fredriks@cbnewse.ATT.COM (lars.fredriksen) (01/03/90)
Hi! Here are some preliminary results that I have gotten from testing 1.5.0 on my system(Sperry PC/HT-300 -- ,really a Mitsubishi machine, XT with 20Mb harddrive, 360K floppy, OMTI HD controller, and the rest seem to be pretty standard, runs at 7.xxMh) Patching and compliling the kernel worked out pretty good. No problems other than those before mentioned by others. I haven't yet recompiled /usr/src/commands, but I don't know if that will change the test retults any. If my memory serves me right, my results looks an awful like those Glen Overby had. Here Goes: test0 - 2 --> ok test3 --> No ok but no error message either test4 --> ok test5 --> Infinite loop test6 - 8 --> ok test9 --> Hangup (Whatever that means) test10 - 14 --> ok, but test10 was awful quick! test20 --> Subtest 3 5 6 7 8 9 10 11 failed with errors like: -No such file -Not enough core -Bad file descriptor -etc (sorry for not bringing all the codes) after killing test20 off (it looped twice), a couple of strange things showed up. ls on '/' showed nothing, ls -l on '/' gave total = 0 as only output; however I could cd to /bin and do an ls there and the files showed up, same for /etc. I had all the test programs on a floppy mounted in /user, and the only thing left on that floppy was an N recursing directory called L5, which the 1.3 version on rmdir couldn't get rid of. All the original files on that floppy was gone. I assumed that most of the tests used only libc calls and not many execs of /bin programs. Howvever, I did notice that test20 tried to exec chmod and rmdir, which failed since I didn't have /usr/bin mounted(May have been my wisest move all last night,.. Thanks a lot for the warnings that have been posted about test20 thrasing filesystems. I ran these tests last night and have not had time to check the source out yet(tonights project). But I thought some people might be interrested in the test results anyhow. If anyone have experienced the same problems, and know what the causes are, please let me know. I am going to recompile all the commands before I run the tests again, to see if that makes any difference. A couple of days ago, I posted a question regardless system(3), and its return code. From what I have found out system(3)s return code is implementation dependent and the way it is right now, look exactly the same as it does in a system 7 offspring called dmert. That is, the return from system looks like that from wait(). Basically system(3) returns status from the child, and programs should check for (127 <<8) as error code, as well as -1. With respect to make never stopping when INTR was hit, that never happened to my 1.4 shell and make. Anyway, hope to get 1.5.0 up and ready soon. Any suggestions are appreciated. Sincerely Lars Fredriksen