mullen@sdsu.UUCP (deborah mullen) (10/18/88)
I have updated to the lasted kernel changes which have the large buffers for rs232. Everything appeared OK, even used the rs232 over my modem ( I had NR_RS_LINES set to 2). In the process of checking my source for lib I found that the command " cat /dev/null > file" just put a bunch of garbage to the file and only stopped when the disk was full. I had this problem when doing some updates ( back in 1.3b) but it went away with a new boot. Now it is back. Then I extracted everything from my library, replace the lib/IBM_PC files and built the library again. The result was a library almost 35K bigger!!! Thinking something strange with the kernel, I tried booting with my old 1.3b boot disk and get unrecoverable disk error on device 3/3 block1 -- panic. I do not have my original 1.3c boot ( lent it to a friend) but will try it with that soon. My lastest boot disk boots fine and gives no errors. I was not under the impression that the format of the filesystems had changed. If anyone has any ideas please let me know. Also, if NR_CONS is changed from 1 to 2, the ( current and old) system will not boot. Just changing this constant should result in more memory being used, but not affect the operation unless other code is changed. Is the DMA overrun bug memtioned by Bruce Evans coming into play here? Any and all comments welcome. Thanks in advance, Deborah Mullen By the way, all my kernel, fs and mm source patch the posted crc ( except bookblok.s which was wrong -- not include the fix and fsck.c -- patched by hand). I have an IBM XT clone (4 & 6MH) with 10M harddisk.