Hey all, So, per Ed's feedback on my first series, I went and spent some time trying to figure out how to have wic know what the UUID would be when updating the fstab. It turns out the easiest answer here is to have WIC make the UUID. Per Otavio's concern last time, I also make sure that the filesystem UUID can be passed in, for reproducibility. One thing to keep in mind here is that FAT filesystem UUIDs are a bit funny to deal with as mkfs.vfat / mkdosfs / etc want to be given a 32bit hexadecimal value. But when we talk mount, it must be split and it must be in uppercase. To make the rest of the code easier I'm encoding the '0x' portion into part.fsuuid rather than doing "-i 0x%s" in a bunch of places.
While preparing all of this, I found a few minor things such as we did not test for squashfs and --use-uuid (not supported) and an incorrect comment around the btrfs support. Since v1, I've added a testcase into wic.Wic.test_qemu for a UUID mount and the UUID that we've given. I think this is cleaner than adding a python function to make a wks file just for this task. Since v2, I've rebased to master again and removed an accidental change that limited qemu to running without graphics. oe-selftest wic continues to have the testcases that I'm modifying work, so I'm not sure about Ross' problem on the autobuilder. Since v3, I've moved the mount point for the test from /uuid to /media as when testing with sysvinit (as is required), /uuid will not be created so we need to test with an existing mount point. -- Tom -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core