Skip to contentSkip to conversation

usb: mtu3: give back request when rx error happens

PATCH Superseded Chunfeng Yun 3 participants 10 messages Apr 7, 2023 – Apr 17, 2023

Subsystems (4)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Chunfeng Yun

author maintainer
#0 lore raw

PATCH v2 1/7 usb: mtu3: give back request when rx error happens

When the Rx enconnter errors, currently, only print error logs, that may cause class driver's RX halt, shall give back the request with error status meanwhile.

---
v2: remove @req suggested by AngeloGioacchino

---
 drivers/usb/mtu3/mtu3_qmu.c | 37 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 36 insertions(+), 1 deletion(-)
drivers/usb/mtu3/mtu3_qmu.c +36 -1
index a2fdab8b63b2..a4da1af0b2c0 100644
@@ -466,6 +466,37 @@ static void qmu_tx_zlp_error_handler(struct mtu3 *mtu, u8 epnum)
 	mtu3_qmu_resume(mep);
 }
 
+/*
+ * when rx error happens (except zlperr), QMU will stop, and RQCPR saves
+ * the GPD encountered error, Done irq will arise after resuming QMU again.
+ */
+static void qmu_error_rx(struct mtu3 *mtu, u8 epnum)
+{
+	struct mtu3_ep *mep = mtu->out_eps + epnum;
+	struct mtu3_gpd_ring *ring = &mep->gpd_ring;
+	struct qmu_gpd *gpd_current = NULL;
+	struct mtu3_request *mreq;
+	dma_addr_t cur_gpd_dma;
+
+	cur_gpd_dma = read_rxq_cur_addr(mtu->mac_base, epnum);
+	gpd_current = gpd_dma_to_virt(ring, cur_gpd_dma);
+
+	mreq = next_request(mep);
+	if (!mreq || mreq->gpd != gpd_current) {
+		dev_err(mtu->dev, "no correct RX req is found\n");
+		return;
+	}
+
+	mreq->request.status = -EAGAIN;
+
+	/* by pass the current GDP */
+	gpd_current->dw0_info |= cpu_to_le32(GPD_FLAGS_BPS | GPD_FLAGS_HWO);
+	mtu3_qmu_resume(mep);
+
+	dev_dbg(mtu->dev, "%s EP%d, current=%p, req=%p\n",
+		__func__, epnum, gpd_current, mreq);
+}
+
 /*
  * NOTE: request list maybe is already empty as following case:
  * queue_tx --> qmu_interrupt(clear interrupt pending, schedule tasklet)-->
@@ -571,14 +602,18 @@ static void qmu_exception_isr(struct mtu3 *mtu, u32 qmu_status)
 
 	if ((qmu_status & RXQ_CSERR_INT) || (qmu_status & RXQ_LENERR_INT)) {
 		errval = mtu3_readl(mbase, U3D_RQERRIR0);
+		mtu3_writel(mbase, U3D_RQERRIR0, errval);
+
 		for (i = 1; i < mtu->num_eps; i++) {
 			if (errval & QMU_RX_CS_ERR(i))
 				dev_err(mtu->dev, "Rx %d CS error!\n", i);
 
 			if (errval & QMU_RX_LEN_ERR(i))
 				dev_err(mtu->dev, "RX %d Length error\n", i);
+
+			if (errval & (QMU_RX_CS_ERR(i) | QMU_RX_LEN_ERR(i)))
+				qmu_error_rx(mtu, i);
 		}
-		mtu3_writel(mbase, U3D_RQERRIR0, errval);
 	}
 
 	if (qmu_status & RXQ_ZLPERR_INT) {
Signature · 1 line
2.18.0
author maintainer
#1 lore raw

PATCH v2 2/7 usb: mtu3: use boolean return value

Prefer to use boolean value due to gpd_ring_empty() return true or false. See "16) Function return values and names" in coding-style.rst

