2021-09-29 12:21:17

by Guo Zhi

Subject: [PATCH] drivers/ata: Fix kernel pointer leak

Pointers should be printed with %p or %px rather than cast to
'unsigned long' and pinted with %lx
Change %lx to %p to print the secured pointer.

Signed-off-by: Guo Zhi <[email protected]>
---
drivers/ata/pata_atp867x.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/ata/pata_atp867x.c b/drivers/ata/pata_atp867x.c
index 2bc5fc81efe3..c32b95f48e50 100644
--- a/drivers/ata/pata_atp867x.c
+++ b/drivers/ata/pata_atp867x.c
@@ -447,11 +447,11 @@ static int atp867x_ata_pci_sff_init_host(struct ata_host *host)
#ifdef ATP867X_DEBUG
atp867x_check_ports(ap, i);
#endif
- ata_port_desc(ap, "cmd 0x%lx ctl 0x%lx",
- (unsigned long)ioaddr->cmd_addr,
- (unsigned long)ioaddr->ctl_addr);
- ata_port_desc(ap, "bmdma 0x%lx",
- (unsigned long)ioaddr->bmdma_addr);
+ ata_port_desc(ap, "cmd 0x%p ctl 0x%p",
+ ioaddr->cmd_addr,
+ ioaddr->ctl_addr);
+ ata_port_desc(ap, "bmdma 0x%p",
+ ioaddr->bmdma_addr);

mask |= 1 << i;
}
--
2.33.0

2021-09-29 14:45:49

by Sergey Shtylyov

Subject: Re: [PATCH] drivers/ata: Fix kernel pointer leak

On 29.09.2021 15:16, Guo Zhi wrote:

I'd recommend the subject prefix to be just "pata_atp867x: "...

> Pointers should be printed with %p or %px rather than cast to
> 'unsigned long' and pinted with %lx

Printed.

> Change %lx to %p to print the secured pointer.
>
> Signed-off-by: Guo Zhi <[email protected]>
> ---
> drivers/ata/pata_atp867x.c | 10 +++++-----
> 1 file changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/ata/pata_atp867x.c b/drivers/ata/pata_atp867x.c
> index 2bc5fc81efe3..c32b95f48e50 100644
> --- a/drivers/ata/pata_atp867x.c
> +++ b/drivers/ata/pata_atp867x.c
> @@ -447,11 +447,11 @@ static int atp867x_ata_pci_sff_init_host(struct ata_host *host)
> #ifdef ATP867X_DEBUG
> atp867x_check_ports(ap, i);
> #endif
> - ata_port_desc(ap, "cmd 0x%lx ctl 0x%lx",
> - (unsigned long)ioaddr->cmd_addr,
> - (unsigned long)ioaddr->ctl_addr);
> - ata_port_desc(ap, "bmdma 0x%lx",
> - (unsigned long)ioaddr->bmdma_addr);
> + ata_port_desc(ap, "cmd 0x%p ctl 0x%p",
> + ioaddr->cmd_addr,

This line shouldn't be broken up, it's not long at all.

> + ioaddr->ctl_addr);
> + ata_port_desc(ap, "bmdma 0x%p",
> + ioaddr->bmdma_addr);

Hmm, I've looked at the driver and got an imperession it only uses the I/O
ports, not MMIO...

[...]

MBR, Sergey