[PATCH] Buglet in Alchemy OHCI driver

[PATCH] Buglet in Alchemy OHCI driver

[Jordan Crouse]

[-- Attachment #1: Type: text/plain, Size: 196 bytes --]

Martin Michlmayr spotted this potentially serious bug.  Please apply.

Jordan

-- 
Jordan Crouse
Senior Linux Engineer
AMD - Personal Connectivity Solutions Group
<www.amd.com/embeddedprocessors>

[-- Attachment #2: ohci-fix.patch --]
[-- Type: text/plain, Size: 827 bytes --]

[PATCH] Buglet in Alchemy OCHI

From: Jordan Crouse <jordan.crouse@amd.com>

Failure to get the right resources should immediately return. 
Current code has the possiblity of running off into the weeds. Spotted by 
Martin Michlmayr.

Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
---

 drivers/usb/host/ohci-au1xxx.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/usb/host/ohci-au1xxx.c b/drivers/usb/host/ohci-au1xxx.c
index aa4d0cd..d8fb1bb 100644
--- a/drivers/usb/host/ohci-au1xxx.c
+++ b/drivers/usb/host/ohci-au1xxx.c
@@ -94,7 +94,7 @@ int usb_hcd_au1xxx_probe (const struct h
 
 	if (dev->resource[1].flags != IORESOURCE_IRQ) {
 		pr_debug ("resource[1] is not IORESOURCE_IRQ");
-		retval = -ENOMEM;
+		return -ENOMEM;
 	}
 
 	hcd = usb_create_hcd(driver, &dev->dev, "Au1xxx");

Re: [PATCH] Buglet in Alchemy OHCI driver

[Martin Michlmayr]

* Jordan Crouse <jordan.crouse@amd.com> [2006-03-01 11:30]:
> Martin Michlmayr spotted this potentially serious bug.  Please apply.

Please don't send patches as MIME attachments.  Here it is again (with
a better summary too):


[PATCH] Alchemy OCHI: return if right resources cannot be obtained

From: Jordan Crouse <jordan.crouse@amd.com>

Failure to get the right resources should immediately return.  Current
code has the possiblity of running off into the weeds. Spotted by
Martin Michlmayr.

Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
Signed-off-by: Martin Michlmayr <tbm@cyrius.com>

---

 drivers/usb/host/ohci-au1xxx.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/usb/host/ohci-au1xxx.c b/drivers/usb/host/ohci-au1xxx.c
index aa4d0cd..d8fb1bb 100644
--- a/drivers/usb/host/ohci-au1xxx.c
+++ b/drivers/usb/host/ohci-au1xxx.c
@@ -94,7 +94,7 @@ int usb_hcd_au1xxx_probe (const struct h
 
 	if (dev->resource[1].flags != IORESOURCE_IRQ) {
 		pr_debug ("resource[1] is not IORESOURCE_IRQ");
-		retval = -ENOMEM;
+		return -ENOMEM;
 	}
 
 	hcd = usb_create_hcd(driver, &dev->dev, "Au1xxx");

-- 
Martin Michlmayr
http://www.cyrius.com/

Re: [PATCH] Buglet in Alchemy OHCI driver

[Greg KH]

On Wed, Mar 01, 2006 at 06:37:35PM +0000, Martin Michlmayr wrote:
> * Jordan Crouse <jordan.crouse@amd.com> [2006-03-01 11:30]:
> > Martin Michlmayr spotted this potentially serious bug.  Please apply.
> 
> Please don't send patches as MIME attachments.  Here it is again (with
> a better summary too):
> 
> 
> [PATCH] Alchemy OCHI: return if right resources cannot be obtained
> 
> From: Jordan Crouse <jordan.crouse@amd.com>
> 
> Failure to get the right resources should immediately return.  Current
> code has the possiblity of running off into the weeds. Spotted by
> Martin Michlmayr.
> 
> Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
> Signed-off-by: Martin Michlmayr <tbm@cyrius.com>

This patch is already in my tree, in the other patch from Jordan, so it
will make it in after 2.6.16-final is out.

thanks,

greg k-h

Re: [PATCH] Buglet in Alchemy OHCI driver

[Ralf Baechle]

On Wed, Mar 01, 2006 at 11:30:26AM -0700, Jordan Crouse wrote:
> Date:	Wed, 1 Mar 2006 11:30:26 -0700
> From:	"Jordan Crouse" <jordan.crouse@amd.com>
> To:	linux-usb-devel@lists.sourceforge.net
> cc:	linux-mips@linux-mips.org, gregkh@suse.de, tbm@cyrius.com
> Subject: [PATCH] Buglet in Alchemy OHCI driver
> Content-Type: multipart/mixed;
>  boundary=5vNYLRcllDrimb99
> 
> Martin Michlmayr spotted this potentially serious bug.  Please apply.

