3.2.11: PCI Express card cannot be re-detected withing cca 60sec timeframe

Yinghai Lu wrote:
> On Wed, Apr 18, 2012 at 10:53 AM, Martin Mokrejs
> <mmokrejs@fold.natur.cuni.cz> wrote:
>>> After you remove USB3 expresscard, you need to
>>>
>>> echo 1 > /sys/devices/pci0000\:00/0000\:00\:1c.7/pcie_link_disable
>>> then
>>> echo 0 > /sys/devices/pci0000\:00/0000\:00\:1c.7/pcie_link_disable
>>
>> So without these two echo commands the is no improvement/fix.
>>
>> [  686.701306] pciehp 0000:00:1c.7:pcie04: pcie_isr: intr_loc 8
>> [  686.701316] pciehp 0000:00:1c.7:pcie04: Presence/Notify input change
>> [  686.701319] pciehp 0000:00:1c.7:pcie04: Card present on Slot(7)
>> [  686.701357] pciehp 0000:00:1c.7:pcie04: Surprise Removal
>> [  686.701390] pciehp 0000:00:1c.7:pcie04: check_link_active: lnk_status = 7011
>> [  686.809763] pciehp 0000:00:1c.7:pcie04: pciehp_check_link_status: lnk_status = 7011
> ...
>> [  686.833678] hub 6-0:1.0: 2 ports detected
>>
>> echo 1 > /sys/devices/pci0000\:00/0000\:00\:1c.7/pcie_link_disable
>>
>> [  716.999938] pcieport 0000:00:1c.7: pcie_link_disable_set: lnk_ctrl = 53
>> [  717.024647] pciehp 0000:00:1c.7:pcie04: pcie_isr: intr_loc 8
>> [  717.024657] pciehp 0000:00:1c.7:pcie04: Presence/Notify input change
>> [  717.024665] pciehp 0000:00:1c.7:pcie04: Card not present on Slot(7)
>> [  717.024723] pciehp 0000:00:1c.7:pcie04: Surprise Removal
>> [  717.024753] pciehp 0000:00:1c.7:pcie04: Disabling domain:bus:device=0000:11:00
>> [  717.024756] pciehp 0000:00:1c.7:pcie04: pciehp_unconfigure_device: domain:bus:dev = 0000:11:00
> ..
>> [  717.045197] pci 0000:11:00.0: freeing pci_dev info
>>
>> echo 0 > /sys/devices/pci0000\:00/0000\:00\:1c.7/pcie_link_disable
>>
>> [  748.456914] pcieport 0000:00:1c.7: pcie_link_disable_set: lnk_ctrl = 40
>>
> 
> so it is chipset problem, it does not detect and report the presence
> of the card after that link is removed.
> 
> Can you try to use acpiphp instead of pciehp?

When I am booted up with CONFIG_HOTPLUG_PCI_ACPI=n and pcie_aspm=off I tested what
happens with several cards. Note that this scenario causes "Unable to request _OSC
control" below and functionally disables my ExpressCard slot:

- Command line: root=/dev/sda5 pciehp.pciehp_debug=1 console=ttyUSB0,115200n8 console=tty0 log_buf_len=409600
+ Command line: root=/dev/sda5 pciehp.pciehp_debug=1 console=ttyUSB0,115200n8 console=tty0 log_buf_len=409600 pcie_aspm=off

-pci0000:00: Requesting ACPI _OSC control (0x1d)   
-pci0000:00: ACPI _OSC control (0x19) granted   
+pci0000:00: Unable to request _OSC control (_OSC support mask:

The express card works although no driver can pickup and therefore nothing was ever logged
through syslog if I plugged in the card,. Still, I could read the lspci output and compare values.

