2022-06-10 10:11:01

by Oleksij Rempel

Subject: [PATCH net-next v1 1/1] net: macb: fix negative max_mtu size for sama5d3

On Microchip SAMA5D3 the JML will return null. So, after header and FCS length
subtraction we will get negative max_mtu size. This issue was directly
affecting DSA drivers with MTU support (for example KSZ9477).

Signed-off-by: Oleksij Rempel <[email protected]>
---
drivers/net/ethernet/cadence/macb_main.c | 20 +++++++++++++++++---
1 file changed, 17 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/cadence/macb_main.c b/drivers/net/ethernet/cadence/macb_main.c
index d89098f4ede8..c7e1c9ac9809 100644
--- a/drivers/net/ethernet/cadence/macb_main.c
+++ b/drivers/net/ethernet/cadence/macb_main.c
@@ -4913,10 +4913,24 @@ static int macb_probe(struct platform_device *pdev)

/* MTU range: 68 - 1500 or 10240 */
dev->min_mtu = GEM_MTU_MIN_SIZE;
- if (bp->caps & MACB_CAPS_JUMBO)
- dev->max_mtu = gem_readl(bp, JML) - ETH_HLEN - ETH_FCS_LEN;
- else
+ if (bp->caps & MACB_CAPS_JUMBO) {
+ u32 val;
+
+ if (bp->jumbo_max_len)
+ val = bp->jumbo_max_len;
+ else
+ val = gem_readl(bp, JML);
+
+ if (val < ETH_DATA_LEN) {
+ dev_warn(&pdev->dev, "Suspicious max MTU size (%u), overwriting to %u\n",
+ val, ETH_DATA_LEN);
+ dev->max_mtu = ETH_DATA_LEN;
+ } else {
+ dev->max_mtu = val - ETH_HLEN - ETH_FCS_LEN;
+ }
+ } else {
dev->max_mtu = ETH_DATA_LEN;
+ }

if (bp->caps & MACB_CAPS_BD_RD_PREFETCH) {
val = GEM_BFEXT(RXBD_RDBUFF, gem_readl(bp, DCFG10));
--
2.30.2

2022-06-10 13:59:55

by Claudiu Beznea

Subject: Re: [PATCH net-next v1 1/1] net: macb: fix negative max_mtu size for sama5d3

On 10.06.2022 12:45, Oleksij Rempel wrote:
> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
>
> On Microchip SAMA5D3 the JML will return null.

Should be because it hasn't been configured yet. It is configured later on
macb_init_hw() which is called on open.

So, after header and FCS length
> subtraction we will get negative max_mtu size. This issue was directly
> affecting DSA drivers with MTU support (for example KSZ9477).
>
> Signed-off-by: Oleksij Rempel <[email protected]>
> ---
> drivers/net/ethernet/cadence/macb_main.c | 20 +++++++++++++++++---
> 1 file changed, 17 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/ethernet/cadence/macb_main.c b/drivers/net/ethernet/cadence/macb_main.c
> index d89098f4ede8..c7e1c9ac9809 100644
> --- a/drivers/net/ethernet/cadence/macb_main.c
> +++ b/drivers/net/ethernet/cadence/macb_main.c
> @@ -4913,10 +4913,24 @@ static int macb_probe(struct platform_device *pdev)
>
> /* MTU range: 68 - 1500 or 10240 */
> dev->min_mtu = GEM_MTU_MIN_SIZE;
> - if (bp->caps & MACB_CAPS_JUMBO)
> - dev->max_mtu = gem_readl(bp, JML) - ETH_HLEN - ETH_FCS_LEN;
> - else
> + if (bp->caps & MACB_CAPS_JUMBO) {
> + u32 val;
> +
> + if (bp->jumbo_max_len)
> + val = bp->jumbo_max_len;
> + else
> + val = gem_readl(bp, JML);
> +
> + if (val < ETH_DATA_LEN) {
> + dev_warn(&pdev->dev, "Suspicious max MTU size (%u), overwriting to %u\n",
> + val, ETH_DATA_LEN);
> + dev->max_mtu = ETH_DATA_LEN;
> + } else {
> + dev->max_mtu = val - ETH_HLEN - ETH_FCS_LEN;
> + }
> + } else {

Maybe, simply would be to also check bp->jumbo_max_len and have something like:
if ((bp->caps & MACB_CAPS_JUMBO) && bp->jumbo_max_len)
dev->max_mtu = bp->jumbo_max_len - ETH_HLEN - ETH_FCS_LEN;

if that works as it is the same condition used when updating JML in
macb_init_hw().

Thank you,
Claudiu Beznea

> dev->max_mtu = ETH_DATA_LEN;
> + }
>
> if (bp->caps & MACB_CAPS_BD_RD_PREFETCH) {
> val = GEM_BFEXT(RXBD_RDBUFF, gem_readl(bp, DCFG10));
> --
> 2.30.2
>