2020-08-26 14:49:30

by Madhuparna Bhowmik

Subject: [PATCH] usb/c67x00/c67x00-drv: Fix Data Race bug

From: Madhuparna Bhowmik <[email protected]>

Currently in c67x00_drv_probe() IRQ is requested before calling
c67x00_probe_sie() and hence if interrupt happens the reading of certain
variables in the handler can race with initialization of the variables,
for e.g. sie->sie_num is written in c67x00_probe_sie() and read in
c67x00_hcd_irq().
Hence, this patch calls c67x00_probe_sie() before requesting IRQ in
probe.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Madhuparna Bhowmik <[email protected]>
---
drivers/usb/c67x00/c67x00-drv.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/usb/c67x00/c67x00-drv.c b/drivers/usb/c67x00/c67x00-drv.c
index 53838e7d4eef..2e816d5ca0eb 100644
--- a/drivers/usb/c67x00/c67x00-drv.c
+++ b/drivers/usb/c67x00/c67x00-drv.c
@@ -146,6 +146,9 @@ static int c67x00_drv_probe(struct platform_device *pdev)
c67x00_ll_init(c67x00);
c67x00_ll_hpi_reg_init(c67x00);

+ for (i = 0; i < C67X00_SIES; i++)
+ c67x00_probe_sie(&c67x00->sie[i], c67x00, i);
+
ret = request_irq(res2->start, c67x00_irq, 0, pdev->name, c67x00);
if (ret) {
dev_err(&pdev->dev, "Cannot claim IRQ\n");
@@ -158,9 +161,6 @@ static int c67x00_drv_probe(struct platform_device *pdev)
goto reset_failed;
}

- for (i = 0; i < C67X00_SIES; i++)
- c67x00_probe_sie(&c67x00->sie[i], c67x00, i);
-
platform_set_drvdata(pdev, c67x00);

return 0;
--
2.17.1

2020-08-28 07:42:17

by Greg Kroah-Hartman

Subject: Re: [PATCH] usb/c67x00/c67x00-drv: Fix Data Race bug

On Wed, Aug 26, 2020 at 08:14:59PM +0530, [email protected] wrote:
> From: Madhuparna Bhowmik <[email protected]>
>
> Currently in c67x00_drv_probe() IRQ is requested before calling
> c67x00_probe_sie() and hence if interrupt happens the reading of certain
> variables in the handler can race with initialization of the variables,
> for e.g. sie->sie_num is written in c67x00_probe_sie() and read in
> c67x00_hcd_irq().
> Hence, this patch calls c67x00_probe_sie() before requesting IRQ in
> probe.
>
> Found by Linux Driver Verification project (linuxtesting.org).
>
> Signed-off-by: Madhuparna Bhowmik <[email protected]>
> ---
> drivers/usb/c67x00/c67x00-drv.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/usb/c67x00/c67x00-drv.c b/drivers/usb/c67x00/c67x00-drv.c
> index 53838e7d4eef..2e816d5ca0eb 100644
> --- a/drivers/usb/c67x00/c67x00-drv.c
> +++ b/drivers/usb/c67x00/c67x00-drv.c
> @@ -146,6 +146,9 @@ static int c67x00_drv_probe(struct platform_device *pdev)
> c67x00_ll_init(c67x00);
> c67x00_ll_hpi_reg_init(c67x00);
>
> + for (i = 0; i < C67X00_SIES; i++)
> + c67x00_probe_sie(&c67x00->sie[i], c67x00, i);
> +
> ret = request_irq(res2->start, c67x00_irq, 0, pdev->name, c67x00);
> if (ret) {
> dev_err(&pdev->dev, "Cannot claim IRQ\n");
> @@ -158,9 +161,6 @@ static int c67x00_drv_probe(struct platform_device *pdev)
> goto reset_failed;
> }
>
> - for (i = 0; i < C67X00_SIES; i++)
> - c67x00_probe_sie(&c67x00->sie[i], c67x00, i);
> -
> platform_set_drvdata(pdev, c67x00);
>
> return 0;

Have you tested this on hardware to verify it works properly?

thanks,

greg k-h

2020-08-28 16:22:24

by Madhuparna Bhowmik

Subject: Re: [PATCH] usb/c67x00/c67x00-drv: Fix Data Race bug

On Fri, Aug 28, 2020 at 1:10 PM Greg KH <[email protected]> wrote:
>
> On Wed, Aug 26, 2020 at 08:14:59PM +0530, [email protected] wrote:
> > From: Madhuparna Bhowmik <[email protected]>
> >
> > Currently in c67x00_drv_probe() IRQ is requested before calling
> > c67x00_probe_sie() and hence if interrupt happens the reading of certain
> > variables in the handler can race with initialization of the variables,
> > for e.g. sie->sie_num is written in c67x00_probe_sie() and read in
> > c67x00_hcd_irq().
> > Hence, this patch calls c67x00_probe_sie() before requesting IRQ in
> > probe.
> >
> > Found by Linux Driver Verification project (linuxtesting.org).
> >
> > Signed-off-by: Madhuparna Bhowmik <[email protected]>
> > ---
> > drivers/usb/c67x00/c67x00-drv.c | 6 +++---
> > 1 file changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/usb/c67x00/c67x00-drv.c b/drivers/usb/c67x00/c67x00-drv.c
> > index 53838e7d4eef..2e816d5ca0eb 100644
> > --- a/drivers/usb/c67x00/c67x00-drv.c
> > +++ b/drivers/usb/c67x00/c67x00-drv.c
> > @@ -146,6 +146,9 @@ static int c67x00_drv_probe(struct platform_device *pdev)
> > c67x00_ll_init(c67x00);
> > c67x00_ll_hpi_reg_init(c67x00);
> >
> > + for (i = 0; i < C67X00_SIES; i++)
> > + c67x00_probe_sie(&c67x00->sie[i], c67x00, i);
> > +
> > ret = request_irq(res2->start, c67x00_irq, 0, pdev->name, c67x00);
> > if (ret) {
> > dev_err(&pdev->dev, "Cannot claim IRQ\n");
> > @@ -158,9 +161,6 @@ static int c67x00_drv_probe(struct platform_device *pdev)
> > goto reset_failed;
> > }
> >
> > - for (i = 0; i < C67X00_SIES; i++)
> > - c67x00_probe_sie(&c67x00->sie[i], c67x00, i);
> > -
> > platform_set_drvdata(pdev, c67x00);
> >
> > return 0;
>
> Have you tested this on hardware to verify it works properly?

No, I haven't, don't think I have the relevant hardware.
It would be great if someone could test this and add a Tested-by tag.

Thanks,
Madhuparna

>
> thanks,
>
> greg k-h