[U-Boot-Users] U-boot with environment in flash now won't use serial console under Linux

[U-Boot-Users] U-boot with environment in flash now won't use serial console under Linux

[Andy Hawkins]

Hi all,

I've recently successfully changed U-Boot such that it's environment is in
flash (on the PPChameleonEVB board). However, now when I boot the kernel I
get no serial console. The serial console works fine under U-Boot up until
the time it transfers control to Linux, at which point I get nothing.

However, if I switch back to the original U-Boot, I *do* get a serial
console (this is with exactly the same kernel). This (to me) tends to point
to some problem with the way I've changed U-Boot.

I've modified the 'evb405ep_early_serial_map' function in
arch/ppc/platforms/PPChameleonEVB.c so that it prints out the parameters
passed in the bd_t structure, and these are identical in both cases (at
least, for the members of the structure that this function uses).

Can anyone offer any advice as to why I'm no longer getting a serial
console? Linux is booting up fine (I can telnet into it), but I'm just not
getting a console.

The changes I've made in include/configs/PPChameleonEVB.h are as follows:

CFG_MONITOR_BASE from 0xFFFC0000 to 0xFFFD0000
CFG_MONITOR_LEN and CFG_MALLOC_LEN from 256k to 192k

Changed the definitions for the environment from

#define CFG_ENV_IS_IN_EEPROM    1       /* use EEPROM for environment vars
*/
#define CFG_ENV_OFFSET          0x100   /* environment starts at the
beginning
#define CFG_ENV_SIZE            0x700   /* 2048 bytes may be used for env
vars*
                                   /* total size of a CAT24WC16 is 2048
bytes *
to

#define CFG_ENV_IS_IN_FLASH     1
#define CFG_ENV_ADDR            0xFFFC0000
#define CFG_ENV_SIZE            0x10000

Thanks for any advice you can offer.

Andy

[U-Boot-Users] U-boot with environment in flash now won't use serial console under Linux

[Wolfgang Denk]

In message <001501c4188e$6bda5570$153335bf@cabletime.com> you wrote:
> 
> I've recently successfully changed U-Boot such that it's environment is in
> flash (on the PPChameleonEVB board). However, now when I boot the kernel I
> get no serial console. The serial console works fine under U-Boot up until
> the time it transfers control to Linux, at which point I get nothing.
> 
> However, if I switch back to the original U-Boot, I *do* get a serial
> console (this is with exactly the same kernel). This (to me) tends to point
> to some problem with the way I've changed U-Boot.

Probably. I recommend to make one change at a time.

> I've modified the 'evb405ep_early_serial_map' function in

I recommend to ONLY make the modification to the environment storage,
and nothing else. Don't modify the serial drivers.

Also, you might want to check the boot  arguments  you  pass  to  the
kernel.  Are  they  really  the  same  in  both cases? Maybe your new
version fails to pass a console= argument?

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
The complexity of software is an essential property, not an  acciden-
tal  one. Hence, descriptions of a software entity that abstract away
its complexity often abstract away its essence.    - Fred Brooks, Jr.

[U-Boot-Users] U-boot with environment in flash now won't use serial console under Linux

[Andy Hawkins]

Hi,

> > I've modified the 'evb405ep_early_serial_map' function in
>
> I recommend to ONLY make the modification to the environment storage,
> and nothing else. Don't modify the serial drivers.

The only modification to the serial driver is to include printk statements
to see what parameters it's getting. This is in the linux kernel, not
u-boot. In both instances, I have been using an identical kernel image, the
only difference is between the two u-boot images. The kernel image has been
working fine using u-boot where the environment is stored in EEPROM.

> Also, you might want to check the boot  arguments  you  pass  to  the
> kernel.  Are  they  really  the  same  in  both cases? Maybe your new
> version fails to pass a console= argument?

I'll verify that the arguments are the same in both cases. However, this is
the command line seen by the non-working kernel:

Kernel command line: root=/dev/nfs rw
nfsroot=191.53.51.22:/home/andyh/target
    ip=191.53.51.200:191.53.51.22::255.255.0.0:PPChameleonEVB:eth0:off
    panic=1 console=ttyS0:115200

This looks fine to me...

However, setserial returns the following:

setserial -a /dev/console
/dev/console, Line 0, UART: 16550A, Port: 0x0000, IRQ: 0
        Baud_base: 694444, close_delay: 50, divisor: 0
        closing_wait: 3000
        Flags: spd_normal skip_test

The baud_base looks odd to me. However, this is the value calculated as
serial_baud_405ep in  the kernel's evb405ep_early_serial_map function (which
is the same in both instances), so is probably correct.

Thanks.

Andy

[U-Boot-Users] U-boot with environment in flash now won't use serial console under Linux

[Andy Hawkins]

<sigh>

I guess my impending wedding is getting to me :)

> Kernel command line: root=/dev/nfs rw
nfsroot=191.53.51.22:/home/andyh/target
>     ip=191.53.51.200:191.53.51.22::255.255.0.0:PPChameleonEVB:eth0:off
>     panic=1 console=ttyS0:115200

This should of course be 'console=ttyS0,115200'

Apologies all, and thanks Wolfgang, you were spot on.

Andy