Praveen Kalamegham wrote:
> I'm trying to get my setup to work with the native PCIe hotplug driver.
>>From my testing, it looks like when the pciehp driver is loaded, it
> expects all the PCIe slots to be powered off. In pciehp_probe(), we
> have:
>
> t_slot->hpc_ops->get_adapter_status(t_slot, &value); /* Check if slot is
> occupied */
> if (value && pciehp_force) {
> rc = pciehp_enable_slot(t_slot);
> if (rc) /* -ENODEV: shouldn't happen, but deal with it */
> value = 0;
> }
> if ((POWER_CTRL(ctrl)) && !value) {
> rc = t_slot->hpc_ops->power_off_slot(t_slot); /* Power off slot if
> not occupied*/
> if (rc)
> goto err_out_free_ctrl_slot;
> }
>
> But pciehp_enable_slot() will return EINVAL if the slot is already
> powered up and ready to go. Consequently pciehp_probe() will attempt to
> power off the slot. If I change the rc conditional after the
> pciehp_enable_slot() call to explicitly specify ENODEV (and thus exclude
> the case of the EINVAL return):
>
> ...
> if (rc == ENODEV) /* no adapter in slot or latch open on slot */
> value = 0;
> }
>
> it seems to work right, the PCIe card in the slot is up initially, I
> modprobe pciehp in, it correctly handles the slot being powered already
> and loads without error, and I'm able to hotplug the card out and back
> in just fine.
>
> But I'm guessing it's like it is for a reason. Can someone explain why
> this is so?
>
As you pointed out, current pciehp_proble() has a bug that causes unexpected
power off if the environment is under the following condition.
- pciehp driver was loaded with pciehp_force option specified, and
- hotplug controller has power controller
I recently posted the patch to fix this bug. Please see.
http://marc.info/?l=linux-pci&m=122871424909137&w=2
I didn't think this fix is not so urgent because I thought the above-mentioned
condition was relatively rare. So it is not scheduled to 2.6.28. Is it urgent
for you?
Thanks,
Kenji Kaneshige
--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
