Skip to contentSkip to conversation

Stop calling request_irq(), etc. with invalid IRQs in the USB drivers`

PATCH In review Sergey Shtylyov 4 participants 9 messages Aug 13, 2021 – Aug 14, 2021

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v3 cover Stop calling request_irq(), etc. with invalid IRQs in the USB drivers`

Here are 2 patches against the 'usb-linus' branch of GregKH's 'usb.git' repo. The affected drivers call platform_get_irq() but largely ignore its result -- they blithely pass the negative error codes to request_irq() (and its ilk) which expects *unsinged* IRQ #s. Stop doing that by checking what exactly platform_get_irq() returns.

I've removed 7 patches that GregKH has applied to the usb-testing' branch (holler if this isn't enough).

[1/2] usb: host: ohci-tmio: add IRQ check
[2/2] usb: phy: tahvo: add IRQ check

#1 lore raw

PATCH v3 1/2 usb: host: ohci-tmio: add IRQ check

The driver neglects to check the result of platform_get_irq()'s call and blithely passes the negative error codes to usb_add_hcd() (which takes *unsigned* IRQ #), causing request_irq() that it calls to fail with -EINVAL, overriding an original error code. Stop calling usb_add_hcd() with the invalid IRQ #s.

Fixes 78c73414f4f6 ("USB: ohci: add support for tmio-ohci cell")

---
Changes in version 3:
- move the IRQ check higher in ohci_hcd_tmio_drv_probe(), to be closer to platfrom_get_irq()'s call.

 drivers/usb/host/ohci-tmio.c |    3 +++
 1 file changed, 3 insertions(+)
drivers/usb/host/ohci-tmio.c +3 -0
Index: usb/drivers/usb/host/ohci-tmio.c
===================================================================
@@ -202,6 +202,9 @@ static int ohci_hcd_tmio_drv_probe(struc
 	if (!cell)
 		return -EINVAL;
 
+	if (irq < 0)
+		return irq;
+
 	hcd = usb_create_hcd(&ohci_tmio_hc_driver, &dev->dev, dev_name(&dev->dev));
 	if (!hcd) {
 		ret = -ENOMEM;
maintainer Acked-by on v3 1/2
#2 lore raw

Re: [PATCH v3 1/2] usb: host: ohci-tmio: add IRQ check

Quoted 10 lines · Sergey Shtylyov

The driver neglects to check the result of platform_get_irq()'s call and blithely passes the negative error codes to usb_add_hcd() (which takes *unsigned* IRQ #), causing request_irq() that it calls to fail with -EINVAL, overriding an original error code. Stop calling usb_add_hcd() with the invalid IRQ #s.

Fixes 78c73414f4f6 ("USB: ohci: add support for tmio-ohci cell")

---

Quoted 21 lines

Changes in version 3:
- move the IRQ check higher in ohci_hcd_tmio_drv_probe(), to be closer to platfrom_get_irq()'s call.

 drivers/usb/host/ohci-tmio.c |    3 +++
 1 file changed, 3 insertions(+)

Index: usb/drivers/usb/host/ohci-tmio.c
===================================================================

drivers/usb/host/ohci-tmio.c +3 -0
@@ -202,6 +202,9 @@ static int ohci_hcd_tmio_drv_probe(struc
 	if (!cell)
 		return -EINVAL;
 
+	if (irq < 0)
+		return irq;
+
 	hcd = usb_create_hcd(&ohci_tmio_hc_driver, &dev->dev, dev_name(&dev->dev));
 	if (!hcd) {
 		ret = -ENOMEM;
#3 lore raw

PATCH v3 2/2 usb: phy: tahvo: add IRQ check

In reply to Sergey Shtylyov

The driver neglects to check the result of platform_get_irq()'s call and blithely passes the negative error codes to request_threaded_irq() (which takes *unsigned* IRQ #), causing it to fail with -EINVAL, overriding an original error code. Stop calling request_threaded_irq() with the invalid IRQ #s.