# dmesg > dmesg_without_CONFIG_HOTPLUG_PCI_ACPI_with_pcie_aspm_off.txt 
# lspci -vvxxx -s 00:1c.7 > 1_fresh_bootup.txt
# echo 7 > /proc/sys/kernel/printk

# # inserted DeLock card
# lspci -vvxxx -s 00:1c.7 > 1_inserted_DeLock_unrecognized.txt
# # card removed
# lspci -vvxxx -s 00:1c.7 > 2_unplugged_DeLock_unrecognized.txt
# # re-inserted the card
# lspci -vvxxx -s 00:1c.7 > 3_re-inserted_DeLock_again_unrecognized.txt
# # unplugged and re-inserted the card
# lspci -vvxxx -s 00:1c.7 > 4_unplug_and_re-plug_DeLock_again_unrecognized.txt
# # unplugged DeLock card
# lspci -vvxxx -s 00:1c.7 > 5_unplugged_DeLock_unrecognized.txt

# ## will now play with Evolve_SpeedCard_HC-303SCX with NEC uPD720200 rev. 3
# # inserting Evolve_SpeedCard_HC-303SCX with NEC uPD720200 rev. 3
# lspci -vvxxx -s 00:1c.7 > 6_inserted_Evolve_SpeedCard_HC-303SCX_with_NEC_uPD720200_rev3.txt
# # popped out, grabbing lspci to proof that later
# lspci -vvxxx -s 00:1c.7 > 7_popped_out_Evolve_SpeedCard_HC-303SCX_with_NEC_uPD720200_rev3.txt
# # re-inserting, sits in now for sure
# lspci -vvxxx -s 00:1c.7 > 8_inserted_Evolve_SpeedCard_HC-303SCX_with_NEC_uPD720200_rev3.txt
# # removed
# lspci -vvxxx -s 00:1c.7 > 9_unplugged_Evolve_SpeedCard_HC-303SCX_with_NEC_uPD720200_rev3.txt
# # re-inserting, 2 attempts to get card held mechanically
# lspci -vvxxx -s 00:1c.7 > 9_inserted_Evolve_SpeedCard_HC-303SCX_with_NEC_uPD720200_rev3.txt
# # unplugging card, nothing appeared in syslog
# lspci -vvxxx -s 00:1c.7 > 10_unplugged_Evolve_SpeedCard_HC-303SCX_with_NEC_uPD720200_rev3.txt

# ## now will play with Zalman ZM-EC302 U3 card having also NEC uPD720200 rev. 3
# # inserting
# lspci -vvxxx -s 00:1c.7 > 11_inserted_Zalman_ZM-EC302_U3.txt
# # removed the card
# lspci -vvxxx -s 00:1c.7 > 12_unplugged_Zalman_ZM-EC302_U3.txt
# # re-inserting
# lspci -vvxxx -s 00:1c.7 > 13_re-inserted_Zalman_ZM-EC302_U3.txt
# # removing
# lspci -vvxxx -s 00:1c.7 > 14_unplugged_Zalman_ZM-EC302_U3.txt

# # inserting Axago-ECF-X1_FireWireCard    
# lspci -vvxxx -s 00:1c.7 > 15_inserted_Axago-ECF-X1_FireWireCard.txt
# # removed Axago-ECF-X1_FireWireCard
# lspci -vvxxx -s 00:1c.7 > 16_unplugged_Axago-ECF-X1_FireWireCard.txt
# # re-inserted
# lspci -vvxxx -s 00:1c.7 > 17_re-inserted_Axago-ECF-X1_FireWireCard.txt
# # unplugged
# lspci -vvxxx -s 00:1c.7 > 18_unplugged_Axago-ECF-X1_FireWireCard.txt
#

# diff -u -w 1_fresh_bootup.txt 1_inserted_DeLock_unrecognized.txt

#

I would get same difference against teh other NEC-based card but so save output I rather show
that when the card is in a slot it is in same state.

# diff -u -w 8_inserted_Evolve_SpeedCard_HC-303SCX_with_NEC_uPD720200_rev3.txt 11_inserted_Zalman_ZM-EC302_U3.txt 
#

First, if hardware does not report Changed PresDet+ and always says "PresDet-" linux could
keep do the comparisons on its own. Second, why does the NEC-chip "ignore?" the pcie_aspm=off
and mangles with the LnkCtl: ASPM? Unlike DeLock eSATA and Axago FireWire cards?
Or, does it make sense at all? ;-)

All of this was with plain 3.4-rc5.
Thanks,
Martin
--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

3.2.11: PCI Express card cannot be re-detected withing cca 60sec timeframe

Commit Message

Patch