[PATCH v2.6.22-rc5] cxgb2: handle possible NULL pointer dereferencing

[PATCH v2.6.22-rc5] cxgb2: handle possible NULL pointer dereferencing

[pradeep singh]

Hi,

Chelsio's in kernel 10G driver does not checks the return value from
t1_get_board_info() in cxgb2.c.
t1_get_board_info may return a NULL and we still go on to dereference
it in the for loop without checking for the NULL.

This patch fixes this.

Signed-off-by: Pradeep Singh <pradeep.rautela@gmail.com>
---
 drivers/net/chelsio/cxgb2.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/net/chelsio/cxgb2.c b/drivers/net/chelsio/cxgb2.c
index 231ce43..edcfeba 100644
--- a/drivers/net/chelsio/cxgb2.c
+++ b/drivers/net/chelsio/cxgb2.c
@@ -1023,7 +1023,7 @@ static int __devinit init_one(struct pci_dev *pdev,
 	mmio_len = pci_resource_len(pdev, 0);
 	bi = t1_get_board_info(ent->driver_data);

-	for (i = 0; i < bi->port_number; ++i) {
+	for (i = 0; bi && i < bi->port_number; ++i) {
 		struct net_device *netdev;

 		netdev = alloc_etherdev(adapter ? 0 : sizeof(*adapter));
-- 
1.4.4.2
Thanks

-- 
Pradeep

Re: [PATCH v2.6.22-rc5] cxgb2: handle possible NULL pointer dereferencing

[Jens Axboe]

On Thu, Jun 21 2007, pradeep singh wrote:
> Hi,
> 
> Chelsio's in kernel 10G driver does not checks the return value from
> t1_get_board_info() in cxgb2.c.
> t1_get_board_info may return a NULL and we still go on to dereference
> it in the for loop without checking for the NULL.
> 
> This patch fixes this.

Patch looks odd - bi is dereferenced a number of times after that loop
anyway, so I don't see your patch fixing much.

-- 
Jens Axboe

Re: [PATCH v2.6.22-rc5] cxgb2: handle possible NULL pointer dereferencing

[pradeep singh]

Hi
On 6/21/07, Jens Axboe <jens.axboe@oracle.com> wrote:
> On Thu, Jun 21 2007, pradeep singh wrote:
> > Hi,
> >
> > Chelsio's in kernel 10G driver does not checks the return value from
> > t1_get_board_info() in cxgb2.c.
> > t1_get_board_info may return a NULL and we still go on to dereference
> > it in the for loop without checking for the NULL.
> >
> > This patch fixes this.
>
> Patch looks odd - bi is dereferenced a number of times after that loop
> anyway, so I don't see your patch fixing much.
Thanks for pointing that out Jens.
Sorry, i pushed it in a haste :(.
Will check again and resubmit it.

Thanks
--pradeep
[snip]
-- 
Pradeep

Re: [PATCH v2.6.22-rc5] cxgb2: handle possible NULL pointer dereferencing

[Jens Axboe]

On Thu, Jun 21 2007, pradeep singh wrote:
> Hi
> On 6/21/07, Jens Axboe <jens.axboe@oracle.com> wrote:
> >On Thu, Jun 21 2007, pradeep singh wrote:
> >> Hi,
> >>
> >> Chelsio's in kernel 10G driver does not checks the return value from
> >> t1_get_board_info() in cxgb2.c.
> >> t1_get_board_info may return a NULL and we still go on to dereference
> >> it in the for loop without checking for the NULL.
> >>
> >> This patch fixes this.
> >
> >Patch looks odd - bi is dereferenced a number of times after that loop
> >anyway, so I don't see your patch fixing much.
> Thanks for pointing that out Jens.
> Sorry, i pushed it in a haste :(.
> Will check again and resubmit it.

You're welcome. The first thing to do is analyze whether a NULL return
from t1_get_board_info() makes any sense. From a quick look, driver_data
should be the index into the t1 pci table. So if t1_get_board_info()
returns NULL, it must be some core bug. So I'd say either don't handle
it, or mark it with BUG_ON(), or do the !bi check and CH_ERR() a warning
and goto out_disable_pdev.

-- 
Jens Axboe

Re: [PATCH v2.6.22-rc5] cxgb2: handle possible NULL pointer dereferencing

[pradeep singh]

On 6/21/07, Jens Axboe <jens.axboe@oracle.com> wrote:
> On Thu, Jun 21 2007, pradeep singh wrote:
> > Hi
> > On 6/21/07, Jens Axboe <jens.axboe@oracle.com> wrote:
> > >On Thu, Jun 21 2007, pradeep singh wrote:
> > >> Hi,
> > >>
> > >> Chelsio's in kernel 10G driver does not checks the return value from
> > >> t1_get_board_info() in cxgb2.c.
> > >> t1_get_board_info may return a NULL and we still go on to dereference
> > >> it in the for loop without checking for the NULL.
> > >>
> > >> This patch fixes this.
> > >
> > >Patch looks odd - bi is dereferenced a number of times after that loop
> > >anyway, so I don't see your patch fixing much.
> > Thanks for pointing that out Jens.
> > Sorry, i pushed it in a haste :(.
> > Will check again and resubmit it.
>
> You're welcome. The first thing to do is analyze whether a NULL return
> from t1_get_board_info() makes any sense. From a quick look, driver_data
> should be the index into the t1 pci table. So if t1_get_board_info()
> returns NULL, it must be some core bug. So I'd say either don't handle
> it, or mark it with BUG_ON(), or do the !bi check and CH_ERR() a warning
> and goto out_disable_pdev.
Thanks for the advice Jens :).

Yeah this is what i am doing right now.
Will resubmit the patch.

Thanks a lot.
--pradeep
>
> --
> Jens Axboe
>
>


-- 
Pradeep