bcx@eeg.UUCP (02/09/88)
I may have detected a possible bug with "mkfs". When mkfs assigns bad
blocks from the raw file system's bad block list, to the bad block
inode 1 in the logical file system, it only assigns individual blocks,
rather than the entire cluster associated with that block. The kernel,
for efficiency, buffers blocks by reading whole clusters (8 blocks
beginning on a mod 8 boundry). As a result, blocks from the bad block
file are read, despite their forbidden status. Continually accessing
bad blocks can, at times, cause a disk to slowly degrade and develope
more bad blocks.
Is this a bug or am I completely misunderstanding the way RTU 3.1
operates?
Bryan Costales, Senior Systems Programmer, EEG Systems Laboratory
1855 Folsom St Rm610, San Francisco, Ca 94103 (415) 621-8343
{ihnp4,sun,bellcore,lll-crg,dual,qantel,pyramid}!ptsfa!eeg!bcx