Re: BUG null dereference in drivers/ata/sata_mv.c

[Robert Hancock <hancockrwd@gmail.com>]

On 12/14/2009 02:51 PM, Alexander Strakh wrote:
> 	KERNEL_VERSION: 2.6.32
> 	SUBJECT: null dereference in function mv_unexpected_intr
> 	DESCRIBE:
> 	In ./drivers/ata/sata_mv.c in function mv_port_intr
>
> 1. If ap == NULL in line 2778, then we goto line 2779.
> 2. In line 2779 function mv_unexpected_intr(ap, 0) is called.
> 3. In line 2538 null dereference: "ap->link.eh_info"
>
> 2773 static void mv_port_intr(struct ata_port *ap, u32 port_cause)
> 2774 {
> ...
> 2778         if (!ap || (ap->flags&  ATA_FLAG_DISABLED)) {
> 2779                 mv_unexpected_intr(ap, 0);
> 2780                 return;
> 2781         }
> ...
> 2809 }
>
> 2536 static void mv_unexpected_intr(struct ata_port *ap, int edma_was_enabled)
> 2537 {
> 2538         struct ata_eh_info *ehi =&ap->link.eh_info;
> ...
> 2555 }
>
> Found by Linux Device Drivers Verification Project (Svace Detector)

I don't think it should be possible for ap to be null at the point the 
check is made. The null check could likely be removed.