Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
soc:2009:pravin:journal:week6 [2009/07/04 06:16]
less1
soc:2009:pravin:journal:week6 [2009/07/04 08:41] (current)
less1
Line 23: Line 23:
   - Now, actual work of understanding /init script and understanding how they use iso image as root starts.   - Now, actual work of understanding /init script and understanding how they use iso image as root starts.
   - Done with modifications in /​sbin/​real-init which will (hopefully) mount the httpfs partition over root.   - Done with modifications in /​sbin/​real-init which will (hopefully) mount the httpfs partition over root.
 +  - Initial runs were not very promising, when tried to boot this modified fedora live cd over BKO using qemu, it dropped me in rescue shell :!:
 +     - There were few error messages in boot sequence
 +        - ''​mount:​ only root can do that''​
 +        - ''​getKernelCmdLine:​ failed to open /​proc/​cmdline:​ No such file or directory''​
 +        - ''​creating initial /dev : only root can do that''​
 +
 +
 +
 +===== 3 July =====
 +  - Pin-pointed the cause 
 +     - when I tried command ''​mount -t proc proc /​proc''​ It gave error saying ''​mount:​ only root can do that''​
 +     - But as this command is executed from within initrd, it is executed as root, so How can I get that error :?:
 +     - With tip from AndyTim, it seemed that the initrd was created by non-root user, so even if those commands are executed by root, they don't have root privilages. ​
 +        - It is strange in many ways, as initrd by non-root user works fine for ubuntu and other distributions I have tried.
 +        - I suspect that selinux security policies are enforcing these extra rules as fedora is having selinux in enforcing mode.
 +        - Anyways, creating initrd by root user does make sense in some ways. So, I will follow this for all distributions.
 +   - The booting process has progressed, but now it is stuck at some other point. It just hangs without giving me rescue shell
 +   - I don't even know if it has used httpfs, all my ''​echo''​ messages are lost in fast scrolling output of qemu, and I am not able to scroll back and see, what was happening. I need to find some way to be able to see previous outputs. of qemu. I miss that feature seriously in qemu.
 +
 +===== 4 July =====
 +  - I will need one of the two things to be able to do debugging properly for problem in hand.
 +     - Scrolling support for qemu (so that I can see old outputs and my debug messages )
 +     - dhcp support in virtual box (<​del>​I presume that it is having some kind of scrolling support</​del>​ I tested and ''​ALT+PG_UP''​ does some scrolling up )
 +  - Started my day with trying to find how to scroll in qemu, but nothing promising came out.
 +  - Both of above ways are time consuming, so I have resolved to use one odd method, taking screen-shots and then checking the old messages, only thing I can say about this method is that, it gets my job done.
 +  - Now, coming back to the problem, I will log the progress with Red Hat [[Fedora11BKO|Here]] ​    
  
  
  

QR Code
QR Code soc:2009:pravin:journal:week6 (generated for current page)