| Message ID | 1460384473-5775-3-git-send-email-okaya@codeaurora.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: > +static int hidma_chan_stats(struct seq_file *s, void *unused) > +{ > + struct hidma_chan *mchan = s->private; > + struct hidma_desc *mdesc; > + struct hidma_dev *dmadev = mchan->dmadev; > + > + pm_runtime_get_sync(dmadev->ddev.dev); debug shouldn't power up device, why do you want to do that
On 2016-04-25 23:30, Vinod Koul wrote: > On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: > >> +static int hidma_chan_stats(struct seq_file *s, void *unused) >> +{ >> + struct hidma_chan *mchan = s->private; >> + struct hidma_desc *mdesc; >> + struct hidma_dev *dmadev = mchan->dmadev; >> + >> + pm_runtime_get_sync(dmadev->ddev.dev); > > debug shouldn't power up device, why do you want to do that Clocks are turned off while the hw is idle. I can’t reach hw registers without restoring power.
On Tue, Apr 26, 2016 at 08:08:16AM -0400, okaya@codeaurora.org wrote: > On 2016-04-25 23:30, Vinod Koul wrote: > >On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: > > > >>+static int hidma_chan_stats(struct seq_file *s, void *unused) > >>+{ > >>+ struct hidma_chan *mchan = s->private; > >>+ struct hidma_desc *mdesc; > >>+ struct hidma_dev *dmadev = mchan->dmadev; > >>+ > >>+ pm_runtime_get_sync(dmadev->ddev.dev); > > > >debug shouldn't power up device, why do you want to do that > > > Clocks are turned off while the hw is idle. I can’t reach hw > registers without restoring power. Hmm, have you thought about using regmap?
On 4/26/2016 12:25 PM, Vinod Koul wrote: > On Tue, Apr 26, 2016 at 08:08:16AM -0400, okaya@codeaurora.org wrote: >> On 2016-04-25 23:30, Vinod Koul wrote: >>> On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: >>> >>>> +static int hidma_chan_stats(struct seq_file *s, void *unused) >>>> +{ >>>> + struct hidma_chan *mchan = s->private; >>>> + struct hidma_desc *mdesc; >>>> + struct hidma_dev *dmadev = mchan->dmadev; >>>> + >>>> + pm_runtime_get_sync(dmadev->ddev.dev); >>> >>> debug shouldn't power up device, why do you want to do that >> >> >> Clocks are turned off while the hw is idle. I can’t reach hw >> registers without restoring power. > > Hmm, have you thought about using regmap? > To be honest, I didn't know what regmap is but I just read some code and looked at how it is used. Feel free to correct me if I got it wrong. Regmap seems to be designed for *slow* speed peripherals to improve frequent accesses by the SW. It looks like it is used by MFD, SPI and I2C drivers. It seems to cache the register contents and flush/invalidate them only when needed. The MMIO version seems to be assuming the presence of device-tree like CLK API which doesn't exist on ACPI systems and is not portable. My reaction is that it is a lot of code with no added functionality to what HIDMA driver is trying to achieve. Given that the use case here is only for debug purposes; I think it is OK to keep this runtime call here. I don't want to add any overhead into the existing code just to support the debug use case. None of my register read/writes are slow. This file will only be used to troubleshoot customer issues.