---
v2: add Reviewed-by AngeloGioacchino

---
 drivers/usb/mtu3/mtu3_qmu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/usb/mtu3/mtu3_qmu.c +1 -1
index a4da1af0b2c0..6be4977a5db5 100644
@@ -221,7 +221,7 @@ static struct qmu_gpd *advance_deq_gpd(struct mtu3_gpd_ring *ring)
 }
 
 /* check if a ring is emtpy */
-static int gpd_ring_empty(struct mtu3_gpd_ring *ring)
+static bool gpd_ring_empty(struct mtu3_gpd_ring *ring)
 {
 	struct qmu_gpd *enq = ring->enqueue;
 	struct qmu_gpd *next;
Signature · 1 line
2.18.0
author maintainer
#2 lore raw

PATCH v2 3/7 usb: mtu3: fix kernel panic at qmu transfer done irq handler

In reply to Chunfeng Yun

When handle qmu transfer irq, it will unlock @mtu->lock before give back request, if another thread handle disconnect event at the same time, and try to disable ep, it may lock @mtu->lock and free qmu ring, then qmu irq hanlder may get a NULL gpd, avoid the KE by checking gpd's value before handling it.

e.g.
qmu done irq on cpu0 thread running on cpu1

qmu_done_tx()
handle gpd [0]

    mtu3_requ_complete()        mtu3_gadget_ep_disable()
      unlock @mtu->lock
        give back request         lock @mtu->lock
                                    mtu3_ep_disable()
                                      mtu3_gpd_ring_free()
                                   unlock @mtu->lock
      lock @mtu->lock
    get next gpd [1]

[1]: goto [0] to handle next gpd, and next gpd may be NULL.

---
v2: fix typo suggested by AngeloGioacchino

---
 drivers/usb/mtu3/mtu3_qmu.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/usb/mtu3/mtu3_qmu.c +3 -2
index 6be4977a5db5..3d77408e3133 100644
@@ -210,6 +210,7 @@ static struct qmu_gpd *advance_enq_gpd(struct mtu3_gpd_ring *ring)
 	return ring->enqueue;
 }
 
+/* @dequeue may be NULL if ring is unallocated or freed */
 static struct qmu_gpd *advance_deq_gpd(struct mtu3_gpd_ring *ring)
 {
 	if (ring->dequeue < ring->end)
@@ -522,7 +523,7 @@ static void qmu_done_tx(struct mtu3 *mtu, u8 epnum)
 	dev_dbg(mtu->dev, "%s EP%d, last=%p, current=%p, enq=%p\n",
 		__func__, epnum, gpd, gpd_current, ring->enqueue);
 
-	while (gpd != gpd_current && !GET_GPD_HWO(gpd)) {
+	while (gpd && gpd != gpd_current && !GET_GPD_HWO(gpd)) {
 
 		mreq = next_request(mep);
 
@@ -561,7 +562,7 @@ static void qmu_done_rx(struct mtu3 *mtu, u8 epnum)
 	dev_dbg(mtu->dev, "%s EP%d, last=%p, current=%p, enq=%p\n",
 		__func__, epnum, gpd, gpd_current, ring->enqueue);
 
-	while (gpd != gpd_current && !GET_GPD_HWO(gpd)) {
+	while (gpd && gpd != gpd_current && !GET_GPD_HWO(gpd)) {
 
 		mreq = next_request(mep);
 
Signature · 1 line
2.18.0
#3 lore raw

Re: [PATCH v2 3/7] usb: mtu3: fix kernel panic at qmu transfer done irq handler

Il 07/04/23 10:29, Chunfeng Yun ha scritto:

Quoted 20 lines

When handle qmu transfer irq, it will unlock @mtu->lock before give back request, if another thread handle disconnect event at the same time, and try to disable ep, it may lock @mtu->lock and free qmu ring, then qmu irq hanlder may get a NULL gpd, avoid the KE by checking gpd's value before handling it.

e.g.
qmu done irq on cpu0 thread running on cpu1

qmu_done_tx()
handle gpd [0]

     mtu3_requ_complete()        mtu3_gadget_ep_disable()
       unlock @mtu->lock
         give back request         lock @mtu->lock
                                     mtu3_ep_disable()
                                       mtu3_gpd_ring_free()
                                    unlock @mtu->lock
       lock @mtu->lock
     get next gpd [1]
Quoted 3 lines

[1]: goto [0] to handle next gpd, and next gpd may be NULL.

