RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
> Hi, > > "Du, Changbin"writes: > >> > At last, comparing with the FIFO/Queue info, I think software transfer > >> > Requests list, TRBs info, EVENTs history are much more useful for > >> debugging > >> > the driver. If you can also add these info to each EP folder, that is > awesome! > >> > :) > >> > >> I'll think about adding these but for the lifetime of requests and trbs > >> and events, etc, we have tracepoints for that. I usually do the > >> following when debugging: > >> > >> # mount -t debugfs none /sys/kernel/debug > >> # cd /sys/kernel/debug/tracing > >> # echo 2048 > buffer_size_kb > >> # echo 1 > events/dwc3/enable > >> > >> (do something to break it) > >> > >> # cp trace /mnt/sdcard # or something like that > >> > >> then read the file. You can make it as large or as small as you like > >> (given some constraints, of course ;-) but I've had no issues allocating > >> 128MiB in the past. > >> > >> -- > >> Balbi > > > > Thanks for the sharing, this is a good approach to capture dynamic > > behaviors. But a dump of current state has below advantages: > > 1. a quick view for the pending transfers. Then we can quickly > > checking the transfer status. > > 2. no side-effect. This is important in some case. We usually > > encounter some transfer issues but very hard to reproduce > > it. But we cannot enable trace all the time since performance > > concern. Then I thought it was so great if I can have a look for > > the trb status. :) > > yeah, okay. We can definitely add "current state" of almost anything, > but if you need history, then debugfs is not the best interface and I'd > point you to tracepoints ;-) > > I'll think about how I can add TRB state, seems like we'd need to dump > the entire endpoint ring, and that's 256 TRBs per endpoint :-p Then we > also need to know endpoint's dequeue and enqueue pointer. Oh well, let > me get this first setup of files out of the way, then we can add more > later much more easily. > > -- > Balbi Okay, things need finish step by step. Thank you, Balbi. ( �b- �b)つロ Best Regards, Du, Changbin
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
> Hi, > > "Du, Changbin" writes: > >> > At last, comparing with the FIFO/Queue info, I think software transfer > >> > Requests list, TRBs info, EVENTs history are much more useful for > >> debugging > >> > the driver. If you can also add these info to each EP folder, that is > awesome! > >> > :) > >> > >> I'll think about adding these but for the lifetime of requests and trbs > >> and events, etc, we have tracepoints for that. I usually do the > >> following when debugging: > >> > >> # mount -t debugfs none /sys/kernel/debug > >> # cd /sys/kernel/debug/tracing > >> # echo 2048 > buffer_size_kb > >> # echo 1 > events/dwc3/enable > >> > >> (do something to break it) > >> > >> # cp trace /mnt/sdcard # or something like that > >> > >> then read the file. You can make it as large or as small as you like > >> (given some constraints, of course ;-) but I've had no issues allocating > >> 128MiB in the past. > >> > >> -- > >> Balbi > > > > Thanks for the sharing, this is a good approach to capture dynamic > > behaviors. But a dump of current state has below advantages: > > 1. a quick view for the pending transfers. Then we can quickly > > checking the transfer status. > > 2. no side-effect. This is important in some case. We usually > > encounter some transfer issues but very hard to reproduce > > it. But we cannot enable trace all the time since performance > > concern. Then I thought it was so great if I can have a look for > > the trb status. :) > > yeah, okay. We can definitely add "current state" of almost anything, > but if you need history, then debugfs is not the best interface and I'd > point you to tracepoints ;-) > > I'll think about how I can add TRB state, seems like we'd need to dump > the entire endpoint ring, and that's 256 TRBs per endpoint :-p Then we > also need to know endpoint's dequeue and enqueue pointer. Oh well, let > me get this first setup of files out of the way, then we can add more > later much more easily. > > -- > Balbi Okay, things need finish step by step. Thank you, Balbi. ( �b- �b)つロ Best Regards, Du, Changbin
