Re: [PATCH] usb: gadget: amd5536udc: fix NULL pointer dereference

[Felipe Balbi <balbi@ti.com>]

[-- Attachment #1: Type: text/plain, Size: 2965 bytes --]

Hi,

On Fri, Sep 11, 2015 at 03:32:56PM +0530, Sudip Mukherjee wrote:
> On Thu, Sep 10, 2015 at 11:03:34AM -0700, David Cohen wrote:
> > Hi Sudip,
> > 
> > On Fri, Sep 04, 2015 at 05:12:23PM +0530, Sudip Mukherjee wrote:
> > > We were checking if dev->regs is NULL but it was done after
> > > dereferencing it. Lets reset the controller and iounmap dev->regs only
> > > if it is not NULL.
> > > free_irq() does not need dev->regs, so unmaping it before freeing the
> > > irq should not matter.
> > > 
> > > Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
> > > ---
> > >  drivers/usb/gadget/udc/amd5536udc.c | 7 ++++---
> > >  1 file changed, 4 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/drivers/usb/gadget/udc/amd5536udc.c b/drivers/usb/gadget/udc/amd5536udc.c
> > > index fdacddb..26066d3 100644
> > > --- a/drivers/usb/gadget/udc/amd5536udc.c
> > > +++ b/drivers/usb/gadget/udc/amd5536udc.c
> > > @@ -3135,11 +3135,12 @@ static void udc_pci_remove(struct pci_dev *pdev)
> <snip>
> > 
> > I'm not familiar with the driver, but you're iounmap'ing before freeing
> > irq. Looks fishy to me.
> Well, I thought you will be able to give me some idea about how fix it. :)
> Then I guess we should be on the safe side and what about the following:
> 
> 
> diff --git a/drivers/usb/gadget/udc/amd5536udc.c b/drivers/usb/gadget/udc/amd5536udc.c
> index fdacddb..82f36f6 100644
> --- a/drivers/usb/gadget/udc/amd5536udc.c
> +++ b/drivers/usb/gadget/udc/amd5536udc.c
> @@ -3134,8 +3134,9 @@ static void udc_pci_remove(struct pci_dev *pdev)
>  		pci_pool_destroy(dev->stp_requests);
>  	}
>  
> -	/* reset controller */
> -	writel(AMD_BIT(UDC_DEVCFG_SOFTRESET), &dev->regs->cfg);
> +	if (dev->regs)
> +		/* reset controller */
> +		writel(AMD_BIT(UDC_DEVCFG_SOFTRESET), &dev->regs->cfg);

no, the check is pointless. Most of these are. Just look at your probe()
and you'll see that if dev->virt_addr is NULL (meaning ioremap_nocache()
failed) you exit from probe() with error. The driver doesn't probe at
all. So you can be sure that by remove, dev->regs is valid.

BTW, if probe fails, you have a TON of leaked resources!! You don't kfree()
dev, you don't pci_disable_device(), you don't release_mem_region(), you
don't iounmap() virt_addr, you don't free_irq().

Also the iounmap() call in remove is wrong. You ioremapped
dev->virt_addr but iounmap() dev->regs. Are you SURE that's ok ?

Man, what a mess! You gotta fix that up.

>  	if (dev->irq_registered)
>  		free_irq(pdev->irq, dev);
>  	if (dev->regs)
> 
> And just for my information: for a device what might happen if I iounmap
> before I free the irq? One thing I can think of is that after iounmap

what happens if an IRQ fires after you iounmap() but before you
free_irq() ?

> just at that moment one interrupt comes and the driver tries to access
> the io memory while servicing the irq.

there you go

-- 
balbi

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]