NACK. You still miss the Fixes tag.

Regards,
Angelo

author maintainer
#4 lore raw

Re: [PATCH v2 3/7] usb: mtu3: fix kernel panic at qmu transfer done irq handler

Quoted 29 lines · AngeloGioacchino Del Regno

External email : Please do not click links or open attachments until you have verified the sender or the content.

Il 07/04/23 10:29, Chunfeng Yun ha scritto:

Quoted 24 lines

When handle qmu transfer irq, it will unlock @mtu->lock before give back
request, if another thread handle disconnect event at the same time, and
try to disable ep, it may lock @mtu->lock and free qmu ring, then qmu
irq hanlder may get a NULL gpd, avoid the KE by checking gpd's value before
handling it.

e.g.
qmu done irq on cpu0 thread running on cpu1

qmu_done_tx()
handle gpd [0]

     mtu3_requ_complete()        mtu3_gadget_ep_disable()
       unlock @mtu->lock
         give back request         lock @mtu->lock
                                     mtu3_ep_disable()
                                       mtu3_gpd_ring_free()
                                    unlock @mtu->lock
       lock @mtu->lock
     get next gpd [1]
Quoted 5 lines
Quoted 3 lines

[1]: goto [0] to handle next gpd, and next gpd may be NULL.

NACK. You still miss the Fixes tag.

Ok, I'll add it, thanks

Quoted 4 lines

Regards,
Angelo

author maintainer
#5 lore raw

PATCH v2 4/7 usb: mtu3: unlock @mtu->lock just before giving back request

In reply to Chunfeng Yun

No need unlock @mtu->lock when unmap request, unlock it just before giving back request, due to we do not lock this spinlock when map the request.

---
v2: add Reviewed-by AngeloGioacchino
---

 drivers/usb/mtu3/mtu3_gadget.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/usb/mtu3/mtu3_gadget.c +1 -1
index c0264d5426bf..ad0eeac4332d 100644
@@ -23,7 +23,6 @@ __acquires(mep->mtu->lock)
 		req->status = status;
 
 	trace_mtu3_req_complete(mreq);
-	spin_unlock(&mtu->lock);
 
 	/* ep0 makes use of PIO, needn't unmap it */
 	if (mep->epnum)
@@ -32,6 +31,7 @@ __acquires(mep->mtu->lock)
 	dev_dbg(mtu->dev, "%s complete req: %p, sts %d, %d/%d\n",
 		mep->name, req, req->status, req->actual, req->length);
 
+	spin_unlock(&mtu->lock);
 	usb_gadget_giveback_request(&mep->ep, req);
 	spin_lock(&mtu->lock);
 }
Signature · 1 line
2.18.0
author maintainer
#6 lore raw

PATCH v2 5/7 usb: mtu3: expose role-switch control to userspace

In reply to Chunfeng Yun

The allow_userspace_control flag enables manual role switch from userspace, turn this feature on like several other USB DRD controller drivers.