On Tue, Apr 26, 2016 at 12:55:18PM -0400, Sinan Kaya wrote: > On 4/26/2016 12:25 PM, Vinod Koul wrote: > > On Tue, Apr 26, 2016 at 08:08:16AM -0400, okaya@codeaurora.org wrote: > >> On 2016-04-25 23:30, Vinod Koul wrote: > >>> On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: > >>> > >>>> +static int hidma_chan_stats(struct seq_file *s, void *unused) > >>>> +{ > >>>> + struct hidma_chan *mchan = s->private; > >>>> + struct hidma_desc *mdesc; > >>>> + struct hidma_dev *dmadev = mchan->dmadev; > >>>> + > >>>> + pm_runtime_get_sync(dmadev->ddev.dev); > >>> > >>> debug shouldn't power up device, why do you want to do that > >> > >> > >> Clocks are turned off while the hw is idle. I can’t reach hw > >> registers without restoring power. > > > > Hmm, have you thought about using regmap? > > > > To be honest, I didn't know what regmap is but I just read some code > and looked at how it is used. Feel free to correct me if I got it > wrong. > > Regmap seems to be designed for *slow* speed peripherals to improve frequent > accesses by the SW. It looks like it is used by MFD, SPI and I2C drivers. > > It seems to cache the register contents and flush/invalidate them only when > needed. > > The MMIO version seems to be assuming the presence of device-tree like CLK > API which doesn't exist on ACPI systems and is not portable. > > My reaction is that it is a lot of code with no added functionality to what > HIDMA driver is trying to achieve. > > Given that the use case here is only for debug purposes; I think it is OK > to keep this runtime call here. I don't want to add any overhead into the > existing code just to support the debug use case. > > None of my register read/writes are slow. This file will only be used to > troubleshoot customer issues. $ is always faster than MMIO. This way you can give reg contents to users without waking up hw. Also we at Intel use regmap on ACPI systems without CLK API
On 27/04/16 09:15, Vinod Koul wrote: > On Tue, Apr 26, 2016 at 12:55:18PM -0400, Sinan Kaya wrote: >> On 4/26/2016 12:25 PM, Vinod Koul wrote: >>> On Tue, Apr 26, 2016 at 08:08:16AM -0400, okaya@codeaurora.org wrote: >>>> On 2016-04-25 23:30, Vinod Koul wrote: >>>>> On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: >>>>> >>>>>> +static int hidma_chan_stats(struct seq_file *s, void *unused) >>>>>> +{ >>>>>> + struct hidma_chan *mchan = s->private; >>>>>> + struct hidma_desc *mdesc; >>>>>> + struct hidma_dev *dmadev = mchan->dmadev; >>>>>> + >>>>>> + pm_runtime_get_sync(dmadev->ddev.dev); >>>>> >>>>> debug shouldn't power up device, why do you want to do that >>>> >>>> >>>> Clocks are turned off while the hw is idle. I can’t reach hw >>>> registers without restoring power. >>> >>> Hmm, have you thought about using regmap? >>> >> >> To be honest, I didn't know what regmap is but I just read some code >> and looked at how it is used. Feel free to correct me if I got it >> wrong. >> >> Regmap seems to be designed for *slow* speed peripherals to improve frequent >> accesses by the SW. It looks like it is used by MFD, SPI and I2C drivers. >> >> It seems to cache the register contents and flush/invalidate them only when >> needed. >> >> The MMIO version seems to be assuming the presence of device-tree like CLK >> API which doesn't exist on ACPI systems and is not portable. >> >> My reaction is that it is a lot of code with no added functionality to what >> HIDMA driver is trying to achieve. >> >> Given that the use case here is only for debug purposes; I think it is OK >> to keep this runtime call here. I don't want to add any overhead into the >> existing code just to support the debug use case. >> >> None of my register read/writes are slow. This file will only be used to >> troubleshoot customer issues. I'd recommend you actually run perf on a a few of your MMIO accesses. I believe the result will be eye opening. On the KVM side, we've trimmed our MMIO access as much as possible, using a memory-based cache (similar to regmap in concept). This has made some code paths about 40% faster. > $ is always faster than MMIO. This way you can give reg contents to users > without waking up hw. Indeed. MMIO access sucks rocks, even on a very fast box. Actually, the faster the box is, the slower MMIO feels (compared to memory). Thanks, M.
