Skip to contentSkip to conversation

USB DWC2 parity fix in isochronous mode

PATCH Superseded Scott Branden 3 participants 14 messages Aug 31, 2015 – Sep 10, 2015

Subsystems (2)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

v2 cover USB DWC2 parity fix in isochronous mode

This patch contains a fix for a real world interop problem found when using the Synopsis DWC2 USB controller with isochronous audio as detailed in the commit message.

Changes from v1:
- Address code review comments as per previous responses:
- renamed parity_set to has_correct_parity and reorder some logic

Roman Bacik (1):
usb: dwc2: gadget: parity fix in isochronous mode

 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51 ++++++++++++++++++++++++++++++++++++++++++++++-
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
Signature · 1 line
2.5.0
#1 lore raw

PATCH v2 1/1 usb: dwc2: gadget: parity fix in isochronous mode

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd microframes.

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is received.

---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51 ++++++++++++++++++++++++++++++++++++++++++++++-
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h +1 -0
index 0ed87620..a5634fd 100644
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c +50 -1

75 lines not on this page · raw message

drivers/usb/dwc2/hw.h +1 -0
index d0a5ed8..553f246 100644
@@ -142,6 +142,7 @@
 #define GINTSTS_RESETDET		(1 << 23)
 #define GINTSTS_FET_SUSP		(1 << 22)
 #define GINTSTS_INCOMPL_IP		(1 << 21)
+#define GINTSTS_INCOMPL_SOOUT		(1 << 21)
 #define GINTSTS_INCOMPL_SOIN		(1 << 20)
 #define GINTSTS_OEPINT			(1 << 19)
 #define GINTSTS_IEPINT			(1 << 18)
Signature · 1 line
2.5.0
#2 lore raw

RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Is there anything else we should address in this patch? Thanks,

Roman

Quoted 143 lines

-----Original Message-----
From: Scott Branden [mailto:[email protected]]
Sent: August-31-15 9:17 AM
To: John Youn; Greg Kroah-Hartman; [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list; Scott Branden
Subject: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd microframes.

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is received.

---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51
++++++++++++++++++++++++++++++++++++++++++++++-
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h index renamed from drivers/usb/dwc2/core.h +0 -0

0ed87620..a5634fd 100644

drivers/usb/dwc2/core.h +1 -0
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c index renamed from drivers/usb/dwc2/gadget.c +0 -0

4d47b7c..fac3e2f 100644

drivers/usb/dwc2/gadget.c +1 -0
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct dwc2_hsotg
 *hsotg, unsigned int idx,
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
drivers/usb/dwc2/gadget.c +2 -1
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void s3c_hsotg_core_init_disconnected(struct
 dwc2_hsotg *hsotg,
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN | GINTSTS_INCOMPL_SOOUT,
 		hsotg->regs + GINTMSK);
 
drivers/usb/dwc2/gadget.c +32 -0
 	if (using_dma(hsotg))
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",
 __func__);
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
+			hs_ep = hsotg->eps_in[idx];
+
+			if (!hs_ep->isochronous || hs_ep-
 >has_correct_parity)
+				continue;
+
+			epctl_reg = DIEPCTL(idx);
+			ctrl = readl(hsotg->regs + epctl_reg);
+
+			if (ctrl & DXEPCTL_EOFRNUM)
+				ctrl |= DXEPCTL_SETEVENFR;
+			else
+				ctrl |= DXEPCTL_SETODDFR;
+			writel(ctrl, hsotg->regs + epctl_reg);
+		}
+		writel(GINTSTS_INCOMPL_SOIN, hsotg->regs + GINTSTS);
+	}
+
+	if (gintsts & GINTSTS_INCOMPL_SOOUT) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOOUT\n",
 __func__);
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
+			hs_ep = hsotg->eps_out[idx];
+
+			if (!hs_ep->isochronous || hs_ep-
Quoted 1 line

has_correct_parity)

drivers/usb/dwc2/gadget.c +15 -1
+				continue;
+
+			epctl_reg = DOEPCTL(idx);
+			ctrl = readl(hsotg->regs + epctl_reg);
+
+			if (ctrl & DXEPCTL_EOFRNUM)
+				ctrl |= DXEPCTL_SETEVENFR;
+			else
+				ctrl |= DXEPCTL_SETODDFR;
+			writel(ctrl, hsotg->regs + epctl_reg);
+		}
+		writel(GINTSTS_INCOMPL_SOOUT, hsotg->regs + GINTSTS);
+	}
+
 	/*
 	 * if we've had fifo events, we should try and go around the
 	 * loop again to see if there's any point in returning yet.
@@ -2667,6 +2715,7 @@ static int s3c_hsotg_ep_enable(struct usb_ep *ep,
 	hs_ep->periodic = 0;
 	hs_ep->halted = 0;
 	hs_ep->interval = desc->bInterval;
+	hs_ep->has_correct_parity = 0;
 
 	if (hs_ep->interval > 1 && hs_ep->mc > 1)
 		dev_err(hsotg->dev, "MC > 1 when interval is not 1\n"); diff -
-git a/drivers/usb/dwc2/hw.h b/drivers/usb/dwc2/hw.h index

d0a5ed8..553f246 100644

drivers/usb/dwc2/hw.h +1 -1
@@ -142,6 +142,7 @@
 #define GINTSTS_RESETDET		(1 << 23)
 #define GINTSTS_FET_SUSP		(1 << 22)
 #define GINTSTS_INCOMPL_IP		(1 << 21)
+#define GINTSTS_INCOMPL_SOOUT		(1 << 21)
 #define GINTSTS_INCOMPL_SOIN		(1 << 20)
 #define GINTSTS_OEPINT			(1 << 19)
 #define GINTSTS_IEPINT			(1 << 18)
--

2.5.0

#3 lore raw

Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

In reply to Scott Branden
Quoted 54 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd microframes.

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is received.

