Re: [PATCH] PCI: pciehp: Fix the problem that the present bit is not cleared though slot is empty

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



 Hello,

On Wed, Feb 12, 2014 at 8:56 AM, Bjorn Helgaas <bhelgaas@xxxxxxxxxx> wrote:
> On Wed, Feb 12, 2014 at 1:00 AM, Izumi, Taku <izumi.taku@xxxxxxxxxxxxxx> wrote:
>> Hi Bjorn,
>>
>>> Hi Taku,
>>>
>>> Thanks a lot for finding and bisecting this problem.  I recently merged
>>> some patches from Rajat that partially revert 2debd9289997.  Can you try
>>> out my pci/pciehp branch and see whether it is enough to fix the problem
>>> for you?
>>>
>>> Here's the branch:
>>> http://git.kernel.org/cgit/linux/kernel/git/helgaas/pci.git/log/?h=pci/pciehp
>>>
>>> and the specific change Rajat made is:
>>> http://git.kernel.org/cgit/linux/kernel/git/helgaas/pci.git/commit/?h=pci/pciehp&id=b1811d2455f32754cc3d8725bf2e961c
>>> 5eda2a72
>>>
>>> Let me know if that isn't enough to fix the problem you're seeing, and we
>>> can work on it some more.
>>
>>  I tested the kernel checkouted from remotes/origin/pci/pciehp.
>>  The original problem I reported was fixed.
>>  However the sequence of hot-plug operation got changed.
>>  The slot power becomes ON the moment when I insert PCIe card on the slot.
>>  This behavior is unacceptable to me.
>>
>>  (before)
>>    - hot-add
>>      1. insert PCIe card
>>      2. echo 1 > power
>>    - hot-remove
>>      1. echo 0 > power
>>      2. eject PCIe card
>>
>>  (after)
>>    - hot-add
>>     1. insert PCIe card (and automatically power-on)
>>
>>    - hot-remove
>>     1. echo 0 > power
>>     2. eject PCIe card
>
> Huh, that doesn't sound good.  Does your slot have an attention
> button?  Can you collect the "lspci -vvv" output for the Downstream
> Port leading to this slot?

If there is no attention button, then the behavior seen, matches with
what I expected of the code.

Seemingly, the HW does not have a power controller (and attention
button), thus the power is automatically turned on (by HW) when card
is inserted and the link comes up. Thus prior to link state based
hot-plug, "echo 1 > power" used to do a little else than initiate a
rescan for this particular HW.

>
> I think that for ExpressCard and similar devices, we want to turn on
> the device and attach a driver as soon as the card is inserted.  But
> in your case, I assume we want a model like that in Table 2-7 of the
> "PCI Standard Hot-Plug Controller and Subsystem Specification," rev
> 1.0, i.e., "Hot Insertion Initiated via Software UI."  So there must
> be some way to differentiate an ExpressCard slot from a slot like
> yours.

Yes, that would help if there was some way.

Thanks,

Rajat

>
> Bjorn
--
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




[Index of Archives]     [DMA Engine]     [Linux Coverity]     [Linux USB]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Greybus]

  Powered by Linux