Message ID | 5283BC93.1090801@koalo.de (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Hi Florian, Seems like I accidentally replied to some old version of this patch. Not sure how many of those comments were still relevant for V3, but let me look at this version and see whether we can still improve things a bit. Please see my comments inline. On Wednesday 13 of November 2013 18:53:23 Florian Meier wrote: > Add support for DMA controller of BCM2835 as used in the Raspberry Pi. > Currently it only supports cyclic DMA. > > Signed-off-by: Florian Meier <florian.meier@koalo.de> > --- > > Besides of many minor improvements (thanks to your helpful comments), > this fourth version does the assignment of the virtual to the hardware > channel already in probe. This simplifies things a lot > (first of all the complex alloc_chan_resources is now gone). > > Regarding the maximum segment size: 0x3FFFFFFF is the maximum possible > DMA transfer length value and I have no evidence that the maximum > segment size is smaller. > > .../devicetree/bindings/dma/bcm2835-dma.txt | 59 ++ > drivers/dma/Kconfig | 6 + > drivers/dma/Makefile | 1 + > drivers/dma/bcm2835-dma.c | 750 ++++++++++++++++++++ > 4 files changed, 816 insertions(+) > create mode 100644 Documentation/devicetree/bindings/dma/bcm2835-dma.txt > create mode 100644 drivers/dma/bcm2835-dma.c > > diff --git a/Documentation/devicetree/bindings/dma/bcm2835-dma.txt b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt > new file mode 100644 > index 0000000..bca5e84 > --- /dev/null > +++ b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt > @@ -0,0 +1,59 @@ > +* BCM2835 DMA controller > + > +Required properties: > +- compatible: Should be "brcm,bcm2835-dma". > +- reg: Should contain DMA registers location and length. > +- interrupts: Should contain the DMA interrupts associated > + to the DMA channels in ascending order. > + First cell is the IRQ bank. > + Second cell is the IRQ number. > +- #dma-cells: Must be <1>, used to represent the number of integer cells in > + the dmas property of client devices. > +- dma-channels: Maximum number of DMA channels available. > +- dma-requests: Number of DMA Requests. The two properties above do not seem to be used anywhere in the driver. > +- brcm,dma-channel-mask: Bit mask representing the channels available. What does the value of this property depend on? Could you describe the structure of this DMA controller? > + > +Example: > + > +dma: dma@7e007000 { > + compatible = "brcm,bcm2835-dma"; > + reg = <0x7e007000 0xf00>; > + interrupts = <1 16 > + 1 17 > + 1 18 > + 1 19 > + 1 20 > + 1 21 > + 1 22 > + 1 23 > + 1 24 > + 1 25 > + 1 26 > + 1 27 > + 1 28>; There are 13 interrupts specified here, but... > + > + #dma-cells = <1>; > + dma-channels = <16>; ...16 channels here... > + dma-requests = <32>; > + brcm,dma-channel-mask = <0x7f35>; ...and 11 set bits here. May I ask you to explain this to me, please? [snip] > diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c > new file mode 100644 > index 0000000..baf072e > --- /dev/null > +++ b/drivers/dma/bcm2835-dma.c [snip] > +static int bcm2835_dma_probe(struct platform_device *pdev) > +{ > + struct bcm2835_dmadev *od; > + struct resource *dma_res = NULL; > + void __iomem *dma_base = NULL; > + int rc = 0; > + int i = 0; > + int irq; > + uint32_t chans_available; [snip] > + if (pdev->dev.of_node) { Is this driver supposed to support non-DT based instantation (aka board files)? If not, maybe it would be cleaner to simply check for !pdev->dev.of_node at the beginning of probe and return an error? > + const void *chan_mask; > + > + /* Device-tree DMA controller registration */ > + rc = of_dma_controller_register(pdev->dev.of_node, > + bcm2835_dma_xlate, od); > + if (rc) { > + dev_err(&pdev->dev, "Failed to register DMA controller\n"); > + dma_async_device_unregister(&od->ddev); > + bcm2835_dma_free(od); > + return rc; > + } > + > + /* Request DMA channel mask from device tree */ > + chan_mask = of_get_property(pdev->dev.of_node, > + "brcm,dma-channel-mask", NULL); > + > + if (!chan_mask) { > + dev_err(&pdev->dev, "Failed to get channel mask\n"); > + dma_async_device_unregister(&od->ddev); > + bcm2835_dma_free(od); > + return -EINVAL; > + } > + > + chans_available = be32_to_cpup(chan_mask); You can use of_property_read_u32(), which will do both of_get_property() and be32_to_cpup() for you. > + > + /* do not use the FIQ and BULK channels */ > + chans_available &= ~0xD; I'm clearly lacking the information about hardware here, but it would be nice to hear what these channels are. Otherwise, the driver looks pretty good. Best regards, Tomasz -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 13.11.2013 19:43, Tomasz Figa wrote: > Hi Florian, > > Seems like I accidentally replied to some old version of this patch. > Not sure how many of those comments were still relevant for V3, but > let me look at this version and see whether we can still improve things > a bit. Please see my comments inline. Some were still helpful (especially the idea to use the chan_init function for setting the irq), thanks. >> .... >> diff --git a/Documentation/devicetree/bindings/dma/bcm2835-dma.txt b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt >> new file mode 100644 >> index 0000000..bca5e84 >> --- /dev/null >> +++ b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt >> @@ -0,0 +1,59 @@ >> +* BCM2835 DMA controller >> + >> +Required properties: >> +- compatible: Should be "brcm,bcm2835-dma". >> +- reg: Should contain DMA registers location and length. >> +- interrupts: Should contain the DMA interrupts associated >> + to the DMA channels in ascending order. >> + First cell is the IRQ bank. >> + Second cell is the IRQ number. >> +- #dma-cells: Must be <1>, used to represent the number of integer cells in >> + the dmas property of client devices. >> +- dma-channels: Maximum number of DMA channels available. >> +- dma-requests: Number of DMA Requests. > > The two properties above do not seem to be used anywhere in the driver. Aren't they necessary for the DMA engine core? >> +- brcm,dma-channel-mask: Bit mask representing the channels available. > > What does the value of this property depend on? Could you describe the > structure of this DMA controller? > >> + >> +Example: >> + >> +dma: dma@7e007000 { >> + compatible = "brcm,bcm2835-dma"; >> + reg = <0x7e007000 0xf00>; >> + interrupts = <1 16 >> + 1 17 >> + 1 18 >> + 1 19 >> + 1 20 >> + 1 21 >> + 1 22 >> + 1 23 >> + 1 24 >> + 1 25 >> + 1 26 >> + 1 27 >> + 1 28>; > > There are 13 interrupts specified here, but... > >> + >> + #dma-cells = <1>; >> + dma-channels = <16>; > > ...16 channels here... > >> + dma-requests = <32>; >> + brcm,dma-channel-mask = <0x7f35>; > > ...and 11 set bits here. May I ask you to explain this to me, please? How I understand this DMA controller: There are 16 DMA channels in the DMA controller, but only 13 interrupts are available at the IRQ controller. Therefore, the upper DMA channels can just not be used. Maybe because there are to many other IRQs and they didn't want to implement another IRQ bank. Furthermore, some of the DMA channels are already used by the VideoCore/GPU/firmware. This is what dma-channel-mask indicates. This should be automatically set by the firmware in the future. Finally, there are some channels with special functionality that should not be used by DMA engine, too. Therefore, these lines: /* do not use the FIQ and BULK channels */ chans_available &= ~0xD; > [snip] >> diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c >> new file mode 100644 >> index 0000000..baf072e >> --- /dev/null >> +++ b/drivers/dma/bcm2835-dma.c > [snip] >> +static int bcm2835_dma_probe(struct platform_device *pdev) >> +{ >> + struct bcm2835_dmadev *od; >> + struct resource *dma_res = NULL; >> + void __iomem *dma_base = NULL; >> + int rc = 0; >> + int i = 0; >> + int irq; >> + uint32_t chans_available; > [snip] >> + if (pdev->dev.of_node) { > > Is this driver supposed to support non-DT based instantation (aka board > files)? If not, maybe it would be cleaner to simply check for > !pdev->dev.of_node at the beginning of probe and return an error? I would like to maintain the possibility for board file based instatiation, because the Raspberry Pi downstream kernel still doesn't support device tree. If this is a no-go, I will accept that. Greetings, Florian -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wednesday 13 of November 2013 20:35:22 Florian Meier wrote: > On 13.11.2013 19:43, Tomasz Figa wrote: > > Hi Florian, > > > > Seems like I accidentally replied to some old version of this patch. > > Not sure how many of those comments were still relevant for V3, but > > let me look at this version and see whether we can still improve things > > a bit. Please see my comments inline. > > Some were still helpful (especially the idea to use the chan_init > function for setting the irq), thanks. Good. You're welcome. > > >> .... > >> diff --git a/Documentation/devicetree/bindings/dma/bcm2835-dma.txt b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt > >> new file mode 100644 > >> index 0000000..bca5e84 > >> --- /dev/null > >> +++ b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt > >> @@ -0,0 +1,59 @@ > >> +* BCM2835 DMA controller > >> + > >> +Required properties: > >> +- compatible: Should be "brcm,bcm2835-dma". > >> +- reg: Should contain DMA registers location and length. > >> +- interrupts: Should contain the DMA interrupts associated > >> + to the DMA channels in ascending