karl@sugar.uu.net (Karl Lehenbauer) (09/15/88)
There has been some discussion of how the filesystem should handle bad blocks. I'd like to note that in Unix System V Release 3.1 (and beyond) the filesystem handles it quite well, to wit: Disk errors are communicated back to the operating system, as always. If it's a soft error (detected and corrected by ECC code) the block is marked bad and the data is written to the new block that has been mapped in its place via the regular bad block handling mechanism. If it is was a hard error, the operating system figures out what the block was in the context of the filesystem, and responds accordingly: If the block is in the free list, it is simply remapped. If it was in a file, a message is printed saying that a file has been corrupted. If it was in /unix (the OS load file), or was really important to the filesystem, the system remaps the block and prints a message to the effect that a critical block has a hard error and the system may not be able to come back up once it is rebooted. -- -- uunet!sugar!karl, Unix BBS (713) 438-5018