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hi, "Du, Changbin"writes: >> > At last, comparing with the FIFO/Queue info, I think software transfer >> > Requests list, TRBs info, EVENTs history are much more useful for >> debugging >> > the driver. If you can also add these info to each EP folder, that is >> > awesome! >> > :) >> >> I'll think about adding these but for the lifetime of requests and trbs >> and events, etc, we have tracepoints for that. I usually do the >> following when debugging: >> >> # mount -t debugfs none /sys/kernel/debug >> # cd /sys/kernel/debug/tracing >> # echo 2048 > buffer_size_kb >> # echo 1 > events/dwc3/enable >> >> (do something to break it) >> >> # cp trace /mnt/sdcard # or something like that >> >> then read the file. You can make it as large or as small as you like >> (given some constraints, of course ;-) but I've had no issues allocating >> 128MiB in the past. >> >> -- >> Balbi > > Thanks for the sharing, this is a good approach to capture dynamic > behaviors. But a dump of current state has below advantages: > 1. a quick view for the pending transfers. Then we can quickly > checking the transfer status. > 2. no side-effect. This is important in some case. We usually > encounter some transfer issues but very hard to reproduce > it. But we cannot enable trace all the time since performance > concern. Then I thought it was so great if I can have a look for > the trb status. :) yeah, okay. We can definitely add "current state" of almost anything, but if you need history, then debugfs is not the best interface and I'd point you to tracepoints ;-) I'll think about how I can add TRB state, seems like we'd need to dump the entire endpoint ring, and that's 256 TRBs per endpoint :-p Then we also need to know endpoint's dequeue and enqueue pointer. Oh well, let me get this first setup of files out of the way, then we can add more later much more easily. -- balbi signature.asc Description: PGP signature
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hi, "Du, Changbin" writes: >> > At last, comparing with the FIFO/Queue info, I think software transfer >> > Requests list, TRBs info, EVENTs history are much more useful for >> debugging >> > the driver. If you can also add these info to each EP folder, that is >> > awesome! >> > :) >> >> I'll think about adding these but for the lifetime of requests and trbs >> and events, etc, we have tracepoints for that. I usually do the >> following when debugging: >> >> # mount -t debugfs none /sys/kernel/debug >> # cd /sys/kernel/debug/tracing >> # echo 2048 > buffer_size_kb >> # echo 1 > events/dwc3/enable >> >> (do something to break it) >> >> # cp trace /mnt/sdcard # or something like that >> >> then read the file. You can make it as large or as small as you like >> (given some constraints, of course ;-) but I've had no issues allocating >> 128MiB in the past. >> >> -- >> Balbi > > Thanks for the sharing, this is a good approach to capture dynamic > behaviors. But a dump of current state has below advantages: > 1. a quick view for the pending transfers. Then we can quickly > checking the transfer status. > 2. no side-effect. This is important in some case. We usually > encounter some transfer issues but very hard to reproduce > it. But we cannot enable trace all the time since performance > concern. Then I thought it was so great if I can have a look for > the trb status. :) yeah, okay. We can definitely add "current state" of almost anything, but if you need history, then debugfs is not the best interface and I'd point you to tracepoints ;-) I'll think about how I can add TRB state, seems like we'd need to dump the entire endpoint ring, and that's 256 TRBs per endpoint :-p Then we also need to know endpoint's dequeue and enqueue pointer. Oh well, let me get this first setup of files out of the way, then we can add more later much more easily. -- balbi signature.asc Description: PGP signature
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
> > At last, comparing with the FIFO/Queue info, I think software transfer > > Requests list, TRBs info, EVENTs history are much more useful for > debugging > > the driver. If you can also add these info to each EP folder, that is > > awesome! > > :) > > I'll think about adding these but for the lifetime of requests and trbs > and events, etc, we have tracepoints for that. I usually do the > following when debugging: > > # mount -t debugfs none /sys/kernel/debug > # cd /sys/kernel/debug/tracing > # echo 2048 > buffer_size_kb > # echo 1 > events/dwc3/enable > > (do something to break it) > > # cp trace /mnt/sdcard # or something like that > > then read the file. You can make it as large or as small as you like > (given some constraints, of course ;-) but I've had no issues allocating > 128MiB in the past. > > -- > Balbi Thanks for the sharing, this is a good approach to capture dynamic behaviors. But a dump of current state has below advantages: 1. a quick view for the pending transfers. Then we can quickly checking the transfer status. 2. no side-effect. This is important in some case. We usually encounter some transfer issues but very hard to reproduce it. But we cannot enable trace all the time since performance concern. Then I thought it was so great if I can have a look for the trb status. :) Best Regards, Du, Changbin
