2007-10-23 10:19:43

by Takashi Iwai

Subject: [PATCH] intel-iommu: Fix array overflow

Fix possible array overflow:

drivers/pci/intel-iommu.c: In function ?dmar_get_fault_reason?:
drivers/pci/intel-iommu.c:753: warning: array subscript is above array bounds
drivers/pci/intel-iommu.c: In function ?iommu_page_fault?:
drivers/pci/intel-iommu.c:753: warning: array subscript is above array bounds

Signed-off-by: Takashi Iwai <[email protected]>

---
drivers/pci/intel-iommu.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index b3d7031..e4b0a0d 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -749,8 +749,8 @@ static char *fault_reason_strings[] =

char *dmar_get_fault_reason(u8 fault_reason)
{
- if (fault_reason > MAX_FAULT_REASON_IDX)
- return fault_reason_strings[MAX_FAULT_REASON_IDX];
+ if (fault_reason >= MAX_FAULT_REASON_IDX)
+ return fault_reason_strings[MAX_FAULT_REASON_IDX - 1];
else
return fault_reason_strings[fault_reason];
}

2007-10-24 23:34:25

by mark gross

Subject: Re: [PATCH] intel-iommu: Fix array overflow

On Tue, Oct 23, 2007 at 10:57:51AM +0200, Takashi Iwai wrote:
> Fix possible array overflow:
>
> drivers/pci/intel-iommu.c: In function ‘dmar_get_fault_reason’:
> drivers/pci/intel-iommu.c:753: warning: array subscript is above array bounds
> drivers/pci/intel-iommu.c: In function ‘iommu_page_fault’:
> drivers/pci/intel-iommu.c:753: warning: array subscript is above array bounds
>
> Signed-off-by: Takashi Iwai <[email protected]>
>
> ---
> drivers/pci/intel-iommu.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
> index b3d7031..e4b0a0d 100644
> --- a/drivers/pci/intel-iommu.c
> +++ b/drivers/pci/intel-iommu.c
> @@ -749,8 +749,8 @@ static char *fault_reason_strings[] =
>
> char *dmar_get_fault_reason(u8 fault_reason)
> {
> - if (fault_reason > MAX_FAULT_REASON_IDX)
> - return fault_reason_strings[MAX_FAULT_REASON_IDX];
> + if (fault_reason >= MAX_FAULT_REASON_IDX)
> + return fault_reason_strings[MAX_FAULT_REASON_IDX - 1];

This looks like what the code meant to implement. I guess future
hardware may be able to generate more types of faults, otherwise I'd put
a BUG here.

--mgross


> else
> return fault_reason_strings[fault_reason];
> }
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/

2007-10-25 09:06:36

by Takashi Iwai

Subject: Re: [PATCH] intel-iommu: Fix array overflow

At Wed, 24 Oct 2007 16:30:37 -0700,
Mark Gross wrote:
>
> On Tue, Oct 23, 2007 at 10:57:51AM +0200, Takashi Iwai wrote:
> > Fix possible array overflow:
> >
> > drivers/pci/intel-iommu.c: In function ?dmar_get_fault_reason?:
> > drivers/pci/intel-iommu.c:753: warning: array subscript is above array bounds
> > drivers/pci/intel-iommu.c: In function ?iommu_page_fault?:
> > drivers/pci/intel-iommu.c:753: warning: array subscript is above array bounds
> >
> > Signed-off-by: Takashi Iwai <[email protected]>
> >
> > ---
> > drivers/pci/intel-iommu.c | 4 ++--
> > 1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
> > index b3d7031..e4b0a0d 100644
> > --- a/drivers/pci/intel-iommu.c
> > +++ b/drivers/pci/intel-iommu.c
> > @@ -749,8 +749,8 @@ static char *fault_reason_strings[] =
> >
> > char *dmar_get_fault_reason(u8 fault_reason)
> > {
> > - if (fault_reason > MAX_FAULT_REASON_IDX)
> > - return fault_reason_strings[MAX_FAULT_REASON_IDX];
> > + if (fault_reason >= MAX_FAULT_REASON_IDX)
> > + return fault_reason_strings[MAX_FAULT_REASON_IDX - 1];
>
> This looks like what the code meant to implement.

I think not. The size of fault_reason_strings[] is
MAX_FAULT_REASON_IDX, not + 1. So gcc warning is correct.
Maybe the main problem is that the constant name is confusing...


Takashi

> I guess future
> hardware may be able to generate more types of faults, otherwise I'd put
> a BUG here.
>
> --mgross
>
>
> > else
> > return fault_reason_strings[fault_reason];
> > }
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to [email protected]
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at http://www.tux.org/lkml/
>

2007-10-25 15:29:39

by Keshavamurthy, Anil S

Subject: Re: [PATCH] intel-iommu: Fix array overflow

On Thu, Oct 25, 2007 at 09:31:02AM +0200, Takashi Iwai wrote:
> At Wed, 24 Oct 2007 16:30:37 -0700,
> Mark Gross wrote:
> >
> > On Tue, Oct 23, 2007 at 10:57:51AM +0200, Takashi Iwai wrote:
> > > Fix possible array overflow:
> > >
> > > drivers/pci/intel-iommu.c: In function ‘dmar_get_fault_reason’:
> > > drivers/pci/intel-iommu.c:753: warning: array subscript is above array bounds
> > > drivers/pci/intel-iommu.c: In function ‘iommu_page_fault’:
> > > drivers/pci/intel-iommu.c:753: warning: array subscript is above array bounds
> > >
> > > Signed-off-by: Takashi Iwai <[email protected]>
> > >
> > > ---
> > > drivers/pci/intel-iommu.c | 4 ++--
> > > 1 file changed, 2 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
> > > index b3d7031..e4b0a0d 100644
> > > --- a/drivers/pci/intel-iommu.c
> > > +++ b/drivers/pci/intel-iommu.c
> > > @@ -749,8 +749,8 @@ static char *fault_reason_strings[] =
> > >
> > > char *dmar_get_fault_reason(u8 fault_reason)
> > > {
> > > - if (fault_reason > MAX_FAULT_REASON_IDX)
> > > - return fault_reason_strings[MAX_FAULT_REASON_IDX];
> > > + if (fault_reason >= MAX_FAULT_REASON_IDX)
> > > + return fault_reason_strings[MAX_FAULT_REASON_IDX - 1];
> >
> > This looks like what the code meant to implement.
>
> I think not. The size of fault_reason_strings[] is
> MAX_FAULT_REASON_IDX, not + 1. So gcc warning is correct.
> Maybe the main problem is that the constant name is confusing...
Yup, GCC warning is correct. the size of fault_reason_strings[]
is MAX_FAULT_REASON_IDX and hence the max array that can be
referenced is [MAX_FAULT_REASON_IDX -1], hence the fix by
Takashi is correct.

-Anil