MFAGKCHR@HMARL5.BITNET (04/09/90)
What I found out about the FAT problems up to now. Problems are known to occur in at least 3 machines. All of them after TOS 1.4 was installed. (One machine had not been used with old ROMs). In 2 cases the ROMs were dutch, in one case US. The symptoms are: - lost and cross-linked clusters on the harddisk. - FAT comparison from dlii or SED (diskeditor from the german `kleisterscheibe' do not give reproducible results. (I.e. the FATs are indicated to be different SOMETIMES). I think wrong reading of the FAT would be enough to explain the problem. But why are the FATs read wrong? Problems stop if harddisk is used with another ST (with TOS 1.2). Thus the problem is not caused by an error in the harddisk itself. Harddisks all were original atari disks. Formatted with HDX and used with AHDI in 2 cases (don't know what version), partitioned with SED and used with CBHD in 1 case. The problem does not seem to be related with any program used. It was suggested that a weak driver chip could cause the problem. I did forward the idea to my dealer, he does say he installed the ROMs as Atari told him to (I don't know whether that means that these chips are changed). Could such a chip really cause the problems indicated here? I did test the ROMs as far as possible. I wrote a little program that reads the ROMs 1000 times and compares the results with what it finds the first time. No errors. Thus reading of the ROMs seems to be reproducible. (This program also calculates a checksum. I would be happy to send it to anyone. I would like to know the checksum somebody else finds with dutch ROMs). Up to now nobody has told me he is using dutch ROMs without problems (which does NOT mean that they are bad, just that they can be bad). It would be nice if some people would want to test dlii's disk check repeatedly. I would like to know whether the problem with sometimes (say 1 in 5-10 checks) wrong FAT reads occurs with others to. Do not use a cache program if you test this. The FATs should really be read again. ============================================================== Dept. of Occup. & Chris Evelo Environm. Health MFAGKCHR@HMARL5 (BITNET) and Toxicology. University of Limburg Tel: 31-43-888635 P.O. Box 616 Fax: 31-43-436080 6200 MD Maastricht The Netherlands ==============================================================