Beechain Thanks for the tip. My Windows driver is 3.1.11, so I believe you meant 3.1.7?
Might try it later but I think there is also a question of having the right firmware in place. A newer firmware might play badly with an older driver, and vice versa.
My current firmware version is 1.5.45 (not 255.255.65535 as above, that was because of the driver failure).
As for Linux… the output of ethtool refers to the kernel (7.2.0 is the last mainline Linux). The driver-specific version is not advertised anywhere as far as I know, at least not on Linux - you simply get whatever version is included in the kernel itself. One can build a different version from source and load that, though. I might try it later and see how it goes.
Meanwhile, I’ve also tested another kernel (6.19.10-300.fc44.x86_64) and the problem persists.
dmesg:
liveuser@localhost-live:~$ sudo dmesg | grep atlantic
[ 2.599859] atlantic: Detect ATL2FW 105002d
[ 2.607833] atlantic 0000:c3:00.0 eno1: renamed from eth0
[ 4.565775] Modules linked in: thunderbolt hid_multitouch amdgpu amdxcp drm_panel_backlight_quirks gpu_sched drm_buddy drm_ttm_helper ttm drm_exec nvme_tcp i2c_algo_bit drm_suballoc_helper drm_display_helper nvme_fabrics nvme atlantic nvme_core video ghash_clmulni_intel cec nvme_keyring nvme_auth macsec hkdf wmi sp5100_tco uas usb_storage sunrpc be2iscsi bnx2i cnic uio cxgb4i cxgb4 tls cxgb3i cxgb3 mdio libcxgbi libcxgb qla4xxx iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi loop fuse i2c_dev nfnetlink
[ 4.565875] Modules linked in: thunderbolt hid_multitouch amdgpu amdxcp drm_panel_backlight_quirks gpu_sched drm_buddy drm_ttm_helper ttm drm_exec nvme_tcp i2c_algo_bit drm_suballoc_helper drm_display_helper nvme_fabrics nvme atlantic nvme_core video ghash_clmulni_intel cec nvme_keyring nvme_auth macsec hkdf wmi sp5100_tco uas usb_storage sunrpc be2iscsi bnx2i cnic uio cxgb4i cxgb4 tls cxgb3i cxgb3 mdio libcxgbi libcxgb qla4xxx iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi loop fuse i2c_dev nfnetlink
[ 4.565915] Modules linked in: thunderbolt hid_multitouch amdgpu amdxcp drm_panel_backlight_quirks gpu_sched drm_buddy drm_ttm_helper ttm drm_exec nvme_tcp i2c_algo_bit drm_suballoc_helper drm_display_helper nvme_fabrics nvme atlantic nvme_core video ghash_clmulni_intel cec nvme_keyring nvme_auth macsec hkdf wmi sp5100_tco uas usb_storage sunrpc be2iscsi bnx2i cnic uio cxgb4i cxgb4 tls cxgb3i cxgb3 mdio libcxgbi libcxgb qla4xxx iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi loop fuse i2c_dev nfnetlink
[ 4.566311] Modules linked in: thunderbolt hid_multitouch amdgpu amdxcp drm_panel_backlight_quirks gpu_sched drm_buddy drm_ttm_helper ttm drm_exec nvme_tcp i2c_algo_bit drm_suballoc_helper drm_display_helper nvme_fabrics nvme atlantic nvme_core video ghash_clmulni_intel cec nvme_keyring nvme_auth macsec hkdf wmi sp5100_tco uas usb_storage sunrpc be2iscsi bnx2i cnic uio cxgb4i cxgb4 tls cxgb3i cxgb3 mdio libcxgbi libcxgb qla4xxx iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi loop fuse i2c_dev nfnetlink
[ 33.674392] atlantic 0000:c3:00.0 eno1: atlantic: link change old 0 new 2500
[ 1952.828020] atlantic 0000:c3:00.0 eno1: atlantic: link change old 2500 new 0
It still fails to come back up:
liveuser@localhost-live:~$ sudo ip link set eno1 down
liveuser@localhost-live:~$ sudo ip link set eno1 up
RTNETLINK answers: Connection timed out
If I unload and reload the driver:
[ 4599.242316] WARNING: drivers/net/ethernet/aquantia/atlantic/hw_atl2/hw_atl2_utils_fw.c:112 at aq_a2_fw_deinit+0xb9/0xd0 [atlantic], CPU#0: ip/14391
[ 4599.242363] snd_acp_legacy_common snd_intel_dspcfg cfg80211 bluetooth kvm snd_intel_sdw_acpi snd_pci_acp6x snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd_pci_acp5x snd_rn_pci_acp3x amdxdna snd spd5118 snd_acp_config snd_soc_acpi i2c_piix4 amd_pmc joydev snd_pci_acp3x rfkill soundcore wmi_bmof k10temp i2c_smbus pcspkr irqbypass rapl zram lz4hc_compress lz4_compress overlay erofs netfs isofs thunderbolt hid_multitouch amdgpu amdxcp drm_panel_backlight_quirks gpu_sched drm_buddy drm_ttm_helper ttm drm_exec nvme_tcp i2c_algo_bit drm_suballoc_helper drm_display_helper nvme_fabrics nvme atlantic nvme_core video ghash_clmulni_intel cec nvme_keyring nvme_auth macsec hkdf wmi sp5100_tco uas usb_storage sunrpc be2iscsi bnx2i cnic uio cxgb4i cxgb4 tls cxgb3i cxgb3 mdio libcxgbi libcxgb qla4xxx iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi loop fuse i2c_dev nfnetlink