On 2016-04-27 04:15, Vinod Koul wrote: > On Tue, Apr 26, 2016 at 12:55:18PM -0400, Sinan Kaya wrote: >> On 4/26/2016 12:25 PM, Vinod Koul wrote: >> > On Tue, Apr 26, 2016 at 08:08:16AM -0400, okaya@codeaurora.org wrote: >> >> On 2016-04-25 23:30, Vinod Koul wrote: >> >>> On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: >> >>> >> >>>> +static int hidma_chan_stats(struct seq_file *s, void *unused) >> >>>> +{ >> >>>> + struct hidma_chan *mchan = s->private; >> >>>> + struct hidma_desc *mdesc; >> >>>> + struct hidma_dev *dmadev = mchan->dmadev; >> >>>> + >> >>>> + pm_runtime_get_sync(dmadev->ddev.dev); >> >>> >> >>> debug shouldn't power up device, why do you want to do that >> >> >> >> >> >> Clocks are turned off while the hw is idle. I can’t reach hw >> >> registers without restoring power. >> > >> > Hmm, have you thought about using regmap? >> > >> >> To be honest, I didn't know what regmap is but I just read some code >> and looked at how it is used. Feel free to correct me if I got it >> wrong. >> >> Regmap seems to be designed for *slow* speed peripherals to improve >> frequent >> accesses by the SW. It looks like it is used by MFD, SPI and I2C >> drivers. >> >> It seems to cache the register contents and flush/invalidate them only >> when >> needed. >> >> The MMIO version seems to be assuming the presence of device-tree like >> CLK >> API which doesn't exist on ACPI systems and is not portable. >> >> My reaction is that it is a lot of code with no added functionality to >> what >> HIDMA driver is trying to achieve. >> >> Given that the use case here is only for debug purposes; I think it is >> OK >> to keep this runtime call here. I don't want to add any overhead into >> the >> existing code just to support the debug use case. >> >> None of my register read/writes are slow. This file will only be used >> to >> troubleshoot customer issues. > > $ is always faster than MMIO. This way you can give reg contents to > users > without waking up hw. > > Also we at Intel use regmap on ACPI systems without CLK API I can try and see the performance impact is. What happens to registers that hw updates like status registers. Those will be most interesting during debug. How does remap get updated for those? Is there a way to tell it not to cache certain registers
On 2016-04-27 04:47, Marc Zyngier wrote: > On 27/04/16 09:15, Vinod Koul wrote: >> On Tue, Apr 26, 2016 at 12:55:18PM -0400, Sinan Kaya wrote: >>> On 4/26/2016 12:25 PM, Vinod Koul wrote: >>>> On Tue, Apr 26, 2016 at 08:08:16AM -0400, okaya@codeaurora.org >>>> wrote: >>>>> On 2016-04-25 23:30, Vinod Koul wrote: >>>>>> On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: >>>>>> >>>>>>> +static int hidma_chan_stats(struct seq_file *s, void *unused) >>>>>>> +{ >>>>>>> + struct hidma_chan *mchan = s->private; >>>>>>> + struct hidma_desc *mdesc; >>>>>>> + struct hidma_dev *dmadev = mchan->dmadev; >>>>>>> + >>>>>>> + pm_runtime_get_sync(dmadev->ddev.dev); >>>>>> >>>>>> debug shouldn't power up device, why do you want to do that >>>>> >>>>> >>>>> Clocks are turned off while the hw is idle. I can’t reach hw >>>>> registers without restoring power. >>>> >>>> Hmm, have you thought about using regmap? >>>> >>> >>> To be honest, I didn't know what regmap is but I just read some code >>> and looked at how it is used. Feel free to correct me if I got it >>> wrong. >>> >>> Regmap seems to be designed for *slow* speed peripherals to improve >>> frequent >>> accesses by the SW. It looks like it is used by MFD, SPI and I2C >>> drivers. >>> >>> It seems to cache the register contents and flush/invalidate them >>> only when >>> needed. >>> >>> The MMIO version seems to be assuming the presence of device-tree >>> like CLK >>> API which doesn't exist on ACPI systems and is not portable. >>> >>> My reaction is that it is a lot of code with no added functionality >>> to what >>> HIDMA driver is trying to achieve. >>> >>> Given that the use case here is only for debug purposes; I think it >>> is OK >>> to keep this runtime call here. I don't want to add any overhead into >>> the >>> existing code just to support the debug use case. >>> >>> None of my register read/writes are slow. This file will only be used >>> to >>> troubleshoot customer issues. > > I'd recommend you actually run perf on a a few of your MMIO accesses. I > believe the result will be eye opening. On the KVM side, we've trimmed > our MMIO access as much as possible, using a memory-based cache > (similar > to regmap in concept). This has made some code paths about 40% faster. > >> $ is always faster than MMIO. This way you can give reg contents to >> users >> without waking up hw. > > Indeed. MMIO access sucks rocks, even on a very fast box. Actually, the > faster the box is, the slower MMIO feels (compared to memory). > > Thanks, > > M. Agreed. However, I need to understand how regmap really works under the covers and whether it is compatible with the hardware.
