We have a new ethernet card that is supported by the atlantic driver:
01:00.0 Ethernet controller [0200]: Aquantia Corp. AQC107 NBase-T/IEEE 802.3bz Ethernet Controller [AQtion] [1d6a:07b1] (rev 02)
But the driver failed to probe the device:
kernel: atlantic: Bad FW version detected: 400001e
kernel: atlantic: probe of 0000:01:00.0 failed with error -95
As a pure guesswork, simply adding the firmware version to the driver
can make it function. Doing iperf3 as a smoketest doesn't show any
abnormality either.
Signed-off-by: Kai-Heng Feng <[email protected]>
---
drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c b/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
index 73c0f41df8d8..0b4cd1c0e022 100644
--- a/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
+++ b/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
@@ -46,6 +46,7 @@
#define HW_ATL_FW_VER_1X 0x01050006U
#define HW_ATL_FW_VER_2X 0x02000000U
#define HW_ATL_FW_VER_3X 0x03000000U
+#define HW_ATL_FW_VER_4X 0x0400001EU
#define FORCE_FLASHLESS 0
@@ -81,6 +82,9 @@ int hw_atl_utils_initfw(struct aq_hw_s *self, const struct aq_fw_ops **fw_ops)
} else if (hw_atl_utils_ver_match(HW_ATL_FW_VER_3X,
self->fw_ver_actual) == 0) {
*fw_ops = &aq_fw_2x_ops;
+ } else if (hw_atl_utils_ver_match(HW_ATL_FW_VER_4X,
+ self->fw_ver_actual) == 0) {
+ *fw_ops = &aq_fw_2x_ops;
} else {
aq_pr_err("Bad FW version detected: %x\n",
self->fw_ver_actual);
--
2.17.1
From: Kai-Heng Feng <[email protected]>
Date: Tue, 7 Jul 2020 14:38:28 +0800
> We have a new ethernet card that is supported by the atlantic driver:
> 01:00.0 Ethernet controller [0200]: Aquantia Corp. AQC107 NBase-T/IEEE 802.3bz Ethernet Controller [AQtion] [1d6a:07b1] (rev 02)
>
> But the driver failed to probe the device:
> kernel: atlantic: Bad FW version detected: 400001e
> kernel: atlantic: probe of 0000:01:00.0 failed with error -95
>
> As a pure guesswork, simply adding the firmware version to the driver
Please don't send "pure guessworks" to net-fixes tree. You should have
reported this as a bug to LKML and/or atlantic team, so we could issue
it.
> can make it function. Doing iperf3 as a smoketest doesn't show any
> abnormality either.
>
> Signed-off-by: Kai-Heng Feng <[email protected]>
> ---
> drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c b/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
> index 73c0f41df8d8..0b4cd1c0e022 100644
> --- a/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
> +++ b/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
> @@ -46,6 +46,7 @@
> #define HW_ATL_FW_VER_1X 0x01050006U
> #define HW_ATL_FW_VER_2X 0x02000000U
> #define HW_ATL_FW_VER_3X 0x03000000U
> +#define HW_ATL_FW_VER_4X 0x0400001EU
>
> #define FORCE_FLASHLESS 0
>
> @@ -81,6 +82,9 @@ int hw_atl_utils_initfw(struct aq_hw_s *self, const struct aq_fw_ops **fw_ops)
> } else if (hw_atl_utils_ver_match(HW_ATL_FW_VER_3X,
> self->fw_ver_actual) == 0) {
> *fw_ops = &aq_fw_2x_ops;
> + } else if (hw_atl_utils_ver_match(HW_ATL_FW_VER_4X,
> + self->fw_ver_actual) == 0) {
> + *fw_ops = &aq_fw_2x_ops;
> } else {
> aq_pr_err("Bad FW version detected: %x\n",
> self->fw_ver_actual);
> --
> 2.17.1
> On Jul 7, 2020, at 16:46, Alexander Lobakin <[email protected]> wrote:
>
> From: Kai-Heng Feng <[email protected]>
> Date: Tue, 7 Jul 2020 14:38:28 +0800
>
>> We have a new ethernet card that is supported by the atlantic driver:
>> 01:00.0 Ethernet controller [0200]: Aquantia Corp. AQC107 NBase-T/IEEE 802.3bz Ethernet Controller [AQtion] [1d6a:07b1] (rev 02)
>>
>> But the driver failed to probe the device:
>> kernel: atlantic: Bad FW version detected: 400001e
>> kernel: atlantic: probe of 0000:01:00.0 failed with error -95
>>
>> As a pure guesswork, simply adding the firmware version to the driver
>
> Please don't send "pure guessworks" to net-fixes tree.
