hcd_alloc_coherent() and usb_alloc_coherent() ends up allocating coherent
memory on behalf of ci_hdrc driver. But as the ci_hdrc is instantiated manually
it will not have any dma_mem or dma_ops assigned, which makes the
dma_alloc_coherent() fail on some platforms (e.g. arm64). This patch solves
this by assigning the dma_mem and dma_ops based on the parent's DeviceTree
node.
Cc: Stephen Boyd <[email protected]>
Signed-off-by: Bjorn Andersson <[email protected]>
---
Hi Peter,
After (once more) debugging why USB doesn't work up on the 64-bit Qualcomm
systems I realized that we never concluded on this patch. Unfortunately I can't
find it in my mailbox either, so resending it to restart the discussion.
Regards,
Bjorn
drivers/usb/chipidea/core.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/usb/chipidea/core.c b/drivers/usb/chipidea/core.c
index 69426e644d17..6218d83cca25 100644
--- a/drivers/usb/chipidea/core.c
+++ b/drivers/usb/chipidea/core.c
@@ -62,6 +62,7 @@
#include <linux/usb/chipidea.h>
#include <linux/usb/of.h>
#include <linux/of.h>
+#include <linux/of_device.h>
#include <linux/phy.h>
#include <linux/regulator/consumer.h>
#include <linux/usb/ehci_def.h>
@@ -837,6 +838,9 @@ struct platform_device *ci_hdrc_add_device(struct device *dev,
pdev->dev.dma_parms = dev->dma_parms;
dma_set_coherent_mask(&pdev->dev, dev->coherent_dma_mask);
+ if (IS_ENABLED(CONFIG_OF) && dev->of_node)
+ of_dma_configure(&pdev->dev, dev->of_node);
+
ret = platform_device_add_resources(pdev, res, nres);
if (ret)
goto err;
--
2.5.0
On 10/21, Bjorn Andersson wrote:
> hcd_alloc_coherent() and usb_alloc_coherent() ends up allocating coherent
> memory on behalf of ci_hdrc driver. But as the ci_hdrc is instantiated manually
> it will not have any dma_mem or dma_ops assigned, which makes the
> dma_alloc_coherent() fail on some platforms (e.g. arm64). This patch solves
> this by assigning the dma_mem and dma_ops based on the parent's DeviceTree
> node.
>
> Cc: Stephen Boyd <[email protected]>
> Signed-off-by: Bjorn Andersson <[email protected]>
> ---
>
> Hi Peter,
>
> After (once more) debugging why USB doesn't work up on the 64-bit Qualcomm
> systems I realized that we never concluded on this patch. Unfortunately I can't
> find it in my mailbox either, so resending it to restart the discussion.
>
I thought we were going to go down the route that Arnd has been
pushing[1]? That should work, but I haven't tried it yet and
there are some more fixes on top from Sriram. I think Sriram is
taking over the patch now?
[1] https://patchwork.kernel.org/patch/9319527/
--
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project
On Fri 21 Oct 10:38 PDT 2016, Stephen Boyd wrote:
> On 10/21, Bjorn Andersson wrote:
> > hcd_alloc_coherent() and usb_alloc_coherent() ends up allocating coherent
> > memory on behalf of ci_hdrc driver. But as the ci_hdrc is instantiated manually
> > it will not have any dma_mem or dma_ops assigned, which makes the
> > dma_alloc_coherent() fail on some platforms (e.g. arm64). This patch solves
> > this by assigning the dma_mem and dma_ops based on the parent's DeviceTree
> > node.
> >
> > Cc: Stephen Boyd <[email protected]>
> > Signed-off-by: Bjorn Andersson <[email protected]>
> > ---
> >
> > Hi Peter,
> >
> > After (once more) debugging why USB doesn't work up on the 64-bit Qualcomm
> > systems I realized that we never concluded on this patch. Unfortunately I can't
> > find it in my mailbox either, so resending it to restart the discussion.
> >
>
> I thought we were going to go down the route that Arnd has been
> pushing[1]? That should work, but I haven't tried it yet and
> there are some more fixes on top from Sriram. I think Sriram is
> taking over the patch now?
>
> [1] https://patchwork.kernel.org/patch/9319527/
Thanks for the pointer, I've heard about it but couldn't find it.
It does make me further wonder about the multi-device model of these
drivers, but I agree with you that it looks like the patch would
solve our issue.
Regards,
Bjorn
>From: Stephen Boyd [mailto:[email protected]]
>On 10/21, Bjorn Andersson wrote:
>> hcd_alloc_coherent() and usb_alloc_coherent() ends up allocating
>> coherent memory on behalf of ci_hdrc driver. But as the ci_hdrc is
>> instantiated manually it will not have any dma_mem or dma_ops
>> assigned, which makes the
>> dma_alloc_coherent() fail on some platforms (e.g. arm64). This patch
>> solves this by assigning the dma_mem and dma_ops based on the parent's
>> DeviceTree node.
>>
>> Cc: Stephen Boyd <[email protected]>
>> Signed-off-by: Bjorn Andersson <[email protected]>
>> ---
>>
>> Hi Peter,
>>
>> After (once more) debugging why USB doesn't work up on the 64-bit
>> Qualcomm systems I realized that we never concluded on this patch.
>> Unfortunately I can't find it in my mailbox either, so resending it to restart the
>discussion.
>>
>
>I thought we were going to go down the route that Arnd has been pushing[1]? That
>should work, but I haven't tried it yet and there are some more fixes on top from
>Sriram. I think Sriram is taking over the patch now?
>
Yes Stephen. I am incorporating the idea from Arnd and working on those patches.
Regards,
Sriram
>[1] https://patchwork.kernel.org/patch/9319527/
>
>--
>Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux
>Foundation Collaborative Project