[ 4599.242410] RIP: 0010:aq_a2_fw_deinit+0xb9/0xd0 [atlantic]
[ 4599.242435] aq_nic_deinit.part.0+0xe2/0x120 [atlantic]
[ 4599.242445] aq_ndev_close+0x2e/0x40 [atlantic]
[ 4602.345094] atlantic: Boot code hanged
[ 4602.463373] WARNING: drivers/net/ethernet/aquantia/atlantic/hw_atl2/hw_atl2_utils_fw.c:112 at aq_a2_fw_deinit+0xb9/0xd0 [atlantic], CPU#0: ip/14408
[ 4602.463428] snd_acp_legacy_common snd_intel_dspcfg cfg80211 bluetooth kvm snd_intel_sdw_acpi snd_pci_acp6x snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd_pci_acp5x snd_rn_pci_acp3x amdxdna snd spd5118 snd_acp_config snd_soc_acpi i2c_piix4 amd_pmc joydev snd_pci_acp3x rfkill soundcore wmi_bmof k10temp i2c_smbus pcspkr irqbypass rapl zram lz4hc_compress lz4_compress overlay erofs netfs isofs thunderbolt hid_multitouch amdgpu amdxcp drm_panel_backlight_quirks gpu_sched drm_buddy drm_ttm_helper ttm drm_exec nvme_tcp i2c_algo_bit drm_suballoc_helper drm_display_helper nvme_fabrics nvme atlantic nvme_core video ghash_clmulni_intel cec nvme_keyring nvme_auth macsec hkdf wmi sp5100_tco uas usb_storage sunrpc be2iscsi bnx2i cnic uio cxgb4i cxgb4 tls cxgb3i cxgb3 mdio libcxgbi libcxgb qla4xxx iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi loop fuse i2c_dev nfnetlink
[ 4602.463476] RIP: 0010:aq_a2_fw_deinit+0xb9/0xd0 [atlantic]
[ 4602.463504] aq_nic_deinit.part.0+0xe2/0x120 [atlantic]
[ 4602.463521] aq_ndev_open+0x5f/0x70 [atlantic]
[ 4662.578136] atlantic 0000:c3:00.0: enabling device (0000 -> 0002)
[ 4662.657361] atlantic: Detect ATL2FW 105002d
[ 4662.675990] atlantic 0000:c3:00.0 eno1: renamed from eth0
[ 4671.002189] atlantic 0000:c3:00.0 eno1: atlantic: link change old 0 new 2500
Driver info:
liveuser@localhost-live:~$ sudo modinfo atlantic
filename: /lib/modules/6.19.10-300.fc44.x86_64/kernel/drivers/net/ethernet/aquantia/atlantic/atlantic.ko.xz
description: Marvell (Aquantia) Corporation(R) Network Driver
author: Marvell
license: GPL v2
alias: pci:v00001D6Ad000011C0sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000034C0sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000012C0sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000014C0sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000004C0sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000093C0sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000094C0sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000000C0sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000092B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000091B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000089B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000088B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000087B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000080B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000012B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000011B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000009B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000008B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000007B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad000000B1sv*sd*bc*sc*i*
alias: pci:v00001D6Ad0000D109sv*sd*bc*sc*i*
alias: pci:v00001D6Ad0000D108sv*sd*bc*sc*i*
alias: pci:v00001D6Ad0000D107sv*sd*bc*sc*i*
alias: pci:v00001D6Ad0000D100sv*sd*bc*sc*i*
alias: pci:v00001D6Ad00000001sv*sd*bc*sc*i*
depends: macsec
intree: Y
name: atlantic
retpoline: Y
vermagic: 6.19.10-300.fc44.x86_64 SMP preempt mod_unload
sig_id: PKCS#7
signer: Fedora kernel signing key
sig_key: 60:7E:32:02:08:6B:F8:92:FA:50:0F:4D:DF:E7:D2:38:3E:54:E2:70
sig_hashalgo: sha256
signature: 09:FA:53:52:78:6D:26:A3:1A:AA:8E:19:95:97:3E:30:AC:86:79:4F:
--- CUT ---
parm: aq_itr:Interrupt throttling mode (uint)
parm: aq_itr_tx:TX interrupt throttle rate (uint)
parm: aq_itr_rx:RX interrupt throttle rate (uint)
```!<
It might be a driver problem, in the sense that it apparently fails to initialize after the interface goes down. But the fact that the interface goes down might be unrelated to the driver and indicate perhaps a hardware or firmware problem.
If the root cause turns out to be a hardware fault (we’ll see what Beelink has to say about it) there is little we can do, I’m afraid, except send our unit back or try get a replacement.
P.S. I’ve also noticed that sometimes the driver causes a hang on shutdown and the machine fails to halt and needs to be forcibly powered down.
