U-boot 3.4.23 - mmcinit has odd output

[mgillespie]

I decided foolishly to pickup this u-boot today, as I noticed mine too had a corrupted enviroment.   I have managed to get the plug into a totally unusable state :-(

I can access u-boot environment, but trying to get sheevainstaller 1.0 to fix things always results in a "Bad Magic Number".

LOGFILE ATTACHED

[samweber]

Since your plug is mucked up, I won't run the risk of making it worse by suggesting the following:

0. bubt <u-boot that ever worked>, y to whack (sort of, see my previous yammering on this subject) the environment.

1. resetenv, reset to really get the environment squared away.

2. Put a good uImage on a USB memory stick (FAT formatted) and plug it into the plug.

3. usb start
    fatload usb 0 0x00800000 uImage
    nand erase clean 0x00100000 0x00400000
    nand write.e 0x00800000 0x00100000 0x00400000

4. Check your environment, set the variables you care about and you should be able to type boot and load the kernel.

5.  My plug refuses to display output after it supposedly decompresses the kernel but I am no doubt angering something by suggesting that the RFS could reside on the SD card.  YMMV.

I'll definitely help you if I can.

[mgillespie]

Found the problem.  u-boot seems to be VERY selective about it's USB drives.  have a 8GB Integral and it was not doing the fatload part properly.   Swapped it for a 2GB Transcend Jetflash, and bingo worked straight away...

Both were VFAT formatted under linux.

[leadog]

mgillespie, how is/was the USB drive attached? Did you have a USB hub in the setup? There is a previously identified bug with uboot and USB device speed (low/full/high) determination when a hub is involved.

[mgillespie]

mgillespie, how is/was the USB drive attached? Did you have a USB hub in the setup? There is a previously identified bug with uboot and USB device speed (low/full/high) determination when a hub is involved.

No, it was connected directly to the Sheevaplug.   The USB drive giving problems was a 8GB Integral USB flash drive, with a single vfat partition created with fdisk under x86 linux.   It identifies itself as; VID_0930&PID_6544

[samweber]

Please post the source for the most recent u-boot.  Time to become an expert on the how u-boot hands environment variables to the kernel.  I still can't boot Linux after upgrading and I don't have the option of giving up on the RFS living on SD.

[marcus]

No joy. 

4. Linux refuses to display output or boot (two incantation recipes tried, including one that displayed output when used with a previous u-boot/kernel).

I can confirm this major problem; the plug gave off a death-like appearance. I wasn't patient enough to wait to see if the final stages of the installer worked, but re-running with an older version of U-Boot confirmed that resuscitation was possible. A version of U-Boot that kills console output is a dangerous beast. (I believe that I am responding to the correct thread in terms of where I downloaded this version, and I believe that 3.4.25 is that version).

[SiG]

samweber: I run into almost the same situation as yours. Now my plug boots from SD, but -and this will sound stupid- I don't know exactly how I made it. I will do further investigation, but let me advise you something you can try. First, here is what happened:
1. With uboot 24 I could not boot from SD.
2. I tried to downgrade to 23, then the plug died after uncompressing the kernel.
(see http://plugcomputer.org/plugforum/index.php?topic=968.0)
3. I loaded the very old uboot from CD/USB_Recovery up.
4. Plug died after uncompressing.
5. I loaded the very old uimage from CD/USB_Recovery up with resetting every env (maybe I reloaded the uboot also, with a 'yes') and set the following envs:

Code:

setenv ipaddr 10.10.0.1
setenv serverip 10.10.0.2
setenv netmask 255.255.0.0
setenv cesvcid ULULULULULULPPULULULULULDA
setenv bootargs_root 'root=/dev/mtdblock2 ro'
setenv console 'console=ttyS0,115200'
setenv ethaddr 'xx:xx:xx:xx:xx:xx'
setenv run_diag no
setenv bootargs 'console=ttyS0,115200 mtdparts=nand_mtd:0x400000@0x100000(uImage),0x1fb00000@0x500000(rootfs) rw root=/dev/mtdblock1 rw ip=10.4.50.4:10.4.50.5:10.4.50.5:255.255.255.0:DB88FXX81:eth0:none'
setenv bootcmd 'nand read.e 0x800000 0x100000 0x400000; bootm 0x800000'

NOTE that to use your actual MAC at ethaddr.
6. Here was the point when my plug booted normally again.
7. Then I upgraded to uboot 23 (with a 'no'). I used the image vioan uploaded on the other forum above (thanks).
8. It worked.

And here is the mysterious part, because I don't know what I had done (eg. tried to load the kernel 31.6 from sheeva.with-linux.com) but my plug failed to read the old recovery kernel from NAND. It said there was a big ERROR and it did not even do the uncompression... I had flashed the NAND again, then it uncompressed the kernel but died after. And as far as I remember I did not do a reloading of uboot (with 'yes'), but only made setting on env, but you can try to reload uboot, if the followings won't work.