---
v2: add Reviewed-by AngeloGioacchino

---
 drivers/usb/mtu3/mtu3_dr.c | 1 +
 1 file changed, 1 insertion(+)
drivers/usb/mtu3/mtu3_dr.c +1 -0
index 9b8aded3d95e..8191b7ed3852 100644
@@ -294,6 +294,7 @@ static int ssusb_role_sw_register(struct otg_switch_mtk *otg_sx)
 	role_sx_desc.get = ssusb_role_sw_get;
 	role_sx_desc.fwnode = dev_fwnode(dev);
 	role_sx_desc.driver_data = ssusb;
+	role_sx_desc.allow_userspace_control = true;
 	otg_sx->role_sw = usb_role_switch_register(dev, &role_sx_desc);
 	if (IS_ERR(otg_sx->role_sw))
 		return PTR_ERR(otg_sx->role_sw);
Signature · 1 line
2.18.0
author maintainer
#7 lore raw

PATCH v2 6/7 dt-bindings: usb: mtu3: add two optional clocks

In reply to Chunfeng Yun

Add optional clock 'xhci_ck' and 'frmcnt_ck';
Add optional property "assigned-clock" and "assigned-clock-parents";

---
v2: remove assigned-clocks* properties suggested by Rob

---
 Documentation/devicetree/bindings/usb/mediatek,mtu3.yaml | 4 ++++
 1 file changed, 4 insertions(+)
Documentation/devicetree/bindings/usb/mediatek,mtu3.yaml +4 -0
index d2655173e108..3d403d944453 100644
@@ -66,6 +66,8 @@ properties:
       - description: Reference clock used by low power mode etc
       - description: Mcu bus clock for register access
       - description: DMA bus clock for data transfer
+      - description: DRD controller clock
+      - description: Frame count clock
 
   clock-names:
     minItems: 1
@@ -74,6 +76,8 @@ properties:
       - const: ref_ck
       - const: mcu_ck
       - const: dma_ck
+      - const: xhci_ck
+      - const: frmcnt_ck
 
   phys:
     description:
Signature · 1 line
2.18.0
maintainer Reviewed-by on v2 6/7
#8 lore raw

Re: [PATCH v2 6/7] dt-bindings: usb: mtu3: add two optional clocks

Quoted 10 lines · Chunfeng Yun

Add optional clock 'xhci_ck' and 'frmcnt_ck';
Add optional property "assigned-clock" and "assigned-clock-parents";

---
v2: remove assigned-clocks* properties suggested by Rob

---
 Documentation/devicetree/bindings/usb/mediatek,mtu3.yaml | 4 ++++
 1 file changed, 4 insertions(+)
author maintainer
#9 lore raw

PATCH v2 7/7 usb: mtu3: add optional clock xhci_ck and frmcnt_ck

In reply to Chunfeng Yun

Add optional clock 'xhci_ck' which is usually the same as sys_ck, but some SoC use two separated clocks when the controller supports dual role mode;
Add optional clock 'frmcnt_ck' used on 4nm or advanced process SoC.

---
v2: no changes

---
 drivers/usb/mtu3/mtu3.h      | 2 +-
 drivers/usb/mtu3/mtu3_plat.c | 2 ++
 2 files changed, 3 insertions(+), 1 deletion(-)
drivers/usb/mtu3/mtu3.h +1 -1
index 2d7b57e07eee..b4a7662dded5 100644
@@ -90,7 +90,7 @@ struct mtu3_request;
  */
 #define EP0_RESPONSE_BUF  6
 
-#define BULK_CLKS_CNT	4
+#define BULK_CLKS_CNT	6
 
 /* device operated link and speed got from DEVICE_CONF register */
 enum mtu3_speed {
drivers/usb/mtu3/mtu3_plat.c +2 -0
index d78ae52b4e26..6f264b129243 100644
@@ -234,6 +234,8 @@ static int get_ssusb_rscs(struct platform_device *pdev, struct ssusb_mtk *ssusb)
 	clks[1].id = "ref_ck";
 	clks[2].id = "mcu_ck";
 	clks[3].id = "dma_ck";
+	clks[4].id = "xhci_ck";
+	clks[5].id = "frmcnt_ck";
 	ret = devm_clk_bulk_get_optional(dev, BULK_CLKS_CNT, clks);
 	if (ret)
 		return ret;
Signature · 1 line
2.18.0