order. > >> + First cell is the IRQ bank. > >> + Second cell is the IRQ number. > >> +- #dma-cells: Must be <1>, used to represent the number of integer cells in > >> + the dmas property of client devices. > >> +- dma-channels: Maximum number of DMA channels available. > >> +- dma-requests: Number of DMA Requests. > > > > The two properties above do not seem to be used anywhere in the driver. > > Aren't they necessary for the DMA engine core? Honestly, looking at the DMA bindings documentation, these two are quite a mystery for me. They don't seem to be used in DMA engine core, but instead several drivers use them for private purposes. So, since they are optional and you don't seem to need them, I wouldn't list them. > > >> +- brcm,dma-channel-mask: Bit mask representing the channels available. > > > > What does the value of this property depend on? Could you describe the > > structure of this DMA controller? > > > >> + > >> +Example: > >> + > >> +dma: dma@7e007000 { > >> + compatible = "brcm,bcm2835-dma"; > >> + reg = <0x7e007000 0xf00>; > >> + interrupts = <1 16 > >> + 1 17 > >> + 1 18 > >> + 1 19 > >> + 1 20 > >> + 1 21 > >> + 1 22 > >> + 1 23 > >> + 1 24 > >> + 1 25 > >> + 1 26 > >> + 1 27 > >> + 1 28>; > > > > There are 13 interrupts specified here, but... > > > >> + > >> + #dma-cells = <1>; > >> + dma-channels = <16>; > > > > ...16 channels here... > > > >> + dma-requests = <32>; > >> + brcm,dma-channel-mask = <0x7f35>; > > > > ...and 11 set bits here. May I ask you to explain this to me, please? > > How I understand this DMA controller: > There are 16 DMA channels in the DMA controller, but only 13 interrupts > are available at the IRQ controller. Therefore, the upper DMA channels > can just not be used. Maybe because there are to many other IRQs and > they didn't want to implement another IRQ bank. > Furthermore, some of the DMA channels are already used by the > VideoCore/GPU/firmware. This is what dma-channel-mask indicates. This > should be automatically set by the firmware in the future. > Finally, there are some channels with special functionality that should > not be used by DMA engine, too. Therefore, these lines: > /* do not use the FIQ and BULK channels */ > chans_available &= ~0xD; OK, this makes it much more clear. So, my only comment remaining here is that you shouldn't include the channels without interrupt signal in the mask. This would allow you to define it as a mask of channels that are operable and then just iterate over all set bits in the driver, instead of using tricks with interrupt resources. What do you think? > > > [snip] > >> diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c > >> new file mode 100644 > >> index 0000000..baf072e > >> --- /dev/null > >> +++ b/drivers/dma/bcm2835-dma.c > > [snip] > >> +static int bcm2835_dma_probe(struct platform_device *pdev) > >> +{ > >> + struct bcm2835_dmadev *od; > >> + struct resource *dma_res = NULL; > >> + void __iomem *dma_base = NULL; > >> + int rc = 0; > >> + int i = 0; > >> + int irq; > >> + uint32_t chans_available; > > [snip] > >> + if (pdev->dev.of_node) { > > > > Is this driver supposed to support non-DT based instantation (aka board > > files)? If not, maybe it would be cleaner to simply check for > > !pdev->dev.of_node at the beginning of probe and return an error? > > I would like to maintain the possibility for board file based > instatiation, because the Raspberry Pi downstream kernel still doesn't > support device tree. If this is a no-go, I will accept that. Sure, you are free to do so. What I meant is that your probe won't call bcm2835_dma_chan_init() at all if there is no pdev->dev.of_node, because the loop iterating over channels is under the if clause. Best regards, Tomasz -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 13.11.2013 21:39, Tomasz Figa wrote: > On Wednesday 13 of November 2013 20:35:22 Florian Meier wrote: >> On 13.11.2013 19:43, Tomasz Figa wrote: >>>> .... >>>> diff --git a/Documentation/devicetree/bindings/dma/bcm2835-dma.txt b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt >>>> new file mode 100644 >>>> index 0000000..bca5e84 >>>> --- /dev/null >>>> +++ b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt >>>> @@ -0,0 +1,59 @@ >>>> +* BCM2835 DMA controller >>>> + >>>> +Required properties: >>>> +- compatible: Should be "brcm,bcm2835-dma". >>>> +- reg: Should contain DMA registers location and length. >>>> +- interrupts: Should contain the DMA interrupts associated >>>> + to the DMA channels in ascending order. >>>> + First cell is the IRQ bank. >>>> + Second cell is the IRQ number. >>>> +- #dma-cells: Must be <1>, used to represent the number of integer cells in >>>> + the dmas property of client devices. >>>> +- dma-channels: Maximum number of DMA channels available. >>>> +- dma-requests: Number of DMA Requests. >>> >>> The two properties above do not seem to be used anywhere in the driver. >> >> Aren't they necessary for the DMA engine core? > > Honestly, looking at the DMA bindings documentation, these two are quite > a mystery for me. They don't seem to be used in DMA engine core, but > instead several drivers use them for private purposes. > > So, since they are optional and you don't seem to need them, I wouldn't > list them. Ok >>>> +- brcm,dma-channel-mask: Bit mask representing the channels available. >>> >>> What does the value of this property depend on? Could you describe the >>> structure of this DMA controller? >>> >>>> + >>>> +Example: >>>> + >>>> +dma: dma@7e007000 { >>>> + compatible = "brcm,bcm2835-dma"; >>>> + reg = <0x7e007000 0xf00>; >>>> + interrupts = <1 16 >>>> + 1 17 >>>> + 1 18 >>>> + 1 19 >>>> + 1 20 >>>> + 1 21 >>>> + 1 22 >>>> + 1 23 >>>> + 1 24 >>>> + 1 25 >>>> + 1 26 >>>> + 1 27 >>>> + 1 28>; >>> >>> There are 13 interrupts specified here, but... >>> >>>> + >>>> + #dma-cells = <1>; >>>> + dma-channels = <16>; >>> >>> ...16 channels here... >>> >>>> + dma-requests = <32>; >>>> + brcm,dma-channel-mask = <0x7f35>; >>> >>> ...and 11 set bits here. May I ask you to explain this to me, please? >> >> How I understand this DMA controller: >> There are 16 DMA channels in the DMA controller, but only 13 interrupts >> are available at the IRQ controller. Therefore, the upper DMA channels >> can just not be used. Maybe because there are to many other IRQs and >> they didn't want to implement another IRQ bank. >> Furthermore, some of the DMA channels are already used by the >> VideoCore/GPU/firmware. This is what dma-channel-mask indicates. This >> should be automatically set by the firmware in the future. >> Finally, there are some channels with special functionality that should >> not be used by DMA engine, too. Therefore, these lines: >> /* do not use the FIQ and BULK channels */ >> chans_available &= ~0xD; > > OK, this makes it much more clear. > > So, my only comment remaining here is that you shouldn't include the > channels without interrupt signal in the mask. This would allow you > to define it as a mask of channels that are operable and then just > iterate over all set bits in the driver, instead of using tricks with > interrupt resources. What do you think? Since the mask will come directly from the firmware, this would require patching the firmware. I think that is not worth the effort. >>> [snip] >>>> diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c >>>> new file mode 100644 >>>> index 0000000..baf072e >>>> --- /dev/null >>>> +++ b/drivers/dma/bcm2835-dma.c >>> [snip] >>>> +static int bcm2835_dma_probe(struct platform_device *pdev) >>>> +{ >>>> + struct bcm2835_dmadev *od; >>>> + struct resource *dma_res = NULL; >>>> + void __iomem *dma_base = NULL; >>>> + int rc = 0; >>>> + int i = 0; >>>> + int irq; >>>> + uint32_t chans_available; >>> [snip] >>>> + if (pdev->dev.of_node) { >>> >>> Is this driver supposed to support non-DT based instantation (aka board >>> files)? If not, maybe it would be cleaner to simply check for >>> !pdev->dev.of_node at the beginning of probe and return an error? >> >> I would like to maintain the possibility for board file based >> instatiation, because the Raspberry Pi downstream kernel still doesn't >> support device tree. If this is a no-go, I will accept that. > > Sure, you are free to do so. > > What I meant is that your probe won't call bcm2835_dma_chan_init() at all > if there is no pdev->dev.of_node, because the loop iterating over channels > is under the if clause. Yes you are right, but I think it will make the patching easier, later. Currently, nothing bad happens without device tree - it just allocates no channels. Greetings, Florian -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thursday 14 of November 2013 08:12:46 Florian Meier wrote: > On 13.11.2013 21:39, Tomasz Figa wrote: > > On Wednesday 13 of November 2013 20:35:22 Florian Meier wrote: > >> On 13.11.2013 19:43, Tomasz Figa wrote: > >>>> .... > >>>> diff --git a/Documentation/devicetree/bindings/dma/bcm2835-dma.txt b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt > >>>> new file mode 100644 > >>>> index 0000000..bca5e84 > >>>> --- /dev/null > >>>> +++ b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt > >>>> @@ -0,0 +1,59 @@ > >>>> +* BCM2835 DMA controller > >>>> + > >>>> +Required properties: > >>>> +- compatible: Should be "brcm,bcm2835-dma". > >>>> +- reg: Should contain DMA registers location and length. > >>>> +- interrupts: Should contain the DMA interrupts associated > >>>> + to the DMA channels in ascending order. > >>>> + First cell is the IRQ bank. > >>>> + Second cell