9. I tried to boot the very new kernel and it worked, then I tried the recovery one, and it died. Then I tried the new kernel with a factory new plug that loaded the factory kernel properly and it died.

Why this last dying happened, I don't know, but I think this one is most mysterious part of the whole thing. Anyway...

10. OK then, I tried to set envs in a nice format, and BOOM it died after loading the new kernel! So I think the uboot-uimage-env trio is extremely sensitive constellation, because I only set bootargs_console, bootargs_root, etc... Then I erased all of the bootargs_xxx=... stuff and used the good old bootargs=..., and voila, it worked again.

So the sum of this all is: I have the latest 2.6.31.6 uimage from http://sheeva.with-linux.com/sheeva/ , I have uboot 3.4.23 from http://plugcomputer.org/plugforum/index.php?action=dlattach;topic=968.0;attach=158 , and the output of printenv is the following:

Code:

Marvell>> printenv
baudrate=115200
loads_echo=0
rootpath=/mnt/ARM_FS/
netmask=255.255.255.0
CASset=min
MALLOC_len=1
ethprime=egiga0
bootargs_end=:::DB88FXX81:eth0:none
image_name=uImage
standalone=fsload 0x2000000 $(image_name);setenv bootargs $(console) root=/dev/mtdblock0 rw ip=$(ipaddr):$(serverip)$(bootargs_end) $(mvPhoneConfig); bootm 0x2000000;
ethmtu=1500
mvPhoneConfig=mv_phone_config=dev0:fxs,dev1:fxs
mvNetConfig=mv_net_config=(00:11:88:0f:62:81,0:1:2:3),mtu=1500
usb0Mode=host
yuk_ethaddr=00:00:00:EE:51:81
nandEcc=1bit
netretry=no
rcvrip=169.254.100.100
loadaddr=0x02000000
autoload=no
ethact=egiga0
cesvcid=ULULULULULULPPULULULULULDA
ethaddr=xx:xx:xx:xx:xx:xx
run_diag=no
ipaddr=192.168.1.11
console=console=ttyS0,115200 mtdparts=nand_mtd:0x100000@0(uboot)ro,0x0@0x100000(uImage),0x1ff00000@0x100000(rootfs)rw
serverip=192.168.1.102
filesize=1F972C
arcNumber=2097
bootcmd_mmc=mmcinit; ext2load mmc 0:1 0x0800000 /boot/uImage
bootcmd=run bootcmd_mmc; bootm 0x0800000
bootargs=console=ttyS0,115200 root=/dev/mmcblk0p1 rw ip=10.4.50.4:10.4.50.5:10.4.50.5:255.255.255.0:DB88FXX81:eth0:none
stdin=serial
stdout=serial
stderr=serial
nandEnvBase=a0000
mainlineLinux=yes
enaMonExt=no
enaCpuStream=no
enaWrAllo=no
pexMode=RC
disL2Cache=no
setL2CacheWT=yes
disL2Prefetch=yes
enaICPref=yes
enaDCPref=yes
sata_dma_mode=yes
netbsd_en=no
vxworks_en=no
bootargs_root=root=/dev/nfs rw
bootdelay=3
disaMvPnp=no
enaAutoRecovery=yes
pcieTune=no

NOTE: ethaddr...

Now it boots the new kernel from the SD card, but does NOT boot the old recovery kernel (CD/USB_Recovery) from NAND or USB or SD.......

Hope this helps.

[samweber]

SiG, thanks for the detailed response.  I can't seem to get to sheeva.with-linux.com.  I get a page that just says "It works" which is actually pretty funny because, of course, it doesn't.

Your willingness to drop back versions and try things is admirable and the mysteries you've uncovered are all interesting.  I've grown weary of my own guesses at what could be going wrong and have decided to figure it out for myself, from source.  Your constellation comment is exactly what I'm interested in.  What do these arguments do to make the kernel hang (or confuse it about what console to use)?  If I can find the answer to that, I'll be in the driver's seat.  But I am stalled out because I don't know where to get the source for the latest u-boot and I'm still not sure which kernel to use (although your latest response shows 2.6.31.6 and that sounds like a good starting point).

Rooster, please post some source for v3.4.25 u-boot.

[SiG]

samweber: sorry i was sloopy with the url. you can you can find all of the kernels here: http://sheeva.with-linux.com/sheeva/

[samweber]

SiG, Thanks man, I'm getting the 2.6.31.6 kernel images now.  Do you happen to know where you can get a System.map file from a kernel build without having to build the kernel yourself?  I have found a technique for debugging the "frozen after kernel decompresses" issue but it requires the address of __log_buf from the map file.  If this technique works, I'll post.

[SiG]

samweber, unfortunately, i don't know. i checked on the plug with find / -name "*ystem*map*" and it doesn't find anything. i will tell you if i find something about it.

[samweber]

Is there an official v3.4.25?  Where is the source for said version?  Please.

[rooster]

Sorry, I can not post any code till this version passes QA 

[samweber]

No drama.  I have my answer.  It's in QA.  Now I'm happy.  Thanks.