Skip to contentSkip to conversation

Strange data behaviour

U-Boot Remco Poelstra 3 participants 4 messages Feb 17, 2009 – Feb 19, 2009

lore
Share & tools
lore
#0 lore raw

Hi,

I'm trying to get my LPC2468 based board to work. I've some problems with the external memory databus.
I would like to know the settings of internal registers to see whether I've initialized them correctly, so I tried making a function that prints the content over the serial link. I'm still in the lowlevel_init function, so I do not have the U-boot puts functions available.
I wrote the following function to convert a long to a HEX string:
-------------------
void print_long(unsigned long data) {
char i;
char
hex_data[]={'0','1','2','3','4','5','6','7','8','9','A','B','C','D','E','F'};

   for(i=0; i<8; i++) {
     while((U0LSR & (1<<5)) == 0); /* Wait for empty U0THR */
//    U0THR = hex_data[(0xDEADBEEF>>i*4)&0xF];
     U0THR = hex_data[(0xDEADBEEF>>4)&0xF];

}
}
---------------
When I run it likes this I get 8 E's. Which is what I expect. When I run it with the commented-out line, I get back 8 0x0's. So it seems that the output is only correct when it is constant.
Does anyone have a clue on why that is?

Kind regards,

Remco Poelstra

#1 lore raw

Dear Remco Poelstra,

Quoted 9 lines · In message <[email protected]> you wrote:

I'm trying to get my LPC2468 based board to work. I've some problems with the external memory databus.
I would like to know the settings of internal registers to see whether I've initialized them correctly, so I tried making a function that prints the content over the serial link. I'm still in the lowlevel_init function, so I do not have the U-boot puts functions available. I wrote the following function to convert a long to a HEX string:
-------------------

Quoted 8 lines
void print_long(unsigned long data) {
   char i;
   char 

hex_data[]={'0','1','2','3','4','5','6','7','8','9','A','B','C','D','E','F'};

   for(i=0; i<8; i++) {
     while((U0LSR & (1<<5)) == 0); /* Wait for empty U0THR */

// U0THR = hex_data[(0xDEADBEEF>>i*4)&0xF];

For clearity of the code I would write that as "...>>(i*4)...", but that's not relevant here.

Quoted 8 lines
     U0THR = hex_data[(0xDEADBEEF>>4)&0xF];

}
}
---------------
When I run it likes this I get 8 E's. Which is what I expect. When I run it with the commented-out line, I get back 8 0x0's. So it seems that the output is only correct when it is constant.
Does anyone have a clue on why that is?

What exactly is U0THR ?

The only definition I can find in the U-Boot sources is here:

include/asm-arm/arch-lpc2292/lpc2292_registers.h:#define U0THR 0xE000C000

but that is obviously not what you are using.

You probably forget the effects of compiler optimization and/or caching here.

I bet you are using a plain pointer access without a "volatile", which is essential here. Maybe even some form of "sync" is needed. In any case, you should use appropriate accessor functions to access device registers.

Best regards,

Wolfgang Denk

Signature · 4 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Never call a man a fool.  Borrow from him.
#2 lore raw

Wolfgang Denk schreef:

Quoted 10 lines
Quoted 8 lines
     U0THR = hex_data[(0xDEADBEEF>>4)&0xF];

}
}
---------------
When I run it likes this I get 8 E's. Which is what I expect. When I run it with the commented-out line, I get back 8 0x0's. So it seems that the output is only correct when it is constant.
Does anyone have a clue on why that is?

What exactly is U0THR ?

Hi,

Thanks for your reply.

For my processor, U0THR is defnied as:
#define U0THR (*(volatile unsigned char *)(UART0_BASE_ADDR +
0x00))

UART0_BASE_ADDR is equal to 0xE000C000

The register itself represents the top of the UART0 transmit FIFO.

Quoted 7 lines

The only definition I can find in the U-Boot sources is here:

include/asm-arm/arch-lpc2292/lpc2292_registers.h:#define U0THR 0xE000C000

but that is obviously not what you are using.

Quoted 8 lines

You probably forget the effects of compiler optimization and/or caching here.

I bet you are using a plain pointer access without a "volatile", which is essential here. Maybe even some form of "sync" is needed. In any case, you should use appropriate accessor functions to access device registers.

I do not know any accessor functions.
What do you mean with a "sync"?

Kind regards,

Remco Poelstra

#3 lore raw
Quoted 7 lines · On Wed, 18 Feb 2009 09:19:08 +0100 Remco Poelstra…

Wolfgang Denk schreef:

Quoted 4 lines

I bet you are using a plain pointer access without a "volatile", which is essential here. Maybe even some form of "sync" is needed. In any case, you should use appropriate accessor functions to access device registers.

I do not know any accessor functions.

things like out_be32 (at least that's what it's called on ppc arch).

Quoted 1 line

What do you mean with a "sync"?

it's a processor instruction that ensures that memory and/or i/o accesses occur exactly when you specify they occur. For this reason, a sync instruction is included as a part of the accessor functions.

Kim