is the IRQ number. > >>>> +- #dma-cells: Must be <1>, used to represent the number of integer cells in > >>>> + the dmas property of client devices. > >>>> +- dma-channels: Maximum number of DMA channels available. > >>>> +- dma-requests: Number of DMA Requests. > >>> > >>> The two properties above do not seem to be used anywhere in the driver. > >> > >> Aren't they necessary for the DMA engine core? > > > > Honestly, looking at the DMA bindings documentation, these two are quite > > a mystery for me. They don't seem to be used in DMA engine core, but > > instead several drivers use them for private purposes. > > > > So, since they are optional and you don't seem to need them, I wouldn't > > list them. > > Ok > > >>>> +- brcm,dma-channel-mask: Bit mask representing the channels available. > >>> > >>> What does the value of this property depend on? Could you describe the > >>> structure of this DMA controller? > >>> > >>>> + > >>>> +Example: > >>>> + > >>>> +dma: dma@7e007000 { > >>>> + compatible = "brcm,bcm2835-dma"; > >>>> + reg = <0x7e007000 0xf00>; > >>>> + interrupts = <1 16 > >>>> + 1 17 > >>>> + 1 18 > >>>> + 1 19 > >>>> + 1 20 > >>>> + 1 21 > >>>> + 1 22 > >>>> + 1 23 > >>>> + 1 24 > >>>> + 1 25 > >>>> + 1 26 > >>>> + 1 27 > >>>> + 1 28>; > >>> > >>> There are 13 interrupts specified here, but... > >>> > >>>> + > >>>> + #dma-cells = <1>; > >>>> + dma-channels = <16>; > >>> > >>> ...16 channels here... > >>> > >>>> + dma-requests = <32>; > >>>> + brcm,dma-channel-mask = <0x7f35>; > >>> > >>> ...and 11 set bits here. May I ask you to explain this to me, please? > >> > >> How I understand this DMA controller: > >> There are 16 DMA channels in the DMA controller, but only 13 interrupts > >> are available at the IRQ controller. Therefore, the upper DMA channels > >> can just not be used. Maybe because there are to many other IRQs and > >> they didn't want to implement another IRQ bank. > >> Furthermore, some of the DMA channels are already used by the > >> VideoCore/GPU/firmware. This is what dma-channel-mask indicates. This > >> should be automatically set by the firmware in the future. > >> Finally, there are some channels with special functionality that should > >> not be used by DMA engine, too. Therefore, these lines: > >> /* do not use the FIQ and BULK channels */ > >> chans_available &= ~0xD; > > > > OK, this makes it much more clear. > > > > So, my only comment remaining here is that you shouldn't include the > > channels without interrupt signal in the mask. This would allow you > > to define it as a mask of channels that are operable and then just > > iterate over all set bits in the driver, instead of using tricks with > > interrupt resources. What do you think? > > Since the mask will come directly from the firmware, this would require > patching the firmware. I think that is not worth the effort. Now I'm slightly confused. Do you already have code in your firmware that adds this property to your device tree? Otherwise in what circumstances such patching would take place? On given hardware (unless it's an FPGA) the configuration of available DMA channels that have interrupt signals should not change. > > >>> [snip] > >>>> diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c > >>>> new file mode 100644 > >>>> index 0000000..baf072e > >>>> --- /dev/null > >>>> +++ b/drivers/dma/bcm2835-dma.c > >>> [snip] > >>>> +static int bcm2835_dma_probe(struct platform_device *pdev) > >>>> +{ > >>>> + struct bcm2835_dmadev *od; > >>>> + struct resource *dma_res = NULL; > >>>> + void __iomem *dma_base = NULL; > >>>> + int rc = 0; > >>>> + int i = 0; > >>>> + int irq; > >>>> + uint32_t chans_available; > >>> [snip] > >>>> + if (pdev->dev.of_node) { > >>> > >>> Is this driver supposed to support non-DT based instantation (aka board > >>> files)? If not, maybe it would be cleaner to simply check for > >>> !pdev->dev.of_node at the beginning of probe and return an error? > >> > >> I would like to maintain the possibility for board file based > >> instatiation, because the Raspberry Pi downstream kernel still doesn't > >> support device tree. If this is a no-go, I will accept that. > > > > Sure, you are free to do so. > > > > What I meant is that your probe won't call bcm2835_dma_chan_init() at all > > if there is no pdev->dev.of_node, because the loop iterating over channels > > is under the if clause. > > Yes you are right, but I think it will make the patching easier, later. > Currently, nothing bad happens without device tree - it just allocates > no channels. But isn't it really an error condition, if no channels are allocated? Anyway, back to my point about leaving non-DT support in a driver, the point is still valid only for drivers, not for platforms/boards. So if there are no boards supported using board files in mainline that could benefit from this driver, then this driver can be safely made DT-only, because no new non-DT platforms/boards can be added. Best regards, Tomasz -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
2013/11/14 Tomasz Figa <tomasz.figa@gmail.com>: > On Thursday 14 of November 2013 08:12:46 Florian Meier wrote: >> On 13.11.2013 21:39, Tomasz Figa wrote: >> > On Wednesday 13 of November 2013 20:35:22 Florian Meier wrote: >> >>>> +- brcm,dma-channel-mask: Bit mask representing the channels available. >> >>> >> >>> What does the value of this property depend on? Could you describe the >> >>> structure of this DMA controller? >> >>> >> >>>> + >> >>>> +Example: >> >>>> + >> >>>> +dma: dma@7e007000 { >> >>>> + compatible = "brcm,bcm2835-dma"; >> >>>> + reg = <0x7e007000 0xf00>; >> >>>> + interrupts = <1 16 >> >>>> + 1 17 >> >>>> + 1 18 >> >>>> + 1 19 >> >>>> + 1 20 >> >>>> + 1 21 >> >>>> + 1 22 >> >>>> + 1 23 >> >>>> + 1 24 >> >>>> + 1 25 >> >>>> + 1 26 >> >>>> + 1 27 >> >>>> + 1 28>; >> >>> >> >>> There are 13 interrupts specified here, but... >> >>> >> >>>> + >> >>>> + #dma-cells = <1>; >> >>>> + dma-channels = <16>; >> >>> >> >>> ...16 channels here... >> >>> >> >>>> + dma-requests = <32>; >> >>>> + brcm,dma-channel-mask = <0x7f35>; >> >>> >> >>> ...and 11 set bits here. May I ask you to explain this to me, please? >> >> >> >> How I understand this DMA controller: >> >> There are 16 DMA channels in the DMA controller, but only 13 interrupts >> >> are available at the IRQ controller. Therefore, the upper DMA channels >> >> can just not be used. Maybe because there are to many other IRQs and >> >> they didn't want to implement another IRQ bank. >> >> Furthermore, some of the DMA channels are already used by the >> >> VideoCore/GPU/firmware. This is what dma-channel-mask indicates. This >> >> should be automatically set by the firmware in the future. >> >> Finally, there are some channels with special functionality that should >> >> not be used by DMA engine, too. Therefore, these lines: >> >> /* do not use the FIQ and BULK channels */ >> >> chans_available &= ~0xD; >> > >> > OK, this makes it much more clear. >> > >> > So, my only comment remaining here is that you shouldn't include the >> > channels without interrupt signal in the mask. This would allow you >> > to define it as a mask of channels that are operable and then just >> > iterate over all set bits in the driver, instead of using tricks with >> > interrupt resources. What do you think? >> >> Since the mask will come directly from the firmware, this would require >> patching the firmware. I think that is not worth the effort. > > Now I'm slightly confused. Do you already have code in your firmware that > adds this property to your device tree? > > Otherwise in what circumstances such patching would take place? On given > hardware (unless it's an FPGA) the configuration of available DMA channels > that have interrupt signals should not change. It is very confusing. I agree. There is already a DMA driver with a proprietary API in the downstream kernel. The firmware already creates this mask and passes it to this proprietary driver. There was already a discussion about this in the first version thread that (as long as I understand it) resulted in "we should pass this mask on to the driver via device tree". So I did that. I have no idea about how this firmware->devicetree interface will take place, but since I didn't want to run in circles I hardcoded it in the device tree. >> >>> [snip] >> >>>> diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c >> >>>> new file mode 100644 >> >>>> index 0000000..baf072e >> >>>> --- /dev/null >> >>>> +++ b/drivers/dma/bcm2835-dma.c >> >>> [snip] >> >>>> +static int bcm2835_dma_probe(struct platform_device *pdev) >> >>>> +{ >> >>>> + struct bcm2835_dmadev *od; >> >>>> + struct resource *dma_res = NULL; >> >>>> + void __iomem *dma_base = NULL; >> >>>> + int rc = 0; >> >>>> + int i = 0; >> >>>> + int irq; >> >>>> + uint32_t chans_available; >> >>> [snip] >> >>>> + if (pdev->dev.of_node) { >> >>> >> >>> Is this driver supposed to support non-DT based instantation (aka board >> >>> files)? If not, maybe it would be cleaner to simply check for >> >>> !pdev->dev.of_node at the beginning of probe and return an error? >> >> >> >> I would like to maintain the possibility for board file based >> >> instatiation, because the Raspberry Pi downstream kernel still doesn't >> >> support device tree. If this is a no-go, I will accept that. >> > >> > Sure, you are free to do so. >> > >> > What I meant is that your probe won't call bcm2835_dma_chan_init() at all >> > if there is no pdev->dev.of_node, because the loop iterating over channels >> > is under the