On 4/27/2016 8:51 AM, okaya@codeaurora.org wrote: > On 2016-04-27 04:15, Vinod Koul wrote: >> On Tue, Apr 26, 2016 at 12:55:18PM -0400, Sinan Kaya wrote: >>> On 4/26/2016 12:25 PM, Vinod Koul wrote: >>> > On Tue, Apr 26, 2016 at 08:08:16AM -0400, okaya@codeaurora.org wrote: >>> >> On 2016-04-25 23:30, Vinod Koul wrote: >>> >>> On Mon, Apr 11, 2016 at 10:21:12AM -0400, Sinan Kaya wrote: >>> >>> >>> >>>> +static int hidma_chan_stats(struct seq_file *s, void *unused) >>> >>>> +{ >>> >>>> + struct hidma_chan *mchan = s->private; >>> >>>> + struct hidma_desc *mdesc; >>> >>>> + struct hidma_dev *dmadev = mchan->dmadev; >>> >>>> + >>> >>>> + pm_runtime_get_sync(dmadev->ddev.dev); >>> >>> >>> >>> debug shouldn't power up device, why do you want to do that >>> >> >>> >> >>> >> Clocks are turned off while the hw is idle. I can’t reach hw >>> >> registers without restoring power. >>> > >>> > Hmm, have you thought about using regmap? >>> > >>> >>> To be honest, I didn't know what regmap is but I just read some code >>> and looked at how it is used. Feel free to correct me if I got it >>> wrong. >>> >>> Regmap seems to be designed for *slow* speed peripherals to improve frequent >>> accesses by the SW. It looks like it is used by MFD, SPI and I2C drivers. >>> >>> It seems to cache the register contents and flush/invalidate them only when >>> needed. >>> >>> The MMIO version seems to be assuming the presence of device-tree like CLK >>> API which doesn't exist on ACPI systems and is not portable. >>> >>> My reaction is that it is a lot of code with no added functionality to what >>> HIDMA driver is trying to achieve. >>> >>> Given that the use case here is only for debug purposes; I think it is OK >>> to keep this runtime call here. I don't want to add any overhead into the >>> existing code just to support the debug use case. >>> >>> None of my register read/writes are slow. This file will only be used to >>> troubleshoot customer issues. >> >> $ is always faster than MMIO. This way you can give reg contents to users >> without waking up hw. >> >> Also we at Intel use regmap on ACPI systems without CLK API > > I can try and see the performance impact is. What happens to registers that hw updates like status registers. Those will be most interesting during debug. How does remap get updated for those? Is there a way to tell it not to cache certain registers My evaluation turned out negative. The regmap code is nice for bus like peripherals like I2C and SPI where everything is bitwise accessed. This is not the case in this code. Regmap is a nice tool if used properly but it doesn't mean that it will work in every single case. It doesn't match with the goal of this driver. As soon as I abstract register accesses, the regmap code writes all MMIO registers with the readl variant functions. Barriers are really expensive on ARM. I paid very special attention in the code to decide when to use relaxed version vs. the readl version. I lose all of this optimization. Since the clocks are restored only during the debug case, I don't see any problems here. It is not worth the effort to do redo the whole thing and introduce errors as I see a lot of tripping points like regmap_sync variants.
