2021-06-04 12:31:41

by Michael Walle

Subject: [PATCH net-next] net: enetc: use get/put_unaligned() for mac address handling

The supplied buffer for the MAC address might not be aligned. Thus
doing a 32bit (or 16bit) access could be on an unaligned address. For
now, enetc is only used on aarch64 which can do unaligned accesses, thus
there is no error. In any case, be correct and use the get/put_unaligned()
helpers.

Signed-off-by: Michael Walle <[email protected]>
---
drivers/net/ethernet/freescale/enetc/enetc_pf.c | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/freescale/enetc/enetc_pf.c b/drivers/net/ethernet/freescale/enetc/enetc_pf.c
index 31274325159a..a96d2acb5e11 100644
--- a/drivers/net/ethernet/freescale/enetc/enetc_pf.c
+++ b/drivers/net/ethernet/freescale/enetc/enetc_pf.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: (GPL-2.0+ OR BSD-3-Clause)
/* Copyright 2017-2019 NXP */

+#include <asm/unaligned.h>
#include <linux/mdio.h>
#include <linux/module.h>
#include <linux/fsl/enetc_mdio.h>
@@ -17,15 +18,15 @@ static void enetc_pf_get_primary_mac_addr(struct enetc_hw *hw, int si, u8 *addr)
u32 upper = __raw_readl(hw->port + ENETC_PSIPMAR0(si));
u16 lower = __raw_readw(hw->port + ENETC_PSIPMAR1(si));

- *(u32 *)addr = upper;
- *(u16 *)(addr + 4) = lower;
+ put_unaligned(upper, (u32 *)addr);
+ put_unaligned(lower, (u16 *)(addr + 4));
}

static void enetc_pf_set_primary_mac_addr(struct enetc_hw *hw, int si,
const u8 *addr)
{
- u32 upper = *(const u32 *)addr;
- u16 lower = *(const u16 *)(addr + 4);
+ u32 upper = get_unaligned((const u32 *)addr);
+ u16 lower = get_unaligned((const u16 *)(addr + 4));

__raw_writel(upper, hw->port + ENETC_PSIPMAR0(si));
__raw_writew(lower, hw->port + ENETC_PSIPMAR1(si));
--
2.20.1

2021-06-04 12:46:10

by Heiner Kallweit

Subject: Re: [PATCH net-next] net: enetc: use get/put_unaligned() for mac address handling

On 04.06.2021 14:30, Michael Walle wrote:
> The supplied buffer for the MAC address might not be aligned. Thus
> doing a 32bit (or 16bit) access could be on an unaligned address. For
> now, enetc is only used on aarch64 which can do unaligned accesses, thus
> there is no error. In any case, be correct and use the get/put_unaligned()
> helpers.
>
> Signed-off-by: Michael Walle <[email protected]>
> ---
> drivers/net/ethernet/freescale/enetc/enetc_pf.c | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/ethernet/freescale/enetc/enetc_pf.c b/drivers/net/ethernet/freescale/enetc/enetc_pf.c
> index 31274325159a..a96d2acb5e11 100644
> --- a/drivers/net/ethernet/freescale/enetc/enetc_pf.c
> +++ b/drivers/net/ethernet/freescale/enetc/enetc_pf.c
> @@ -1,6 +1,7 @@
> // SPDX-License-Identifier: (GPL-2.0+ OR BSD-3-Clause)
> /* Copyright 2017-2019 NXP */
>
> +#include <asm/unaligned.h>
> #include <linux/mdio.h>
> #include <linux/module.h>
> #include <linux/fsl/enetc_mdio.h>
> @@ -17,15 +18,15 @@ static void enetc_pf_get_primary_mac_addr(struct enetc_hw *hw, int si, u8 *addr)
> u32 upper = __raw_readl(hw->port + ENETC_PSIPMAR0(si));
> u16 lower = __raw_readw(hw->port + ENETC_PSIPMAR1(si));
>
> - *(u32 *)addr = upper;
> - *(u16 *)(addr + 4) = lower;
> + put_unaligned(upper, (u32 *)addr);
> + put_unaligned(lower, (u16 *)(addr + 4));

I think you want to write little endian, therefore on a BE platform
this code may be wrong. Better use put_unaligned_le32?
By using these versions of the unaligned helpers you could also
remove the pointer cast.

