Yinghai,
Does this workaround have any known side effect to 'other'
platform ? for example, some old platforms don't have such buggy
chipset/FPPA logic. have you tested it on those hotplug capable box ?
If the issue is caused by commit
2debd9289997fc5d1c0043b41201a8b40d5e11d0, should we revert it
first ?
And if the workaround is specific to any vendor's platform, could
we introduce a method of only workaround the specific vendor id /
device id instead of patching generic code ?
Thanks,
Ethan
On Sat, Dec 7, 2013 at 4:22 AM, Yinghai Lu <yinghai@xxxxxxxxxx> wrote:
> During hotplug test on platform, found slot status register does not
> report present status correctly. That present bit does not get cleared
> even after that card is removed.
>
> That problem is caused by commit:
> | commit 2debd9289997fc5d1c0043b41201a8b40d5e11d0
> |
> | PCI: pciehp: Disable/enable link during slot power off/on
>
> It looks like chipset bug, that PresDet bit is "OR" operation between
> sideband input from FPGA, and chipset inband detection from pcie link.
> It does not like if we disable pcie link at first and power off
> other side device, and it has input from inband detection
> always 1.
>
> Workaround: Try turn on link a while after power off.
>
> After this patch, PresDet report correct status when removing or adding
> card later.
>
> Signed-off-by: Yinghai Lu <yinghai@xxxxxxxxxx>
> Cc: Kenji Kaneshige <kaneshige.kenji@xxxxxxxxxxxxxx>
> Cc: <stable@xxxxxxxxxxxxxxx> 3.4+
>
> ---
> drivers/pci/hotplug/pciehp_hpc.c | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
> Index: linux-2.6/drivers/pci/hotplug/pciehp_hpc.c
> ===================================================================
> --- linux-2.6.orig/drivers/pci/hotplug/pciehp_hpc.c
> +++ linux-2.6/drivers/pci/hotplug/pciehp_hpc.c
> @@ -637,6 +637,15 @@ int pciehp_power_off_slot(struct slot *
> }
> ctrl_dbg(ctrl, "%s: SLOTCTRL %x write cmd %x\n", __func__,
> pci_pcie_cap(ctrl->pcie->port) + PCI_EXP_SLTCTL, slot_cmd);
> +
> + /*
> + * Enable link for a while so chipset could settle down
> + * inband presence detection logic
> + */
> + pciehp_link_enable(ctrl);
> + msleep(20);
> + pciehp_link_disable(ctrl);
> +
> return 0;
> }
>
> --
> 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
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html