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
> > At last, comparing with the FIFO/Queue info, I think software transfer > > Requests list, TRBs info, EVENTs history are much more useful for > debugging > > the driver. If you can also add these info to each EP folder, that is > > awesome! > > :) > > I'll think about adding these but for the lifetime of requests and trbs > and events, etc, we have tracepoints for that. I usually do the > following when debugging: > > # mount -t debugfs none /sys/kernel/debug > # cd /sys/kernel/debug/tracing > # echo 2048 > buffer_size_kb > # echo 1 > events/dwc3/enable > > (do something to break it) > > # cp trace /mnt/sdcard # or something like that > > then read the file. You can make it as large or as small as you like > (given some constraints, of course ;-) but I've had no issues allocating > 128MiB in the past. > > -- > Balbi Thanks for the sharing, this is a good approach to capture dynamic behaviors. But a dump of current state has below advantages: 1. a quick view for the pending transfers. Then we can quickly checking the transfer status. 2. no side-effect. This is important in some case. We usually encounter some transfer issues but very hard to reproduce it. But we cannot enable trace all the time since performance concern. Then I thought it was so great if I can have a look for the trb status. :) Best Regards, Du, Changbin
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hi, "Du, Changbin"writes: > Hi, Balbi. > > Feel free to change it, I may not have enough time on this currently. > "per-endpoint directory" is great idea, then we do not need find out > wanted info from one big file, but just go to specific dir. that was the idea, glad you liked it ;-) > Btw, I'd mention that not all out ep has a rx fifo. So in my original patch, yeah, rx fifos are dynamically allocated by the HW itself, AFAICT. > not all FIFO/Queue info are valid. We need pick out the real info we need. > And I didn't find any method to read the FIFO map. > > At last, comparing with the FIFO/Queue info, I think software transfer > Requests list, TRBs info, EVENTs history are much more useful for debugging > the driver. If you can also add these info to each EP folder, that is awesome! > :) I'll think about adding these but for the lifetime of requests and trbs and events, etc, we have tracepoints for that. I usually do the following when debugging: # mount -t debugfs none /sys/kernel/debug # cd /sys/kernel/debug/tracing # echo 2048 > buffer_size_kb # echo 1 > events/dwc3/enable (do something to break it) # cp trace /mnt/sdcard # or something like that then read the file. You can make it as large or as small as you like (given some constraints, of course ;-) but I've had no issues allocating 128MiB in the past. -- balbi signature.asc Description: PGP signature
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hi, "Du, Changbin" writes: > Hi, Balbi. > > Feel free to change it, I may not have enough time on this currently. > "per-endpoint directory" is great idea, then we do not need find out > wanted info from one big file, but just go to specific dir. that was the idea, glad you liked it ;-) > Btw, I'd mention that not all out ep has a rx fifo. So in my original patch, yeah, rx fifos are dynamically allocated by the HW itself, AFAICT. > not all FIFO/Queue info are valid. We need pick out the real info we need. > And I didn't find any method to read the FIFO map. > > At last, comparing with the FIFO/Queue info, I think software transfer > Requests list, TRBs info, EVENTs history are much more useful for debugging > the driver. If you can also add these info to each EP folder, that is awesome! > :) I'll think about adding these but for the lifetime of requests and trbs and events, etc, we have tracepoints for that. I usually do the following when debugging: # mount -t debugfs none /sys/kernel/debug # cd /sys/kernel/debug/tracing # echo 2048 > buffer_size_kb # echo 1 > events/dwc3/enable (do something to break it) # cp trace /mnt/sdcard # or something like that then read the file. You can make it as large or as small as you like (given some constraints, of course ;-) but I've had no issues allocating 128MiB in the past. -- balbi signature.asc Description: PGP signature