Fixes 9ba96ae5074c ("usb: omap1: Tahvo USB transceiver driver")
---
 drivers/usb/phy/phy-tahvo.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/usb/phy/phy-tahvo.c +3 -1
Index: usb/drivers/usb/phy/phy-tahvo.c
===================================================================
@@ -393,7 +393,9 @@ static int tahvo_usb_probe(struct platfo
 
 	dev_set_drvdata(&pdev->dev, tu);
 
-	tu->irq = platform_get_irq(pdev, 0);
+	tu->irq = ret = platform_get_irq(pdev, 0);
+	if (ret < 0)
+		return ret;
 	ret = request_threaded_irq(tu->irq, NULL, tahvo_usb_vbus_interrupt,
 				   IRQF_ONESHOT,
 				   "tahvo-vbus", tu);
Acked-by on v3 2/2
#4 lore raw

Re: [PATCH v3 2/2] usb: phy: tahvo: add IRQ check

Quoted 8 lines · Sergey Shtylyov

The driver neglects to check the result of platform_get_irq()'s call and blithely passes the negative error codes to request_threaded_irq() (which takes *unsigned* IRQ #), causing it to fail with -EINVAL, overriding an original error code. Stop calling request_threaded_irq() with the invalid IRQ #s.

Fixes 9ba96ae5074c ("usb: omap1: Tahvo USB transceiver driver")
Signature · 1 line
balbi
#5 lore raw
In reply to Sergey Shtylyov
Quoted 1 line · Sergey Shtylyov

Here are 2 patches against the 'usb-linus' branch of GregKH's 'usb.git' repo.

Wait, why that branch? Please make them against the branch you want them applied to. Hopefully they will apply to the usb-next branch...

#6 lore raw
Quoted 3 lines · Greg Kroah-Hartman
Quoted 1 line

Here are 2 patches against the 'usb-linus' branch of GregKH's 'usb.git' repo.

Wait, why that branch?

What branch I'd use for the fixes then?

Quoted 2 lines

Please make them against the branch you want
them applied to. Hopefully they will apply to the usb-next branch...

I didn't intend them for usb-next but looks like they apply there too.

MBR, Sergey

#7 lore raw
Quoted 7 lines · Sergey Shtylyov
Quoted 3 lines · Greg Kroah-Hartman
Quoted 1 line

Here are 2 patches against the 'usb-linus' branch of GregKH's 'usb.git' repo.

Wait, why that branch?

What branch I'd use for the fixes then?

Ah, you really want this in for 5.14-final? People are hitting this issue now?

Quoted 4 lines
Quoted 2 lines

Please make them against the branch you want
them applied to. Hopefully they will apply to the usb-next branch...

I didn't intend them for usb-next but looks like they apply there too.

I think it belongs there as a "nice cleanup to have", right?

thanks,

greg k-h

#8 lore raw
Quoted 7 lines · Greg Kroah-Hartman
Quoted 5 lines
Quoted 3 lines
Quoted 1 line

Here are 2 patches against the 'usb-linus' branch of GregKH's 'usb.git' repo.

Wait, why that branch?

What branch I'd use for the fixes then?

Ah, you really want this in for 5.14-final?

Not necessarily, it's your call. But all the patches are fixes.

Quoted 1 line

People are hitting this issue now?

No, the patches ware all the result of the code reviews...

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

Please make them against the branch you want
them applied to. Hopefully they will apply to the usb-next branch...

I didn't intend them for usb-next but looks like they apply there too.

I think it belongs there as a "nice cleanup to have", right?

No, they're definitely not cleanups and all have the "Fixes:" tags, so going
to end up in the stable trees (some already have).
There's going to be 10-patch series soon, all fixing the deferred probing due
to platfrorm_get_irq()...
Luckily, the USB tree doesn't shave the 3rd kind of platfrorm_get_irq() bugs:
treating 0 as error and returning it immediately along with the negative values, without doing the remaining part of probe...

Quoted 3 lines

thanks,

greg k-h

MBR, Sergey