Re: undefined instruction

[Steve Poulsen <spoulsen@css-design.us>]

Mohammed,

Are you using NFS to boot, then mounting to /mnt/flash?   This is a good 
test, but I just want to be sure it is what you are doing.

Can you post the output from:
#  cat /proc/mtd

Also, you need to check and verify that the data in the flash matches 
your file.   Using a hex editor, examine your file (rootfs-jffs2.img).   
Record a few bytes at offset 0 and offset 0xc00000 (where you split 
it).   Make sure the bytes just before and after the split point are 
recorded.

Now go to you board and examine the flash.  There are several ways to do 
this.  One way is to copy the whole sector to a file (assuming you boot 
NFS and have the space).  

cp /dev/mtd3 somefile.img

Compare the image with the one you started with.   The new one will be 
larger because you dump the whole flash section, so ignore the extra 
data.  

This is not really a solution to your problem, but can give you some 
techniques to help figure out how to determine what is wrong.

Good luck!

Steve


mohammed shareef wrote:
> Dear Steve,
>
> i did as you said, transfered both the images one after the other.
> then i made nodes for mtd3 and mtdblock3
>
> and i tried to test whether the fs is able to mount:
>
> i did:
>
> mount -t jffs2 /dev/mtdblock3 /mnt/flash
>
> then i got the following error:
>
> jffs2)scan)eraseblock():magic bitmask 0x1985 not found at 0x00000000:
> 0x05cf instead
> ..
> ..
> Old JFFS2 bitmask found at 0x00012cd8
> You cannot use older JFFS2 filesystems with newer kernels
>
> and it doesnt mount in /mnt/flash
>
> could you please tell me whats wrong.
> thank you,
> regards,
> Shareef
>
>
>
> On Fri, Jun 27, 2008 at 12:03 AM, Steve Poulsen <spoulsen@css-design.us> wrote:
>   
>> Mohammed,
>>
>> This is the correct approach.  As long as you avoid anything below 0x240000,
>> you avoid touching u-boot and the kernel.   I suggest to turn protect off
>> only for the sectors needed.  The sectors/addresses depends upon your flash.
>>  If you are using the OSK, then these are told in the OSK newbie guide (I
>> can't recall and our flash is different).
>> For the OSK, the filesystem is at 0x240000.  Therefore, you transfer the
>> first file to 0x240000 and the next file transfers to 0x240000 + 0xc00000
>> (E40000).   This is purely a memory copy.   You don't need to combine.
>> When you place the two pieces next to each other, they are basically
>> combined.
>> I'm not sure you understand this, but when you use tftp, you are copying the
>> file to SDRAM.   This address is fixed for the OSK at 0x10000000.
>> Therefore the steps are:
>>
>> 1) tftp part 1 into memory 0x10000000
>> 2) copy the memory of 0x10000000 to 0x240000
>> 3) tftp part 2 into memory 0x10000000
>> 4) copy the memory of 0x10000000 to 0xE40000
>> 5) Reset the board (You could boot from here if you wish, but a reset is
>> simpler and puts the flash back to protected)
>>
>> Steve
>>
>> mohammed shareef wrote:
>>     
>>> i erased location 0x240000 to 0x1ffffff;
>>> then i transfered the first file to location 0x240000, the size of the
>>> first image was c00000 (12Mb)
>>> but now i have the second image on RAM of osk.
>>> the flash segment for ffs2 filesystem runs from 0x01000000 till
>>> 0x0fffffff(please correct me if i am wrong).  i dont know to which
>>> location on flash to transfer this to. could someone please help me on
>>> this. thank you. and i also have doubts on how to combine the two
>>> images on the flash and make it tun. thannk you.
>>> regards,
>>> Shareef
>>>
>>> On Thu, Jun 26, 2008 at 9:59 PM, mohammed shareef <mdshareef@gmail.com>
>>> wrote:
>>>
>>>       
>>>> Dear Steve,
>>>>
>>>> i split the file into two pieces:
>>>>
>>>> split rootfs-jffs2.img --bytes=12m
>>>>
>>>> so i have two files with xaa(12Mb) and xab(11.5Mb)
>>>>
>>>> i was ablt to transfer the first file completely with any problem.
>>>>  but i dont know what to do next. should i transfer the first image in
>>>> RAM to flash? could you please tell me how many sectors i need erase