---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51 ++++++++++++++++++++++++++++++++++++++++++++++-
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h +1 -0
index 0ed87620..a5634fd 100644
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c +3 -1
index 4d47b7c..fac3e2f 100644
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct dwc2_hsotg *hsotg, unsigned int idx,
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void s3c_hsotg_core_init_disconnected(struct dwc2_hsotg *hsotg,
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN | GINTSTS_INCOMPL_SOOUT,
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 9 lines
drivers/usb/dwc2/gadget.c +6 -0
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n", __func__);
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in the databook?

drivers/usb/dwc2/gadget.c irq_retry: · line 2604

Quoted 13 lines
drivers/usb/dwc2/gadget.c +13 -0
+			hs_ep = hsotg->eps_in[idx];
+
+			if (!hs_ep->isochronous || hs_ep->has_correct_parity)
+				continue;
+
+			epctl_reg = DIEPCTL(idx);
+			ctrl = readl(hsotg->regs + epctl_reg);
+
+			if (ctrl & DXEPCTL_EOFRNUM)
+				ctrl |= DXEPCTL_SETEVENFR;
+			else
+				ctrl |= DXEPCTL_SETODDFR;
+			writel(ctrl, hsotg->regs + epctl_reg);

This toggling code could be moved out to a helper function to reduce redundancy as it is in three places now.

Regards,
John

#4 lore raw

RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 76 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 5 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

microframes.

Quoted 11 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is received.

---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

drivers/usb/dwc2/gadget.c static void s3c_hsotg_epint(struct dwc2_hsotg

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h index renamed from drivers/usb/dwc2/core.h +0 -0

0ed87620..a5634fd 100644

drivers/usb/dwc2/core.h +1 -0
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c +0 -0
index 4d47b7c..fac3e2f 100644
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct dwc2_hsotg

*hsotg, unsigned int idx,

drivers/usb/dwc2/gadget.c void s3c_hsotg_core_init_disconnected(struct

Quoted 8 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void s3c_hsotg_core_init_disconnected(struct

dwc2_hsotg *hsotg,

Quoted 10 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN | GINTSTS_INCOMPL_SOOUT,
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))

drivers/usb/dwc2/gadget.c irq_retry: · line 2590

Quoted 8 lines
drivers/usb/dwc2/gadget.c +5 -0
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",

__func__);

Quoted 8 lines

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in the databook?

We are using procedure based on description from this source:

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints in DWC_otg Slave Mode

Synopsys databook is not in a public domain to quote the exact paragraph here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have described.

Please, provide an exact reference and possibly enough information that we can turn it into a working code.

Quoted 20 lines
Quoted 4 lines
drivers/usb/dwc2/gadget.c +3 -0
+			hs_ep = hsotg->eps_in[idx];
+
+			if (!hs_ep->isochronous || hs_ep-
 has_correct_parity)

drivers/usb/dwc2/gadget.c irq_retry: · line 2604

Quoted 10 lines
drivers/usb/dwc2/gadget.c +10 -0
+				continue;
+
+			epctl_reg = DIEPCTL(idx);
+			ctrl = readl(hsotg->regs + epctl_reg);
+
+			if (ctrl & DXEPCTL_EOFRNUM)
+				ctrl |= DXEPCTL_SETEVENFR;
+			else
+				ctrl |= DXEPCTL_SETODDFR;
+			writel(ctrl, hsotg->regs + epctl_reg);

This toggling code could be moved out to a helper function to reduce redundancy as it is in three places now.

Regards,
John

#5 lore raw

Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 93 lines · Roman Bacik
Quoted 76 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 5 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

microframes.

Quoted 11 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is received.

---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

drivers/usb/dwc2/gadget.c static void s3c_hsotg_epint(struct dwc2_hsotg

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h index renamed from drivers/usb/dwc2/core.h +0 -0

0ed87620..a5634fd 100644

drivers/usb/dwc2/core.h +1 -0
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c +0 -0
index 4d47b7c..fac3e2f 100644
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct dwc2_hsotg

*hsotg, unsigned int idx,

drivers/usb/dwc2/gadget.c void s3c_hsotg_core_init_disconnected(struct

Quoted 8 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void s3c_hsotg_core_init_disconnected(struct

dwc2_hsotg *hsotg,

Quoted 10 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN | GINTSTS_INCOMPL_SOOUT,
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))

drivers/usb/dwc2/gadget.c irq_retry: · line 2590

Quoted 8 lines
drivers/usb/dwc2/gadget.c +5 -0
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",

__func__);

Quoted 8 lines

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in the databook?

We are using procedure based on description from this source:

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints in DWC_otg Slave Mode

Quoted 4 lines

Synopsys databook is not in a public domain to quote the exact paragraph here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have described.

Please, provide an exact reference and possibly enough information that we can turn it into a working code.

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

John

#6 lore raw

RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 112 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:11 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 12 lines · Roman Bacik
Quoted 12 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 2 lines · Scott Branden

From: Roman Bacik <[email protected]>

Quoted 8 lines
Quoted 8 lines
Quoted 4 lines

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

microframes.

Quoted 3 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is

received.

Quoted 69 lines
Quoted 58 lines
Quoted 8 lines
---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

drivers/usb/dwc2/gadget.c static void s3c_hsotg_epint(struct dwc2_hsotg

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h index renamed from drivers/usb/dwc2/core.h +0 -0

0ed87620..a5634fd 100644

drivers/usb/dwc2/core.h +1 -0
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c +0 -0
index 4d47b7c..fac3e2f 100644
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct dwc2_hsotg

*hsotg, unsigned int idx,

drivers/usb/dwc2/gadget.c void s3c_hsotg_core_init_disconnected(struct

Quoted 8 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void s3c_hsotg_core_init_disconnected(struct

dwc2_hsotg *hsotg,

Quoted 10 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN | GINTSTS_INCOMPL_SOOUT,
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))

drivers/usb/dwc2/gadget.c irq_retry: · line 2590

Quoted 8 lines
drivers/usb/dwc2/gadget.c +5 -0
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",

__func__);

Quoted 8 lines

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in the databook?

