U-Boot for Nokia RX-51 with BootMenu (updated version 2012.10-rc3-1) - Page 11 - maemo.org

Active Topics

more...

Page 11 of 85

Thread Tools

pali	2012-02-02 , 13:36
Posts: 2,154 \| Thanked: 8,464 times \| Joined on May 2010	#101

Originally Posted by Netweaver

I've found this info on the error, talking about a kernel bug in the same kernel version as Nemo is using, also on armel architecture

http://groups.google.com/group/beagleboard/browse_thread/thread/87f191353f55ecc2?pli=1

Smoking gun ???

Something for Carsten and his team I guess.

try reporting bug

Quote & Reply |

Netweaver	2012-02-02 , 13:37
Posts: 323 \| Thanked: 180 times \| Joined on Oct 2009 @ Gent, Belgium	#102

from the Nemo Kernel config :

# CONFIG_PREEMPT_NONE is not set
# CONFIG_PREEMPT_VOLUNTARY is not set
CONFIG_PREEMPT=y

According to the kernel bug report, the problem can be avoided by compiling the kernel with CONFIG_PREEMPT_NONE set (and of course the CONFIG_PREEMP not set).

I'm not sure what side effects that has. Need to do some more digging for that...

UPDATE : I see why this option was chosen, to reduce latency.
http://www.hotaboutlinux.com/2010/01...air-scheduler/
Might be a proper kernel patch solution is the best one, if existing ... I'll dig up the link for the Nemo/Mer bug report.

Last edited by Netweaver; 2012-02-02 at 13:42.

Quote & Reply |

The Following 2 Users Say Thank You to Netweaver For This Useful Post:
Estel, Hossein

Fabry	2012-02-02 , 13:46
Posts: 362 \| Thanked: 426 times \| Joined on Nov 2010 @ Italy, Lombardia	#103

I have found a bug in bootmenu generation.

I have used this profile:

Code:

ITEM_NAME="Nemo from boot.scr"
ITEM_ID="Nemo2"
ITEM_DEVICE="${INT_CARD}p1"
ITEM_FSTYPE="fat"
ITEM_USEATAG="1"
ITEM_SCRIPT="boot.scr"

to load from MyDocs my old (and working) bootscript for Nemo.

It not works and if i examine the generated menu entry, I find :

Code:

setenv bootmenu_3 'Nemo from boot.scr=
setenv mmcnum 1;
setenv mmcpart 1;
setenv mmctype fat;
setenv bootargs;
setenv atagaddr;
setenv mmckernfile;
setenv mmcinitrdfile;
setenv mmcscriptfile boot.scr;
run trymmcscriptboot;'

Error: Atag not set

Last edited by Fabry; 2012-02-02 at 14:17.

Quote & Reply |

The Following 3 Users Say Thank You to Fabry For This Useful Post:
Estel, Netweaver, Sourav.dubey

Netweaver	2012-02-02 , 14:14
Posts: 323 \| Thanked: 180 times \| Joined on Oct 2009 @ Gent, Belgium	#104

bug entered on the Mer/Nemo bugzilla for the External SD card initialization failure : https://bugs.merproject.org/show_bug.cgi?id=150

Quote & Reply |

The Following User Says Thank You to Netweaver For This Useful Post:
Estel

Fabry	2012-02-02 , 14:32
Posts: 362 \| Thanked: 426 times \| Joined on Nov 2010 @ Italy, Lombardia	#105

I have the same problem (error -110) with eMMC but with previous u-boot instead it is working fine

Quote & Reply |

The Following User Says Thank You to Fabry For This Useful Post:
tanpoaran

Netweaver	2012-02-02 , 15:24
Posts: 323 \| Thanked: 180 times \| Joined on Oct 2009 @ Gent, Belgium	#106

Hi Pali,
can u have a look at your code again ?

Specifically compare with the patches that were used by Carsten to create his uboot-pr13 package, as they seem to initialize everything fine : http://al.robotfuzz.com/~al/maemo/u-...-support.patch

We have to avoid ending up in the fingerpointing : it's the Nemo kernel's fault <-> it's Pali's u-boot fault

Also Fabry, if you say it was working fine with the previous one, is that a previous Pali version (with the battery drain) or the 'official' uboot-pr13 version ? Just to narrow down where the differences can be looked for.

Thanks again. I hope u have some time, as I know your pretty busy recently.

