With USB drives, it's not a good idea to depend on any kind of kernel-assigned host/drive number. The kernel assigns those dynamically, depending on the order it recognises them in, the order it deigns to process them in, the locking order of internal kernel locks, the phase of the moon, and how you hold your mouth. It is usually the same order from a clean bootup with no cabling changes, but nothing is guaranteed. That's why I thought that mount-by-label was vital to Tomato.freddyspam,
The swapon command as you use it will not work if kernel assigns different host number to your drive after a reboot. However, it doesn't explain why the partition name was displayed incorrectly in the GUI.
Yup. No bugs in Tomato (probably). What there is, is a kernel bug. Finding it will be a bitch. I have it on my todo list-----but way down---after jffs improvements and webcam support. To do it, I'm planning on building the kernel and Tomato's USB code on a PC, and using the PC console & kernel debug facilities. This is not something that can be easily done on router hardware.I don't see any obvious flaws in the code that can cause it, but it looks like reading from procfs files (/proc/scsi/scsi or /proc/partitions etc) while kernel is writing to these files can make scsi/usb subsystem to go wild - I'll do more research on that.
BTW, the kernel doesn't "write to these files". The /proc files are a window into the kernel. The contents are generated on-the-fly as you read them. That's why the sizes of the /proc files always display as zero.