| Summary: | No space left on device | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] BSPs | Reporter: | Manjunath <manjunath> |
| Component: | bsps-runtime | Assignee: | Saul Wold <sgw> |
| Status: | RESOLVED NOTABUG | QA Contact: | |
| Severity: | normal | ||
| Priority: | Medium | CC: | dvhart, manjunath, sgw, yp.kernel.watcher, yp.watcher |
| Version: | 1.1-beta | ||
| Target Milestone: | 1.3 | ||
| Hardware: | mpc8315e-rdb | ||
| OS: | ppc | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | --- | |
|
Description
Manjunath
2012-01-07 00:45:28 UTC
Please provide details about how you mount the ramdisk and the kernel parameters, for example:
bzImage initrd=initrd.img root=/dev/ram0
What does "mount" report, just run it and provide the output. Please include the complete output of "df" as well.
changing status to reflect the ask for details (In reply to comment #2) > changing status to reflect the ask for details Hi Darren,Bruce and Saul, This issue is resolved. The file system ext2 was created by genext2fs, which used to create sufficient number of inodes for rootfs by default. Now we added -i option to mention the required number of inodes and this has solved the issue as of now. Thanks for your effort. Thanks & Regards, Manjunath We'd still like to automate this to determine the proper inode count. But with the workaround in place, we'll postpone to post 1.2. I will pick this up and review the inode count set up. Can you provide me with a df -i (inodes) on your first boot of the orignal ppc image that you had problems with. I am not sure what you changed the -i to, but it's a bytes/inode value vs setting the # of inodes via the -N options. Manjunath, I found a problem with inode computation in genext2fs, but I am not sure it will address your issue, I need to get more details about your image creation, are you building any of the core-image-* targets when this occurs? Do you have any other settings for changing the size of the image? As I asked before, what does df -i show on first boot and did you really just change the -i parameter to genext2fs or did you use the -N, this information will greatly help me create a reproducer for this issue. Thanks (In reply to comment #6) > Can you provide me with a df -i (inodes) on your first boot of the orignal ppc > image that you had problems with. I am not sure what you changed the -i to, > but it's a bytes/inode value vs setting the # of inodes via the -N options. The command df -i does not work on my target, only df command works (In reply to comment #7) > Manjunath, > > I found a problem with inode computation in genext2fs, but I am not sure it > will address your issue, I need to get more details about your image creation, > are you building any of the core-image-* targets when this occurs? Do you have > any other settings for changing the size of the image? As I asked before, what > does df -i show on first boot and did you really just change the -i parameter > to genext2fs or did you use the -N, this information will greatly help me > create a reproducer for this issue. > > Thanks Yes we create rootfs image using core-image-*, it is compatible with u-boot. No we did not tried with -N option. We also modified rootfs size. When free command is issued the target used to mention between 60-70% usage of rootfs which means still we had space ( memory ) to add files, but if we add any file it used to display "No space left on device", but if i remove a file and add another file whose size is 2 times of the size of original file size ( file which was removed ) it was able to add ( replace file was 2 times bigger than the removed file ). This means the error was not due to space, it has to do with the number of files that can be placed on file system. With addition of -i option it worked. Manjunath, Thank you for this information, could you please run the following and provide me the output: df -i That output will tell us about the inode count, can you please include that along with a regular "df -h" output. That information will help me also, can you send me a diff of the changes that you did make. Thanks Manjunath, is this still an issue for you? Since we have not heard any updates on this, closing as not a bug, if this issue still exists, please re-open. |