Heads up - symlinked GotoBLAS in F14 - R-SIG-Fedora

Sun, Nov 7, 2010 11:54 PM #

Hi,

I installed F14 on release, but was diquieted to note that my symlinked 
GotoBLAS (http://cran.r-project.org/doc/manuals/R-admin.html#Shared-BLAS, 
foot of paragraph) prevented R (compiled from source, no special configure 
options) from loading - everything had been fine on F < 14. It seems that:

cat /selinux/booleans/allow_execstack

is 1 1 on my RHEL systems, but became (?) 0 0 in F14. Could anyone with 
F < 14 check its value there?

The "fix" on F14 is:

System -> Administration -> SELinux Management -> (root password, wait), 
then Boolean -> global | allow_execstack | Allow unconfirmed executables 
to make their stack executable [tick]

The comments suggest that this is not a good policy! Following the 
adjustment, the booleans are 1 1, and R loads with symlinked GotoBLAS on 
F14.

Roger

Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no

Gavin Simpson

Mon, Nov 8, 2010 12:07 AM #

On Mon, 2010-11-08 at 08:54 +0100, Roger Bivand wrote:

It is 1 1 on both my F13 and F12 boxes (Time to upgrade at least one of
those I guess...) but I haven't used any BLAS other than the one
supplied in the R sources.

HTH

G

%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
 Dr. Gavin Simpson             [t] +44 (0)20 7679 0522
 ECRC, UCL Geography,          [f] +44 (0)20 7679 0565
 Pearson Building,             [e] gavin.simpsonATNOSPAMucl.ac.uk
 Gower Street, London          [w] http://www.ucl.ac.uk/~ucfagls/
 UK. WC1E 6BT.                 [w] http://www.freshwaters.org.uk
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%

Roger Bivand

Mon, Nov 8, 2010 12:14 AM #

On Mon, 8 Nov 2010, Gavin Simpson wrote:

Thanks. So the SELinux policy on this changed for F14, and likely will 
change for RHEL 6. Using GotoBLAS isn't necessary and doesn't help in many 
settings, where the standard R BLAS is absolutely OK. But for larger 
matrices on multicore machines, it can make things go faster when the 
conditions are right, because it can share the operation across cores.

Roger

Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no