Closing this bug with Won't fix as Hardy is no longer supported.
Please feel free to open a new bug report if you're still experiencing this on
a newer release (Bionic 18.04.3 / Disco 19.04)
Thanks!
** Changed in: linux (Ubuntu)
Status: Incomplete => Won't Fix
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/207422
Title:
/sysfs problems with SDIO/MMC and sd8686 Wlan driver
Status in acton:
New
Status in linux package in Ubuntu:
Won't Fix
Bug description:
Binary package hint: linux-source-2.6.24
We've run into a problem on Ubuntu Mobile ( based on Hardy ) with the
Marvell sd8686 driver which seems to be caused by the SDIO core
driver.
The short description is that when NetworkManager looks up the
net.physical_device ( or net.originating_device ) using HAL, the node
it finds is missing a driver property. NetworkManager sometimes uses
the value of this property to gate driver-specific logic. Sometimes,
in the past, patches have crept in which don't check for NULL, and NM
crashes. We've had to patch NM in the past due to this problem.
I've looked at the sd8686 driver source, and it just sets the value
"sdio_8686" in the sdio_driver struct ( see
LUM/ubuntu/wireless/marvell/8686_wlan/if/if_sdio.c ) and passes it to
the sdio code via the call sdio_register_driver().
In order to follow this bug, please refer to the attached lshal.log
file.
The physical device that NM is returned is a property of the following
UDI ( note - I also include the corresponding sysfs_path ):
'/org/freedesktop/Hal/devices/net_00_10_61_64_10_10' [
/sys/devices/pci0000:00/0000:00:1e.2/mmc_host/mmc2/mmc2:0001/net/wlan0
]
The net.physical_device ( or net.originating_device ) property has the
following value:
'/org/freedesktop/Hal/devices/net/pci_8086_811e_mmc_card_rca1'
The abbreviated sysfs_path is: ../mmc_host/mmc2/mmc2:0001
This node in the Hal output has not property ( info.linux_driver ),
nor does the above sysfs_path have a link called "driver".
You'll note however that there's another node that has parent equal to
the net.physical_device ( ie. mmc_card_rca1 ). It's UDI is:
'/org/freedesktop/Hal/devices/net/pci_8086_811e_mmc_card_rca1_sdio1'
...and this node has a driver property. In fact, it's sysfs_path is a
subdirectory of the net.physical_device's sysfs_path:
/sys/devices/pci0000:00/0000:00:1e.2/mmc_host/mmc2/mmc2:0001/mmc2:0001:1
In fact, all of the driver specific values are present in this
directory ( ie. vendor, class, type, etc... ).
So, from all the evidence, if the top-level sysfs directory (
../mmc_host/mmc2/mmc2:0001/net/wlan0 ) had it's driver link updated to
point the right place ( ../mmc_host/mmc2/mmc2:0001/mmc2:0001:1 ),
things would just work.
I have a hunch this problem may be due to the fact that SDIO cards can
be multi-function, but that's just a guess.
By the way, the kernel and LUM involved are the latest on the abi-12
branches of these git trees:
http://kernel.ubuntu.com/git?p=mid-team/hardy-jax.git;a=summary
http://kernel.ubuntu.com/git?p=mid-team/hardy-lum-jax.git;a=summary
To manage notifications about this bug go to:
https://bugs.launchpad.net/acton/+bug/207422/+subscriptions
--
Mailing list: https://launchpad.net/~kernel-packages
Post to : kernel-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~kernel-packages
More help : https://help.launchpad.net/ListHelp
