Re: [PATCH net 4/5] net: hns3: differentiate autoneg default values between copper and fiber
From: Simon Horman <horms@kernel.org>
Date: 2026-06-18 15:34:56
Also in:
lkml
This is an AI-generated review of your patch. The human sending this email has considered the AI review valid, or at least plausible. Full review at: https://sashiko.dev --- net: hns3: differentiate autoneg default values between copper and fiber This commit addresses a link loss issue during driver initialization on optical ports connected to forced-mode remote switches. It modifies hclge_init_ae_dev to set autonegotiation based on media type. Copper ports retain the AUTONEG_ENABLE default, while optical ports inherit the firmware pre-configured autoneg setting.
quoted hunk ↗ jump to hunk
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c --- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c +++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c@@ -11916,6 +11916,9 @@ static int hclge_init_ae_dev(struct hnae3_ae_dev *ae_dev) if (ret) goto err_ptp_uninit; + if (hdev->hw.mac.media_type != HNAE3_MEDIA_TYPE_COPPER) + hdev->hw.mac.req_autoneg = hdev->hw.mac.autoneg; +
[Severity: High]
Does this apply the wrong speed when autonegotiation is disabled on fiber
ports?
During driver initialization, hclge_update_port_info() dynamically probes
the SFP module and updates mac->speed, but does not update mac->req_speed,
which retains its statically configured default capability.
When hdev->hw.mac.autoneg is 0 (forced-mode), req_autoneg becomes 0, and
then hclge_set_autoneg_speed_dup() uses req_speed:
drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c:hclge_set_autoneg_speed_dup() {
...
if (!hdev->hw.mac.req_autoneg) {
ret = hclge_cfg_mac_speed_dup_hw(hdev, hdev->hw.mac.req_speed,
...
}
Could this override the dynamically probed SFP speed with the static default
speed?
[Severity: High]
Does this unconditionally disable autonegotiation for V1 optical hardware?
Looking at hclge_update_port_info(), it only queries SFP information via
hclge_get_sfp_info() for V2 hardware. For V1 hardware, it falls back to
hclge_get_sfp_speed(), which leaves mac->autoneg uninitialized (0):
drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c:hclge_update_port_info() {
...
if (hdev->ae_dev->dev_version >= HNAE3_DEVICE_VERSION_V2) {
speed = mac->speed;
ret = hclge_get_sfp_info(hdev, mac);
} else {
speed = HCLGE_MAC_SPEED_UNKNOWN;
ret = hclge_get_sfp_speed(hdev, &speed);
}
...
}
If mac->autoneg is left as 0, wouldn't req_autoneg unconditionally be set
to 0 here, overriding the speed and breaking connectivity for V1 devices?
ret = hclge_set_autoneg_speed_dup(hdev);
if (ret) {
dev_err(&pdev->dev,