We are using procedure based on description from this source:

Quoted 8 lines

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints in DWC_otg Slave Mode

Synopsys databook is not in a public domain to quote the exact paragraph

here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have described.

Quoted 2 lines

Please, provide an exact reference and possibly enough information that

Quoted 9 lines

we can turn it into a working code.

Quoted 1 line

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

John

Yes, that is exactly what we are doing what is described in the last bullet of 526 and the first two bullets of page 527. So what is the problem?

#7 lore raw

Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 115 lines · Roman Bacik
Quoted 115 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:11 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 12 lines · Roman Bacik
Quoted 12 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 2 lines · Scott Branden

From: Roman Bacik <[email protected]>

Quoted 8 lines
Quoted 8 lines
Quoted 4 lines

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

microframes.

Quoted 3 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is

received.

Quoted 69 lines
Quoted 58 lines
Quoted 8 lines
---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

drivers/usb/dwc2/gadget.c static void s3c_hsotg_epint(struct dwc2_hsotg

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h index renamed from drivers/usb/dwc2/core.h +0 -0

0ed87620..a5634fd 100644

drivers/usb/dwc2/core.h +1 -0
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c +0 -0
index 4d47b7c..fac3e2f 100644
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct dwc2_hsotg

*hsotg, unsigned int idx,

drivers/usb/dwc2/gadget.c void s3c_hsotg_core_init_disconnected(struct

Quoted 8 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void s3c_hsotg_core_init_disconnected(struct

dwc2_hsotg *hsotg,

Quoted 10 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN | GINTSTS_INCOMPL_SOOUT,
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))

drivers/usb/dwc2/gadget.c irq_retry: · line 2590

Quoted 8 lines
drivers/usb/dwc2/gadget.c +5 -0
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",

__func__);

Quoted 8 lines

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in the databook?

We are using procedure based on description from this source:

Quoted 8 lines

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints in DWC_otg Slave Mode

Synopsys databook is not in a public domain to quote the exact paragraph

here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have described.

Quoted 2 lines

Please, provide an exact reference and possibly enough information that

we can turn it into a working code.

Quoted 1 line
Quoted 9 lines
Quoted 6 lines

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

John

Yes, that is exactly what we are doing what is described in the last bullet of 526 and the first two bullets of page 527. So what is the problem?

Maybe I missed it. Where exactly are you doing it like it says in these two paragraphs?

Quote:

GINTSTS.incompISOIN is a global interrupt. Therefore, when more than one isochronous endpoints is active, then the application must determine which isochronous IN endpoints have incomplete data.

To accomplish this, read DSTS and DIEPCTLn for all isochronous endpoints. If DSTS.SOFFN[0] =DIEPCTLn.EO_FrNum (Bit 16),then that endpoint has an incomplete transfer

John

#8 lore raw

RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 153 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:25 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 63 lines · Roman Bacik
Quoted 63 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:11 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 16 lines · Roman Bacik
Quoted 16 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 6 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

Quoted 5 lines
Quoted 5 lines

microframes.

Quoted 4 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is

received.

Quoted 31 lines
Quoted 31 lines
Quoted 8 lines
---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h +1 -0
index 0ed87620..a5634fd 100644
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c +0 -0
index 4d47b7c..fac3e2f 100644
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct
Quoted 1 line

dwc2_hsotg

*hsotg, unsigned int idx,

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

drivers/usb/dwc2/gadget.c void

Quoted 8 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void

s3c_hsotg_core_init_disconnected(struct

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

dwc2_hsotg *hsotg,

drivers/usb/dwc2/gadget.c void s3c_hsotg_core_init_disconnected(struct dwc2_hsotg *hsotg,

Quoted 6 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN |

GINTSTS_INCOMPL_SOOUT,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
drivers/usb/dwc2/gadget.c +0 -0
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

drivers/usb/dwc2/gadget.c irq_retry:

Quoted 5 lines
drivers/usb/dwc2/gadget.c +5 -0
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s:

GINTSTS_INCOMPL_SOIN\n",

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

__func__);

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in

the databook?

Quoted 22 lines
Quoted 22 lines
Quoted 11 lines
Quoted 1 line

We are using procedure based on description from this source:

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints in DWC_otg Slave Mode

Quoted 2 lines

Synopsys databook is not in a public domain to quote the exact paragraph

here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have described.

Quoted 3 lines

Please, provide an exact reference and possibly enough information that

we can turn it into a working code.

Quoted 1 line
Quoted 8 lines
Quoted 6 lines

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

John

Yes, that is exactly what we are doing what is described in the last bullet of

526 and the first two bullets of page 527. So what is the problem?

Quoted 1 line

Maybe I missed it. Where exactly are you doing it like it says in these two paragraphs?

Quote:

GINTSTS.incompISOIN is a global interrupt. Therefore, when more than one isochronous endpoints is active, then the application must determine which isochronous IN endpoints have incomplete data.

Quoted 7 lines

To accomplish this, read DSTS and DIEPCTLn for all isochronous endpoints. If DSTS.SOFFN[0] =DIEPCTLn.EO_FrNum (Bit 16),then that endpoint has an incomplete transfer

John

That is the content of the two for loops. I will paste the code here again for your convenience:

Quoted 50 lines
drivers/usb/dwc2/gadget.c +5 -0
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",

__func__);

drivers/usb/dwc2/gadget.c +4 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
+			hs_ep = hsotg->eps_in[idx];
+
+			if (!hs_ep->isochronous || hs_ep-
Quoted 1 line

has_correct_parity)

drivers/usb/dwc2/gadget.c +19 -0
+				continue;
+
+			epctl_reg = DIEPCTL(idx);
+			ctrl = readl(hsotg->regs + epctl_reg);
+
+			if (ctrl & DXEPCTL_EOFRNUM)
+				ctrl |= DXEPCTL_SETEVENFR;
+			else
+				ctrl |= DXEPCTL_SETODDFR;
+			writel(ctrl, hsotg->regs + epctl_reg);
+		}
+		writel(GINTSTS_INCOMPL_SOIN, hsotg->regs + GINTSTS);
+	}
+
+	if (gintsts & GINTSTS_INCOMPL_SOOUT) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOOUT\n",

