netdev - Re: AMD XGBE "phy irq request failed" kernel v5.17-rc2 on V1500B based board

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <57cf8ba6-98a7-5d4a-76d0-4b533da06819@amd.com>
Date:   Fri, 11 Feb 2022 09:48:48 -0600
From:   Tom Lendacky <thomas.lendacky@....com>
To:     Shyam Sundar S K <Shyam-sundar.S-k@....com>,
        Thomas Kupper <thomas@...per.org>
Cc:     netdev@...r.kernel.org
Subject: Re: AMD XGBE "phy irq request failed" kernel v5.17-rc2 on V1500B
 based board

On 2/11/22 03:49, Shyam Sundar S K wrote:
> On 2/11/2022 3:03 PM, Thomas Kupper wrote:
>> Am 08.02.22 um 17:24 schrieb Tom Lendacky:
>>> On 2/7/22 11:59, Thomas Kupper wrote:
>>>> Am 07.02.22 um 16:19 schrieb Shyam Sundar S K:
>>>>> On 2/7/2022 8:02 PM, Tom Lendacky wrote:
>>>>>> On 2/5/22 12:14, Thomas Kupper wrote:
>>>>>>> Am 05.02.22 um 16:51 schrieb Tom Lendacky:
>>>>>>>> On 2/5/22 04:06, Thomas Kupper wrote:

>>
>> Thanks Tom, I now got time to update to 5.17-rc3 and add the 'debug'
>> module parameter. I assume that parameter works with the non-debug
>> kernel? I don't really see any new messages related to the amd-xgbe driver:
>>
>> dmesg right after boot:
>>
>> [    0.000000] Linux version 5.17.0-rc3-tk (jane@...0q-ubu21) (gcc
>> (Ubuntu 11.2.0-7ubuntu2) 11.2.0, GNU ld (GNU Binutils for Ubuntu) 2.37)
>> #12 SMP PREEMPT Tue Feb 8 19:52:19 CET 2022
>> [    0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-5.17.0-rc3-tk
>> root=UUID=8e462830-8ba0-4061-8f23-6f29ce751792 ro console=tty0
>> console=ttyS0,115200n8 amd_xgbe.dyndbg=+p amd_xgbe.debug=0x37
>> ...
>> [    5.275730] amd-xgbe 0000:06:00.1 eth0: net device enabled
>> [    5.277766] amd-xgbe 0000:06:00.2 eth1: net device enabled
>> [    5.665315] amd-xgbe 0000:06:00.2 enp6s0f2: renamed from eth1
>> [    5.696665] amd-xgbe 0000:06:00.1 enp6s0f1: renamed from eth0

Hmmm... that's strange. There should have been some messages issued by the
xgbe-phy-v2.c file from the xgbe_phy_init() routine.

Thomas, if you're up for a bit of kernel hacking, can you remove the
"if (netif_msg_probe(pdata)) {" that wrap the dev_dbg() calls in the
xgbe-phy-v2.c file? There are 5 locations.

>>
>> dmesg right after 'ifconfig enp6s0f2 up'
>>
>> [   88.843454] amd_xgbe:xgbe_alloc_channels: amd-xgbe 0000:06:00.2
>> enp6s0f2: channel-0: cpu=0, node=0


> Can you add this change and see if it solves the problem?
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=68c2d6af1f1e

I would imagine that patch has nothing to do with the real issue. Given
the previous messages of:

> [  648.038655] genirq: Flags mismatch irq 59. 00000000 (enp6s0f2-pcs) vs. 00000000 (enp6s0f2-pcs)
> [  648.048303] amd-xgbe 0000:06:00.2 enp6s0f2: phy irq request failed

There should be no reason for not being able to obtain the IRQ.

I suspect it is something in the BIOS setup that is not correct and thus
the Linux driver is not working properly because of bad input/setup from
the BIOS. This was probably worked around by the driver used in the
OPNsense DEC740 firewall.

Shyam has worked more closely with the embedded area of this device, I'll
let him take it from here.

Thanks,
Tom