Hi,
I had exactly the same problem here. In my case I found a solution to get R initialized which is reported below. (Unfortunately, I got stuck with a subsequent problem: http://tech.knime.org/forum/r-statistics-nodes-and-integration/r-interactive-error-in-identicalsimplify-false. Maybe you can help there?)
I'm running a Linux 64bit system, Knime 2.9.2, R 3.0.3, and followed the steps in the FAQ: http://tech.knime.org/faq#q25.
As mentioned in #8 above, my $R_HOME/library/rJava/jri/ only contains the two archives JRIEngine.jar and REngine.jar, but no *.so or *.jnilib and no x64 subfolder. Within the rJava folder, I only found one *.so file, which is rJava.so inside the $R_HOME/library/rJava/libs/ folder. However, pointing the knime.ini to this path doesn't make any difference. Whenever I tried to run (or configure) an R node, I got an error that R could not be initialized.
The important point was that I compiled R from the sources! When installing rJava within R, I found the following log messages:
checking whether JRI should be compiled (autodetect)... no
...
checking whether JRI is requested... no
So I tried to explicitly request JRI using the following command:
> install.packages("rJava", configure.args=c("--enable-jri"))
which in turn required the R system to be build as a shared library:
$ ./configure --enable-R-shlib
$ make
Finally, after re-building R with "enable-R-shlib" option and subsequently installing rJava (which than automatically enabled jri), I got a "libjri.so" in the folder $R_HOME/library/rJava/jri/ to which I pointed the knime.ini, and Knime is able to initialize R 3.0.3.
Best, Andreas