2.6.25 / mainline tracking effort

Andy Green andy at openmoko.com
Sun Apr 13 11:54:56 CEST 2008

Hash: SHA1

Hi folks -

Now we changed to git and froze mokopatches for the time being, it can
be easier to track mainline.  We only managed to struggle up to 2.6.24
because of a one-shot effort "beyond the call of duty" from Harald to
uplevel his stuff, otherwise we would be stuck at right now.
So we don't want to waste this chance.

Also, there is no possibility to get stuff into upstream unless we track
current upstream.  It doesn't mean our stable branch is tracking
upstream HEAD but it means we have to have a branch that tries to keep
pretty current stuff from our side on top of pretty current stuff from
their side, and deal with breakage.  This is important right now because
there is a guy considering to help us with 24xx mokopatches upstream at
the moment.

Therefore I introduce two new branches in git:

~ mokopatches-tracking
~ andy-tracking

these are upleveled versions of mokopatches and andy branches to track
upstream mainline kernel.  You only need to care about andy-tracking if
you want to build it.

Basing against Linux 2.6.25-rc9 we are in fairly good shape after some
mokopatch kickings-around to deal with breakage.  There are three
obvious breakages I saw:

~ - GTA01 ocf50606 compatability symlink is broken, see


for the details, if we can find the right device name it should get
fixed OK IIUI.

~ - SDIO / WLAN driver fails to probe, complaining about DMA

[    3.345000] S3c2440 SDIO Host controller
[    3.450000] SDIO BusDriver - SDIO_GetBusOSDevice, registering driver:
sdio_s3c24xx_hcd DMAmask: 0x0
[    3.455000] pnp: the driver 'sdio_s3c24xx_hcd' has been registered
[    3.465000] mapped channel 0 to 0
[    3.470000] unable to get DMA channel.
[    3.475000] HW Init failed
[    3.480000] sdio_s3c24xx_hcd 00:00: disabled
[    3.485000] sdio_s3c24xx_hcd: probe of 00:00 failed with error -14
[    3.490000] pnp 00:00: SDIO device, IDs SD_0008 (disabled)
[    3.500000] pnp: the driver 'sdio_wlan' has been registered
[    3.505000] sdio_wlan: probe of 00:00 failed with error -1

SDIO helper thread then blocks and makes trouble

[ 1551.560000] INFO: task SDIO Helper:286 blocked for more than 120 seconds.
[ 1551.565000] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[ 1551.570000] SDIO Helper   D c02bf450     0   286      2
[ 1551.580000] [<c02bf204>] (schedule+0x0/0x2b0) from [<c02bed2c>]
[ 1551.590000] [<c02bec48>] (__down+0x0/0x100) from [<c02beb4c>]
[ 1551.595000]  r7:00000000 r6:c01dcb20 r5:c7d94f14 r4:c7da6000
[ 1551.600000] [<c01db518>] (CardDetectHelperFunction+0x0/0x1a8) from
[<c01dcb3c>] (HelperLaunch+0x1c/0x28)
[ 1551.620000]  r5:c7d94f14 r4:c7d94f14
[ 1551.630000] [<c01dcb20>] (HelperLaunch+0x0/0x28) from [<c005dcc0>]
[ 1551.635000]  r4:c7da6000
[ 1551.640000] [<c005dc60>] (kthread+0x0/0x94) from [<c004b9ec>]
[ 1551.645000]  r6:00000000 r5:00000000 r4:00000000
[ 1551.650000] no locks held by SDIO Helper/286.

~ - can't enter suspend, echo mem > /sys/power/state does not do it but
sits there, the console still echos CRLF.  Maybe the SDIO/WLAN thing
breaks it.

Overall we are in fairly good shape but the WLAN and suspend
particularly need to be resolved.

- -Andy

Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org


More information about the openmoko-kernel mailing list