if clause. >> >> Yes you are right, but I think it will make the patching easier, later. >> Currently, nothing bad happens without device tree - it just allocates >> no channels. > > But isn't it really an error condition, if no channels are allocated? A fridge is still a working fridge, even if no beer is inside ;-) Ok, bad example, but you will get an error message anyway when you try to get a channel. > Anyway, back to my point about leaving non-DT support in a driver, the > point is still valid only for drivers, not for platforms/boards. So if > there are no boards supported using board files in mainline that could > benefit from this driver, then this driver can be safely made DT-only, > because no new non-DT platforms/boards can be added. I don't have a telling argument against this, but just thought writing it this way will make the migration of the downstream kernel to upstream easier, but if you say I should change it, I will of course do that. I am becoming desperate anyway that this migration will ever fully take place.... Greetings, Florian -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thursday 14 of November 2013 15:44:05 Florian Meier wrote: > 2013/11/14 Tomasz Figa <tomasz.figa@gmail.com>: > > On Thursday 14 of November 2013 08:12:46 Florian Meier wrote: > >> On 13.11.2013 21:39, Tomasz Figa wrote: > >> > On Wednesday 13 of November 2013 20:35:22 Florian Meier wrote: > >> >>>> +- brcm,dma-channel-mask: Bit mask representing the channels available. > >> >>> > >> >>> What does the value of this property depend on? Could you describe the > >> >>> structure of this DMA controller? > >> >>> > >> >>>> + > >> >>>> +Example: > >> >>>> + > >> >>>> +dma: dma@7e007000 { > >> >>>> + compatible = "brcm,bcm2835-dma"; > >> >>>> + reg = <0x7e007000 0xf00>; > >> >>>> + interrupts = <1 16 > >> >>>> + 1 17 > >> >>>> + 1 18 > >> >>>> + 1 19 > >> >>>> + 1 20 > >> >>>> + 1 21 > >> >>>> + 1 22 > >> >>>> + 1 23 > >> >>>> + 1 24 > >> >>>> + 1 25 > >> >>>> + 1 26 > >> >>>> + 1 27 > >> >>>> + 1 28>; > >> >>> > >> >>> There are 13 interrupts specified here, but... > >> >>> > >> >>>> + > >> >>>> + #dma-cells = <1>; > >> >>>> + dma-channels = <16>; > >> >>> > >> >>> ...16 channels here... > >> >>> > >> >>>> + dma-requests = <32>; > >> >>>> + brcm,dma-channel-mask = <0x7f35>; > >> >>> > >> >>> ...and 11 set bits here. May I ask you to explain this to me, please? > >> >> > >> >> How I understand this DMA controller: > >> >> There are 16 DMA channels in the DMA controller, but only 13 interrupts > >> >> are available at the IRQ controller. Therefore, the upper DMA channels > >> >> can just not be used. Maybe because there are to many other IRQs and > >> >> they didn't want to implement another IRQ bank. > >> >> Furthermore, some of the DMA channels are already used by the > >> >> VideoCore/GPU/firmware. This is what dma-channel-mask indicates. This > >> >> should be automatically set by the firmware in the future. > >> >> Finally, there are some channels with special functionality that should > >> >> not be used by DMA engine, too. Therefore, these lines: > >> >> /* do not use the FIQ and BULK channels */ > >> >> chans_available &= ~0xD; > >> > > >> > OK, this makes it much more clear. > >> > > >> > So, my only comment remaining here is that you shouldn't include the > >> > channels without interrupt signal in the mask. This would allow you > >> > to define it as a mask of channels that are operable and then just > >> > iterate over all set bits in the driver, instead of using tricks with > >> > interrupt resources. What do you think? > >> > >> Since the mask will come directly from the firmware, this would require > >> patching the firmware. I think that is not worth the effort. > > > > Now I'm slightly confused. Do you already have code in your firmware that > > adds this property to your device tree? > > > > Otherwise in what circumstances such patching would take place? On given > > hardware (unless it's an FPGA) the configuration of available DMA channels > > that have interrupt signals should not change. > > It is very confusing. I agree. > There is already a DMA driver with a proprietary API in the downstream > kernel. The firmware already creates this mask and passes it to this > proprietary driver. > There was already a discussion about this in the first version thread > that (as long as I understand it) resulted in "we should pass this > mask on to the driver via device tree". So I did that. I have no idea > about how this firmware->devicetree interface will take place, but > since I didn't want to run in circles I hardcoded it in the device > tree. > OK. So the firmware defines what set and clear bits in the mask mean. It's fine then. > > >> >>> [snip] > >> >>>> diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c > >> >>>> new file mode 100644 > >> >>>> index 0000000..baf072e > >> >>>> --- /dev/null > >> >>>> +++ b/drivers/dma/bcm2835-dma.c > >> >>> [snip] > >> >>>> +static int bcm2835_dma_probe(struct platform_device *pdev) > >> >>>> +{ > >> >>>> + struct bcm2835_dmadev *od; > >> >>>> + struct resource *dma_res = NULL; > >> >>>> + void __iomem *dma_base = NULL; > >> >>>> + int rc = 0; > >> >>>> + int i = 0; > >> >>>> + int irq; > >> >>>> + uint32_t chans_available; > >> >>> [snip] > >> >>>> + if (pdev->dev.of_node) { > >> >>> > >> >>> Is this driver supposed to support non-DT based instantation (aka board > >> >>> files)? If not, maybe it would be cleaner to simply check for > >> >>> !pdev->dev.of_node at the beginning of probe and return an error? > >> >> > >> >> I would like to maintain the possibility for board file based > >> >> instatiation, because the Raspberry Pi downstream kernel still doesn't > >> >> support device tree. If this is a no-go, I will accept that. > >> > > >> > Sure, you are free to do so. > >> > > >> > What I meant is that your probe won't call bcm2835_dma_chan_init() at all > >> > if there is no pdev->dev.of_node, because the loop iterating over channels > >> > is under the if clause. > >> > >> Yes you are right, but I think it will make the patching easier, later. > >> Currently, nothing bad happens without device tree - it just allocates > >> no channels. > > > > But isn't it really an error condition, if no channels are allocated? > > A fridge is still a working fridge, even if no beer is inside ;-) > Ok, bad example, but you will get an error message anyway when you try > to get a channel. > > > Anyway, back to my point about leaving non-DT support in a driver, the > > point is still valid only for drivers, not for platforms/boards. So if > > there are no boards supported using board files in mainline that could > > benefit from this driver, then this driver can be safely made DT-only, > > because no new non-DT platforms/boards can be added. > > I don't have a telling argument against this, but just thought writing > it this way will > make the migration of the downstream kernel to upstream easier, but if you say I > should change it, I will of course do that. I'm just presenting you the possible options. You are still free to have non-DT support in the driver, but if you don't need it (because you can't have any new non-DT platforms in mainline) then you can simplify some code. However the driver shouldn't be left with illusionary support for non-DT platforms until you decide to implement that. Instead, if you don't want to add non-DT support now, just make the driver DT-only, while keeping its design in a way allowing you to add non-DT support in future. In other words, a driver should not be able to probe using board files if support for such probing method is not available in it yet. > I am becoming desperate anyway that this migration will ever fully > take place.... Why not? It's just a matter of people like you working on this (and addressing some review comments ;)). Best regards, Tomasz -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
2013/11/14 Tomasz Figa <tomasz.figa@gmail.com>: > On Thursday 14 of November 2013 15:44:05 Florian Meier wrote: >> 2013/11/14 Tomasz Figa <tomasz.figa@gmail.com>: >> > On Thursday 14 of November 2013 08:12:46 Florian Meier wrote: >> >> On 13.11.2013 21:39, Tomasz Figa wrote: >> >> > On Wednesday 13 of November 2013 20:35:22 Florian Meier wrote: >> >> >>>> +- brcm,dma-channel-mask: Bit mask representing the channels available. >> >> >>> >> >> >>> What does the value of this property depend on? Could you describe the >> >> >>> structure of this DMA controller? >> >> >>> >> >> >>>> + >> >> >>>> +Example: >> >> >>>> + >> >> >>>> +dma: dma@7e007000 { >> >> >>>> + compatible = "brcm,bcm2835-dma"; >> >> >>>> + reg = <0x7e007000 0xf00>; >> >> >>>> + interrupts = <1 16 >> >> >>>> + 1 17 >> >> >>>> + 1 18 >> >> >>>> + 1 19 >> >> >>>> + 1 20 >> >> >>>> + 1 21 >> >> >>>> + 1 22 >> >> >>>> + 1 23 >> >> >>>> + 1 24 >> >> >>>> + 1 25 >> >> >>>> + 1 26 >> >> >>>> + 1 27 >> >> >>>> + 1 28>; >> >> >>> >> >> >>> There are 13 interrupts specified here, but... >> >> >>> >> >> >>>> + >> >> >>>> + #dma-cells = <1>; >> >> >>>> + dma-channels = <16>; >> >> >>> >> >> >>> ...16 channels here... >> >> >>> >> >> >>>> + dma-requests = <32>; >> >> >>>> + brcm,dma-channel-mask = <0x7f35>; >> >> >>> >> >> >>> ...and 11 set bits here. May I ask you to explain this to me, please? >> >> >> >> >> >> How I understand this DMA controller: >> >> >> There are 16 DMA channels in the DMA controller, but only 13 interrupts >> >> >> are available at the IRQ controller. Therefore, the upper DMA channels >> >> >> can just not be