On Sun, May 01, 2016 at 12:35:37AM -0400, Sinan Kaya wrote: > >>> > > >>> > Hmm, have you thought about using regmap? > >>> > >>> To be honest, I didn't know what regmap is but I just read some code > >>> and looked at how it is used. Feel free to correct me if I got it > >>> wrong. > >>> > >>> Regmap seems to be designed for *slow* speed peripherals to improve frequent > >>> accesses by the SW. It looks like it is used by MFD, SPI and I2C drivers. > >>> > >>> It seems to cache the register contents and flush/invalidate them only when > >>> needed. > >>> > >>> The MMIO version seems to be assuming the presence of device-tree like CLK > >>> API which doesn't exist on ACPI systems and is not portable. > >>> > >>> My reaction is that it is a lot of code with no added functionality to what > >>> HIDMA driver is trying to achieve. > >>> > >>> Given that the use case here is only for debug purposes; I think it is OK > >>> to keep this runtime call here. I don't want to add any overhead into the > >>> existing code just to support the debug use case. > >>> > >>> None of my register read/writes are slow. This file will only be used to > >>> troubleshoot customer issues. > >> > >> $ is always faster than MMIO. This way you can give reg contents to users > >> without waking up hw. > >> > >> Also we at Intel use regmap on ACPI systems without CLK API > > > > I can try and see the performance impact is. What happens to registers that hw updates like status registers. Those will be most interesting during debug. How does remap get updated for those? Is there a way to tell it not to cache certain registers > > My evaluation turned out negative. The regmap code is nice for bus like peripherals > like I2C and SPI where everything is bitwise accessed. This is not the case > in this code. I do not entirely agree with the statements here, it does give big benefit on our systems with MMIO. I am going to ask Mark to comment on this, he know better and understands ARM. I am probably going to be okay with this not using regmap and it is debug but you should give that a try in future for better performance and ofcourse you can add to regmap to get a better model for your device > Regmap is a nice tool if used properly but it doesn't mean that it will work > in every single case. It doesn't match with the goal of this driver. > > As soon as I abstract register accesses, the regmap code writes all MMIO registers > with the readl variant functions. > > Barriers are really expensive on ARM. I paid very special attention in the > code to decide when to use relaxed version vs. the readl version. I lose > all of this optimization. > > Since the clocks are restored only during the debug case, I don't see any > problems here. It is not worth the effort to do redo the whole thing and > introduce errors as I see a lot of tripping points like regmap_sync variants.
On Mon, May 02, 2016 at 02:55:52PM +0530, Vinod Koul wrote: > On Sun, May 01, 2016 at 12:35:37AM -0400, Sinan Kaya wrote: > > My evaluation turned out negative. The regmap code is nice for bus like peripherals > > like I2C and SPI where everything is bitwise accessed. This is not the case > > in this code. > I do not entirely agree with the statements here, it does give big benefit > on our systems with MMIO. I am going to ask Mark to comment on this, he > know better and understands ARM. > I am probably going to be okay with this not using regmap and it is debug > but you should give that a try in future for better performance and ofcourse > you can add to regmap to get a better model for your device I've no idea what "this" is, sorry. All I've got here is an enormous backtrace with a bunch of the messages not even word wrapped. Please be aware that I get CCed on so much irrelevant crap that copying me into the middle of a thread about some other subsystem is likely to get missed - almost always it's review of some patch I've got no interest in. > > Barriers are really expensive on ARM. I paid very special attention in the > > code to decide when to use relaxed version vs. the readl version. I lose > > all of this optimization. Drivers should not be using the relaxed accessors, they are there for the generic code to build on not for drivers and they really need the cache flush operations. Getting the cache flush operations right, especially on ARM, isn't easy and needs detailed review.