Quote & Reply |

The Following User Says Thank You to Netweaver For This Useful Post:
Estel

pali	2012-02-02 , 15:27
Posts: 2,154 \| Thanked: 8,464 times \| Joined on May 2010	#107

Originally Posted by Fabry

I have found a bug in bootmenu generation.

I have used this profile:
Code:
ITEM_NAME="Nemo from boot.scr"
ITEM_ID="Nemo2"
ITEM_DEVICE="${INT_CARD}p1"
ITEM_FSTYPE="fat"
ITEM_USEATAG="1"
ITEM_SCRIPT="boot.scr"
to load from MyDocs my old (and working) bootscript for Nemo.

It not works and if i examine the generated menu entry, I find :
Code:
setenv bootmenu_3 'Nemo from boot.scr=
setenv mmcnum 1;
setenv mmcpart 1;
setenv mmctype fat;
setenv bootargs;
setenv atagaddr;
setenv mmckernfile;
setenv mmcinitrdfile;
setenv mmcscriptfile boot.scr;
run trymmcscriptboot;'
Error: Atag not set

This is not error I think. You are not booting kernel, but starting other u-boot script. So if you need to reuse atags from nolo, your script should set all needed variables for booting.

I checked update script and ITEM_USEATAG is relevant only for booting kernel.

... But what other people think? Should external u-boot script set all needed variables for booting? Or should only some and some should set update-bootmenu?
I think that - if user/developer/... decide to use own u-boot script for booting (not prepaired u-boot variables for kernel/initrd) it should also handle passing atags to bootm command.

Quote & Reply |

The Following User Says Thank You to pali For This Useful Post:
Estel

pali	2012-02-02 , 15:42
Posts: 2,154 \| Thanked: 8,464 times \| Joined on May 2010	#108

Originally Posted by Netweaver

Hi Pali,
can u have a look at your code again ?

Specifically compare with the patches that were used by Carsten to create his uboot-pr13 package, as they seem to initialize everything fine : http://al.robotfuzz.com/~al/maemo/u-...-support.patch

We have to avoid ending up in the fingerpointing : it's the Nemo kernel's fault <-> it's Pali's u-boot fault

Also Fabry, if you say it was working fine with the previous one, is that a previous Pali version (with the battery drain) or the 'official' uboot-pr13 version ? Just to narrow down where the differences can be looked for.

Thanks again. I hope u have some time, as I know your pretty busy recently.

My u-boot patches are based on that you pointing. Also my patches are based on recent u-boot version and ali's are based on version 2 years old. Also problem can be in upstream u-boot - but we have al least one kernel which booting without problem (stock kernel and kernel-power). My question is why some kernel booting without problem? Also you can see that in ali's patch is nothing about mmc and my code has added twl4030 regulator (which enable eMMC). Old u-boot has no (correct) eMMC support.

Quote & Reply |

The Following 2 Users Say Thank You to pali For This Useful Post:
Estel, Mentalist Traceur

Netweaver	2012-02-02 , 18:25
Posts: 323 \| Thanked: 180 times \| Joined on Oct 2009 @ Gent, Belgium	#109

Whoever is interested, some more reading on the same issue, encountered by other people on Armel platform:

http://permalink.gmane.org/gmane.lin...arm.omap/52674

same issue, fixed by dissect between 2 kernel trees to find the troubled sourcecode in ubuntu on beagleboard :
https://bugs.launchpad.net/ubuntu/+s...ux/+bug/591941

issue still stays why some kernels work. Basically the 2.6.28 kernels (stock + derived kernel-power line) all work. Are they maybe configured with PREEMPT_NONE=y ?

That could explain the difference in behavior with the 2.6.37 Nemo kernel line.

Minute timing differences in the boot/init process can already influence this race condition, making it hard to pinpoint the issue.

Reading back at the Mer/Nemo install wiki, it almost looks like this issue was known/encountered before but was never really bottomed out, maybe because other more important issue were present.at the time.

Just my 2 cents, to be confirmed.

Quote & Reply |

The Following 3 Users Say Thank You to Netweaver For This Useful Post:
don_falcone, Estel, fw190

pali	2012-02-02 , 18:34
Posts: 2,154 \| Thanked: 8,464 times \| Joined on May 2010	#110

kernel-power has this config:

Code:

$ grep PREEMPT .config 
CONFIG_PREEMPT=y
# CONFIG_DEBUG_PREEMPT is not set
# CONFIG_PREEMPT_TRACER is not set

Quote & Reply |

The Following 4 Users Say Thank You to pali For This Useful Post:
don_falcone, Estel, Mentalist Traceur, Netweaver

Page 11 of 85