On Wed, Aug 20, 2025 at 10:30:17PM +0200, Andrew Lunn wrote:

quoted

+int mucse_mbx_get_capability(struct mucse_hw *hw)
+{
+	struct hw_abilities ability = {};
+	int try_cnt = 3;
+	int err = -EIO;
+
+	while (try_cnt--) {
+		err = mucse_fw_get_capability(hw, &ability);
+		if (err)
+			continue;
+		hw->pfvfnum = le16_to_cpu(ability.pfnum);
+		hw->fw_version = le32_to_cpu(ability.fw_version);
+		hw->usecstocount = le32_to_cpu(ability.axi_mhz);

If you can get it from the hardware, why do you need to initialise it
in the earlier patch?

I guess you have a bootstrap problem, you need it to get it. But
cannot you just initialise it to a single pessimistic value which will
work well enough for all hardware variants until you can actually ask
the hardware?

    Andrew

It is a problem related with fw version. Older fw may return with axi_mhz
0, So I init a no-zero default value first. Also, I missed to check the axi_mhz
here. The 'usecstocount' is removed in v6, I will update here like this in
the patch which truely use 'usecstocount':

if (le32_to_cpu(ability.axi_mhz))
	hw->usecstocount = le32_to_cpu(ability.axi_mhz);
/* else keep use the default value */

Thanks for your feedback.