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hi, Balbi. Feel free to change it, I may not have enough time on this currently. "per-endpoint directory" is great idea, then we do not need find out wanted info from one big file, but just go to specific dir. Btw, I'd mention that not all out ep has a rx fifo. So in my original patch, not all FIFO/Queue info are valid. We need pick out the real info we need. And I didn't find any method to read the FIFO map. At last, comparing with the FIFO/Queue info, I think software transfer Requests list, TRBs info, EVENTs history are much more useful for debugging the driver. If you can also add these info to each EP folder, that is awesome! :) Best Regards, Du, Changbin > -Original Message- > From: Felipe Balbi [mailto:ba...@kernel.org] > Sent: Thursday, April 14, 2016 4:03 PM > To: Du, Changbin <changbin...@intel.com> > Cc: gre...@linuxfoundation.org; linux-...@vger.kernel.org; linux- > ker...@vger.kernel.org; Du, Changbin <changbin...@intel.com> > Subject: Re: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump > FIFO/Queue available space > > > Hi, > > changbin...@intel.com writes: > > From: "Du, Changbin" <changbin...@intel.com> > > > > For DWC3 USB controller, the Global Debug Queue/FIFO Space Available > > Register(GDBGFIFOSPACE) can be used to dump FIFO/Queue available > space. > > This can be used to check some special issues, like whether data is > > successfully copied from memory to fifo when a trb is blocked. > > > > Signed-off-by: Du, Changbin <changbin...@intel.com> > > --- > > v4: > > Do not fail silently, but print error. > > > > --- > > drivers/usb/dwc3/core.h| 5 + > > drivers/usb/dwc3/debugfs.c | 44 > > > 2 files changed, 49 insertions(+) > > > > diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h > > index 6254b2f..899cf76 100644 > > --- a/drivers/usb/dwc3/core.h > > +++ b/drivers/usb/dwc3/core.h > > @@ -348,6 +348,11 @@ > > #define DWC3_DSTS_LOWSPEED (2 << 0) > > #define DWC3_DSTS_FULLSPEED1 (3 << 0) > > > > +/* Global Debug Queue/FIFO Space Available Register */ > > +#define DWC3_GDBGFIFOSPACE_NUM(x) (((x) << 0) & 0x1F) > > +#define DWC3_GDBGFIFOSPACE_TYPE(x) (((x) << 5) & 0xE0) > > +#define DWC3_GDBGFIFOSPACE_GET_SPACE(x)(((x) >> 16) & 0x) > > we always use lower case hex numbers. Also, databook refers to top 16 > bits as "Space Avaiable" so I'd prefer that you called this macro: > > DWC3_GDBGFIFOSPACE_SPACE_AVAILABLE(x) > > as that will aid with grepping > > > diff --git a/drivers/usb/dwc3/debugfs.c b/drivers/usb/dwc3/debugfs.c > > index 615d4dc..83e5bc1 100644 > > --- a/drivers/usb/dwc3/debugfs.c > > +++ b/drivers/usb/dwc3/debugfs.c > > @@ -426,6 +426,45 @@ static const struct file_operations > dwc3_mode_fops = { > > .release= single_release, > > }; > > > > +static int dwc3_fifo_show(struct seq_file *s, void *unused) > > you call this file 'fifo' however you print more than just the > fifos. You printk the request queues, info queue, descriptor fetch > queue, event queue and protocol status queue. > > It seems, to me, you're trying to do way too much in a single file. > > > +{ > > + struct dwc3 *dwc = s->private; > > + unsigned long flags; > > + unsigned inttype, index; > > + const char *name; > > + u32 reg; > > + > > + static const char * const fifo_names[] = { > > + "TxFIFO", "RxFIFO", "TxReqQ", "RxReqQ", "RxInfoQ", > > + "DescFetchQ", "EventQ", "ProtocolStatusQ"}; > > + spin_lock_irqsave(>lock, flags); > > + for (type = 0; type < 8; type++) { > > type < ARRAY_SIZE(fifo_names) ?? > > > + name = fifo_names[type]; > > + for (index = 0; index < 32; index++) { > > not *all* dwc3 implementations enable all 32 endpoints, that's why we > have dwc->num_endpoints > > > + dwc3_writel(dwc->regs, DWC3_GDBGFIFOSPACE, > > + DWC3_GDBGFIFOSPACE_NUM(index) | > > + DWC3_GDBGFIFOSPACE_TYPE(type)); > > + reg = dwc3_readl(dwc->regs, > DWC3_GDBGFIFOSPACE); > > this writel() followed by a readl() could be a nice little helper > function in core.c. Remember that we need the FIFO Space and link state