__func__);

drivers/usb/dwc2/gadget.c +4 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
+			hs_ep = hsotg->eps_out[idx];
+
+			if (!hs_ep->isochronous || hs_ep-
Quoted 1 line

has_correct_parity)

drivers/usb/dwc2/gadget.c +14 -0
+				continue;
+
+			epctl_reg = DOEPCTL(idx);
+			ctrl = readl(hsotg->regs + epctl_reg);
+
+			if (ctrl & DXEPCTL_EOFRNUM)
+				ctrl |= DXEPCTL_SETEVENFR;
+			else
+				ctrl |= DXEPCTL_SETODDFR;
+			writel(ctrl, hsotg->regs + epctl_reg);
+		}
+		writel(GINTSTS_INCOMPL_SOOUT, hsotg->regs + GINTSTS);
+	}
+
#9 lore raw

RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

In reply to John Youn
Quoted 212 lines

-----Original Message-----
From: Roman Bacik
Sent: September-09-15 7:36 PM
To: 'John Youn'; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 109 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:25 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 63 lines · Roman Bacik
Quoted 63 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:11 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 16 lines · Roman Bacik
Quoted 16 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 6 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

Quoted 5 lines
Quoted 5 lines

microframes.

Quoted 4 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is

received.

Quoted 31 lines
Quoted 31 lines
Quoted 8 lines
---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h +1 -0
index 0ed87620..a5634fd 100644
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
(quoted diff) +0 -0

b/drivers/usb/dwc2/gadget.c index 4d47b7c..fac3e2f 100644

drivers/usb/dwc2/gadget.c +0 -0
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct
Quoted 1 line

dwc2_hsotg

*hsotg, unsigned int idx,

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

drivers/usb/dwc2/gadget.c void

Quoted 8 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void

s3c_hsotg_core_init_disconnected(struct

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

dwc2_hsotg *hsotg,

drivers/usb/dwc2/gadget.c void s3c_hsotg_core_init_disconnected(struct dwc2_hsotg *hsotg,

Quoted 6 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN |

GINTSTS_INCOMPL_SOOUT,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
drivers/usb/dwc2/gadget.c +0 -0
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

drivers/usb/dwc2/gadget.c irq_retry:

Quoted 5 lines
drivers/usb/dwc2/gadget.c +5 -0
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s:

GINTSTS_INCOMPL_SOIN\n",

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

__func__);

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Quoted 13 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in

the databook?

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 1 line

We are using procedure based on description from this source:

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints

in

Quoted 34 lines
Quoted 17 lines
Quoted 17 lines
Quoted 4 lines

DWC_otg Slave Mode

Synopsys databook is not in a public domain to quote the exact paragraph

here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have described.

Quoted 3 lines

Please, provide an exact reference and possibly enough information that

we can turn it into a working code.

Quoted 1 line

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

Quoted 5 lines
Quoted 2 lines

John

Yes, that is exactly what we are doing what is described in the last bullet of

526 and the first two bullets of page 527. So what is the problem?

Quoted 1 line

Maybe I missed it. Where exactly are you doing it like it says in these two paragraphs?

Quote:

GINTSTS.incompISOIN is a global interrupt. Therefore, when more than one isochronous endpoints is active, then the application must determine which isochronous IN endpoints have incomplete data.

Quoted 7 lines

To accomplish this, read DSTS and DIEPCTLn for all isochronous endpoints. If DSTS.SOFFN[0] =DIEPCTLn.EO_FrNum (Bit 16),then that endpoint has an incomplete transfer

John

That is the content of the two for loops. I will paste the code here again for your convenience:

Quoted 36 lines
drivers/usb/dwc2/gadget.c +5 -0
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",

__func__);

drivers/usb/dwc2/gadget.c +4 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
+			hs_ep = hsotg->eps_in[idx];
+
+			if (!hs_ep->isochronous || hs_ep-
Quoted 1 line

has_correct_parity)

drivers/usb/dwc2/gadget.c +19 -0
+				continue;
+
+			epctl_reg = DIEPCTL(idx);
+			ctrl = readl(hsotg->regs + epctl_reg);
+
+			if (ctrl & DXEPCTL_EOFRNUM)
+				ctrl |= DXEPCTL_SETEVENFR;
+			else
+				ctrl |= DXEPCTL_SETODDFR;
+			writel(ctrl, hsotg->regs + epctl_reg);
+		}
+		writel(GINTSTS_INCOMPL_SOIN, hsotg->regs + GINTSTS);
+	}
+
+	if (gintsts & GINTSTS_INCOMPL_SOOUT) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOOUT\n",

__func__);

drivers/usb/dwc2/gadget.c +4 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
+			hs_ep = hsotg->eps_out[idx];
+
+			if (!hs_ep->isochronous || hs_ep-
Quoted 1 line

has_correct_parity)

Quoted 16 lines
Quoted 4 lines
drivers/usb/dwc2/gadget.c +4 -0
+				continue;
+
+			epctl_reg = DOEPCTL(idx);
+			ctrl = readl(hsotg->regs + epctl_reg);

drivers/usb/dwc2/gadget.c irq_retry: · line 2631

Quoted 10 lines
drivers/usb/dwc2/gadget.c +10 -0
+
+			if (ctrl & DXEPCTL_EOFRNUM)
+				ctrl |= DXEPCTL_SETEVENFR;
+			else
+				ctrl |= DXEPCTL_SETODDFR;
+			writel(ctrl, hsotg->regs + epctl_reg);
+		}
+		writel(GINTSTS_INCOMPL_SOOUT, hsotg->regs + GINTSTS);
+	}
+

I think I am starting to understand your objection. When the parity is not set on an isochronous endpoint, we flip the parity on that endpoint. I think you would prefer if we flip the parity only on isochronous endpoints, which do not have parity set, and got the actual interrupt. Is this interpretation correct?

