> The remedy sounds like it solved your problem, but I don't understand > why there mere use of the drive cache, which should be transparent, > should cause CRC errors directly. > > Maybe there is an underlying issue with cable length or quality vs the > UDMA speed in use with it... when the cache was enabled the drive could > spew back data fast enough to show the problem? Just a guess that maybe > dropping down a UDMA level or two might also make the problem go away. > Although it being a VIA chipset, anything is possible or even likely. > I had found that solution over the net (in some obscure link after searching for days). I had tried changing the cable also, but it did not work. Anyhow, my system is as stable as it should be after I disabled the cached flushes (sorry I wrote cached writes earlier). My drive reports the following in the output of dmesg hde: cache flushes supported i run hdparm -W0 /dev/hde in /etc/rc.local to disable that. Vijay G.
