmac driver was earlier adapted to do 64-bit transfers on the memory side (https://lkml.org/lkml/2012/1/18/52) This works on ARM platforms but for AVR32 (AP700x) the maximum allowed transfer size is 32-bits. A new 'automatic' config item 'DW_DMAC_MEM_64_BIT' determines if 64-bit memory transfers are allowed. This boolean will be defined on non-AVR32 platforms No arch code will have to be changed to make this patch work. Signed-off-by: Hein Tibosch <hein_tibosch@yahoo.es> --- drivers/dma/Kconfig | 8 ++++++++ drivers/dma/dw_dmac.c | 15 ++++++++++++--- 2 files changed, 20 insertions(+), 3 deletions(-) diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig index 3635daf..82e958f 100644 --- a/drivers/dma/Kconfig +++ b/drivers/dma/Kconfig @@ -97,6 +97,14 @@ config DW_DMAC_BE Say yes if access to the Synopsys DesignWare AHB DMA controller should be big endian, such as for Atmel AT32ap7000 +config DW_DMAC_MEM_64_BIT + bool "Allow Synopsys DesignWare AHB DMA to do 64-bit mem transfers" + default y if !AVR32 + depends on DW_DMAC + help + Say yes if the Synopsys DesignWare AHB DMA controller may do + 64-bit memory transfers + config AT_HDMAC tristate "Atmel AHB DMA support" depends on ARCH_AT91 diff --git a/drivers/dma/dw_dmac.c b/drivers/dma/dw_dmac.c index 7212961..2be010c 100644 --- a/drivers/dma/dw_dmac.c +++ b/drivers/dma/dw_dmac.c @@ -639,9 +639,12 @@ dwc_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src, * We can be a lot more clever here, but this should take care * of the most common optimization. */ +#ifdef DW_DMAC_MEM_64_BIT if (!((src | dest | len) & 7)) src_width = dst_width = 3; - else if (!((src | dest | len) & 3)) + else +#endif + if (!((src | dest | len) & 3)) src_width = dst_width = 2; else if (!((src | dest | len) & 1)) src_width = dst_width = 1; @@ -746,9 +749,12 @@ dwc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, mem = sg_dma_address(sg); len = sg_dma_len(sg); +#ifdef DW_DMAC_MEM_64_BIT if (!((mem | len) & 7)) mem_width = 3; - else if (!((mem | len) & 3)) + else +#endif + if (!((mem | len) & 3)) mem_width = 2; else if (!((mem | len) & 1)) mem_width = 1; @@ -813,9 +819,12 @@ slave_sg_todev_fill_desc: mem = sg_dma_address(sg); len = sg_dma_len(sg); +#ifdef DW_DMAC_MEM_64_BIT if (!((mem | len) & 7)) mem_width = 3; - else if (!((mem | len) & 3)) + else +#endif + if (!((mem | len) & 3)) mem_width = 2; else if (!((mem | len) & 1)) mem_width = 1; -- 1.7.8.0 ^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH 2/2] dw_dmac: max_mem_width limits value for SRC/DST_TR_WID register 2012-08-27 7:48 ` Hein Tibosch @ 2012-08-27 8:26 ` Viresh Kumar 0 siblings, 0 replies; 4+ messages in thread From: Viresh Kumar @ 2012-08-27 8:26 UTC (permalink / raw) To: Hein Tibosch Cc: spear-devel, Linux Kernel Mailing List, ludovic.desroches, Havard Skinnemoen, Nicolas Ferre, egtvedt, Andrew Morton, Arnd Bergmann, vinod.koul, dan.j.williams On 27 August 2012 13:18, Hein Tibosch <hein_tibosch@yahoo.es> wrote: > Viresh, > > If you're OK with the version below, I'll send both dw_dmac patches > again as v2 Yes, they look fine. viresh ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH 2/2] dw_dmac: max_mem_width limits value for SRC/DST_TR_WID register 2012-08-26 20:41 [PATCH 2/2] dw_dmac: max_mem_width limits value for SRC/DST_TR_WID register Hein Tibosch [not found] ` <CAKohpontcFpMiju1LwO3fL43Oud0jAixYf8ZNN+6-hTj35ownA@mail.gmail.com> @ 2012-08-31 23:49 ` Andrew Morton 1 sibling, 0 replies; 4+ messages in thread From: Andrew Morton @ 2012-08-31 23:49 UTC (permalink / raw) To: Hein Tibosch Cc: viresh kumar, spear-devel, Linux Kernel Mailing List, ludovic.desroches, Havard Skinnemoen, Nicolas Ferre, egtvedt, Arnd Bergmann On Mon, 27 Aug 2012 04:41:04 +0800 Hein Tibosch <hein_tibosch@yahoo.es> wrote: > The dw_dmac driver was earlier adapted to do 64-bit transfers > on the memory side (https://lkml.org/lkml/2012/1/18/52) > This works on ARM platforms but for AVR32 (AP700x) the maximum > allowed transfer size is 32-bits. > This patch allows the arch to set a new slave property > max_mem_width to limit the size. > Allowable values for dw_dma_slave::max_mem_width are: > This patch fails to apply to mainline and to linux-next and I can't work out how to make it to apply to either. And "avr32-linux: at32ap700x: set DMA slave properties for MCI dw_dmac" is said to depend on this patch. All confused, giving up. ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2012-08-31 23:49 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-08-26 20:41 [PATCH 2/2] dw_dmac: max_mem_width limits value for SRC/DST_TR_WID register Hein Tibosch
[not found] ` <CAKohpontcFpMiju1LwO3fL43Oud0jAixYf8ZNN+6-hTj35ownA@mail.gmail.com>
2012-08-27 7:48 ` Hein Tibosch
2012-08-27 8:26 ` Viresh Kumar
2012-08-31 23:49 ` Andrew Morton
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox