Results for haifa-enabled egcs-2.93.20 19990427 (gcc2 ss-980929 experimental) testsuite on rs6000-ibm-aix4.1.5.0

[Donn Terry]

I mentioned by private mail to Craig that I am seeing the
same thing, but couldn't remember the exact details; your note
reminded me of them.

I'm pretty sure this is the file group inheretence thing.
Some systems create files with the group owner of the group
owner of the directory, some with the group owner of the process.
Both are permitted by POSIX.

The easy fix to this one is to bag that clause in the test.

The right fix is to find the directory, stat it, and get its
group owner and compare the two.

I'll let someone else make that judgement call.

Donn

David W. Schuler wrote:
> 
> I've done some work in debugging the following errors under AIX 4.1.5:
> 
> >FAIL: g77.f-torture/execute/u77-test.f execution,  -O0
> >FAIL: g77.f-torture/execute/u77-test.f execution,  -O1
> >FAIL: g77.f-torture/execute/u77-test.f execution,  -O2
> >FAIL: g77.f-torture/execute/u77-test.f execution,  -O2 -fomit-frame-pointer -finline-functions
> >FAIL: g77.f-torture/execute/u77-test.f execution,  -O2 -fomit-frame-pointer -finline-functions -funroll-loops
> >FAIL: g77.f-torture/execute/u77-test.f execution,  -O2 -fomit-frame-pointer -finline-functions -funroll-all-loops
> >FAIL: g77.f-torture/execute/u77-test.f execution,  -Os
> 
> First, it's probably very important to explain my environment:
> 
>     Home directory is in AFS, with a dual-client AFS/DFS system
>     EGCS code is located in an area that is NFS-mounted to yet another box
> 
> That said, when I run the u77-test.f code, it aborts at line 225 with the
> following output:
> 
>     ...
>      Current directory is "/usr/local/x86srvr1/WORKSPACE5/schuld/EGCS/egcs-19990428-aix41/gcc/tmp"
>      Creating file "foo" for testing...
>      with stat array  -2147483629 2900650 33206 1 2359 3 2913149 2 925407554
>       925407554 925407554 4096 8
>      *** FSTAT uid, gid or nlink is wrong
> 
> The code in this area is:
> 
>       if (statb(5).ne.getuid () .or. statb(6).ne.getgid() .or. statb(4)
>      +     .ne. 1) then
>         write (6,*) '*** FSTAT uid, gid or nlink is wrong'
>         call abort
>       end if
> 
> I did some further investigation, and found the statb(6) result the stat()
> call (executed a few lines prior to the above code) is returning a gid value
> of "1597", but get getgid() call returns "3".  The "1597" is the gid value in
> the /etc/passwd file, and "3" is the entry for "sys".  I notice that all of my
> files are created with a gid of "3".  I believe this to be an "AFS-ism", and
> not an error in the g77 code itself.
> 
> Perhaps someone with more knowledge of AFS (and DFS) could think about how
> this code could/should be restructured to take AFS and/or DFS into account
> when running this test.  Is there anything unique that should be added to
> the testcases specifically for AFS or DFS enabled systems???
> 
> -------------------------------------------------------------------------------
> David W. Schuler - Advisory Engineer
> Semiconductor Contract Mfg            Internet: schuld@btv.ibm.com
> IBM Microelectronics                     Phone: 802-769-7636
> 1000 River Street - MS/862D                FAX: 802-769-6206
> Essex Junction, VT   05452-4299          Pager: 888-438-4343
> For IBM Foundry information:         http://www.chips.ibm.com/services/foundry/

-- 

===================================================
Donn Terry                  mailto:donn@interix.com
Softway Systems, Inc.        http://www.interix.com
2850 McClelland Dr, Ste. 1800   Ft.Collins CO 80525
Tel: +1-970-204-9900           Fax: +1-970-204-9951
===================================================