Ehh...  This problem doesn't exist on kernel.org.  Greg, can you ignore
this one, please?

  Ralf

Re: [PATCH] Buglet in Alchemy OHCI driver

[Greg KH]

On Thu, Mar 02, 2006 at 03:33:44PM +0000, Ralf Baechle wrote:
> On Wed, Mar 01, 2006 at 11:30:26AM -0700, Jordan Crouse wrote:
> > Date:	Wed, 1 Mar 2006 11:30:26 -0700
> > From:	"Jordan Crouse" <jordan.crouse@amd.com>
> > To:	linux-usb-devel@lists.sourceforge.net
> > cc:	linux-mips@linux-mips.org, gregkh@suse.de, tbm@cyrius.com
> > Subject: [PATCH] Buglet in Alchemy OHCI driver
> > Content-Type: multipart/mixed;
> >  boundary=5vNYLRcllDrimb99
> > 
> > Martin Michlmayr spotted this potentially serious bug.  Please apply.
> 
> Ehh...  This problem doesn't exist on kernel.org.  Greg, can you ignore
> this one, please?

Consider it ignored :)

thanks,

greg k-h

RE: [PATCH] Buglet in Alchemy OHCI driver

[Hamilton, Ian]

Hi Greg,

I spotted this in the current latest version of ohci-au1xxx.c (accessed
via git web interface):

94
95 		if (dev->resource[1].flags != IORESOURCE_IRQ) {
96 		pr_debug ("resource[1] is not IORESOURCE_IRQ");
97 		retval -ENOMEM;
98 	}
99  

instead this from Martin's patch:

diff --git a/drivers/usb/host/ohci-au1xxx.c
b/drivers/usb/host/ohci-au1xxx.c
index aa4d0cd..d8fb1bb 100644
--- a/drivers/usb/host/ohci-au1xxx.c
+++ b/drivers/usb/host/ohci-au1xxx.c
@@ -94,7 +94,7 @@ int usb_hcd_au1xxx_probe (const struct h
 
        if (dev->resource[1].flags != IORESOURCE_IRQ) {
                pr_debug ("resource[1] is not IORESOURCE_IRQ");
-               retval = -ENOMEM;
+               return -ENOMEM;
        }

Line 97 produces a warning, but doesn't stop the build, so may have been
missed.

Cheers,
Ian Hamilton.

-----Original Message-----
From: linux-mips-bounce@linux-mips.org
[mailto:linux-mips-bounce@linux-mips.org] On Behalf Of Greg KH
Sent: 01 March 2006 23:22
To: Martin Michlmayr
Cc: Jordan Crouse; linux-usb-devel@lists.sourceforge.net;
linux-mips@linux-mips.org; gregkh@suse.de
Subject: Re: [PATCH] Buglet in Alchemy OHCI driver

On Wed, Mar 01, 2006 at 06:37:35PM +0000, Martin Michlmayr wrote:
> * Jordan Crouse <jordan.crouse@amd.com> [2006-03-01 11:30]:
> > Martin Michlmayr spotted this potentially serious bug.  Please
apply.
> 
> Please don't send patches as MIME attachments.  Here it is again (with
> a better summary too):
> 
> 
> [PATCH] Alchemy OCHI: return if right resources cannot be obtained
> 
> From: Jordan Crouse <jordan.crouse@amd.com>
> 
> Failure to get the right resources should immediately return.  Current
> code has the possiblity of running off into the weeds. Spotted by
> Martin Michlmayr.
> 
> Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
> Signed-off-by: Martin Michlmayr <tbm@cyrius.com>

This patch is already in my tree, in the other patch from Jordan, so it
will make it in after 2.6.16-final is out.

thanks,

greg k-h

RE: [PATCH] Buglet in Alchemy OHCI driver

[Hamilton, Ian]