used. Maybe because there are to many other IRQs and >> >> >> they didn't want to implement another IRQ bank. >> >> >> Furthermore, some of the DMA channels are already used by the >> >> >> VideoCore/GPU/firmware. This is what dma-channel-mask indicates. This >> >> >> should be automatically set by the firmware in the future. >> >> >> Finally, there are some channels with special functionality that should >> >> >> not be used by DMA engine, too. Therefore, these lines: >> >> >> /* do not use the FIQ and BULK channels */ >> >> >> chans_available &= ~0xD; >> >> > >> >> > OK, this makes it much more clear. >> >> > >> >> > So, my only comment remaining here is that you shouldn't include the >> >> > channels without interrupt signal in the mask. This would allow you >> >> > to define it as a mask of channels that are operable and then just >> >> > iterate over all set bits in the driver, instead of using tricks with >> >> > interrupt resources. What do you think? >> >> >> >> Since the mask will come directly from the firmware, this would require >> >> patching the firmware. I think that is not worth the effort. >> > >> > Now I'm slightly confused. Do you already have code in your firmware that >> > adds this property to your device tree? >> > >> > Otherwise in what circumstances such patching would take place? On given >> > hardware (unless it's an FPGA) the configuration of available DMA channels >> > that have interrupt signals should not change. >> >> It is very confusing. I agree. >> There is already a DMA driver with a proprietary API in the downstream >> kernel. The firmware already creates this mask and passes it to this >> proprietary driver. >> There was already a discussion about this in the first version thread >> that (as long as I understand it) resulted in "we should pass this >> mask on to the driver via device tree". So I did that. I have no idea >> about how this firmware->devicetree interface will take place, but >> since I didn't want to run in circles I hardcoded it in the device >> tree. >> > > OK. So the firmware defines what set and clear bits in the mask mean. > It's fine then. Yes, this mask is actually an "I am not using this channel"-mask. >> >> >>> [snip] >> >> >>>> diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c >> >> >>>> new file mode 100644 >> >> >>>> index 0000000..baf072e >> >> >>>> --- /dev/null >> >> >>>> +++ b/drivers/dma/bcm2835-dma.c >> >> >>> [snip] >> >> >>>> +static int bcm2835_dma_probe(struct platform_device *pdev) >> >> >>>> +{ >> >> >>>> + struct bcm2835_dmadev *od; >> >> >>>> + struct resource *dma_res = NULL; >> >> >>>> + void __iomem *dma_base = NULL; >> >> >>>> + int rc = 0; >> >> >>>> + int i = 0; >> >> >>>> + int irq; >> >> >>>> + uint32_t chans_available; >> >> >>> [snip] >> >> >>>> + if (pdev->dev.of_node) { >> >> >>> >> >> >>> Is this driver supposed to support non-DT based instantation (aka board >> >> >>> files)? If not, maybe it would be cleaner to simply check for >> >> >>> !pdev->dev.of_node at the beginning of probe and return an error? >> >> >> >> >> >> I would like to maintain the possibility for board file based >> >> >> instatiation, because the Raspberry Pi downstream kernel still doesn't >> >> >> support device tree. If this is a no-go, I will accept that. >> >> > >> >> > Sure, you are free to do so. >> >> > >> >> > What I meant is that your probe won't call bcm2835_dma_chan_init() at all >> >> > if there is no pdev->dev.of_node, because the loop iterating over channels >> >> > is under the if clause. >> >> >> >> Yes you are right, but I think it will make the patching easier, later. >> >> Currently, nothing bad happens without device tree - it just allocates >> >> no channels. >> > >> > But isn't it really an error condition, if no channels are allocated? >> >> A fridge is still a working fridge, even if no beer is inside ;-) >> Ok, bad example, but you will get an error message anyway when you try >> to get a channel. >> >> > Anyway, back to my point about leaving non-DT support in a driver, the >> > point is still valid only for drivers, not for platforms/boards. So if >> > there are no boards supported using board files in mainline that could >> > benefit from this driver, then this driver can be safely made DT-only, >> > because no new non-DT platforms/boards can be added. >> >> I don't have a telling argument against this, but just thought writing >> it this way will >> make the migration of the downstream kernel to upstream easier, but if you say I >> should change it, I will of course do that. > > I'm just presenting you the possible options. You are still free to have > non-DT support in the driver, but if you don't need it (because you can't > have any new non-DT platforms in mainline) then you can simplify some > code. > > However the driver shouldn't be left with illusionary support for non-DT > platforms until you decide to implement that. Instead, if you don't want > to add non-DT support now, just make the driver DT-only, while keeping > its design in a way allowing you to add non-DT support in future. This statement hits the nail on the head. Thank you! > In other words, a driver should not be able to probe using board files > if support for such probing method is not available in it yet. That is meaningful. So would the following be ok? + ..... + if (pdev->dev.of_node) { + ..... + } else { + dev_err(&pdev->dev, + "Failed to initialize channels, because device tree not available: %d\n", rc); + bcm2835_dma_free(od); + return rc; + } + + dev_dbg(&pdev->dev, "Initialized %i DMA channels\n", i); + + ..... >> I am becoming desperate anyway that this migration will ever fully >> take place.... > > Why not? It's just a matter of people like you working on this (and > addressing some review comments ;)). The most common comment about this is that people will not put effort in the upstream kernel as long as there is no comfortable way for debugging in the upstream kernel (i.e. at least USB support)....... Greetings, Florian -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thursday 14 of November 2013 16:41:56 Florian Meier wrote: > 2013/11/14 Tomasz Figa <tomasz.figa@gmail.com>: > > On Thursday 14 of November 2013 15:44:05 Florian Meier wrote: > >> 2013/11/14 Tomasz Figa <tomasz.figa@gmail.com>: > >> > On Thursday 14 of November 2013 08:12:46 Florian Meier wrote: > >> >> On 13.11.2013 21:39, Tomasz Figa wrote: > >> >> > On Wednesday 13 of November 2013 20:35:22 Florian Meier wrote: > >> >> >>>> diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c > >> >> >>>> new file mode 100644 > >> >> >>>> index 0000000..baf072e > >> >> >>>> --- /dev/null > >> >> >>>> +++ b/drivers/dma/bcm2835-dma.c > >> >> >>> [snip] > >> >> >>>> +static int bcm2835_dma_probe(struct platform_device *pdev) > >> >> >>>> +{ > >> >> >>>> + struct bcm2835_dmadev *od; > >> >> >>>> + struct resource *dma_res = NULL; > >> >> >>>> + void __iomem *dma_base = NULL; > >> >> >>>> + int rc = 0; > >> >> >>>> + int i = 0; > >> >> >>>> + int irq; > >> >> >>>> + uint32_t chans_available; > >> >> >>> [snip] > >> >> >>>> + if (pdev->dev.of_node) { > >> >> >>> > >> >> >>> Is this driver supposed to support non-DT based instantation (aka board > >> >> >>> files)? If not, maybe it would be cleaner to simply check for > >> >> >>> !pdev->dev.of_node at the beginning of probe and return an error? > >> >> >> > >> >> >> I would like to maintain the possibility for board file based > >> >> >> instatiation, because the Raspberry Pi downstream kernel still doesn't > >> >> >> support device tree. If this is a no-go, I will accept that. > >> >> > > >> >> > Sure, you are free to do so. > >> >> > > >> >> > What I meant is that your probe won't call bcm2835_dma_chan_init() at all > >> >> > if there is no pdev->dev.of_node, because the loop iterating over channels > >> >> > is under the if clause. > >> >> > >> >> Yes you are right, but I think it will make the patching easier, later. > >> >> Currently, nothing bad happens without device tree - it just allocates > >> >> no channels. > >> > > >> > But isn't it really an error condition, if no channels are allocated? > >> > >> A fridge is still a working fridge, even if no beer is inside ;-) > >> Ok, bad example, but you will get an error message anyway when you try > >> to get a channel. > >> > >> > Anyway, back to my point about leaving non-DT support in a driver, the > >> > point is still valid only for drivers, not for platforms/boards. So if > >> > there are no boards supported using board files in mainline that could > >> > benefit from this driver, then this driver can be safely made DT-only, > >> > because no new non-DT platforms/boards can be added. > >> > >> I don't have a telling argument against this, but just thought writing > >> it this way will > >> make the migration of the downstream kernel to upstream easier, but if you say I > >> should change it, I will of course do that. > > > > I'm just presenting you the possible options. You are still free to have > > non-DT support in the driver, but if you don't need it (because you can't > > have any new non-DT platforms in mainline) then you can simplify some > > code. > > > > However the driver shouldn't be left with illusionary support for non-DT > > platforms until you decide to implement that. Instead, if you don't want > > to add non-DT support now, just make the driver DT-only, while keeping > > its design in a way allowing you to add non-DT support in future. > > This statement hits the nail on the head. Thank you! > > > In other words, a