> }
>
> static void enetc_pf_set_primary_mac_addr(struct enetc_hw *hw, int si,
> const u8 *addr)
> {
> - u32 upper = *(const u32 *)addr;
> - u16 lower = *(const u16 *)(addr + 4);
> + u32 upper = get_unaligned((const u32 *)addr);
> + u16 lower = get_unaligned((const u16 *)(addr + 4));
>
> __raw_writel(upper, hw->port + ENETC_PSIPMAR0(si));
> __raw_writew(lower, hw->port + ENETC_PSIPMAR1(si));
>

2021-06-04 13:31:02

by Michael Walle

Subject: Re: [PATCH net-next] net: enetc: use get/put_unaligned() for mac address handling

Am 2021-06-04 14:44, schrieb Heiner Kallweit:
> On 04.06.2021 14:30, Michael Walle wrote:
>> The supplied buffer for the MAC address might not be aligned. Thus
>> doing a 32bit (or 16bit) access could be on an unaligned address. For
>> now, enetc is only used on aarch64 which can do unaligned accesses,
>> thus
>> there is no error. In any case, be correct and use the
>> get/put_unaligned()
>> helpers.
>>
>> Signed-off-by: Michael Walle <[email protected]>
>> ---
>> drivers/net/ethernet/freescale/enetc/enetc_pf.c | 9 +++++----
>> 1 file changed, 5 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/net/ethernet/freescale/enetc/enetc_pf.c
>> b/drivers/net/ethernet/freescale/enetc/enetc_pf.c
>> index 31274325159a..a96d2acb5e11 100644
>> --- a/drivers/net/ethernet/freescale/enetc/enetc_pf.c
>> +++ b/drivers/net/ethernet/freescale/enetc/enetc_pf.c
>> @@ -1,6 +1,7 @@
>> // SPDX-License-Identifier: (GPL-2.0+ OR BSD-3-Clause)
>> /* Copyright 2017-2019 NXP */
>>
>> +#include <asm/unaligned.h>
>> #include <linux/mdio.h>
>> #include <linux/module.h>
>> #include <linux/fsl/enetc_mdio.h>
>> @@ -17,15 +18,15 @@ static void enetc_pf_get_primary_mac_addr(struct
>> enetc_hw *hw, int si, u8 *addr)
>> u32 upper = __raw_readl(hw->port + ENETC_PSIPMAR0(si));
>> u16 lower = __raw_readw(hw->port + ENETC_PSIPMAR1(si));
>>
>> - *(u32 *)addr = upper;
>> - *(u16 *)(addr + 4) = lower;
>> + put_unaligned(upper, (u32 *)addr);
>> + put_unaligned(lower, (u16 *)(addr + 4));
>
> I think you want to write little endian, therefore on a BE platform
> this code may be wrong. Better use put_unaligned_le32?
> By using these versions of the unaligned helpers you could also
> remove the pointer cast.

I wasn't sure about the endianness. Might be the case, that on BE
platforms, the endianess of the register will swap too. (OTOH I
could use aarch64 with BE..)

But I'm fine with changing it. I'd presume this being the only
platform for now it doesn't really matter.

-michael

2021-06-04 13:36:33

by Claudiu Manoil

Subject: RE: [PATCH net-next] net: enetc: use get/put_unaligned() for mac address handling

> -----Original Message-----
> From: Heiner Kallweit <[email protected]>
> Sent: Friday, June 4, 2021 3:44 PM
[...]
>
> On 04.06.2021 14:30, Michael Walle wrote:
> > The supplied buffer for the MAC address might not be aligned. Thus
> > doing a 32bit (or 16bit) access could be on an unaligned address. For
> > now, enetc is only used on aarch64 which can do unaligned accesses, thus
> > there is no error. In any case, be correct and use the get/put_unaligned()
> > helpers.
> >
> > Signed-off-by: Michael Walle <[email protected]>
> > ---
> > drivers/net/ethernet/freescale/enetc/enetc_pf.c | 9 +++++----
> > 1 file changed, 5 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/freescale/enetc/enetc_pf.c
> b/drivers/net/ethernet/freescale/enetc/enetc_pf.c
> > index 31274325159a..a96d2acb5e11 100644
> > --- a/drivers/net/ethernet/freescale/enetc/enetc_pf.c
> > +++ b/drivers/net/ethernet/freescale/enetc/enetc_pf.c
> > @@ -1,6 +1,7 @@
> > // SPDX-License-Identifier: (GPL-2.0+ OR BSD-3-Clause)
> > /* Copyright 2017-2019 NXP */
> >
> > +#include <asm/unaligned.h>
> > #include <linux/mdio.h>
> > #include <linux/module.h>
> > #include <linux/fsl/enetc_mdio.h>
> > @@ -17,15 +18,15 @@ static void enetc_pf_get_primary_mac_addr(struct
> enetc_hw *hw, int si, u8 *addr)
> > u32 upper = __raw_readl(hw->port + ENETC_PSIPMAR0(si));
> > u16 lower = __raw_readw(hw->port + ENETC_PSIPMAR1(si));
> >
> > - *(u32 *)addr = upper;
> > - *(u16 *)(addr + 4) = lower;
> > + put_unaligned(upper, (u32 *)addr);
> > + put_unaligned(lower, (u16 *)(addr + 4));
>
> I think you want to write little endian, therefore on a BE platform
> this code may be wrong. Better use put_unaligned_le32?
> By using these versions of the unaligned helpers you could also
> remove the pointer cast.
>

+1