From: John Doe <jdmls at yahoo.com> > 4. OT: fs corruption ont he USB keys... > I copy the iso file on the key (no error message), then compare it to the > original and > some random differences appear. I noticed in the logs: > "ext2_new_block: Allocating block in system zone - block = 65536" > "ext2_free_blocks: bit already cleared for block 88678" > "ext2_free_blocks: Freeing blocks not in datazone - block = 3003818537, > count = 1" About the corruption problems, I noticed: kernel: attempt to access beyond end of device kernel: sdg2: rw=0, want=27039680368, limit=15746760 kernel: attempt to access beyond end of device kernel: sdg2: rw=0, want=7948148456, limit=15746760 kernel: attempt to access beyond end of device kernel: sdg2: rw=0, want=7948148456, limit=15746760 kernel: attempt to access beyond end of device kernel: sdg2: rw=0, want=10942330464, limit=15746760 kernel: attempt to access beyond end of device kernel: sdg2: rw=0, want=10942330464, limit=15746760 etc... The limit would look correct: 15746760*512=8062341120=8GB The usb-storage seems to have a little problem with the keys geometry... I am not even entirely filling the key; I fill it up to 60% It happens with 2 different (but same model) keys. # fdisk -l /dev/sdg Disk /dev/sdg: 8086 MB, 8086618112 bytes 249 heads, 62 sectors/track, 1023 cylinders Units = cylinders of 15438 * 512 = 7904256 bytes Device Boot Start End Blocks Id System /dev/sdg1 * 1 3 23126 6 FAT16 /dev/sdg2 4 1023 7873380 83 Linux Do USB keys need special geometries (not a boot problem)? Or could it be the corruption problem that was reported on the list with the new kernel...? Thx, JD