>>>> and from which bank? i am afraid that i may end up erasing the u-boot.
>>>> thank you.
>>>> regards,
>>>> Shareef
>>>>
>>>> On Thu, Jun 26, 2008 at 9:41 PM, Steve Poulsen <spoulsen@css-design.us>
>>>> wrote:
>>>>
>>>>         
>>>>> Mohammed,
>>>>>
>>>>> When you tftp the file to memory, you need to make sure the filesize
>>>>> fits in
>>>>> the memory available.  Since you have experimentally done that and now
>>>>> want
>>>>> to flash the pieces, I suggest you look at the "split" command under
>>>>> Linux.
>>>>>  You will need to split the file into pieces that fit into RAM and flash
>>>>> at
>>>>> the proper address.  If you split the file into two pieces, then you
>>>>> will
>>>>> need to flash the first piece at address X and the second piece at
>>>>> address X
>>>>> + 16meg.  You should make sure you split the file on a sector boundary.
>>>>>   If
>>>>> you don't want to think about this, then you should erase/unprotect the
>>>>> whole area you will need first, then transfer and flash the pieces. You
>>>>> may
>>>>> want to look at the omapfl utility.   With some modification, you could
>>>>> flash your image more easily via USB.
>>>>>
>>>>> Steve
>>>>>
>>>>> mohammed shareef wrote:
>>>>>
>>>>>           
>>>>>> Hi,
>>>>>>
>>>>>> I tried to do the same procedure with a small filesystem image <
>>>>>> 16Mb... it worked. i didnt have such problems. so could someone please
>>>>>> tell me how to divide the filesystem image in to two and flash it?
>>>>>> thank you,
>>>>>> regards,
>>>>>> shareef
>>>>>>
>>>>>> On Thu, Jun 26, 2008 at 4:36 PM, mohammed shareef <mdshareef@gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>
>>>>>>             
>>>>>>> Hi,
>>>>>>>
>>>>>>> i did the below. i got an image. but i am still having the same
>>>>>>> problem
>>>>>>>
>>>>>>> my file size is 23Mb
>>>>>>>
>>>>>>> [root@localhost tftpboot]# mkfs.jffs2 --squash -r /data/rootfs2.6
>>>>>>> -e131072 > /data/rootfs-jffs2.img
>>>>>>> [root@localhost tftpboot]# cp /data/rootfs-jffs2.img
>>>>>>> /tftpboot/rootfs-jffs2.img
>>>>>>>
>>>>>>> \0x09
>>>>>>> #################################################################
>>>>>>> \0x09 #############undefined instruction
>>>>>>> pc : [<e0000004>]    lr : [<00000002>]
>>>>>>> sp : 1103fca4  ip : 11095dd8  fp : 00000001
>>>>>>> r10: 10963410  r9 : 1103fd24  r8 : 1103ffdc
>>>>>>> r7 : 270a30a1  r6 : 8695632d  r5 : 08016ffa  r4 : 5aebcc39
>>>>>>> r3 : 00000032  r2 : 11095dd4  r1 : 000000a0  r0 : 00000000
>>>>>>> Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
>>>>>>> Resetting CPU ...
>>>>>>>
>>>>>>> could you please tell me what i should do. thank you.
>>>>>>> regards,
>>>>>>> Shareef
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Wed, Jun 25, 2008 at 9:52 PM, Hunter, Jon <jon-hunter@ti.com>
>>>>>>> wrote:
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>>>>> then i changed the filename and the
>>>>>>>>> tftpboot transfer started. But on the mid-way it complains
>>>>>>>>>  "undefined
>>>>>>>>> instruction".
>>>>>>>>>
>>>>>>>>> could some one please tell me where the problem is? thank you.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>> How big is the file that you are attempting to download over tftp?
>>>>>>>>
>>>>>>>> U-boot executes in the upper part of the RAM and so if your file is
>>>>>>>> too
>>>>>>>> big, then there is a good chance you are overwriting u-boot which
>>>>>>>> would
>>>>>>>> cause u-boot to crash eventually. U-boot does not protect against
>>>>>>>> this. This
>>>>>>>> would be a potential cause of an undefined instruction exception.
>>>>>>>>
>>>>>>>> Jon
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>> --
>>>>>> To unsubscribe from this list: send the line "unsubscribe linux-omap"
>>>>>> in
>>>>>> the body of a message to majordomo@vger.kernel.org
>>>>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>>>>>
>>>>>>
>>>>>>
>>>>>>             
>>>>>           
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-omap" in
>>> the body of a message to majordomo@vger.kernel.org
>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>>
>>>
>>>       
>>     
>
>