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hi, Balbi. Feel free to change it, I may not have enough time on this currently. "per-endpoint directory" is great idea, then we do not need find out wanted info from one big file, but just go to specific dir. Btw, I'd mention that not all out ep has a rx fifo. So in my original patch, not all FIFO/Queue info are valid. We need pick out the real info we need. And I didn't find any method to read the FIFO map. At last, comparing with the FIFO/Queue info, I think software transfer Requests list, TRBs info, EVENTs history are much more useful for debugging the driver. If you can also add these info to each EP folder, that is awesome! :) Best Regards, Du, Changbin > -Original Message- > From: Felipe Balbi [mailto:ba...@kernel.org] > Sent: Thursday, April 14, 2016 4:03 PM > To: Du, Changbin > Cc: gre...@linuxfoundation.org; linux-...@vger.kernel.org; linux- > ker...@vger.kernel.org; Du, Changbin > Subject: Re: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump > FIFO/Queue available space > > > Hi, > > changbin...@intel.com writes: > > From: "Du, Changbin" > > > > For DWC3 USB controller, the Global Debug Queue/FIFO Space Available > > Register(GDBGFIFOSPACE) can be used to dump FIFO/Queue available > space. > > This can be used to check some special issues, like whether data is > > successfully copied from memory to fifo when a trb is blocked. > > > > Signed-off-by: Du, Changbin > > --- > > v4: > > Do not fail silently, but print error. > > > > --- > > drivers/usb/dwc3/core.h| 5 + > > drivers/usb/dwc3/debugfs.c | 44 > > > 2 files changed, 49 insertions(+) > > > > diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h > > index 6254b2f..899cf76 100644 > > --- a/drivers/usb/dwc3/core.h > > +++ b/drivers/usb/dwc3/core.h > > @@ -348,6 +348,11 @@ > > #define DWC3_DSTS_LOWSPEED (2 << 0) > > #define DWC3_DSTS_FULLSPEED1 (3 << 0) > > > > +/* Global Debug Queue/FIFO Space Available Register */ > > +#define DWC3_GDBGFIFOSPACE_NUM(x) (((x) << 0) & 0x1F) > > +#define DWC3_GDBGFIFOSPACE_TYPE(x) (((x) << 5) & 0xE0) > > +#define DWC3_GDBGFIFOSPACE_GET_SPACE(x)(((x) >> 16) & 0x) > > we always use lower case hex numbers. Also, databook refers to top 16 > bits as "Space Avaiable" so I'd prefer that you called this macro: > > DWC3_GDBGFIFOSPACE_SPACE_AVAILABLE(x) > > as that will aid with grepping > > > diff --git a/drivers/usb/dwc3/debugfs.c b/drivers/usb/dwc3/debugfs.c > > index 615d4dc..83e5bc1 100644 > > --- a/drivers/usb/dwc3/debugfs.c > > +++ b/drivers/usb/dwc3/debugfs.c > > @@ -426,6 +426,45 @@ static const struct file_operations > dwc3_mode_fops = { > > .release= single_release, > > }; > > > > +static int dwc3_fifo_show(struct seq_file *s, void *unused) > > you call this file 'fifo' however you print more than just the > fifos. You printk the request queues, info queue, descriptor fetch > queue, event queue and protocol status queue. > > It seems, to me, you're trying to do way too much in a single file. > > > +{ > > + struct dwc3 *dwc = s->private; > > + unsigned long flags; > > + unsigned inttype, index; > > + const char *name; > > + u32 reg; > > + > > + static const char * const fifo_names[] = { > > + "TxFIFO", "RxFIFO", "TxReqQ", "RxReqQ", "RxInfoQ", > > + "DescFetchQ", "EventQ", "ProtocolStatusQ"}; > > + spin_lock_irqsave(>lock, flags); > > + for (type = 0; type < 8; type++) { > > type < ARRAY_SIZE(fifo_names) ?? > > > + name = fifo_names[type]; > > + for (index = 0; index < 32; index++) { > > not *all* dwc3 implementations enable all 32 endpoints, that's why we > have dwc->num_endpoints > > > + dwc3_writel(dwc->regs, DWC3_GDBGFIFOSPACE, > > + DWC3_GDBGFIFOSPACE_NUM(index) | > > + DWC3_GDBGFIFOSPACE_TYPE(type)); > > + reg = dwc3_readl(dwc->regs, > DWC3_GDBGFIFOSPACE); > > this writel() followed by a readl() could be a nice little helper > function in core.c. Remember that we need the FIFO Space and link state > to make sure we're allowed to start a transfer ;-) > > I'd suggesting adding the following to core.c: >
Re: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hi, changbin...@intel.com writes: > From: "Du, Changbin"> > For DWC3 USB controller, the Global Debug Queue/FIFO Space Available > Register(GDBGFIFOSPACE) can be used to dump FIFO/Queue available space. > This can be used to check some special issues, like whether data is > successfully copied from memory to fifo when a trb is blocked. > > Signed-off-by: Du, Changbin > --- > v4: > Do not fail silently, but print error. > > --- > drivers/usb/dwc3/core.h| 5 + > drivers/usb/dwc3/debugfs.c | 44 > 2 files changed, 49 insertions(+) > > diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h > index 6254b2f..899cf76 100644 > --- a/drivers/usb/dwc3/core.h > +++ b/drivers/usb/dwc3/core.h > @@ -348,6 +348,11 @@ > #define