We believe the way it is now gets correct parity on all isochronous endpoints faster and it was tested working well. But if you prefer adding another condition before flipping the endpoints we can try it out and retest.

Regards,

Roman

#10 lore raw

RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

In reply to John Youn
Quoted 141 lines

-----Original Message-----
From: Roman Bacik
Sent: September-09-15 7:59 PM
To: 'John Youn'; Scott Branden; 'Greg Kroah-Hartman'; 'linux- [email protected]'
Cc: '[email protected]'; bcm-kernel-feedback-list
Subject: RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 55 lines

-----Original Message-----
From: Roman Bacik
Sent: September-09-15 7:36 PM
To: 'John Youn'; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 46 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:25 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 36 lines · Roman Bacik
Quoted 36 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:11 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 16 lines · Roman Bacik
Quoted 16 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 6 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

Quoted 5 lines
Quoted 5 lines

microframes.

Quoted 4 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is

received.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Reviewed-by Srinath Mannam

<[email protected]>

Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 4 lines
---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h +1 -0
index 0ed87620..a5634fd 100644
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
(quoted diff) +0 -0

b/drivers/usb/dwc2/gadget.c index 4d47b7c..fac3e2f 100644

drivers/usb/dwc2/gadget.c +0 -0
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct
Quoted 1 line

dwc2_hsotg

*hsotg, unsigned int idx,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

drivers/usb/dwc2/gadget.c static void s3c_hsotg_epint(struct dwc2_hsotg *hsotg, unsigned int idx,

Quoted 7 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR; @@ -

2316,7 +2317,8 @@ void

Quoted 42 lines
Quoted 28 lines

s3c_hsotg_core_init_disconnected(struct

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

dwc2_hsotg *hsotg,

drivers/usb/dwc2/gadget.c void s3c_hsotg_core_init_disconnected(struct dwc2_hsotg *hsotg,

Quoted 6 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN |

GINTSTS_INCOMPL_SOOUT,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
drivers/usb/dwc2/gadget.c +0 -0
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

drivers/usb/dwc2/gadget.c irq_retry:

Quoted 5 lines
drivers/usb/dwc2/gadget.c +5 -0
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s:

GINTSTS_INCOMPL_SOIN\n",

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

__func__);

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Quoted 13 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in

the databook?

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 1 line

We are using procedure based on description from this source:

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints

in

Quoted 43 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 4 lines

DWC_otg Slave Mode

Synopsys databook is not in a public domain to quote the exact paragraph

here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have

described.

Quoted 35 lines
Quoted 34 lines
Quoted 10 lines
Quoted 10 lines
Quoted 3 lines

Please, provide an exact reference and possibly enough information that

we can turn it into a working code.

Quoted 1 line

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

Quoted 5 lines
Quoted 2 lines

John

Yes, that is exactly what we are doing what is described in the last bullet of

526 and the first two bullets of page 527. So what is the problem?

Quoted 1 line

Maybe I missed it. Where exactly are you doing it like it says in these two paragraphs?

Quote:

GINTSTS.incompISOIN is a global interrupt. Therefore, when more than one isochronous endpoints is active, then the application must determine which isochronous IN endpoints have incomplete data.

To accomplish this, read DSTS and DIEPCTLn for all isochronous endpoints. If DSTS.SOFFN[0] =DIEPCTLn.EO_FrNum (Bit 16),then that endpoint has an incomplete transfer

John

We have replaced the original code with the following code for both IN and OUT endpoints and retested:

	if (gintsts & GINTSTS_INCOMPL_SOIN) {
		u32 idx, epctl_reg, ctrl, dsts;
		struct s3c_hsotg_ep *hs_ep;
+		dsts = readl(hsotg->regs + DSTS);
		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n", __func__);
		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
			hs_ep = hsotg->eps_in[idx];

			if (!hs_ep->isochronous || hs_ep->has_correct_parity)
				continue;

			epctl_reg = DIEPCTL(idx);
			ctrl = readl(hsotg->regs + epctl_reg);

+ if (((dsts >> DSTS_SOFFN_SHIFT) & 1) !=
+ ((ctrl >> DXEPCTL_EOFRNUM) & 1))
+ continue;

			if (ctrl & DXEPCTL_EOFRNUM)
				ctrl |= DXEPCTL_SETEVENFR;
			else
				ctrl |= DXEPCTL_SETODDFR;
			writel(ctrl, hsotg->regs + epctl_reg);
			break;
		}
		writel(GINTSTS_INCOMPL_SOIN, hsotg->regs + GINTSTS);
	}

It still works. However, based on our experience with the audio gadget, USB host would either always use odd microframes or even microframes. It does not change the parity and hence gadget either always works or never works without this patch. So we can call hosts either odd or even.

When a gadget connects to even host, since it has even parity by default, it would receive transfer complete packets. So it recognizes it has correct parity and stop flipping.

When a gadget connects to an odd host, it would flip all parities to the correct one on all endpoints after receiving an incomplete interrupt from any of the endpoints connected to this odd host using the original code. On the other hand it would require an interrupt for each connected endpoint in order to get the parity right using the new code above.

Even when connected to multiple PCs with different parities, the worst case number of swaps (and hence lost packets) would be the number of connected hosts when using the original code, and the total number of endpoints, which is bigger than the number of connected hosts, when using the new code.

Please, let us know if you would prefer the new code anyway.

Regards,

Roman

#11 lore raw

Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 206 lines · Roman Bacik
Quoted 141 lines

-----Original Message-----
From: Roman Bacik
Sent: September-09-15 7:59 PM
To: 'John Youn'; Scott Branden; 'Greg Kroah-Hartman'; 'linux- [email protected]'
Cc: '[email protected]'; bcm-kernel-feedback-list
Subject: RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 55 lines

-----Original Message-----
From: Roman Bacik
Sent: September-09-15 7:36 PM
To: 'John Youn'; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 46 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:25 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 36 lines · Roman Bacik
Quoted 36 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:11 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 16 lines · Roman Bacik
Quoted 16 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 6 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

Quoted 5 lines
Quoted 5 lines

microframes.