driver should not be able to probe using board files > > if support for such probing method is not available in it yet. > > That is meaningful. So would the following be ok? > > + ..... > + if (pdev->dev.of_node) { > + ..... > + } else { > + dev_err(&pdev->dev, > + "Failed to initialize channels, because device > tree not available: %d\n", rc); > + bcm2835_dma_free(od); > + return rc; > + } > + > + dev_dbg(&pdev->dev, "Initialized %i DMA channels\n", i); > + > + ..... > I mean, something closer to: static int bcm2835_dma_probe(struct platform_device *pdev) { uint32_t chan_mask; // ... if (pdev->dev.of_node) { rc = of_property_read_u32(..., &chan_mask); if (rc) { // Error out } } else { // Error message return -EINVAL; } // ... // Channel registration loop (independent of any OF functions) // ... if (pdev->dev.of_node) { rc = of_dma_controller_register(...); // ... } // ... } > >> I am becoming desperate anyway that this migration will ever fully > >> take place.... > > > > Why not? It's just a matter of people like you working on this (and > > addressing some review comments ;)). > > The most common comment about this is that people will not put effort in > the upstream kernel as long as there is no comfortable way for debugging > in the upstream kernel (i.e. at least USB support)....... Right, you need USB to have ethernet working. Still, IMHO UART (for console) + ethernet (for file transfer or NFS root) is the reasonable setup allowing you to debug further drivers in a comfortable way. Best regards, Tomasz -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
> I mean, something closer to: > ..... Ok :-) >> >> I am becoming desperate anyway that this migration will ever fully >> >> take place.... >> > >> > Why not? It's just a matter of people like you working on this (and >> > addressing some review comments ;)). >> >> The most common comment about this is that people will not put effort in >> the upstream kernel as long as there is no comfortable way for debugging >> in the upstream kernel (i.e. at least USB support)....... > > Right, you need USB to have ethernet working. Still, IMHO UART (for > console) + ethernet (for file transfer or NFS root) is the reasonable > setup allowing you to debug further drivers in a comfortable way. I am fine with UART and switching SD cards :-) But apparently this is not generally accepted. Greetings, Florian -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thursday 14 of November 2013 17:14:31 Florian Meier wrote: > >> >> I am becoming desperate anyway that this migration will ever fully > >> >> take place.... > >> > > >> > Why not? It's just a matter of people like you working on this (and > >> > addressing some review comments ;)). > >> > >> The most common comment about this is that people will not put effort in > >> the upstream kernel as long as there is no comfortable way for debugging > >> in the upstream kernel (i.e. at least USB support)....... > > > > Right, you need USB to have ethernet working. Still, IMHO UART (for > > console) + ethernet (for file transfer or NFS root) is the reasonable > > setup allowing you to debug further drivers in a comfortable way. > > I am fine with UART and switching SD cards :-) > But apparently this is not generally accepted. Still, isn't some work on USB support for RPi already going on? I believe Matt Porter (now on Cc) has been working on unifying dwc2 host-mode driver with Samsung s3c-hsotg device-mode driver (for the same IP) to get full OTG support on all applicable platforms. Best regards, Tomasz -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
2013/11/14 Tomasz Figa <tomasz.figa@gmail.com>: > On Thursday 14 of November 2013 17:14:31 Florian Meier wrote: >> >> >> I am becoming desperate anyway that this migration will ever fully >> >> >> take place.... >> >> > >> >> > Why not? It's just a matter of people like you working on this (and >> >> > addressing some review comments ;)). >> >> >> >> The most common comment about this is that people will not put effort in >> >> the upstream kernel as long as there is no comfortable way for debugging >> >> in the upstream kernel (i.e. at least USB support)....... >> > >> > Right, you need USB to have ethernet working. Still, IMHO UART (for >> > console) + ethernet (for file transfer or NFS root) is the reasonable >> > setup allowing you to debug further drivers in a comfortable way. >> >> I am fine with UART and switching SD cards :-) >> But apparently this is not generally accepted. > > Still, isn't some work on USB support for RPi already going on? I believe > Matt Porter (now on Cc) has been working on unifying dwc2 host-mode > driver with Samsung s3c-hsotg device-mode driver (for the same IP) to > get full OTG support on all applicable platforms. That sounds great! I never heard about that. Maybe some day it really happens that Raspbian uses the upstream kernel :-) -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, Nov 14, 2013 at 05:23:08PM +0100, Tomasz Figa wrote: > On Thursday 14 of November 2013 17:14:31 Florian Meier wrote: > > >> >> I am becoming desperate anyway that this migration will ever fully > > >> >> take place.... > > >> > > > >> > Why not? It's just a matter of people like you working on this (and > > >> > addressing some review comments ;)). > > >> > > >> The most common comment about this is that people will not put effort in > > >> the upstream kernel as long as there is no comfortable way for debugging > > >> in the upstream kernel (i.e. at least USB support)....... > > > > > > Right, you need USB to have ethernet working. Still, IMHO UART (for > > > console) + ethernet (for file transfer or NFS root) is the reasonable > > > setup allowing you to debug further drivers in a comfortable way. > > > > I am fine with UART and switching SD cards :-) > > But apparently this is not generally accepted. > > Still, isn't some work on USB support for RPi already going on? I believe > Matt Porter (now on Cc) has been working on unifying dwc2 host-mode > driver with Samsung s3c-hsotg device-mode driver (for the same IP) to > get full OTG support on all applicable platforms. Baby steps in that direction, yes. I need to finish the basic s3c-hsotg and dwc2 host only enablement for my platform and then will be focused on combining them. -Matt -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/Documentation/devicetree/bindings/dma/bcm2835-dma.txt b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt new file mode 100644 index 0000000..bca5e84 --- /dev/null +++ b/Documentation/devicetree/bindings/dma/bcm2835-dma.txt @@ -0,0 +1,59 @@ +* BCM2835 DMA controller + +Required properties: +- compatible: Should be "brcm,bcm2835-dma". +- reg: Should contain DMA registers location and length. +- interrupts: Should contain the DMA interrupts associated + to the DMA channels in ascending order. + First cell is the IRQ bank. + Second cell is the IRQ number. +- #dma-cells: Must be <1>, used to represent the number of integer cells in + the dmas property of client devices. +- dma-channels: Maximum number of DMA channels available. +- dma-requests: Number of DMA Requests. +- brcm,dma-channel-mask: Bit mask representing the channels available. + +Example: + +dma: dma@7e007000 { + compatible = "brcm,bcm2835-dma"; + reg = <0x7e007000 0xf00>; + interrupts = <1 16 + 1 17 + 1 18 + 1 19 + 1 20 + 1 21 + 1 22 + 1 23 + 1 24 + 1 25 + 1 26 + 1 27 + 1 28>; + + #dma-cells = <1>; + dma-channels = <16>; + dma-requests = <32>; + brcm,dma-channel-mask = <0x7f35>; +}; + +DMA clients connected to the BCM2835 DMA controller must use the format +described in the dma.txt file, using a two-cell specifier for each channel: +a phandle plus one integer cells. +The two cells in order are: + +1. A phandle pointing to the DMA controller. +2. The DREQ number. + +Example: + +bcm2835_i2s: i2s@7e203000 { + compatible = "brcm,bcm2835-i2s"; + reg = < 0x7e203000 0x20 + 0x7e101098 0x02>; + + dmas = <&dma 2 + &dma 3>; + dma-names = "tx", "rx"; +}; diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig index c61a6ec..880e723 100644 --- a/drivers/dma/Kconfig +++ b/drivers/dma/Kconfig @@ -300,6 +300,12 @@ config DMA_OMAP select DMA_ENGINE select DMA_VIRTUAL_CHANNELS +config DMA_BCM2835 + tristate "BCM2835 DMA engine support" + depends on (ARCH_BCM2835 || MACH_BCM2708) + select DMA_ENGINE + select DMA_VIRTUAL_CHANNELS + config TI_CPPI41 tristate "AM33xx CPPI41 DMA support" depends on ARCH_OMAP diff --git a/drivers/dma/Makefile b/drivers/dma/Makefile index 0ce2da9..0a6f08e 100644 --- a/drivers/dma/Makefile +++ b/drivers/dma/Makefile @@ -38,6 +38,7 @@ obj-$(CONFIG_EP93XX_DMA) += ep93xx_dma.o obj-$(CONFIG_DMA_SA11X0) += sa11x0-dma.o obj-$(CONFIG_MMP_TDMA) += mmp_tdma.o obj-$(CONFIG_DMA_OMAP) += omap-dma.o +obj-$(CONFIG_DMA_BCM2835) += bcm2835-dma.o obj-$(CONFIG_MMP_PDMA) += mmp_pdma.o obj-$(CONFIG_DMA_JZ4740) += dma-jz4740.o obj-$(CONFIG_TI_CPPI41) += cppi41.o diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c new file mode 100644 index 0000000..baf072e --- /dev/null +++ b/drivers/dma/bcm2835-dma.c @@ -0,0 +1,750 @@ +/* + * BCM2835 DMA engine support + * + * This driver only supports cyclic DMA transfers + * as needed for the I2S module. + * + * Author: Florian Meier, <florian.meier@koalo.de> + * Copyright 2013 + * + * based on + * OMAP DMAengine support by Russell King + * + * BCM2708 DMA Driver + * Copyright (C) 2010 Broadcom + * + * Raspberry Pi PCM I2S ALSA Driver + * Copyright (c) by Phil Poole 2013 + * + * MARVELL MMP Peripheral DMA Driver + * Copyright 2012 Marvell International Ltd. