Hi Yao, I found your test useful for confirming that the extra logic of my patch [1] has no measurable impact on SO unloading. > + for (i = 0; i < number; i++) > + dlclose (handles[i]); The loop above closes SOs in FIFO style, which might be optimal for GDB. You could alternate closing from the front and closing from the back. I found a bit odd that "make check-perf" is not recognized in the top gdb folder, as "make check" is. But again, it's a minor point to me. Regards, Nicolas [1] http://sourceware.org/ml/gdb-patches/2013-07/msg00684.html Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen, Deutschland Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Christian Lamprechter, Hannes Schwaderer, Douglas Lusk Registergericht: Muenchen HRB 47456 Ust.-IdNr./VAT Registration No.: DE129385895 Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052 &j!z޶N