DWC3_DSTS_LOWSPEED (2 << 0) > #define DWC3_DSTS_FULLSPEED1 (3 << 0) > > +/* Global Debug Queue/FIFO Space Available Register */ > +#define DWC3_GDBGFIFOSPACE_NUM(x)(((x) << 0) & 0x1F) > +#define DWC3_GDBGFIFOSPACE_TYPE(x) (((x) << 5) & 0xE0) > +#define DWC3_GDBGFIFOSPACE_GET_SPACE(x) (((x) >> 16) & 0x) we always use lower case hex numbers. Also, databook refers to top 16 bits as "Space Avaiable" so I'd prefer that you called this macro: DWC3_GDBGFIFOSPACE_SPACE_AVAILABLE(x) as that will aid with grepping > diff --git a/drivers/usb/dwc3/debugfs.c b/drivers/usb/dwc3/debugfs.c > index 615d4dc..83e5bc1 100644 > --- a/drivers/usb/dwc3/debugfs.c > +++ b/drivers/usb/dwc3/debugfs.c > @@ -426,6 +426,45 @@ static const struct file_operations dwc3_mode_fops = { > .release= single_release, > }; > > +static int dwc3_fifo_show(struct seq_file *s, void *unused) you call this file 'fifo' however you print more than just the fifos. You printk the request queues, info queue, descriptor fetch queue, event queue and protocol status queue. It seems, to me, you're trying to do way too much in a single file. > +{ > + struct dwc3 *dwc = s->private; > + unsigned long flags; > + unsigned inttype, index; > + const char *name; > + u32 reg; > + > + static const char * const fifo_names[] = { > + "TxFIFO", "RxFIFO", "TxReqQ", "RxReqQ", "RxInfoQ", > + "DescFetchQ", "EventQ", "ProtocolStatusQ"}; > + spin_lock_irqsave(>lock, flags); > + for (type = 0; type < 8; type++) { type < ARRAY_SIZE(fifo_names) ?? > + name = fifo_names[type]; > + for (index = 0; index < 32; index++) { not *all* dwc3 implementations enable all 32 endpoints, that's why we have dwc->num_endpoints > + dwc3_writel(dwc->regs, DWC3_GDBGFIFOSPACE, > + DWC3_GDBGFIFOSPACE_NUM(index) | > + DWC3_GDBGFIFOSPACE_TYPE(type)); > + reg = dwc3_readl(dwc->regs, DWC3_GDBGFIFOSPACE); this writel() followed by a readl() could be a nice little helper function in core.c. Remember that we need the FIFO Space and link state to make sure we're allowed to start a transfer ;-) I'd suggesting adding the following to core.c: int dwc3_core_fifo_space(struct dwc3_ep *dep, unsigned int type) { u32 reg; dwc3_writel(dwc->regs, DWC3_GDBGFIFOSPACE, DWC3_GDBGFIFOSPACE_NUM(dep->number) | DWC3_GDBGFIFOSPACE_TYPE(type)); reg = dwc3_readl(dwc->regs, DWC3_GDBGFIFOSPACE); return DWC3_GDBGFIFOSPACE_SPACE_AVAILABLE(reg); } then, add the prototype to core.h. (both of these in a separate patch). Then $subject just uses that helper to print out the data you want to print. I still think we might be better off with one file per type (iow TX fifo file, RX fifo file, TX request queue file, and so on). In fact, we could actually have a per-endpoint directory where all of these details are printed out in a much more structured form. If you want, I can work on that, no problem, but you're also welcome to implement it yourself and I'd be around to help as needed. In summary, we should have our debugfs looking like so: ./ep0 ./ep0/descriptor_fetch_queue ./ep0/event_queue ./ep0/info_queue ./ep0/protocol_status_queue ./ep0/request_queue ./ep0/rx_fifo ./ep0/rx_request_queue ./ep0/tx_fifo ./ep0/tx_request_queue ./ep1 ./ep10 ./ep10/descriptor_fetch_queue ./ep10/event_queue ./ep10/info_queue ./ep10/protocol_status_queue ./ep10/request_queue ./ep10/rx_fifo ./ep10/rx_request_queue ./ep10/tx_fifo ./ep10/tx_request_queue ./ep11 ./ep11/descriptor_fetch_queue ./ep11/event_queue ./ep11/info_queue ./ep11/protocol_status_queue ./ep11/request_queue ./ep11/rx_fifo ./ep11/rx_request_queue ./ep11/tx_fifo ./ep11/tx_request_queue ./ep12 ./ep12/descriptor_fetch_queue ./ep12/event_queue ./ep12/info_queue ./ep12/protocol_status_queue ./ep12/request_queue ./ep12/rx_fifo ./ep12/rx_request_queue ./ep12/tx_fifo ./ep12/tx_request_queue
Re: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hi, changbin...@intel.com writes: > From: "Du, Changbin" > > For DWC3 USB controller, the Global Debug Queue/FIFO Space Available > Register(GDBGFIFOSPACE) can be used to dump FIFO/Queue available space. > This can be used to check some special issues, like whether data is > successfully copied from memory to fifo when a trb is blocked. > > Signed-off-by: Du, Changbin > --- > v4: > Do not fail silently, but print error. > > --- > drivers/usb/dwc3/core.h| 5 + > drivers/usb/dwc3/debugfs.c | 44 > 2 files changed, 49 insertions(+) > > diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h > index 6254b2f..899cf76 100644 > --- a/drivers/usb/dwc3/core.h > +++ b/drivers/usb/dwc3/core.h > @@ -348,6 +348,11 @@ > #define DWC3_DSTS_LOWSPEED (2 << 0) > #define DWC3_DSTS_FULLSPEED1 (3 << 0) > > +/* Global Debug Queue/FIFO Space Available Register */ > +#define DWC3_GDBGFIFOSPACE_NUM(x)(((x) << 0) & 0x1F) > +#define DWC3_GDBGFIFOSPACE_TYPE(x) (((x) << 5) & 0xE0) > +#define DWC3_GDBGFIFOSPACE_GET_SPACE(x) (((x) >> 16) & 0x) we always use lower case hex numbers. Also, databook refers to top 16 bits as "Space Avaiable" so I'd prefer that you called this macro: DWC3_GDBGFIFOSPACE_SPACE_AVAILABLE(x) as that will aid with grepping > diff --git a/drivers/usb/dwc3/debugfs.c b/drivers/usb/dwc3/debugfs.c > index 615d4dc..83e5bc1 100644 > --- a/drivers/usb/dwc3/debugfs.c > +++ b/drivers/usb/dwc3/debugfs.c > @@ -426,6 +426,45 @@ static const struct file_operations dwc3_mode_fops = { > .release= single_release, > }; > > +static int dwc3_fifo_show(struct seq_file *s, void *unused) you call this file 'fifo' however you print more than just the fifos. You printk the request queues, info queue, descriptor fetch queue, event queue and protocol status queue. It seems, to me, you're trying to do way too much in a single file. > +{ > + struct dwc3 *dwc = s->private; > + unsigned long flags; > + unsigned inttype, index; > + const char *name; > + u32 reg; > + > + static const char * const fifo_names[] = { > + "TxFIFO", "RxFIFO", "TxReqQ", "RxReqQ", "RxInfoQ", > + "DescFetchQ", "EventQ", "ProtocolStatusQ"}; > + spin_lock_irqsave(>lock, flags); > + for (type = 0; type < 8; type++) { type < ARRAY_SIZE(fifo_names) ?? > + name = fifo_names[type]; > + for (index = 0; index < 32; index++) { not *all* dwc3 implementations enable all 32 endpoints, that's why we have dwc->num_endpoints > + dwc3_writel(dwc->regs, DWC3_GDBGFIFOSPACE, > + DWC3_GDBGFIFOSPACE_NUM(index) | > + DWC3_GDBGFIFOSPACE_TYPE(type)); > + reg = dwc3_readl(dwc->regs, DWC3_GDBGFIFOSPACE); this writel() followed by a readl() could be a nice little helper function in core.c. Remember that we need the FIFO Space and link state to make sure we're allowed to start a transfer ;-) I'd suggesting adding the following to core.c: int dwc3_core_fifo_space(struct dwc3_ep *dep, unsigned int type) { u32 reg; dwc3_writel(dwc->regs, DWC3_GDBGFIFOSPACE, DWC3_GDBGFIFOSPACE_NUM(dep->number) | DWC3_GDBGFIFOSPACE_TYPE(type)); reg = dwc3_readl(dwc->regs, DWC3_GDBGFIFOSPACE); return DWC3_GDBGFIFOSPACE_SPACE_AVAILABLE(reg); } then, add the prototype to core.h. (both of these in a separate patch). Then $subject just uses that helper to print out the data you want to print. I still think we might be better off with one file per type (iow TX fifo file, RX fifo file, TX request queue file, and so on). In fact, we could actually have a per-endpoint directory where all of these details are printed out in a much more structured form. If you want, I can work on that, no problem, but you're also welcome to implement it yourself and I'd be around to help as needed. In summary, we should have our debugfs looking like so: ./ep0 ./ep0/descriptor_fetch_queue ./ep0/event_queue ./ep0/info_queue ./ep0/protocol_status_queue ./ep0/request_queue ./ep0/rx_fifo ./ep0/rx_request_queue ./ep0/tx_fifo ./ep0/tx_request_queue ./ep1 ./ep10 ./ep10/descriptor_fetch_queue ./ep10/event_queue ./ep10/info_queue ./ep10/protocol_status_queue ./ep10/request_queue ./ep10/rx_fifo ./ep10/rx_request_queue ./ep10/tx_fifo ./ep10/tx_request_queue ./ep11 ./ep11/descriptor_fetch_queue ./ep11/event_queue ./ep11/info_queue ./ep11/protocol_status_queue ./ep11/request_queue ./ep11/rx_fifo ./ep11/rx_request_queue ./ep11/tx_fifo ./ep11/tx_request_queue ./ep12 ./ep12/descriptor_fetch_queue ./ep12/event_queue ./ep12/info_queue ./ep12/protocol_status_queue ./ep12/request_queue ./ep12/rx_fifo ./ep12/rx_request_queue ./ep12/tx_fifo ./ep12/tx_request_queue ./ep13 ./ep13/descriptor_fetch_queue
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hello, Sergei, > > From: "Du, Changbin"> > > > For DWC3 USB controller, the Global Debug Queue/FIFO Space Available > > Register(GDBGFIFOSPACE) can be used to dump FIFO/Queue available > space. > > Space needed before (. Okay. > > > This can be used to check some special issues, like whether data is > > successfully copied from memory to fifo when a trb is blocked. > > > > Signed-off-by: Du, Changbin > > --- > > v4: > >Do not fail silently, but print error. > [...] > > diff --git a/drivers/usb/dwc3/debugfs.c b/drivers/usb/dwc3/debugfs.c > > index 615d4dc..83e5bc1 100644 > > --- a/drivers/usb/dwc3/debugfs.c > > +++ b/drivers/usb/dwc3/debugfs.c > [...] > > @@ -642,10 +681,15 @@ void dwc3_debugfs_init(struct dwc3 *dwc) > > dwc->regset->nregs = ARRAY_SIZE(dwc3_regs); > > dwc->regset->base = dwc->regs; > > > > + > > Why? Thanks for point out, I will remove this additional line. > > > file = debugfs_create_regset32("regdump", S_IRUGO, root, dwc- > >regset); > > if (!file) > > dev_err(dwc->dev, "Can't create debugfs regdump\n"); > > > > + file = debugfs_create_file("fifo", S_IRUGO, root, dwc, > _fifo_fops); > > + if (!file) > > + dev_err(dwc->dev, "Can't create debugfs fifo\n"); > > + > > if (IS_ENABLED(CONFIG_USB_DWC3_DUAL_ROLE)) { > > file = debugfs_create_file("mode", S_IRUGO | S_IWUSR, > root, > > dwc, _mode_fops); > > MBR, Sergei