diff --git a/drivers/dma/qcom/Makefile b/drivers/dma/qcom/Makefile index 6bf9267..4bfc38b 100644 --- a/drivers/dma/qcom/Makefile +++ b/drivers/dma/qcom/Makefile @@ -2,4 +2,4 @@ obj-$(CONFIG_QCOM_BAM_DMA) += bam_dma.o obj-$(CONFIG_QCOM_HIDMA_MGMT) += hdma_mgmt.o hdma_mgmt-objs := hidma_mgmt.o hidma_mgmt_sys.o obj-$(CONFIG_QCOM_HIDMA) += hdma.o -hdma-objs := hidma_ll.o hidma.o +hdma-objs := hidma_ll.o hidma.o hidma_dbg.o diff --git a/drivers/dma/qcom/hidma.c b/drivers/dma/qcom/hidma.c index f8960f1..8972508 100644 --- a/drivers/dma/qcom/hidma.c +++ b/drivers/dma/qcom/hidma.c @@ -1,7 +1,7 @@ /* * Qualcomm Technologies HIDMA DMA engine interface * - * Copyright (c) 2015, The Linux Foundation. All rights reserved. + * Copyright (c) 2015-2016, The Linux Foundation. All rights reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 and @@ -681,6 +681,7 @@ static int hidma_probe(struct platform_device *pdev) dmadev->irq = chirq; tasklet_init(&dmadev->task, hidma_issue_task, (unsigned long)dmadev); + hidma_debug_init(dmadev); dev_info(&pdev->dev, "HI-DMA engine driver registration complete\n"); platform_set_drvdata(pdev, dmadev); pm_runtime_mark_last_busy(dmadev->ddev.dev); @@ -689,6 +690,7 @@ static int hidma_probe(struct platform_device *pdev) return 0; uninit: + hidma_debug_uninit(dmadev); hidma_ll_uninit(dmadev->lldev); dmafree: if (dmadev) @@ -706,6 +708,7 @@ static int hidma_remove(struct platform_device *pdev) pm_runtime_get_sync(dmadev->ddev.dev); dma_async_device_unregister(&dmadev->ddev); devm_free_irq(dmadev->ddev.dev, dmadev->irq, dmadev->lldev); + hidma_debug_uninit(dmadev); hidma_ll_uninit(dmadev->lldev); hidma_free(dmadev); diff --git a/drivers/dma/qcom/hidma.h b/drivers/dma/qcom/hidma.h index c5eea65..22806a2 100644 --- a/drivers/dma/qcom/hidma.h +++ b/drivers/dma/qcom/hidma.h @@ -157,4 +157,6 @@ int hidma_ll_uninit(struct hidma_lldev *llhndl); irqreturn_t hidma_ll_inthandler(int irq, void *arg); void hidma_cleanup_pending_tre(struct hidma_lldev *llhndl, u8 err_info, u8 err_code); +int hidma_debug_init(struct hidma_dev *dmadev); +void hidma_debug_uninit(struct hidma_dev *dmadev); #endif diff --git a/drivers/dma/qcom/hidma_dbg.c b/drivers/dma/qcom/hidma_dbg.c new file mode 100644 index 0000000..68e779e --- /dev/null +++ b/drivers/dma/qcom/hidma_dbg.c @@ -0,0 +1,219 @@ +/* + * Qualcomm Technologies HIDMA debug file + * + * Copyright (c) 2015-2016, The Linux Foundation. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 and + * only version 2 as published by the Free Software Foundation. + * + * 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/debugfs.h> +#include <linux/device.h> +#include <linux/list.h> +#include <linux/pm_runtime.h> + +#include "hidma.h" + +static void hidma_ll_chstats(struct seq_file *s, void *llhndl, u32 tre_ch) +{ + struct hidma_lldev *lldev = llhndl; + struct hidma_tre *tre; + u32 length; + dma_addr_t src_start; + dma_addr_t dest_start; + u32 *tre_local; + + if (tre_ch >= lldev->nr_tres) { + dev_err(lldev->dev, "invalid TRE number in chstats:%d", tre_ch); + return; + } + tre = &lldev->trepool[tre_ch]; + seq_printf(s, "------Channel %d -----\n", tre_ch); + seq_printf(s, "allocated=%d\n", atomic_read(&tre->allocated)); + seq_printf(s, "queued = 0x%x\n", tre->queued); + seq_printf(s, "err_info = 0x%x\n", + lldev->tx_status_list[tre->idx].err_info); + seq_printf(s, "err_code = 0x%x\n", + lldev->tx_status_list[tre->idx].err_code); + seq_printf(s, "status = 0x%x\n", tre->status); + seq_printf(s, "idx = 0x%x\n", tre->idx); + seq_printf(s, "dma_sig = 0x%x\n", tre->dma_sig); + seq_printf(s, "dev_name=%s\n", tre->dev_name); + seq_printf(s, "callback=%p\n", tre->callback); + seq_printf(s, "data=%p\n", tre->data); + seq_printf(s, "tre_index = 0x%x\n", tre->tre_index); + + tre_local = &tre->tre_local[0]; + src_start = tre_local[HIDMA_TRE_SRC_LOW_IDX]; + src_start = ((u64) (tre_local[HIDMA_TRE_SRC_HI_IDX]) << 32) + src_start; + dest_start = tre_local[HIDMA_TRE_DEST_LOW_IDX]; + dest_start += ((u64) (tre_local[HIDMA_TRE_DEST_HI_IDX]) << 32); + length = tre_local[HIDMA_TRE_LEN_IDX]; + + seq_printf(s, "src=%pap\n", &src_start); + seq_printf(s, "dest=%pap\n", &dest_start); + seq_printf(s, "length = 0x%x\n", length); +} + +static void hidma_ll_devstats(struct seq_file *s, void *llhndl) +{ + struct hidma_lldev *lldev = llhndl; + + seq_puts(s, "------Device -----\n"); + seq_printf(s, "lldev init = 0x%x\n", lldev->initialized); + seq_printf(s, "trch_state = 0x%x\n", lldev->trch_state); + seq_printf(s, "evch_state = 0x%x\n", lldev->evch_state); + seq_printf(s, "chidx = 0x%x\n", lldev->chidx); + seq_printf(s, "nr_tres = 0x%x\n", lldev->nr_tres); + seq_printf(s, "trca=%p\n", lldev->trca); + seq_printf(s, "tre_ring=%p\n", lldev->tre_ring); + seq_printf(s, "tre_ring_handle=%pap\n", &lldev->tre_ring_handle); + seq_printf(s, "tre_ring_size = 0x%x\n", lldev->tre_ring_size); + seq_printf(s, "tre_processed_off = 0x%x\n", lldev->tre_processed_off); + seq_printf(s, "pending_tre_count=%d\n", lldev->pending_tre_count); + seq_printf(s, "evca=%p\n", lldev->evca); + seq_printf(s, "evre_ring=%p\n", lldev->evre_ring); + seq_printf(s, "evre_ring_handle=%pap\n", &lldev->evre_ring_handle); + seq_printf(s, "evre_ring_size = 0x%x\n", lldev->evre_ring_size); + seq_printf(s, "evre_processed_off = 0x%x\n", lldev->evre_processed_off); + seq_printf(s, "tre_write_offset = 0x%x\n", lldev->tre_write_offset); +} + +/* + * hidma_chan_stats: display HIDMA channel statistics + * + * Display the statistics for the current HIDMA virtual channel device. + */ +static int hidma_chan_stats(struct seq_file *s, void *unused) +{ + struct hidma_chan *mchan = s->private; + struct hidma_desc *mdesc; + struct hidma_dev *dmadev = mchan->dmadev; + + pm_runtime_get_sync(dmadev->ddev.dev); + seq_printf(s, "paused=%u\n", mchan->paused); + seq_printf(s, "dma_sig=%u\n", mchan->dma_sig); + seq_puts(s, "prepared\n"); + list_for_each_entry(mdesc, &mchan->prepared, node) + hidma_ll_chstats(s, mchan->dmadev->lldev, mdesc->tre_ch); + + seq_puts(s, "active\n"); + list_for_each_entry(mdesc, &mchan->active, node) + hidma_ll_chstats(s, mchan->dmadev->lldev, mdesc->tre_ch); + + seq_puts(s, "completed\n"); + list_for_each_entry(mdesc, &mchan->completed, node) + hidma_ll_chstats(s, mchan->dmadev->lldev, mdesc->tre_ch); + + hidma_ll_devstats(s, mchan->dmadev->lldev); + pm_runtime_mark_last_busy(dmadev->ddev.dev); + pm_runtime_put_autosuspend(dmadev->ddev.dev); + return 