Re: [PATCH] i386/x86-64: Return defined error value for bad PCI config space accesses

Re: [PATCH] i386/x86-64: Return defined error value for bad PCI config space accesses

[Jeff Garzik]

Linux Kernel Mailing List wrote:
> -	if (!value || (bus > 255) || (devfn > 255) || (reg > 255))
> +	if (!value || (bus > 255) || (devfn > 255) || (reg > 255)) {
> +		*value = -1;
>  		return -EINVAL;
> +	}
>  
>  	spin_lock_irqsave(&pci_config_lock, flags);
>  
> diff --git a/arch/i386/pci/mmconfig.c b/arch/i386/pci/mmconfig.c
> index 2002c74..f77d7f8 100644
> --- a/arch/i386/pci/mmconfig.c
> +++ b/arch/i386/pci/mmconfig.c
> @@ -80,8 +80,10 @@ static int pci_mmcfg_read(unsigned int s
>  	unsigned long flags;
>  	u32 base;
>  
> -	if (!value || (bus > 255) || (devfn > 255) || (reg > 4095))
> +	if (!value || (bus > 255) || (devfn > 255) || (reg > 4095)) {
> +		*value = -1;
>  		return -EINVAL;
> +	}
>  
>  	base = get_base_addr(seg, bus, devfn);
>  	if (!base)
> diff --git a/arch/x86_64/pci/mmconfig.c b/arch/x86_64/pci/mmconfig.c
> index d4e25f3..b493ed9 100644
> --- a/arch/x86_64/pci/mmconfig.c
> +++ b/arch/x86_64/pci/mmconfig.c
> @@ -75,8 +75,10 @@ static int pci_mmcfg_read(unsigned int s
>  	char __iomem *addr;
>  
>  	/* Why do we have this when nobody checks it. How about a BUG()!? -AK */
> -	if (unlikely(!value || (bus > 255) || (devfn > 255) || (reg > 4095)))
> +	if (unlikely(!value || (bus > 255) || (devfn > 255) || (reg > 4095))) {
> +		*value = -1;

As the code check indicates, value might be NULL.

Please fix.

	Jeff

Re: [PATCH] i386/x86-64: Return defined error value for bad PCI config space accesses

[Andi Kleen]

On Sunday 09 April 2006 21:17, Jeff Garzik wrote:

> As the code check indicates, value might be NULL.
> 
> Please fix.

It should never be NULL.  If anything that's a BUG_ON, but crashing on it is ok
too. 

But I'll change it to a BUG_ON in the next patch thanks.

-Andi