RE: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hello, Sergei, > > From: "Du, Changbin" > > > > For DWC3 USB controller, the Global Debug Queue/FIFO Space Available > > Register(GDBGFIFOSPACE) can be used to dump FIFO/Queue available > space. > > Space needed before (. Okay. > > > This can be used to check some special issues, like whether data is > > successfully copied from memory to fifo when a trb is blocked. > > > > Signed-off-by: Du, Changbin > > --- > > v4: > >Do not fail silently, but print error. > [...] > > diff --git a/drivers/usb/dwc3/debugfs.c b/drivers/usb/dwc3/debugfs.c > > index 615d4dc..83e5bc1 100644 > > --- a/drivers/usb/dwc3/debugfs.c > > +++ b/drivers/usb/dwc3/debugfs.c > [...] > > @@ -642,10 +681,15 @@ void dwc3_debugfs_init(struct dwc3 *dwc) > > dwc->regset->nregs = ARRAY_SIZE(dwc3_regs); > > dwc->regset->base = dwc->regs; > > > > + > > Why? Thanks for point out, I will remove this additional line. > > > file = debugfs_create_regset32("regdump", S_IRUGO, root, dwc- > >regset); > > if (!file) > > dev_err(dwc->dev, "Can't create debugfs regdump\n"); > > > > + file = debugfs_create_file("fifo", S_IRUGO, root, dwc, > _fifo_fops); > > + if (!file) > > + dev_err(dwc->dev, "Can't create debugfs fifo\n"); > > + > > if (IS_ENABLED(CONFIG_USB_DWC3_DUAL_ROLE)) { > > file = debugfs_create_file("mode", S_IRUGO | S_IWUSR, > root, > > dwc, _mode_fops); > > MBR, Sergei
Re: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hello. On 4/12/2016 2:10 PM, changbin...@intel.com wrote: From: "Du, Changbin"For DWC3 USB controller, the Global Debug Queue/FIFO Space Available Register(GDBGFIFOSPACE) can be used to dump FIFO/Queue available space. Space needed before (. This can be used to check some special issues, like whether data is successfully copied from memory to fifo when a trb is blocked. Signed-off-by: Du, Changbin --- v4: Do not fail silently, but print error. [...] diff --git a/drivers/usb/dwc3/debugfs.c b/drivers/usb/dwc3/debugfs.c index 615d4dc..83e5bc1 100644 --- a/drivers/usb/dwc3/debugfs.c +++ b/drivers/usb/dwc3/debugfs.c [...] @@ -642,10 +681,15 @@ void dwc3_debugfs_init(struct dwc3 *dwc) dwc->regset->nregs = ARRAY_SIZE(dwc3_regs); dwc->regset->base = dwc->regs; + Why? file = debugfs_create_regset32("regdump", S_IRUGO, root, dwc->regset); if (!file) dev_err(dwc->dev, "Can't create debugfs regdump\n"); + file = debugfs_create_file("fifo", S_IRUGO, root, dwc, _fifo_fops); + if (!file) + dev_err(dwc->dev, "Can't create debugfs fifo\n"); + if (IS_ENABLED(CONFIG_USB_DWC3_DUAL_ROLE)) { file = debugfs_create_file("mode", S_IRUGO | S_IWUSR, root, dwc, _mode_fops); MBR, Sergei
Re: [PATCH v4 2/2] usb: dwc3: add debugfs node to dump FIFO/Queue available space
Hello. On 4/12/2016 2:10 PM, changbin...@intel.com wrote: From: "Du, Changbin" For DWC3 USB controller, the Global Debug Queue/FIFO Space Available Register(GDBGFIFOSPACE) can be used to dump FIFO/Queue available space. Space needed before (. This can be used to check some special issues, like whether data is successfully copied from memory to fifo when a trb is blocked. Signed-off-by: Du, Changbin --- v4: Do not fail silently, but print error. [...] diff --git a/drivers/usb/dwc3/debugfs.c b/drivers/usb/dwc3/debugfs.c index 615d4dc..83e5bc1 100644 --- a/drivers/usb/dwc3/debugfs.c +++ b/drivers/usb/dwc3/debugfs.c [...] @@ -642,10 +681,15 @@ void dwc3_debugfs_init(struct dwc3 *dwc) dwc->regset->nregs = ARRAY_SIZE(dwc3_regs); dwc->regset->base = dwc->regs; + Why? file = debugfs_create_regset32("regdump", S_IRUGO, root, dwc->regset); if (!file) dev_err(dwc->dev, "Can't create debugfs regdump\n"); + file = debugfs_create_file("fifo", S_IRUGO, root, dwc, _fifo_fops); + if (!file) + dev_err(dwc->dev, "Can't create debugfs fifo\n"); + if (IS_ENABLED(CONFIG_USB_DWC3_DUAL_ROLE)) { file = debugfs_create_file("mode", S_IRUGO | S_IWUSR, root, dwc, _mode_fops); MBR, Sergei