0; +} + +/* + * hidma_dma_info: display HIDMA device info + * + * Display the info for the current HIDMA device. + */ +static int hidma_dma_info(struct seq_file *s, void *unused) +{ + struct hidma_dev *dmadev = s->private; + resource_size_t sz; + + seq_printf(s, "nr_descriptors=%d\n", dmadev->nr_descriptors); + seq_printf(s, "dev_trca=%p\n", &dmadev->dev_trca); + seq_printf(s, "dev_trca_phys=%pa\n", &dmadev->trca_resource->start); + sz = resource_size(dmadev->trca_resource); + seq_printf(s, "dev_trca_size=%pa\n", &sz); + seq_printf(s, "dev_evca=%p\n", &dmadev->dev_evca); + seq_printf(s, "dev_evca_phys=%pa\n", &dmadev->evca_resource->start); + sz = resource_size(dmadev->evca_resource); + seq_printf(s, "dev_evca_size=%pa\n", &sz); + return 0; +} + +static int hidma_chan_stats_open(struct inode *inode, struct file *file) +{ + return single_open(file, hidma_chan_stats, inode->i_private); +} + +static int hidma_dma_info_open(struct inode *inode, struct file *file) +{ + return single_open(file, hidma_dma_info, inode->i_private); +} + +static const struct file_operations hidma_chan_fops = { + .open = hidma_chan_stats_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static const struct file_operations hidma_dma_fops = { + .open = hidma_dma_info_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +void hidma_debug_uninit(struct hidma_dev *dmadev) +{ + debugfs_remove_recursive(dmadev->debugfs); + debugfs_remove_recursive(dmadev->stats); +} + +int hidma_debug_init(struct hidma_dev *dmadev) +{ + int rc = 0; + int chidx = 0; + struct list_head *position = NULL; + + dmadev->debugfs = debugfs_create_dir(dev_name(dmadev->ddev.dev), NULL); + if (!dmadev->debugfs) { + rc = -ENODEV; + return rc; + } + + /* walk through the virtual channel list */ + list_for_each(position, &dmadev->ddev.channels) { + struct hidma_chan *chan; + + chan = list_entry(position, struct hidma_chan, + chan.device_node); + sprintf(chan->dbg_name, "chan%d", chidx); + chan->debugfs = debugfs_create_dir(chan->dbg_name, + dmadev->debugfs); + if (!chan->debugfs) { + rc = -ENOMEM; + goto cleanup; + } + chan->stats = debugfs_create_file("stats", S_IRUGO, + chan->debugfs, chan, + &hidma_chan_fops); + if (!chan->stats) { + rc = -ENOMEM; + goto cleanup; + } + chidx++; + } + + dmadev->stats = debugfs_create_file("stats", S_IRUGO, + dmadev->debugfs, dmadev, + &hidma_dma_fops); + if (!dmadev->stats) { + rc = -ENOMEM; + goto cleanup; + } + + return 0; +cleanup: + hidma_debug_uninit(dmadev); + return rc; +}
Add debugfs hooks for debugging the execution behavior of the DMA channel. The debugfs hooks get initialized by the probe function and uninitialized by the remove function. A stats file is created in debugfs. The stats file will show the information about each HIDMA channel as well as each asynchronous job queued and completed at a given time. Signed-off-by: Sinan Kaya <okaya@codeaurora.org> --- drivers/dma/qcom/Makefile | 2 +- drivers/dma/qcom/hidma.c | 5 +- drivers/dma/qcom/hidma.h | 2 + drivers/dma/qcom/hidma_dbg.c | 219 +++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 226 insertions(+), 2 deletions(-) create mode 100644 drivers/dma/qcom/hidma_dbg.c