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ +#include <linux/dmaengine.h> +#include <linux/dma-mapping.h> +#include <linux/err.h> +#include <linux/init.h> +#include <linux/interrupt.h> +#include <linux/list.h> +#include <linux/module.h> +#include <linux/platform_device.h> +#include <linux/slab.h> +#include <linux/io.h> +#include <linux/spinlock.h> +#include <linux/of.h> +#include <linux/of_dma.h> + +#include "virt-dma.h" + +struct bcm2835_dmadev { + struct dma_device ddev; + spinlock_t lock; + void __iomem *dma_base; + struct device_dma_parameters dma_parms; +}; + +struct bcm2835_dma_cb { + uint32_t info; + uint32_t src; + uint32_t dst; + uint32_t length; + uint32_t stride; + uint32_t next; + uint32_t pad[2]; +}; + +struct bcm2835_chan { + struct virt_dma_chan vc; + struct list_head node; + + struct dma_slave_config cfg; + bool cyclic; + unsigned dreq; + + int dma_ch; + struct bcm2835_desc *desc; + + void __iomem *dma_chan_base; + int dma_irq_number; +}; + +struct bcm2835_desc { + struct virt_dma_desc vd; + enum dma_transfer_direction dir; + + unsigned int control_block_size; + struct bcm2835_dma_cb *control_block_base; + dma_addr_t control_block_base_phys; + + unsigned frames; + size_t size; +}; + +#define BCM2835_DMA_CS 0x00 +#define BCM2835_DMA_ADDR 0x04 +#define BCM2835_DMA_SOURCE_AD 0x0c +#define BCM2835_DMA_DEST_AD 0x10 +#define BCM2835_DMA_NEXTCB 0x1C + +/* DMA CS Control and Status bits */ +#define BCM2835_DMA_ACTIVE (1 << 0) +#define BCM2835_DMA_INT (1 << 2) +#define BCM2835_DMA_ISPAUSED (1 << 4) /* Pause requested or not active */ +#define BCM2835_DMA_ISHELD (1 << 5) /* Is held by DREQ flow control */ +#define BCM2835_DMA_ERR (1 << 8) +#define BCM2835_DMA_ABORT (1 << 30) /* stop current CB, go to next, WO */ +#define BCM2835_DMA_RESET (1 << 31) /* WO, self clearing */ + +#define BCM2835_DMA_INT_EN (1 << 0) +#define BCM2835_DMA_D_INC (1 << 4) +#define BCM2835_DMA_D_DREQ (1 << 6) +#define BCM2835_DMA_S_INC (1 << 8) +#define BCM2835_DMA_S_DREQ (1 << 6) + +#define BCM2835_DMA_PER_MAP(x) ((x) << 16) + +#define BCM2835_DMA_DATA_TYPE_S8 1 +#define BCM2835_DMA_DATA_TYPE_S16 2 +#define BCM2835_DMA_DATA_TYPE_S32 4 +#define BCM2835_DMA_DATA_TYPE_S128 16 + +/* valid only for channels 0 - 14, 15 has its own base address */ +#define BCM2835_DMA_CHAN(n) ((n) << 8) /* base address */ +#define BCM2835_DMA_CHANIO(dma_base, n) ((dma_base) + BCM2835_DMA_CHAN(n)) + +static inline struct bcm2835_dmadev *to_bcm2835_dma_dev(struct dma_device *d) +{ + return container_of(d, struct bcm2835_dmadev, ddev); +} + +static inline struct bcm2835_chan *to_bcm2835_dma_chan(struct dma_chan *c) +{ + return container_of(c, struct bcm2835_chan, vc.chan); +} + +static inline struct bcm2835_desc *to_bcm2835_dma_desc( + struct dma_async_tx_descriptor *t) +{ + return container_of(t, struct bcm2835_desc, vd.tx); +} + +static void bcm2835_dma_desc_free(struct virt_dma_desc *vd) +{ + struct bcm2835_desc *desc = container_of(vd, struct bcm2835_desc, vd); + dma_free_coherent(desc->vd.tx.chan->device->dev, + desc->control_block_size, + desc->control_block_base, + desc->control_block_base_phys); + kfree(desc); +} + +static int bcm2835_dma_abort(void __iomem *dma_chan_base) +{ + unsigned long int cs; + int rc = 0; + + cs = readl(dma_chan_base + BCM2835_DMA_CS); + + if (BCM2835_DMA_ACTIVE & cs) { + long int timeout = 10000; + + /* write 0 to the active bit - pause the DMA */ + writel(0, dma_chan_base + BCM2835_DMA_CS); + + /* wait for any current AXI transfer to complete */ + while ((cs & BCM2835_DMA_ISPAUSED) && --timeout >= 0) + cs = readl(dma_chan_base + BCM2835_DMA_CS); + + if (cs & BCM2835_DMA_ISPAUSED) { + /* we'll un-pause when we set of our next DMA */ + rc = -ETIMEDOUT; + + } else if (BCM2835_DMA_ACTIVE & cs) { + /* terminate the control block chain */ + writel(0, dma_chan_base + BCM2835_DMA_NEXTCB); + + /* abort the whole DMA */ + writel(BCM2835_DMA_ABORT | BCM2835_DMA_ACTIVE, + dma_chan_base + BCM2835_DMA_CS); + } + } + + return rc; +} + +static void bcm2835_dma_start_desc(struct bcm2835_chan *c) +{ + struct virt_dma_desc *vd = vchan_next_desc(&c->vc); + struct bcm2835_desc *d; + + if (!vd) { + c->desc = NULL; + return; + } + + list_del(&vd->node); + + c->desc = d = to_bcm2835_dma_desc(&vd->tx); + + dsb(); /* ARM data synchronization (push) operation */ + + writel(d->control_block_base_phys, c->dma_chan_base + BCM2835_DMA_ADDR); + writel(BCM2835_DMA_ACTIVE, c->dma_chan_base + BCM2835_DMA_CS); +} + +static irqreturn_t bcm2835_dma_callback(int irq, void *data) +{ + struct bcm2835_chan *c = data; + struct bcm2835_desc *d; + unsigned long flags; + + spin_lock_irqsave(&c->vc.lock, flags); + + /* acknowledge interrupt */ + writel(BCM2835_DMA_INT, c->dma_chan_base + BCM2835_DMA_CS); + + d = c->desc; + + if (d) { + /* TODO Only works for cyclic DMA */ + vchan_cyclic_callback(&d->vd); + } + + /* keep the DMA engine running */ + dsb(); /* ARM synchronization barrier */ + writel(BCM2835_DMA_ACTIVE, c->dma_chan_base + BCM2835_DMA_CS); + + spin_unlock_irqrestore(&c->vc.lock, flags); + + return IRQ_HANDLED; +} + +static int bcm2835_dma_alloc_chan_resources(struct dma_chan *chan) +{ + struct bcm2835_chan *c = to_bcm2835_dma_chan(chan); + + dev_dbg(c->vc.chan.device->dev, + "Allocating DMA channel %i\n", c->dma_ch); + + return request_irq(c->dma_irq_number, + bcm2835_dma_callback, 0, "DMA IRQ", c); +} + +static void bcm2835_dma_free_chan_resources(struct dma_chan *chan) +{ + struct bcm2835_chan *c = to_bcm2835_dma_chan(chan); + + vchan_free_chan_resources(&c->vc); + free_irq(c->dma_irq_number, c); + + dev_dbg(c->vc.chan.device->dev, "Freeing DMA channel %u\n", c->dma_ch); +} + +static size_t bcm2835_dma_desc_size(struct bcm2835_desc *d) +{ + return d->size; +} + +static size_t bcm2835_dma_desc_size_pos(struct bcm2835_desc *d, dma_addr_t addr) +{ + unsigned i; + size_t size; + + for (size = i = 0; i < d->frames; i++) { + struct bcm2835_dma_cb *control_block = + &d->control_block_base[i]; + size_t this_size = control_block->length; + dma_addr_t dma; + + if (d->dir == DMA_DEV_TO_MEM) + dma = control_block->dst; + else + dma = control_block->src; + + if (size) + size += this_size; + else if (addr >= dma && addr < dma + this_size) + size += dma + this_size - addr; + } + + return size; +} + +static enum dma_status bcm2835_dma_tx_status(struct dma_chan *chan, + dma_cookie_t cookie, struct dma_tx_state *txstate) +{ + struct bcm2835_chan *c = to_bcm2835_dma_chan(chan); + struct virt_dma_desc *vd; + enum dma_status ret; + unsigned long flags; + + ret = dma_cookie_status(chan, cookie, txstate); + if (ret == DMA_SUCCESS || !txstate) + return ret; + + spin_lock_irqsave(&c->vc.lock, flags); + vd = vchan_find_desc(&c->vc, cookie); + if (vd) { + txstate->residue = + bcm2835_dma_desc_size(to_bcm2835_dma_desc(&vd->tx)); + } else if (c->desc && c->desc->vd.tx.cookie == cookie) { + struct bcm2835_desc *d = c->desc; + dma_addr_t pos; + + if (d->dir == DMA_MEM_TO_DEV) + pos = readl(c->dma_chan_base + BCM2835_DMA_SOURCE_AD); + else if (d->dir == DMA_DEV_TO_MEM) + pos = readl(c->dma_chan_base + BCM2835_DMA_DEST_AD); + else + pos = 0; + + txstate->residue = bcm2835_dma_desc_size_pos(d, pos); + } else { + txstate->residue = 0; + } + + spin_unlock_irqrestore(&c->vc.lock, flags); + + return ret; +} + +static void bcm2835_dma_issue_pending(struct dma_chan *chan) +{ + struct bcm2835_chan *c = to_bcm2835_dma_chan(chan); + unsigned long flags; + + c->cyclic = true; /* nothing else is implemented */ + + spin_lock_irqsave(&c->vc.lock, flags); + if (vchan_issue_pending(&c->vc) && !c->desc) + bcm2835_dma_start_desc(c); + + spin_unlock_irqrestore(&c->vc.lock, flags); +} + + +static struct dma_async_tx_descriptor *bcm2835_dma_prep_dma_cyclic( + struct dma_chan *chan, dma_addr_t buf_addr, size_t buf_len, + size_t period_len, enum dma_transfer_direction direction, + unsigned long flags, void *context) +{ + struct bcm2835_chan *c = to_bcm2835_dma_chan(chan); + enum dma_slave_buswidth dev_width; + struct bcm2835_desc *d; + dma_addr_t dev_addr; + unsigned es, sync_type; + unsigned frame; + + /* Grab configuration */ + if (direction == DMA_DEV_TO_MEM) { + dev_addr = c->cfg.src_addr; + dev_width = c->cfg.src_addr_width; + sync_type = BCM2835_DMA_S_DREQ; + } else if (direction == DMA_MEM_TO_DEV) { + dev_addr = c->cfg.dst_addr; + dev_width = c->cfg.dst_addr_width; + sync_type = BCM2835_DMA_D_DREQ; + } else { + dev_err(chan->device->dev, "%s: bad direction?