[U-Boot] beaglxm/bootz: kernel boot fail due to align err: cache handling issues?

Nishanth Menon nm at ti.com
Fri Aug 23 16:54:00 CEST 2013


On 08/23/2013 09:32 AM, Robert Nelson wrote:
> On Fri, Aug 23, 2013 at 9:21 AM, Nishanth Menon <nm at ti.com> wrote:
>> Hi,
>> Noticed an interesting behavior where adding a dcache flush kernel boot
>> fail due to unaligned-accesses seems to go away.
>> Full details: http://pastebin.com/kVBRWsbE
>> Without dcache flush being added to my uenvcmd:
>>
>> Kernel image @ 0x80200000 [ 0x000000 - 0x3e2690 ]
>> data abort
>>
>> MAYBE you should read doc/README.arm-unaligned-accesses
>>
>> pc : [<9ff6f6b4>]          lr : [<9ff6ef6c>]
>> sp : 9fefaaf0  ip : 9fefe8ec     fp : 00000000
>> r10: 9ffa6544  r9 : 9ffa5e5c     r8 : 9fefaf30
>> r7 : 0f800000  r6 : 9ffa6544     r5 : 9ffa65b0  r4 : 9ffa65b4
>> r3 : 00000049  r2 : 00000010     r1 : 00000000  r0 : 0f800000
>> Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
>> Resetting CPU ...
>>
>> With dcache flush being added to my uenvcmd: complete boot
>>
>>
>> I think something of the form was highlighted here:
>> http://marc.info/?l=u-boot&m=137276534624656&w=2
>>
>> Not sure if there was any conclusion to the effect.
> 
> There may still be an underline bug, but the "bootz ${loadaddr}" issue
> i showed in that post was fixed very late in the v2013.07, i forget
> the commit, it may have been a day or two right before.
> 
> So looking at your pastebin.com log: "U-Boot
> 2013.07-rc2-00044-g1d28a6a" it might be worthwhile to re-test with
> v2013.07 final, specially since your using bootz...

Arrgh... I must have been completely blind :(
you are right,
v2013.07 - still failed, BUT,
v2013.10-rc1 just works fine.
2013.10-rc1-00029-g6612ab also works fine.

-- 
Regards,
Nishanth Menon


More information about the U-Boot mailing list