Is matching a new firmware version really that different to matching a new PCI ID or USB ID?
> You should have
> reported this as a bug to LKML and/or atlantic team, so we could issue
> it.
Right, I'll wait for a new patch from atlantic team.
Kai-Heng
>
>> can make it function. Doing iperf3 as a smoketest doesn't show any
>> abnormality either.
>>
>> Signed-off-by: Kai-Heng Feng <[email protected]>
>> ---
>> drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c | 4 ++++
>> 1 file changed, 4 insertions(+)
>>
>> diff --git a/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c b/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
>> index 73c0f41df8d8..0b4cd1c0e022 100644
>> --- a/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
>> +++ b/drivers/net/ethernet/aquantia/atlantic/hw_atl/hw_atl_utils.c
>> @@ -46,6 +46,7 @@
>> #define HW_ATL_FW_VER_1X 0x01050006U
>> #define HW_ATL_FW_VER_2X 0x02000000U
>> #define HW_ATL_FW_VER_3X 0x03000000U
>> +#define HW_ATL_FW_VER_4X 0x0400001EU
>>
>> #define FORCE_FLASHLESS 0
>>
>> @@ -81,6 +82,9 @@ int hw_atl_utils_initfw(struct aq_hw_s *self, const struct aq_fw_ops **fw_ops)
>> } else if (hw_atl_utils_ver_match(HW_ATL_FW_VER_3X,
>> self->fw_ver_actual) == 0) {
>> *fw_ops = &aq_fw_2x_ops;
>> + } else if (hw_atl_utils_ver_match(HW_ATL_FW_VER_4X,
>> + self->fw_ver_actual) == 0) {
>> + *fw_ops = &aq_fw_2x_ops;
>> } else {
>> aq_pr_err("Bad FW version detected: %x\n",
>> self->fw_ver_actual);
>> --
>> 2.17.1
From: Alexander Lobakin <[email protected]>
Date: Tue, 7 Jul 2020 11:46:57 +0300
> From: Kai-Heng Feng <[email protected]>
> Date: Tue, 7 Jul 2020 14:38:28 +0800
>
>> We have a new ethernet card that is supported by the atlantic driver:
>> 01:00.0 Ethernet controller [0200]: Aquantia Corp. AQC107 NBase-T/IEEE 802.3bz Ethernet Controller [AQtion] [1d6a:07b1] (rev 02)
>>
>> But the driver failed to probe the device:
>> kernel: atlantic: Bad FW version detected: 400001e
>> kernel: atlantic: probe of 0000:01:00.0 failed with error -95
>>
>> As a pure guesswork, simply adding the firmware version to the driver
>
> Please don't send "pure guessworks" to net-fixes tree. You should have
> reported this as a bug to LKML and/or atlantic team, so we could issue
> it.
Production hardware is shipping to customers and the driver
maintainers didn't add support for this ID yet? What is that
"atlantic team" waiting for?
Honestly I don't blame someone for posting a patch like this to get it
to work.
>>> We have a new ethernet card that is supported by the atlantic driver:
>>> 01:00.0 Ethernet controller [0200]: Aquantia Corp. AQC107 NBase-T/IEEE
> 802.3bz Ethernet Controller [AQtion] [1d6a:07b1] (rev 02)
>>>
>>> But the driver failed to probe the device:
>>> kernel: atlantic: Bad FW version detected: 400001e
>>> kernel: atlantic: probe of 0000:01:00.0 failed with error -95
>>>
>>> As a pure guesswork, simply adding the firmware version to the driver
>>
>> Please don't send "pure guessworks" to net-fixes tree. You should have
>> reported this as a bug to LKML and/or atlantic team, so we could issue
>> it.
>
> Production hardware is shipping to customers and the driver
> maintainers didn't add support for this ID yet? What is that
> "atlantic team" waiting for?
>
> Honestly I don't blame someone for posting a patch like this to get it
> to work.
Me too ;)
We've discussed this with Kai Heng internally, that was really an engineering
sample.
We'll repost this patch soon with some more changes required for that new FW.
Thanks,
Igor