[daip] New Ticket - [!XAQ-182450]: Text installation of AIPS on Ubuntu 11.04 (gcc, gforgran 4.5.2)

[Michael Bietenholz]

New Ticket: Text installation of AIPS on Ubuntu 11.04 (gcc,gforgran 4.5.2)

This is probably not so much a ticket per se, but some of the
information may be useful to you, and there are a few questions at the
end that I would be happy to have an expert answer to.

I think I have a text installation on a 64-bit Ubuntu 11.04 system to
work.  (Linux 2.6.38-11-generic #48-Ubuntu SMP x86_64 x86_64 x86_64
GNU/Linux; gcc and gfortran 4.5.2)

Here's what I found is needed:

1) fix the Makefile for XAS (remove the -axWPT compiler flag).

2) here's the compiler options I used (.AIPSRC)

FORT = /usr/bin/gfortran
FORTOPT = -c -fno-automatic -Wall -U_FORTIFY_SOURCE
CCOM = /usr/bin/gcc
CCOMOPT = -c -O3 -fomit-frame-pointer -funroll-loops -I/home/aips/31DEC11/INC -D_FILE_OFFSET_BITS=64 -DHAVE_LINUX_GLIBC -U_FORTIFY_SOURCE
LINK = /usr/bin/gfortran
READLINE = /usr/lib/libreadline.a

3) After that INSTEP2,4 worked except for two tasks, PRTAB and RSTOR,
which had to be compiled manually with "NOOPT" set (the compiler
segfault's if run with standard optimization).

Now I first tried withouth the '-U_FORTIFY_SOURCE' flag, and got very
strange results.  Some tasks died with "buffer overflow" messages (see
attached messages from PRTUV), but then, later on (withouth afaik any
changes to either AIPS or the system), PRTUV ran fine.

I don't really know anything about "fortify" - it seems to include
some form of run-time checks designed to prevent buffer overflow
exploits into the executables.  

so here's my questions:

Q1) is turning off the "fortify" checks with -U_FORTIFY_SOURCE
reasonable?

Q2) why does "fortify" cause AIPS tasks to crash - is AIPS exploiting
buffer overflows? why did PRTUV first crash and then subsequently run
(same data set etc)

Q2) in Y2K I got only 7 and 10 bits correct (for APRES), which seems
rather low I've attached the results from Y2K-LARGE - are they
reasonable?

Q4) should I worry about having to compile PRTAB and RSTOR without
optimization (the resulting loss in speed for these two rarely used
tasks is certainly not a problem; but the compiler segfaulting
does not lend confidence in the rest of its output!)


Ticket Details
===================
Ticket ID: XAQ-182450
Department: AIPS Data Processing
Priority: Default
Status: Open
Link:  https://help.nrao.edu/staff/index.php?_m=tickets&_a=viewticket&ticketid=1066
-------------- next part --------------
A non-text attachment was scrubbed...
Name: y2k-large
Type: application/octet-stream
Size: 57924 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20110821/a2fac5a6/attachment.obj>