\n", __func__); + return NULL; + } + + /* Bus width translates to the element size (ES) */ + switch (dev_width) { + case DMA_SLAVE_BUSWIDTH_4_BYTES: + es = BCM2835_DMA_DATA_TYPE_S32; + break; + default: + return NULL; + } + + /* Now allocate and setup the descriptor. */ + d = kzalloc(sizeof(*d), GFP_NOWAIT); + if (!d) + return NULL; + + d->dir = direction; + d->frames = buf_len / period_len; + + /* Allocate memory for control blocks */ + d->control_block_size = d->frames * sizeof(struct bcm2835_dma_cb); + d->control_block_base = dma_alloc_coherent(chan->device->dev, + d->control_block_size, &d->control_block_base_phys, + GFP_NOWAIT); + + if (!d->control_block_base) { + kfree(d); + dev_err(chan->device->dev, + "%s: Memory allocation error\n", __func__); + return NULL; + } + + memset(d->control_block_base, 0, d->control_block_size); + + /* + * Iterate over all frames, create a control block + * for each frame and link them together. + */ + for (frame = 0; frame < d->frames; frame++) { + struct bcm2835_dma_cb *control_block = + &d->control_block_base[frame]; + + /* Setup adresses */ + if (d->dir == DMA_DEV_TO_MEM) { + control_block->info = BCM2835_DMA_D_INC; + control_block->src = dev_addr; + control_block->dst = buf_addr + frame * period_len; + } else { + control_block->info = BCM2835_DMA_S_INC; + control_block->src = buf_addr + frame * period_len; + control_block->dst = dev_addr; + } + + /* Enable interrupt */ + control_block->info |= BCM2835_DMA_INT_EN; + + /* Setup synchronization */ + if (sync_type != 0) + control_block->info |= sync_type; + + /* Setup DREQ channel */ + if (c->dreq != 0) + control_block->info |= + BCM2835_DMA_PER_MAP(c->dreq); + + /* Length of a frame */ + control_block->length = period_len; + d->size += control_block->length; + + /* + * Next block is the next frame. + * This DMA engine driver currently only supports cyclic DMA. + * Therefore, wrap around at number of frames. + */ + control_block->next = d->control_block_base_phys + + sizeof(struct bcm2835_dma_cb) + * ((frame + 1) % d->frames); + + /* The following fields are not used here */ + control_block->stride = 0; + control_block->pad[0] = 0; + control_block->pad[1] = 0; + } + + return vchan_tx_prep(&c->vc, &d->vd, flags); +} + +static int bcm2835_dma_slave_config(struct bcm2835_chan *c, + struct dma_slave_config *cfg) +{ + if ((cfg->direction == DMA_DEV_TO_MEM && + cfg->src_addr_width != DMA_SLAVE_BUSWIDTH_4_BYTES) || + (cfg->direction == DMA_MEM_TO_DEV && + cfg->dst_addr_width != DMA_SLAVE_BUSWIDTH_4_BYTES) || + (cfg->direction != DMA_DEV_TO_MEM && + cfg->direction != DMA_MEM_TO_DEV)) { + return -EINVAL; + } + + c->cfg = *cfg; + + return 0; +} + +static int bcm2835_dma_terminate_all(struct bcm2835_chan *c) +{ + struct bcm2835_dmadev *d = to_bcm2835_dma_dev(c->vc.chan.device); + unsigned long flags; + int timeout = 1000; + LIST_HEAD(head); + + spin_lock_irqsave(&c->vc.lock, flags); + + /* Prevent this channel being scheduled */ + spin_lock(&d->lock); + list_del_init(&c->node); + spin_unlock(&d->lock); + + /* + * Stop DMA activity: we assume the callback will not be called + * after bcm_dma_abort() returns (even if it does, it will see + * c->desc is NULL and exit.) + */ + if (c->desc) { + c->desc = NULL; + bcm2835_dma_abort(c->dma_chan_base); + + /* Wait for stopping */ + while (timeout > 0) { + timeout--; + if (!(readl(c->dma_chan_base + BCM2835_DMA_CS) & + BCM2835_DMA_ACTIVE)) + break; + + cpu_relax(); + } + + if (timeout <= 0) + dev_err(d->ddev.dev, "DMA transfer could not be terminated\n"); + } + + vchan_get_all_descriptors(&c->vc, &head); + spin_unlock_irqrestore(&c->vc.lock, flags); + vchan_dma_desc_free_list(&c->vc, &head); + + return 0; +} + +static int bcm2835_dma_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd, + unsigned long arg) +{ + struct bcm2835_chan *c = to_bcm2835_dma_chan(chan); + int ret; + + switch (cmd) { + case DMA_SLAVE_CONFIG: + return bcm2835_dma_slave_config(c, + (struct dma_slave_config *)arg); + + case DMA_TERMINATE_ALL: + bcm2835_dma_terminate_all(c); + break; + + default: + ret = -ENXIO; + break; + } + + return ret; +} + +static int bcm2835_dma_chan_init(struct bcm2835_dmadev *d, int chan_id, int irq) +{ + struct bcm2835_chan *c; + + c = kzalloc(sizeof(*c), GFP_KERNEL); + if (!c) + return -ENOMEM; + + c->vc.desc_free = bcm2835_dma_desc_free; + vchan_init(&c->vc, &d->ddev); + INIT_LIST_HEAD(&c->node); + + d->ddev.chancnt++; + + c->dma_chan_base = BCM2835_DMA_CHANIO(d->dma_base, chan_id); + c->dma_ch = chan_id; + c->dma_irq_number = irq; + + return 0; +} + +static void bcm2835_dma_free(struct bcm2835_dmadev *od) +{ + while (!list_empty(&od->ddev.channels)) { + struct bcm2835_chan *c = list_first_entry(&od->ddev.channels, + struct bcm2835_chan, vc.chan.device_node); + + list_del(&c->vc.chan.device_node); + tasklet_kill(&c->vc.task); + kfree(c); + } +} + +#if defined(CONFIG_OF) +static const struct of_device_id bcm2835_dma_of_match[] = { + { .compatible = "brcm,bcm2835-dma", }, + {}, +}; +MODULE_DEVICE_TABLE(of, bcm2835_dma_of_match); +#endif + +static struct dma_chan *bcm2835_dma_xlate(struct of_phandle_args *dma_spec, + struct of_dma *ofdma) +{ + struct bcm2835_dmadev *d = ofdma->of_dma_data; + struct dma_chan *chan, *candidate; + +retry: + candidate = NULL; + + /* walk the list of channels registered with the current instance and + * find one that is currently unused */ + list_for_each_entry(chan, &d->ddev.channels, device_node) + if (chan->client_count == 0) { + candidate = chan; + break; + } + + if (!candidate) + return NULL; + + /* dma_get_slave_channel will return NULL if we lost a race between + * the lookup and the reservation */ + chan = dma_get_slave_channel(candidate); + + if (chan) { + struct bcm2835_chan *c = to_bcm2835_dma_chan(chan); + + /* Set DREQ from param */ + c->dreq = dma_spec->args[0]; + + return chan; + } + + goto retry; +} + +static int bcm2835_dma_device_slave_caps(struct dma_chan *dchan, + struct dma_slave_caps *caps) +{ + caps->src_addr_widths = BIT(DMA_SLAVE_BUSWIDTH_4_BYTES); + caps->dstn_addr_widths = BIT(DMA_SLAVE_BUSWIDTH_4_BYTES); + caps->directions = BIT(DMA_DEV_TO_MEM) | BIT(DMA_MEM_TO_DEV); + caps->cmd_pause = false; + caps->cmd_terminate = true; + + return 0; +} + +static int bcm2835_dma_probe(struct platform_device *pdev) +{ + struct bcm2835_dmadev *od; + struct resource *dma_res = NULL; + void __iomem *dma_base = NULL; + int rc = 0; + int i = 0; + int irq; + uint32_t chans_available; + + if (!pdev->dev.dma_mask) + pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask; + + rc = dma_set_mask(&pdev->dev, DMA_BIT_MASK(32)); + if (rc) + return rc; + dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32)); + + od = devm_kzalloc(&pdev->dev, sizeof(*od), GFP_KERNEL); + if (!od) + return -ENOMEM; + + pdev->dev.dma_parms = &od->dma_parms; + dma_set_max_seg_size(&pdev->dev, 0x3FFFFFFF); + + dma_res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + dma_base = devm_ioremap_resource(&pdev->dev, dma_res); + if (IS_ERR(dma_base)) + return PTR_ERR(dma_base); + + od->dma_base = dma_base; + + dma_cap_set(DMA_SLAVE, od->ddev.cap_mask); + dma_cap_set(DMA_CYCLIC, od->ddev.cap_mask); + od->ddev.device_alloc_chan_resources = bcm2835_dma_alloc_chan_resources; + od->ddev.device_free_chan_resources = bcm2835_dma_free_chan_resources; + od->ddev.device_tx_status = bcm2835_dma_tx_status; + od->ddev.device_issue_pending = bcm2835_dma_issue_pending; + od->ddev.device_slave_caps = bcm2835_dma_device_slave_caps; + od->ddev.device_prep_dma_cyclic = bcm2835_dma_prep_dma_cyclic; + od->ddev.device_control = bcm2835_dma_control; + od->ddev.dev = &pdev->dev; + INIT_LIST_HEAD(&od->ddev.channels); + spin_lock_init(&od->lock); + + platform_set_drvdata(pdev, od); + + if (pdev->dev.of_node) { + const void *chan_mask; + + /* Device-tree DMA controller registration */ + rc = of_dma_controller_register(pdev->dev.of_node, + bcm2835_dma_xlate, od); + if (rc) { + dev_err(&pdev->dev, "Failed to register DMA controller\n"); + dma_async_device_unregister(&od->ddev); + bcm2835_dma_free(od); + return rc; + } + + /* Request DMA channel mask from device tree */ + chan_mask = of_get_property(pdev->dev.of_node, + "brcm,dma-channel-mask", NULL); + + if (!chan_mask) { + dev_err(&pdev->dev, "Failed to get channel mask\n"); + dma_async_device_unregister(&od->ddev); + bcm2835_dma_free(od); + return -EINVAL; + } + + chans_available = be32_to_cpup(chan_mask); + + /* do not use the FIQ and BULK channels */ + chans_available &= ~0xD; + + for (i = 0; i < pdev->num_resources; i++) { + irq = platform_get_irq(pdev, i); + if (irq < 0) + break; + + if (chans_available & (1 << i)) { + rc = bcm2835_dma_chan_init(od, i, irq); + if (rc) { + bcm2835_dma_free(od); + return rc; + } + } + } + } + + dev_dbg(&pdev->dev, "Initialized %i DMA channels\n", i); + + rc = dma_async_device_register(&od->ddev); + if (rc) { + dev_err(&pdev->dev, + "Failed to register slave DMA engine device: %d\n", rc); + bcm2835_dma_free(od); + return rc; + } + + dev_dbg(&pdev->dev, "Load BCM2835 DMA engine driver\n"); + + return rc; +} + +static int bcm2835_dma_remove(struct platform_device *pdev) +{ + struct bcm2835_dmadev *od = platform_get_drvdata(pdev); + + dma_async_device_unregister(&od->ddev); + bcm2835_dma_free(od); + + return 0; +} + +static struct platform_driver bcm2835_dma_driver = { + .probe = bcm2835_dma_probe, + .remove = bcm2835_dma_remove, + .driver = { + .name = "bcm2835-dma", + .owner = THIS_MODULE, + .of_match_table = of_match_ptr(bcm2835_dma_of_match), + }, +}; + +module_platform_driver(bcm2835_dma_driver); + +MODULE_ALIAS("platform:bcm2835-dma"); +MODULE_DESCRIPTION("BCM2835 DMA engine driver"); +MODULE_AUTHOR("Florian Meier <florian.meier@koalo.de>"); +MODULE_LICENSE("GPL v2");
Add support for DMA controller of BCM2835 as used in the Raspberry Pi. Currently it only supports cyclic DMA. Signed-off-by: Florian Meier <florian.meier@koalo.de> --- Besides of many minor improvements (thanks to your helpful comments), this fourth version does the assignment of the virtual to the hardware channel already in probe. This simplifies things a lot (first of all the complex alloc_chan_resources is now gone). Regarding the maximum segment size: 0x3FFFFFFF is the maximum possible DMA transfer length value and I have no evidence that the maximum segment size is smaller. .../devicetree/bindings/dma/bcm2835-dma.txt | 59 ++ drivers/dma/Kconfig | 6 + drivers/dma/Makefile | 1 + drivers/dma/bcm2835-dma.c | 750 ++++++++++++++++++++ 4 files changed, 816 insertions(+) create mode 100644 Documentation/devicetree/bindings/dma/bcm2835-dma.txt create mode 100644 drivers/dma/bcm2835-dma.c