[PATCH] fixup! net: designware_eqs: set upper 32bit address for DMA descriptors to support 64-bit addressing

[PATCH] fixup! net: designware_eqs: set upper 32bit address for DMA descriptors to support 64-bit addressing

[Ahmad Fatoum]

net: designware_eqos: fix left shift count >= width of type

CI builds with -Werror, including -Werror=shift-count-overflow, which
fails for STM32MP enabled platform that use EQOS on a 32-bit platform.

As des1 will have always been programmed to zero on a 32-bit platform,
we can just suppress the warning with a cast.

Cc: zhengxiaojun <zheng.xiaojun@foxmail.com>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
---
 drivers/net/designware_eqos.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/designware_eqos.c b/drivers/net/designware_eqos.c
index 332dc7eb7366..4011b006dd70 100644
--- a/drivers/net/designware_eqos.c
+++ b/drivers/net/designware_eqos.c
@@ -955,8 +955,8 @@ void eqos_remove(struct device *dev)
 
 	mdiobus_unregister(&eqos->miibus);
 
-	dma = eqos->rx_descs[0].des1;
-	dma = (dma << 32) + eqos->rx_descs[0].des0;
+	dma = (u64)eqos->rx_descs[0].des1 << 32;
+	dma |= eqos->rx_descs[0].des0;
 	dma_free(phys_to_virt(dma));
 	dma_free_coherent(DMA_DEVICE_BROKEN,
 			  eqos->tx_descs, 0, EQOS_DESCRIPTORS_SIZE);
-- 
2.47.3

Re: [PATCH] fixup! net: designware_eqs: set upper 32bit address for DMA descriptors to support 64-bit addressing

[Sascha Hauer]

On 2026-08-05 16:21, Ahmad Fatoum wrote:
> net: designware_eqos: fix left shift count >= width of type
> 
> CI builds with -Werror, including -Werror=shift-count-overflow, which
> fails for STM32MP enabled platform that use EQOS on a 32-bit platform.
> 
> As des1 will have always been programmed to zero on a 32-bit platform,
> we can just suppress the warning with a cast.
> 
> Cc: zhengxiaojun <zheng.xiaojun@foxmail.com>
> Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
> ---
>  drivers/net/designware_eqos.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/designware_eqos.c b/drivers/net/designware_eqos.c
> index 332dc7eb7366..4011b006dd70 100644
> --- a/drivers/net/designware_eqos.c
> +++ b/drivers/net/designware_eqos.c
> @@ -955,8 +955,8 @@ void eqos_remove(struct device *dev)
>  
>  	mdiobus_unregister(&eqos->miibus);
>  
> -	dma = eqos->rx_descs[0].des1;
> -	dma = (dma << 32) + eqos->rx_descs[0].des0;
> +	dma = (u64)eqos->rx_descs[0].des1 << 32;
> +	dma |= eqos->rx_descs[0].des0;

I also stumbled upon this and was halfway through implementing an
alternative: keep the virtual address around in private data so that
we can directly use it instead of reconstructing it from the descriptor
ring. I've sent it out now.

Sascha

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |