Date: Wed, 26 Feb 1997 20:41:13 -0500 (EST) From: Thomas David Rivers <ponds!rivers@dg-rtp.dg.com> To: ponds!freefall.cdrom.com!freebsd-hackers Subject: disksort (bad dir/dup alloc research) Message-ID: <199702270141.UAA24345@lakes.water.net>
next in thread | raw e-mail | index | archive | help
While researching the dup alloc/bad dir problem; I thought perhaps disksort() was loosing something (which would explain why newfs thinks it's written 0's and no 0's appear on the disk.) I noticed that in 2.0, disksort() works (mostly) on cylinder values, but in 2.0.5 and later this was rewritten to simply use the pblkno [physical block number] field. Just glancing through disksort(), I didn't see anything that would have caused a I/O to "go missing" - but I plan on adding some printf's just to make sure... [The reason for this is, I believe, my research indicates that, sometimes, I/O does seem to "go missing."] - Dave Rivers -
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199702270141.UAA24345>