booting uclinux from u-boot

Altera_Forum · ‎05-29-2006

hi all

I am trying to boot uclinux from u-boot on 2s60 board.

I have one doubt regarding Exception vector table entry point address.

What I have observed in uclinux (head.S) that exception vector table entry point jump instruction sits at Base Addr of RAM + 0x20 offset, same information I have given in Hardware configuration file in sopc builder (.sof).

And same location is used for u-boot's Execption vector table (specified in the u-boot config file).

My SDRAM start address is 0x0100_0000.

So exception vector table sits at 0x0100_0020 in both the cases of u-boot and uclinux.

The kernel image has entry_address at 0x0100_0000.

Now if I download kernel image through u-boot, then Exception table gets overwritten by kernel image and u-boot hangs.

Please share your experience with how you deal with this issue. I mean when you boot kernel from u-boot then which address is given for exception vector table. Or it's ok to use same addr there is some other issue ?

thanks

Ankur

Altera_Forum · ‎05-29-2006

--- Quote Start ---

originally posted by ankur@May 29 2006, 04:24 AM

hi all

i am trying to boot uclinux from u-boot on 2s60 board.

i have one doubt regarding exception vector table entry point address.

what i have observed in uclinux (head.s) that exception vector table entry point jump instruction sits at base addr of ram + 0x20 offset, same information i have given in hardware configuration file in sopc builder (.sof).

and same location is used for u-boot's execption vector table (specified in the u-boot config file).

my sdram start address is 0x0100_0000.

so exception vector table sits at 0x0100_0020 in both the cases of u-boot and uclinux.

the kernel image has entry_address at 0x0100_0000.

now if i download kernel image through u-boot, then exception table gets overwritten by kernel image and u-boot hangs.

please share your experience with how you deal with this issue. i mean when you boot kernel from u-boot then which address is given for exception vector table. or it's ok to use same addr there is some other issue ?

thanks

ankur

<div align='right'><{post_snapback}> (index.php?act=findpost&pid=15652)

--- quote end ---

--- Quote End ---

hi all,

I can download my kernel image file in the sdram. There were some misunderstanding with CFG_NIOS_CPU_VEC_BASE and CFG_EXCEPTION_ADDR.

Now both I made equal to 0x0100_0020.

But still I can't boot my kernel.

I am downloading uncompressed kernel through tftpboot at 0x0100_0000.

Then I give bootm 0x0100_0000

Is it fine or should I change the location ??

thanks

Ankur

Altera_Forum · ‎05-30-2006

--- Quote Start ---

originally posted by ankur+may 29 2006, 08:18 am--><div class='quotetop'>quote (ankur @ may 29 2006, 08:18 am)</div>

--- quote start ---

<!--quotebegin-ankur@May 29 2006, 04:24 AM

hi all

i am trying to boot uclinux from u-boot on 2s60 board.

i have one doubt regarding exception vector table entry point address.

what i have observed in uclinux (head.s) that exception vector table entry point jump instruction sits at base addr of ram + 0x20 offset, same information i have given in hardware configuration file in sopc builder (.sof).

and same location is used for u-boot's execption vector table (specified in the u-boot config file).

my sdram start address is 0x0100_0000.

so exception vector table sits at 0x0100_0020 in both the cases of u-boot and uclinux.

the kernel image has entry_address at 0x0100_0000.

now if i download kernel image through u-boot, then exception table gets overwritten by kernel image and u-boot hangs.

please share your experience with how you deal with this issue. i mean when you boot kernel from u-boot then which address is given for exception vector table. or it's ok to use same addr there is some other issue ?

thanks

ankur

<div align='right'><{post_snapback}> (index.php?act=findpost&pid=15652)

--- quote end ---

--- Quote End ---

hi all,

I can download my kernel image file in the sdram. There were some misunderstanding with CFG_NIOS_CPU_VEC_BASE and CFG_EXCEPTION_ADDR.

Now both I made equal to 0x0100_0020.

But still I can't boot my kernel.

I am downloading uncompressed kernel through tftpboot at 0x0100_0000.

Then I give bootm 0x0100_0000

Is it fine or should I change the location ??

thanks

Ankur

<div align='right'><{post_snapback}> (index.php?act=findpost&pid=15664)</div>

[/b]

--- Quote End ---

Altera_Forum · ‎05-31-2006

Hi Ankur,

> There were some misunderstanding with CFG_NIOS_CPU_VEC_BASE and

> CFG_EXCEPTION_ADDR.

> Now both I made equal to 0x0100_0020.

First, don't use CFG_NIOS_CPU_VEC_BASE ... this is an old (and totally brain-damaged) obfuscation

that was foolishly added to some Nios-32 u-boot board configurations -- it is not used in the

Nios II sources. And please don't ask for any details about CFG_NIOS_CPU_VEC_BASE

-- I'll just end up embarrassing myself with an endless rant ;-)

Second, please use the code from: http://www.psyent.com/download (http://www.psyent.com/download) -- it contains all of the patches

that have been submitted to the u-boot project (and are still pending). If you're using code directly

from the u-boot project you will likely have problems without the patches.

> am downloading uncompressed kernel through tftpboot at 0x0100_0000.

> Then I give bootm 0x0100_0000

> Is it fine or should I change the location ??

Change this -- you are overwriting your exception trampoline during download. The first

interrupt will put the CPU in the weeds. Load the image somewhere higher in memory.

The 'bootm' command will copy (and optionally decompress) the image to the load

address you specified with mkimage, after disabling interrupts. Try downloading to

0140_0000 for example (but leave the load address and entry point at 0100_0000). Then,

==> bootm 1400000

Regards,

--Scott

Altera_Forum · ‎05-31-2006

--- Quote Start ---

originally posted by smcnutt@May 30 2006, 07:39 PM

hi ankur,

> there were some misunderstanding with cfg_nios_cpu_vec_base and

> cfg_exception_addr.

> now both i made equal to 0x0100_0020.

first, don't use cfg_nios_cpu_vec_base ... this is an old (and totally brain-damaged) obfuscation

that was foolishly added to some nios-32 u-boot board configurations -- it is not used in the

nios ii sources. and please don't ask for any details about cfg_nios_cpu_vec_base

-- i'll just end up embarrassing myself with an endless rant ;-)

second, please use the code from: http://www.psyent.com/download (http://www.psyent.com/download) -- it contains all of the patches

that have been submitted to the u-boot project (and are still pending). if you're using code directly

from the u-boot project you will likely have problems without the patches.

> am downloading uncompressed kernel through tftpboot at 0x0100_0000.

> then i give bootm 0x0100_0000

> is it fine or should i change the location ??

change this -- you are overwriting your exception trampoline during download. the first

interrupt will put the cpu in the weeds. load the image somewhere higher in memory.

the 'bootm' command will copy (and optionally decompress) the image to the load

address you specified with mkimage, after disabling interrupts.[/b] Try downloading to

0140_0000 for example (but leave the load address and entry point at 0100_0000). Then,

==> bootm 1400000

Regards,

--Scott

<div align='right'><{post_snapback}> (index.php?act=findpost&pid=15700)

--- Quote End ---

[/b]

--- Quote End ---

hi Scott,

thanks, for the reply. Actually I was using u-boot -1.1.2 taken from source forge. (I know its old now). But after following your instructions uclinux is booting now.

thanks once again

Ankur