Hi Greg,

I spotted this in the current latest version of ohci-au1xxx.c (accessed
via git web interface):

94
95 		if (dev->resource[1].flags != IORESOURCE_IRQ) {
96 		pr_debug ("resource[1] is not IORESOURCE_IRQ");
97 		retval -ENOMEM;
98 	}
99  

instead this from Martin's patch:

diff --git a/drivers/usb/host/ohci-au1xxx.c
b/drivers/usb/host/ohci-au1xxx.c
index aa4d0cd..d8fb1bb 100644
--- a/drivers/usb/host/ohci-au1xxx.c
+++ b/drivers/usb/host/ohci-au1xxx.c
@@ -94,7 +94,7 @@ int usb_hcd_au1xxx_probe (const struct h
 
        if (dev->resource[1].flags != IORESOURCE_IRQ) {
                pr_debug ("resource[1] is not IORESOURCE_IRQ");
-               retval = -ENOMEM;
+               return -ENOMEM;
        }

Line 97 produces a warning, but doesn't stop the build, so may have been
missed.

Cheers,
Ian Hamilton.

-----Original Message-----
From: linux-mips-bounce@linux-mips.org
[mailto:linux-mips-bounce@linux-mips.org] On Behalf Of Greg KH
Sent: 01 March 2006 23:22
To: Martin Michlmayr
Cc: Jordan Crouse; linux-usb-devel@lists.sourceforge.net;
linux-mips@linux-mips.org; gregkh@suse.de
Subject: Re: [PATCH] Buglet in Alchemy OHCI driver

On Wed, Mar 01, 2006 at 06:37:35PM +0000, Martin Michlmayr wrote:
> * Jordan Crouse <jordan.crouse@amd.com> [2006-03-01 11:30]:
> > Martin Michlmayr spotted this potentially serious bug.  Please
apply.
> 
> Please don't send patches as MIME attachments.  Here it is again (with
> a better summary too):
> 
> 
> [PATCH] Alchemy OCHI: return if right resources cannot be obtained
> 
> From: Jordan Crouse <jordan.crouse@amd.com>
> 
> Failure to get the right resources should immediately return.  Current
> code has the possiblity of running off into the weeds. Spotted by
> Martin Michlmayr.
> 
> Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
> Signed-off-by: Martin Michlmayr <tbm@cyrius.com>

This patch is already in my tree, in the other patch from Jordan, so it
will make it in after 2.6.16-final is out.

thanks,

greg k-h

Re: [PATCH] Buglet in Alchemy OHCI driver

[Greg KH]

On Tue, Mar 07, 2006 at 12:17:21PM -0000, Hamilton, Ian wrote:
> Hi Greg,
> 
> I spotted this in the current latest version of ohci-au1xxx.c (accessed
> via git web interface):
> 
> 94
> 95 		if (dev->resource[1].flags != IORESOURCE_IRQ) {
> 96 		pr_debug ("resource[1] is not IORESOURCE_IRQ");
> 97 		retval -ENOMEM;
> 98 	}
> 99  
> 
> instead this from Martin's patch:
> 
> diff --git a/drivers/usb/host/ohci-au1xxx.c
> b/drivers/usb/host/ohci-au1xxx.c
> index aa4d0cd..d8fb1bb 100644
> --- a/drivers/usb/host/ohci-au1xxx.c
> +++ b/drivers/usb/host/ohci-au1xxx.c
> @@ -94,7 +94,7 @@ int usb_hcd_au1xxx_probe (const struct h
>  
>         if (dev->resource[1].flags != IORESOURCE_IRQ) {
>                 pr_debug ("resource[1] is not IORESOURCE_IRQ");
> -               retval = -ENOMEM;
> +               return -ENOMEM;
>         }
> 
> Line 97 produces a warning, but doesn't stop the build, so may have been
> missed.

Can you check the -mm tree to see if it is fixed there or not?  That has
the patches that are pending from my usb tree in it.

And if not, can you send me a patch against the latest -mm?

thanks,

greg k-h

Re: [PATCH] Buglet in Alchemy OHCI driver

[Ralf Baechle]

On Tue, Mar 07, 2006 at 08:54:45AM -0800, Greg KH wrote:

Again that's been a bug report against the wrong tree - it was broken in
the MIPS tree only and is fixed since this morning.

  Ralf