Created attachment 20 [details] qemumips poky-image-sato/sdk fail to boot Testing Env: Host: Ubuntu 10.04 LTS Kernel: 2.6.32-21-generic-pae Poky gitinfo(used to build qemu-native): 0c3b1aa0a5461c9b5f3833feb091f7c8a459ca9b (git://git.pokylinux.org/poky.git) all the images from weekly build (20100721) on autobuilder, http://yoctobuild01.jf.intel.com/build-areas/weekly-release/build/build/tmp/deploy/images/. With the qemumips poky-image-sato/sdk images from weekly builds 20100721, they fail to boot up with "No space" error. Pls. see attachment.
this perhaps due to inode limitation, but busybox 'df' doesn't report that information now... need more work to confirm
this does come from inode limitation: (sato pre boot) Inode count: 6864 Block count: 270308 Reserved block count: 13515 Free blocks: 70938 Free inodes: 33 (sato after boot) Inode count: 6864 Block count: 270308 Reserved block count: 13515 Free blocks: 69909 Free inodes: 0 Since free inode is empty now, "no space on device" error appears. SDK image has more available inodes: (sdk) Inode count: 17608 Block count: 581092 Reserved block count: 29054 Free blocks: 55675 Free inodes: 263 But it still suffers from potential "no space" risk soon. Actually I also checked no MIPS image generated at Jul.10, which looks more reasonable: (07/10 sato) Inode count: 28672 Block count: 229377 Reserved block count: 2049 Free blocks: 20559 Free inodes: 21816 It looks over 20,000 inodes are missing in current images. Even current x86 image has similar lackage risk: (x86) Inode count: 6216 Block count: 296704 Reserved block count: 14835 Free blocks: 59231 Free inodes: 153 Appears that the reason is it. I'll find appropriate fix soon
it appears that this relates to recent genext2fs upgrade...
I've sent out a fix to poky mailing list for this bug
Kevin, I think a bug can be marked as FIXED only after the fixing patch is picked in tree and built in weekly/milestone build. BTW, for the inode bug, I think you may need to ask Saul put it into M2 brunch if we want it fixed in M2.
This has been integrated into the M2 branch
Verified with M2-20100728 build, the bug is fixed.
*** Bug 130 has been marked as a duplicate of this bug. ***