Beaglebone latest image observations - lock order reversal from ufs

[Ian Lepore]

On Fri, 2013-02-15 at 23:48 +0100, Werner Thie wrote:
> Hi
> 
> today, after several failed attempts (build world failing), I managed 
> to build a new kernel with Tim's script for the Bone.
> 
>  From what I see it started spewing the 'ti_mmchs0: Error: current
> cmd 
> null' again.
> 
> ti_mmchs0: Error: current cmd NULL, already done?
> 
> I'm now also getting the occasional
> 
> lock order reversal:
>   1st 0xc2c8ad84 ufs (ufs) @ /usr/local/src/sys/kern/vfs_subr.c:2176
>   2nd 0xc97a2378 bufwait (bufwait) @ 
> /usr/local/src/sys/ufs/ffs/ffs_vnops.c:261
>   3rd 0xc2d807f8 ufs (ufs) @ /usr/local/src/sys/kern/vfs_subr.c:2176
> KDB: stack backtrace:
>        :
>        :
> Anybody out there being able to enlighten me about this? Does the
> first 
> error relate or is triggered by second one?
> 
This shouldn't be at all related to the "current cmd NULL" (that's a
form of unexpected-interrupt error; it shouldn't be happening, but it
should be fairly harmless).

This LOR probably does need to be looked at more closely, though.

> Seeping through dmesg I see
>        :
> cpsw0: <3-port Switch Ethernet Subsystem> mem 0xcf619000-0xcf61cfff
> irq 
> 40,41,42,43 on simplebus0
> cpsw0: CPSW SS Version 1.12 (0)
> _bus_dmamap_sync: wrong user map: 0 4

This is most likely the same problem Thomas Skibo just submitted a fix
for.  If so, it's committed as r246859.

-- Ian

> cpsw0: Initial queue size TX=128 RX=384
> cpsw0: Ethernet address: 00:18:31:8e:0e:1d
>        :
> 
> I also observed that Python27 built fresh from ports aborts on
> importing 
> ctypes.
> 
> I keep on building and reporting, the progress is still amazing!
> 
> If there's anything specific to try, please let me know
> 
> Thxs, Werner
> 
>