Quoted 4 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is

received.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Reviewed-by Srinath Mannam

<[email protected]>

Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 4 lines
---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h +1 -0
index 0ed87620..a5634fd 100644
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
(quoted diff) +0 -0

b/drivers/usb/dwc2/gadget.c index 4d47b7c..fac3e2f 100644

drivers/usb/dwc2/gadget.c +0 -0
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct
Quoted 1 line

dwc2_hsotg

*hsotg, unsigned int idx,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

drivers/usb/dwc2/gadget.c

Quoted 7 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR; @@ -

2316,7 +2317,8 @@ void

Quoted 42 lines
Quoted 28 lines

s3c_hsotg_core_init_disconnected(struct

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

dwc2_hsotg *hsotg,

drivers/usb/dwc2/gadget.c

Quoted 6 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN |

GINTSTS_INCOMPL_SOOUT,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
drivers/usb/dwc2/gadget.c +0 -0
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

drivers/usb/dwc2/gadget.c irq_retry: · line 2590

Quoted 5 lines
drivers/usb/dwc2/gadget.c +5 -0
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s:

GINTSTS_INCOMPL_SOIN\n",

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

__func__);

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Quoted 13 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in

the databook?

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 1 line

We are using procedure based on description from this source:

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints

in

Quoted 43 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 4 lines

DWC_otg Slave Mode

Synopsys databook is not in a public domain to quote the exact paragraph

here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have

described.

Quoted 35 lines
Quoted 34 lines
Quoted 10 lines
Quoted 10 lines
Quoted 3 lines

Please, provide an exact reference and possibly enough information that

we can turn it into a working code.

Quoted 1 line

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

Quoted 5 lines
Quoted 2 lines

John

Yes, that is exactly what we are doing what is described in the last bullet of

526 and the first two bullets of page 527. So what is the problem?

Quoted 1 line

Maybe I missed it. Where exactly are you doing it like it says in these two paragraphs?

Quote:

GINTSTS.incompISOIN is a global interrupt. Therefore, when more than one isochronous endpoints is active, then the application must determine which isochronous IN endpoints have incomplete data.

To accomplish this, read DSTS and DIEPCTLn for all isochronous endpoints. If DSTS.SOFFN[0] =DIEPCTLn.EO_FrNum (Bit 16),then that endpoint has an incomplete transfer

John

We have replaced the original code with the following code for both IN and OUT endpoints and retested:

	if (gintsts & GINTSTS_INCOMPL_SOIN) {
		u32 idx, epctl_reg, ctrl, dsts;
		struct s3c_hsotg_ep *hs_ep;
(quoted diff) +1 -0
+		dsts = readl(hsotg->regs + DSTS);
		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n", __func__);
		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
			hs_ep = hsotg->eps_in[idx];

			if (!hs_ep->isochronous || hs_ep->has_correct_parity)
				continue;

			epctl_reg = DIEPCTL(idx);
			ctrl = readl(hsotg->regs + epctl_reg);
(quoted diff) +3 -0
+			if (((dsts >> DSTS_SOFFN_SHIFT) & 1) !=
+				((ctrl >> DXEPCTL_EOFRNUM) & 1))
+				continue;
Quoted 10 lines
			if (ctrl & DXEPCTL_EOFRNUM)
				ctrl |= DXEPCTL_SETEVENFR;
			else
				ctrl |= DXEPCTL_SETODDFR;
			writel(ctrl, hsotg->regs + epctl_reg);
			break;
		}
		writel(GINTSTS_INCOMPL_SOIN, hsotg->regs + GINTSTS);
	}

Yes that is what I was asking about, checking for the source of the interrupt.

But I'm convinced that the original code should be ok even with multiple iso endpoints given that all you are doing is toggling to find the correct parity the first time.

However a full implementation of this interrupt would need to know the exact source because, even with parity set, you could get this interrupt and will need to handle it properly. For example if you are slow reading or writing the fifo.

Quoted 6 lines

It still works. However, based on our experience with the audio gadget, USB host would either always use odd microframes or even microframes. It does not change the parity and hence gadget either always works or never works without this patch. So we can call hosts either odd or even.

When a gadget connects to even host, since it has even parity by default, it would receive transfer complete packets. So it recognizes it has correct parity and stop flipping.

When a gadget connects to an odd host, it would flip all parities to the correct one on all endpoints after receiving an incomplete interrupt from any of the endpoints connected to this odd host using the original code. On the other hand it would require an interrupt for each connected endpoint in order to get the parity right using the new code above.

You would still get this interrupt just once per microframe. And you would determine all the endpoints that missed that microframe at once in the loop. The result should be the same as before, the only overhead being the register read/check per iteration.

Quoted 3 lines

Even when connected to multiple PCs with different parities, the worst case number of swaps (and hence lost packets) would be the number of connected hosts when using the original code, and the total number of endpoints, which is bigger than the number of connected hosts, when using the new code.

Please, let us know if you would prefer the new code anyway.

The original code is fine. But we do plan to update the gadget code for isochronous so we may change how this is handled in the future.

If you resubmit with my other feedback I will review and test on our platforms.

Regards,
John

#12 lore raw

RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 273 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-10-15 12:08 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 134 lines · Roman Bacik
Quoted 134 lines

-----Original Message-----
From: Roman Bacik
Sent: September-09-15 7:59 PM
To: 'John Youn'; Scott Branden; 'Greg Kroah-Hartman'; 'linux- [email protected]'
Cc: '[email protected]'; bcm-kernel-feedback-list
Subject: RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 55 lines

-----Original Message-----
From: Roman Bacik
Sent: September-09-15 7:36 PM
To: 'John Youn'; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 46 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:25 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 36 lines · Roman Bacik
Quoted 36 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:11 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 16 lines · Roman Bacik
Quoted 16 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 6 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

Quoted 5 lines
Quoted 5 lines

microframes.

