Bug 6177

Summary: emgd.ko module is missing in target system (core-image-sato image with the meta-crownbay layer)
Product: [Yocto Project Subprojects] Kernel Reporter: Maksym Zhelieznyi <mv.zhelieznyi>
Component: kernel-configurationAssignee: Nitin Kamble <nitin.a.kamble>
Status: RESOLVED FIXED QA Contact:
Severity: major    
Priority: Undecided CC: dvhart, yp.kernel.watcher, yp.watcher
Version: 1.5.1   
Target Milestone: ---   
Hardware: Crownbay   
OS: x86   
Whiteboard:
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: Don't know
Attachments:
Description Flags
log kernel configcheck
none
kernel config
none
invalid
none
mismatch
none
non hdw none

Description Maksym Zhelieznyi 2014-04-18 13:31:00 UTC 
Created attachment 1920 [details]
log kernel configcheck

Host OS:
Ubuntu 12.04.4 LTS x86_64, linux 3.8.0-38

Steps:
1. Download necessary archives:
- Yocto Project 1.5.1 (http://downloads.yoctoproject.org/releases/yocto/yocto-1.5.1/poky-dora-10.0.1.tar.bz2)
- Intel® Atom™ E6xx Processor With Intel EG20T Controller Hub Development Kit (Crown Bay) (http://downloads.yoctoproject.org/releases/yocto/yocto-1.5/machines/crownbay/meta-crownbay-dora-10.0.0.tar.bz2)
2. Extract tarballs, copy meta-intel directory to top of yocto directory
3. Add to the BBLAYERS in the conf/bblayers.conf next directories:
yocto/meta-intel \
yocto/meta-intel/meta-crownbay \
4. Change conf/local.conf:
MACHINE ?= "crownbay"
LICENSE_FLAGS_WHITELIST = "license_emgd-driver-bin_1.18"
5. cd to yocto directory and execute:
source oe-init-build-env
bitbake core-image-sato
6. Run image on the target hardware 
OR 
   mount image and check if emgd.ko module is present in image

During booting system on target hardware, system raises next message:
Please wait: booting...
bootlogd.
xinit: giving up
xinit: unable to connect to X server: Connection refused
xinit: server error

From Xorg logs:
[drm] failed to load kernel module "emgd"

Previous release of the Yocto works good on the same system.
Also see logs of kernel config validation and final kernel config file in attachments.
Comment 1 Maksym Zhelieznyi 2014-04-18 13:31:48 UTC 
Created attachment 1921 [details]
kernel config
Comment 2 Maksym Zhelieznyi 2014-04-18 13:32:09 UTC 
Created attachment 1922 [details]
invalid
Comment 3 Maksym Zhelieznyi 2014-04-18 13:32:26 UTC 
Created attachment 1923 [details]
mismatch
Comment 4 Maksym Zhelieznyi 2014-04-18 13:32:49 UTC 
Created attachment 1924 [details]
non hdw
Comment 5 Darren Hart 2014-04-18 14:23:42 UTC 
Thanks for the detailed report Maksym. The "invalid" log suggests linux-yocto tooling failed to merge the emgd branch. Is there anything to suggest that in the logs?
Comment 6 Darren Hart 2014-04-18 14:24:35 UTC 
Nitin, can you please have a look into this with Maksym. I think there might be a SRCREV issue here related to the LTSI 3.10 updates.
Comment 7 Nitin Kamble 2014-04-18 16:15:02 UTC 
This is for Dora release 1.5.1. So it is using LY kernel 3.8. So it should not be LTSI 3.10 kernel issue. I will build it myself to see what is going on.

Nitin
Comment 8 Nitin Kamble 2014-04-18 17:06:25 UTC 
This bug is with the poky 1.5.0 version of meta-intel BSP. Looks like poky 1.5.1 (dora 10.0.1) source tarballs for meta-intel BSPs are missing. Even with the latest dora branch I am seeing the EGD config being disabled. I am updating SRCREVS for the v3.10 kernel which should fix the issue. 

Note that, the Poky 1.5.2 release will move the kernel for meta-intel BSPs from 3.8 to the latest 3.10+LTSI.

Nitin
Comment 9 Nitin Kamble 2014-04-18 17:27:10 UTC 
Updating to 3.10.35-ltsi is fixing the config issue of the kernel. It is keeping  CONFIG_DRM_EGD=m in the final .config. I will send the commits for dora branch soon.

Nitin
Comment 10 Nitin Kamble 2014-04-22 15:37:21 UTC 
commit 9e27e3f734f5210b18c14607bb85810afa3ce027
Author: Nitin A Kamble <nitin.a.kamble@intel.com>
Date:   Fri Apr 18 10:23:06 2014 -0700

    crownbay: Update linux-yocto 3.10 SRCREV (3.10.35-LTSI)
    
    Update to the HEAD of standard/crownbay (3.10.35-LTSI). This also works
    around an open issue with do_validate_branches where feature branches
    are reset the HEAD of the machine branch if they contain that commit.
    
    Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
    Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>