[PATCH] tg3: make depend on ZONE_DMA

[PATCH] tg3: make depend on ZONE_DMA

[Rui Salvaterra]

Since 614f4d166eee, on systems with ZONE_DMA unavailable, the Tigon3 driver
fails to initialize, at least on systems with BCM57766 controllers. On a late
2012 Mac Mini:

[    1.226623] tg3 0000:01:00.0: Unable to obtain 64 bit DMA for consistent allocations
[    1.226718] tg3 0000:01:00.0: probe with driver tg3 failed with error -5

Fixes: 614f4d166eee "tg3: Set coherent DMA mask bits to 31 for BCM57766 chipsets"

Cc: stable@vger.kernel.org

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
---
 drivers/net/ethernet/broadcom/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/ethernet/broadcom/Kconfig b/drivers/net/ethernet/broadcom/Kconfig
index f0bac0dd1439..c9875daa043b 100644
--- a/drivers/net/ethernet/broadcom/Kconfig
+++ b/drivers/net/ethernet/broadcom/Kconfig
@@ -124,6 +124,7 @@ config TIGON3
 	tristate "Broadcom Tigon3 support"
 	depends on PCI
 	depends on PTP_1588_CLOCK_OPTIONAL
+	depends on ZONE_DMA
 	select CRC32
 	select PHYLIB
 	help
-- 
2.55.0

Re: [PATCH] tg3: make depend on ZONE_DMA

[Pavan Chebbi]

[-- Attachment #1: Type: text/plain, Size: 783 bytes --]

On Sat, Oct 31, 2026 at 7:05 PM Rui Salvaterra <rsalvaterra@gmail.com> wrote:
>
> Since 614f4d166eee, on systems with ZONE_DMA unavailable, the Tigon3 driver
> fails to initialize, at least on systems with BCM57766 controllers. On a late
> 2012 Mac Mini:
>
> [    1.226623] tg3 0000:01:00.0: Unable to obtain 64 bit DMA for consistent allocations
> [    1.226718] tg3 0000:01:00.0: probe with driver tg3 failed with error -5
>
> Fixes: 614f4d166eee "tg3: Set coherent DMA mask bits to 31 for BCM57766 chipsets"
>
> Cc: stable@vger.kernel.org
>
> Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
> ---
>  drivers/net/ethernet/broadcom/Kconfig | 1 +
>  1 file changed, 1 insertion(+)
>

No! This is not the solution. Please wait while I am trying to fix it.

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 5469 bytes --]

Re: [PATCH] tg3: make depend on ZONE_DMA

[Jason Xing]

On Sat, Oct 31, 2026 at 9:36 PM Rui Salvaterra <rsalvaterra@gmail.com> wrote:
>
> Since 614f4d166eee, on systems with ZONE_DMA unavailable, the Tigon3 driver
> fails to initialize, at least on systems with BCM57766 controllers. On a late
> 2012 Mac Mini:
>
> [    1.226623] tg3 0000:01:00.0: Unable to obtain 64 bit DMA for consistent allocations
> [    1.226718] tg3 0000:01:00.0: probe with driver tg3 failed with error -5
>
> Fixes: 614f4d166eee "tg3: Set coherent DMA mask bits to 31 for BCM57766 chipsets"

The format is not right. And please target net branch if it's a fix by
changing your title to ']PATCH net]'.

>
> Cc: stable@vger.kernel.org
>
> Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
> ---
>  drivers/net/ethernet/broadcom/Kconfig | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/drivers/net/ethernet/broadcom/Kconfig b/drivers/net/ethernet/broadcom/Kconfig
> index f0bac0dd1439..c9875daa043b 100644
> --- a/drivers/net/ethernet/broadcom/Kconfig
> +++ b/drivers/net/ethernet/broadcom/Kconfig
> @@ -124,6 +124,7 @@ config TIGON3
>         tristate "Broadcom Tigon3 support"
>         depends on PCI
>         depends on PTP_1588_CLOCK_OPTIONAL
> +       depends on ZONE_DMA

Did you verify it possibly affects those kernels without DMA config?

The blamed patch seems correct, so I think you need to root cause it
instead of adding a global dependency here.

Thanks,
Jason

>         select CRC32
>         select PHYLIB
>         help
> --
> 2.55.0
>
>

Re: [PATCH] tg3: make depend on ZONE_DMA

[Rui Salvaterra]

Hi, Jason,


On Sun, 1 Nov 2026 at 05:48, Jason Xing <kerneljasonxing@gmail.com> wrote:
>
[snipped]
>
> The blamed patch seems correct, so I think you need to root cause it
> instead of adding a global dependency here.

I'm sorry, perhaps I misunderstood, but *I* need to root cause it? I
never claimed to fully understand the implications of the patch I
sent, nor the intricacies of the tg3 driver. What I *do* understand is
that a supposed "fix" from Broadcom has left the Ethernet controller
on my late 2012 Mac Mini inoperable since the 24th November 2024, when
absolutely nothing changed in my kconfig. That is, by definition, a
regression. I'm entirely available for testing any patches people send
my way, if whoever fully understands the driver and the hardware is
unable to reproduce the issue. However, asking the user to fix a
problem created by a patch from the hardware manufacturer is… an
interesting approach, at best.


Kind regards,

Rui Salvaterra.

Re: [PATCH] tg3: make depend on ZONE_DMA

[Jason Xing]

On Mon, Nov 2, 2026 at 3:41 AM Rui Salvaterra <rsalvaterra@gmail.com> wrote:
>
> Hi, Jason,
>
>
> On Sun, 1 Nov 2026 at 05:48, Jason Xing <kerneljasonxing@gmail.com> wrote:
> >
> [snipped]
> >
> > The blamed patch seems correct, so I think you need to root cause it
> > instead of adding a global dependency here.
>
> I'm sorry, perhaps I misunderstood, but *I* need to root cause it? I
> never claimed to fully understand the implications of the patch I

This is not how things are going here. As long as someone proposes a
patch, they should at least know what they are doing.

> sent, nor the intricacies of the tg3 driver. What I *do* understand is
> that a supposed "fix" from Broadcom has left the Ethernet controller
> on my late 2012 Mac Mini inoperable since the 24th November 2024, when
> absolutely nothing changed in my kconfig. That is, by definition, a
> regression. I'm entirely available for testing any patches people send
> my way, if whoever fully understands the driver and the hardware is
> unable to reproduce the issue. However, asking the user to fix a
> problem created by a patch from the hardware manufacturer is… an
> interesting approach, at best.

I see what you meant; I thought you were trying to make your first
contribution to the kernel :)

With that said, maintainers would pick up this and probably send a fix
with your credit (reported-by tag). Thanks for your report :)

Thanks,
Jason