Skip to contentSkip to conversation

staging: emxx_udc: Fix passing of NULL to dma_alloc_coherent()

PATCH Superseded Alex Dewar 2 participants 6 messages Aug 24, 2020 – Aug 25, 2020 1 message not in the archive Staging

lore
Share & tools
lore

Alex Dewar

author
#1 lore raw

PATCH 1/1 staging: emxx_udc: Fix passing of NULL to dma_alloc_coherent()

In reply to a message not in the archive

In nbu2ss_eq_queue() memory is allocated with dma_alloc_coherent(), though, strangely, NULL is passed as the struct device* argument. Pass the UDC's device instead.

Build-tested on x86 only.

Fixes 33aa8d45a4fe ("staging: emxx_udc: Add Emma Mobile USB Gadget driver")

---

So I *think* this is the right fix, but I don't have the hardware so I've only been able to build-test it. My worry is that I could be passing in the wrong struct device* here, which would squelch the warning without fixing the breakage.

Can someone cleverer than me tell me if this makes sense?

- Alex

---
 drivers/staging/emxx_udc/emxx_udc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/emxx_udc/emxx_udc.c +1 -1
index 03929b9d3a8b..09e91449c08c 100644
@@ -2593,7 +2593,7 @@ static int nbu2ss_ep_queue(struct usb_ep *_ep,
 
 	if (req->unaligned) {
 		if (!ep->virt_buf)
-			ep->virt_buf = dma_alloc_coherent(NULL, PAGE_SIZE,
+			ep->virt_buf = dma_alloc_coherent(udc->dev, PAGE_SIZE,
 							  &ep->phys_buf,
 							  GFP_ATOMIC | GFP_DMA);
 		if (ep->epnum > 0)  {
Signature · 1 line
2.28.0
author
#2 lore raw
Quoted 10 lines · Alex Dewar

In nbu2ss_eq_queue() memory is allocated with dma_alloc_coherent(), though, strangely, NULL is passed as the struct device* argument. Pass the UDC's device instead.

Build-tested on x86 only.

Fixes 33aa8d45a4fe ("staging: emxx_udc: Add Emma Mobile USB Gadget driver")

---

Quoted 8 lines

So I *think* this is the right fix, but I don't have the hardware so I've only been able to build-test it. My worry is that I could be passing in the wrong struct device* here, which would squelch the warning without fixing the breakage.

Can someone cleverer than me tell me if this makes sense?

- Alex

PS -- I meant to put an RFC in the subject line and an extra tag:

Quoted 20 lines
---
 drivers/staging/emxx_udc/emxx_udc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/emxx_udc/emxx_udc.c +1 -1
index 03929b9d3a8b..09e91449c08c 100644
@@ -2593,7 +2593,7 @@ static int nbu2ss_ep_queue(struct usb_ep *_ep,
 
 	if (req->unaligned) {
 		if (!ep->virt_buf)
-			ep->virt_buf = dma_alloc_coherent(NULL, PAGE_SIZE,
+			ep->virt_buf = dma_alloc_coherent(udc->dev, PAGE_SIZE,
 							  &ep->phys_buf,
 							  GFP_ATOMIC | GFP_DMA);
 		if (ep->epnum > 0)  {
Signature · 1 line
2.28.0
#3 lore raw
Quoted 4 lines · Alex Dewar
Quoted 3 lines · Alex Dewar

In nbu2ss_eq_queue() memory is allocated with dma_alloc_coherent(), though, strangely, NULL is passed as the struct device* argument. Pass the UDC's device instead.

I think passing NULL was always wrong, but it used to not cause an Oops. This was changed a year or two ago.

Quoted 12 lines
Quoted 12 lines

Build-tested on x86 only.

Fixes 33aa8d45a4fe ("staging: emxx_udc: Add Emma Mobile USB Gadget driver")

---

So I *think* this is the right fix, but I don't have the hardware so I've only been able to build-test it. My worry is that I could be passing in the wrong struct device* here, which would squelch the warning without fixing the breakage.

Quoted 7 lines
Quoted 3 lines

Can someone cleverer than me tell me if this makes sense?

- Alex

PS -- I meant to put an RFC in the subject line and an extra tag:

I don't know which dev pointer we're supposed to pass... It would be good to find someone to test the patch but if not then applying your patch is reasonable.

But could you search through the file and update the rest as well. The dma_free_coherent() needs to be updated and there was a second dma_alloc_coherent() in the bug report.

regards,
dan carpenter

author
#4 lore raw
Quoted 33 lines · Dan Carpenter
Quoted 4 lines · Alex Dewar
Quoted 3 lines · Alex Dewar

In nbu2ss_eq_queue() memory is allocated with dma_alloc_coherent(), though, strangely, NULL is passed as the struct device* argument. Pass the UDC's device instead.

I think passing NULL was always wrong, but it used to not cause an Oops. This was changed a year or two ago.

Quoted 12 lines
Quoted 12 lines

Build-tested on x86 only.

Fixes 33aa8d45a4fe ("staging: emxx_udc: Add Emma Mobile USB Gadget driver")

---

So I *think* this is the right fix, but I don't have the hardware so I've only been able to build-test it. My worry is that I could be passing in the wrong struct device* here, which would squelch the warning without fixing the breakage.

Quoted 7 lines
Quoted 3 lines

Can someone cleverer than me tell me if this makes sense?

- Alex

PS -- I meant to put an RFC in the subject line and an extra tag:

I don't know which dev pointer we're supposed to pass... It would be good to find someone to test the patch but if not then applying your patch is reasonable.

Quoted 7 lines

But could you search through the file and update the rest as well. The dma_free_coherent() needs to be updated and there was a second dma_alloc_coherent() in the bug report.

regards,
dan carpenter

I can only find the one instance of dma_alloc_coherent(). Was it by any chance a different warning about the same call....?

#5 lore raw
Quoted 38 lines · Alex Dewar
Quoted 37 lines · Dan Carpenter
Quoted 4 lines · Alex Dewar
Quoted 3 lines · Alex Dewar

In nbu2ss_eq_queue() memory is allocated with dma_alloc_coherent(), though, strangely, NULL is passed as the struct device* argument. Pass the UDC's device instead.

I think passing NULL was always wrong, but it used to not cause an Oops. This was changed a year or two ago.

Quoted 12 lines
Quoted 12 lines

Build-tested on x86 only.

Fixes 33aa8d45a4fe ("staging: emxx_udc: Add Emma Mobile USB Gadget driver")

---

So I *think* this is the right fix, but I don't have the hardware so I've only been able to build-test it. My worry is that I could be passing in the wrong struct device* here, which would squelch the warning without fixing the breakage.

Quoted 7 lines
Quoted 3 lines

Can someone cleverer than me tell me if this makes sense?

- Alex

PS -- I meant to put an RFC in the subject line and an extra tag:

I don't know which dev pointer we're supposed to pass... It would be good to find someone to test the patch but if not then applying your patch is reasonable.

But could you search through the file and update the rest as well. The dma_free_coherent() needs to be updated and there was a second dma_alloc_coherent() in the bug report.

Quoted 6 lines
Quoted 3 lines

regards,
dan carpenter

I can only find the one instance of dma_alloc_coherent(). Was it by any chance a different warning about the same call....?

Ah.... It was warning about the dma_free_coherent().

regards,
dan carpenter

author
#6 lore raw

PATCH v2 1/1 staging: emxx_udc: Fix passing of NULL to dma_alloc_coherent()

In nbu2ss_eq_queue() memory is allocated with dma_alloc_coherent(), though, strangely, NULL is passed as the struct device* argument. Pass the UDC's device instead. Fix up the corresponding call to dma_free_coherent() in the same way.

Build-tested on x86 only.

Fixes 33aa8d45a4fe ("staging: emxx_udc: Add Emma Mobile USB Gadget driver")

---
v2: also fix call to dma_free_coherent() (Dan)

@Magnus: I noticed you contributed this code back in 2014... I don't suppose you happen to have the hardware lying around to test this so we can get a Tested-by....? :)

---
 drivers/staging/emxx_udc/emxx_udc.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/staging/emxx_udc/emxx_udc.c +2 -2
index 03929b9d3a8bc..d0725bc8b48a4 100644
@@ -2593,7 +2593,7 @@ static int nbu2ss_ep_queue(struct usb_ep *_ep,
 
 	if (req->unaligned) {
 		if (!ep->virt_buf)
-			ep->virt_buf = dma_alloc_coherent(NULL, PAGE_SIZE,
+			ep->virt_buf = dma_alloc_coherent(udc->dev, PAGE_SIZE,
 							  &ep->phys_buf,
 							  GFP_ATOMIC | GFP_DMA);
 		if (ep->epnum > 0)  {
@@ -3148,7 +3148,7 @@ static int nbu2ss_drv_remove(struct platform_device *pdev)
 	for (i = 0; i < NUM_ENDPOINTS; i++) {
 		ep = &udc->ep[i];
 		if (ep->virt_buf)
-			dma_free_coherent(NULL, PAGE_SIZE, (void *)ep->virt_buf,
+			dma_free_coherent(udc->dev, PAGE_SIZE, (void *)ep->virt_buf,
 					  ep->phys_buf);
 	}
 
Signature · 1 line
2.28.0