Quoted 4 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is

received.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Reviewed-by Srinath Mannam

<[email protected]>

Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 4 lines
---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h +1 -0
index 0ed87620..a5634fd 100644
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
(quoted diff) +0 -0

b/drivers/usb/dwc2/gadget.c index 4d47b7c..fac3e2f 100644

drivers/usb/dwc2/gadget.c +0 -0
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct
Quoted 1 line

dwc2_hsotg

*hsotg, unsigned int idx,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

drivers/usb/dwc2/gadget.c

Quoted 7 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR; @@ -

2316,7 +2317,8 @@ void

Quoted 34 lines
Quoted 34 lines

s3c_hsotg_core_init_disconnected(struct

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

dwc2_hsotg *hsotg,

drivers/usb/dwc2/gadget.c

Quoted 6 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN |

GINTSTS_INCOMPL_SOOUT,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
drivers/usb/dwc2/gadget.c +0 -0
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

drivers/usb/dwc2/gadget.c irq_retry: · line 2590

Quoted 5 lines
drivers/usb/dwc2/gadget.c +5 -0
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s:

GINTSTS_INCOMPL_SOIN\n",

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines

__func__);

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in

the databook?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

We are using procedure based on description from this source:

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous

Endpoints

Quoted 40 lines
Quoted 40 lines
Quoted 8 lines

in

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 4 lines

DWC_otg Slave Mode

Synopsys databook is not in a public domain to quote the exact paragraph

here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have

described.

Quoted 31 lines
Quoted 31 lines
Quoted 10 lines
Quoted 10 lines
Quoted 3 lines

Please, provide an exact reference and possibly enough information that

we can turn it into a working code.

Quoted 1 line

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

Quoted 5 lines
Quoted 2 lines

John

Yes, that is exactly what we are doing what is described in the last bullet of

526 and the first two bullets of page 527. So what is the problem?

Quoted 1 line

Maybe I missed it. Where exactly are you doing it like it says in these two paragraphs?

Quote:

GINTSTS.incompISOIN is a global interrupt. Therefore, when more than one isochronous endpoints is active, then the application must determine which isochronous IN endpoints have incomplete data.

To accomplish this, read DSTS and DIEPCTLn for all isochronous endpoints. If DSTS.SOFFN[0] =DIEPCTLn.EO_FrNum (Bit 16),then that endpoint has an incomplete transfer

Quoted 6 lines
Quoted 4 lines
Quoted 4 lines
Quoted 3 lines

John

We have replaced the original code with the following code for both IN and

OUT endpoints and retested:

Quoted 7 lines
	if (gintsts & GINTSTS_INCOMPL_SOIN) {
		u32 idx, epctl_reg, ctrl, dsts;
		struct s3c_hsotg_ep *hs_ep;
(quoted diff) +1 -0
+		dsts = readl(hsotg->regs + DSTS);
		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",

__func__);

Quoted 14 lines
		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {
			hs_ep = hsotg->eps_in[idx];

			if (!hs_ep->isochronous || hs_ep-
has_correct_parity)
				continue;

			epctl_reg = DIEPCTL(idx);
			ctrl = readl(hsotg->regs + epctl_reg);
(quoted diff) +3 -0
+			if (((dsts >> DSTS_SOFFN_SHIFT) & 1) !=
+				((ctrl >> DXEPCTL_EOFRNUM) & 1))
+				continue;
Quoted 10 lines
			if (ctrl & DXEPCTL_EOFRNUM)
				ctrl |= DXEPCTL_SETEVENFR;
			else
				ctrl |= DXEPCTL_SETODDFR;
			writel(ctrl, hsotg->regs + epctl_reg);
			break;
		}
		writel(GINTSTS_INCOMPL_SOIN, hsotg->regs + GINTSTS);
	}

Yes that is what I was asking about, checking for the source of the interrupt.

But I'm convinced that the original code should be ok even with multiple iso endpoints given that all you are doing is toggling to find the correct parity the first time.

However a full implementation of this interrupt would need to know the exact source because, even with parity set, you could get this interrupt and will need to handle it properly. For example if you are slow reading or writing the fifo.

Quoted 1 line

It still works. However, based on our experience with the audio gadget,

USB host would either always use odd microframes or even microframes. It does not change the parity and hence gadget either always works or never works without this patch. So we can call hosts either odd or even.

Quoted 2 lines

When a gadget connects to even host, since it has even parity by default, it

would receive transfer complete packets. So it recognizes it has correct parity and stop flipping.

Quoted 2 lines

When a gadget connects to an odd host, it would flip all parities to the

correct one on all endpoints after receiving an incomplete interrupt from any of the endpoints connected to this odd host using the original code. On the other hand it would require an interrupt for each connected endpoint in order to get the parity right using the new code above.

Quoted 1 line

You would still get this interrupt just once per microframe. And you would determine all the endpoints that missed that microframe at once in the loop. The result should be the same as before, the only overhead being the register read/check per iteration.

Quoted 1 line

Even when connected to multiple PCs with different parities, the worst

case number of swaps (and hence lost packets) would be the number of connected hosts when using the original code, and the total number of endpoints, which is bigger than the number of connected hosts, when using the new code.

Quoted 2 lines

Please, let us know if you would prefer the new code anyway.

Quoted 10 lines

The original code is fine. But we do plan to update the gadget code for isochronous so we may change how this is handled in the future.

If you resubmit with my other feedback I will review and test on our platforms.

Regards,
John

I will resubmit with the change to the for loop bound replacing MAX_EPS_CHANNELS with hsotg->num_of_eps and add a method for parity change for three times repeated code:

			if (ctrl & DXEPCTL_EOFRNUM)
				ctrl |= DXEPCTL_SETEVENFR;
			else
				ctrl |= DXEPCTL_SETODDFR;
			writel(ctrl, hsotg->regs + epctl_reg);

Regards,

Roman

#13 lore raw

RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

In reply to John Youn
Quoted 134 lines

-----Original Message-----
From: Roman Bacik
Sent: September-09-15 7:17 PM
To: 'John Youn'; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: RE: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 63 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-09-15 7:11 PM
To: Roman Bacik; John Youn; Scott Branden; Greg Kroah-Hartman; linux- [email protected]
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 16 lines · Roman Bacik
Quoted 16 lines

-----Original Message-----
From: John Youn [mailto:[email protected]]
Sent: September-03-15 11:53 PM
To: Scott Branden; John Youn; Greg Kroah-Hartman; linux- [email protected]; Roman Bacik
Cc: [email protected]; bcm-kernel-feedback-list
Subject: Re: [PATCH v2 1/1] usb: dwc2: gadget: parity fix in isochronous mode

Quoted 6 lines · Scott Branden

From: Roman Bacik <[email protected]>

USB OTG driver in isochronous mode has to set the parity of the receiving microframe. The parity is set to even by default. This causes problems for an audio gadget, if the host starts transmitting on odd

Quoted 5 lines
Quoted 5 lines

microframes.

Quoted 4 lines

This fix uses Incomplete Periodic Transfer interrupt to toggle between even and odd parity until the Transfer Complete interrupt is

received.

Quoted 31 lines
Quoted 31 lines
Quoted 8 lines
---
 drivers/usb/dwc2/core.h   |  1 +
 drivers/usb/dwc2/gadget.c | 51

++++++++++++++++++++++++++++++++++++++++++++++-

Quoted 20 lines
 drivers/usb/dwc2/hw.h     |  1 +
 3 files changed, 52 insertions(+), 1 deletion(-)
drivers/usb/dwc2/core.h +1 -0
index 0ed87620..a5634fd 100644
@@ -150,6 +150,7 @@ struct s3c_hsotg_ep {
 	unsigned int            periodic:1;
 	unsigned int            isochronous:1;
 	unsigned int            send_zlp:1;
+	unsigned int            has_correct_parity:1;
 
 	char                    name[10];
 };
drivers/usb/dwc2/gadget.c +0 -0
index 4d47b7c..fac3e2f 100644
@@ -1954,6 +1954,7 @@ static void s3c_hsotg_epint(struct
Quoted 1 line

dwc2_hsotg

*hsotg, unsigned int idx,

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

drivers/usb/dwc2/gadget.c void

Quoted 8 lines
drivers/usb/dwc2/gadget.c +1 -0
 		ints &= ~DXEPINT_XFERCOMPL;
 
 	if (ints & DXEPINT_XFERCOMPL) {
+		hs_ep->has_correct_parity = 1;
 		if (hs_ep->isochronous && hs_ep->interval == 1) {
 			if (ctrl & DXEPCTL_EOFRNUM)
 				ctrl |= DXEPCTL_SETEVENFR;
@@ -2316,7 +2317,8 @@ void

s3c_hsotg_core_init_disconnected(struct

Quoted 20 lines
Quoted 20 lines
Quoted 20 lines

dwc2_hsotg *hsotg,

Quoted 10 lines
drivers/usb/dwc2/gadget.c +2 -1
 		GINTSTS_CONIDSTSCHNG | GINTSTS_USBRST |
 		GINTSTS_RESETDET | GINTSTS_ENUMDONE |
 		GINTSTS_OTGINT | GINTSTS_USBSUSP |
-		GINTSTS_WKUPINT,
+		GINTSTS_WKUPINT |
+		GINTSTS_INCOMPL_SOIN | GINTSTS_INCOMPL_SOOUT,
 		hsotg->regs + GINTMSK);
 
 	if (using_dma(hsotg))

drivers/usb/dwc2/gadget.c irq_retry: · line 2590

Quoted 8 lines
drivers/usb/dwc2/gadget.c +5 -0
@@ -2581,6 +2583,52 @@ irq_retry:
 		s3c_hsotg_dump(hsotg);
 	}
 
+	if (gintsts & GINTSTS_INCOMPL_SOIN) {
+		u32 idx, epctl_reg, ctrl;
+		struct s3c_hsotg_ep *hs_ep;
+
+		dev_dbg(hsotg->dev, "%s: GINTSTS_INCOMPL_SOIN\n",

__func__);

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

drivers/usb/dwc2/gadget.c irq_retry: · line 2591

Quoted 1 line
drivers/usb/dwc2/gadget.c +1 -0
+		for (idx = 1; idx < MAX_EPS_CHANNELS; idx++) {

Valid endpoints are only up to hsotg->num_of_eps so this might crash on certain configurations.

Also, have you tried to find the endpoint which caused the incomplete interrupt by reading the control registers as described in the

databook?

Quoted 26 lines
Quoted 11 lines
Quoted 1 line

We are using procedure based on description from this source:

Synopsys, Inc. SolvNet 527
DesignWare.com
3.00a
April 2012
USB 2.0 Hi-Speed On-The-Go (OTG) Databook Isochronous Endpoints in DWC_otg Slave Mode

Quoted 2 lines

Synopsys databook is not in a public domain to quote the exact paragraph

here. You can find it in Chapter E, pp 526-527. There is no register in this databook, which would provide information about the source endpoint of the incomplete interrupt, as you have described.

Quoted 3 lines

Please, provide an exact reference and possibly enough information that

we can turn it into a working code.

Quoted 1 line

You can check the source of the interrupt by reading the DSTS and ep control registers as described in the databook page 526, last bullet, and 527 first bullet.

Quoted 5 lines
Quoted 2 lines

John

Yes, that is exactly what we are doing what is described in the last bullet of 526 and the first two bullets of page 527. So what is the problem?

In order to avoid any further confusion, here is the quote of the three bullets:

"- GINTSTS.incompISOIN is a global interrupt. Therefore, when more than one isochronous endpoints is active, then the application must determine which isochronous IN endpoints have incomplete data.
- To accomplish this, read DSTS and DIEPCTLn for all isochronous endpoints. If DSTS.SOFFN[0] =DIEPCTLn.EO_FrNum (Bit 16), then that endpoint has an incomplete transfer.
- When there is an incomplete transfer for an endpoint, then toggle the Even/Odd bit."

We believe that this is exactly what we coded in this patch.

Regards,

Roman