OK, thanks for the update. partitioning would be very important for my current work so I'd like to understand what is so special with ARM that it's the only target that can't achieve that (on the V7 at least ). Christophe, Mathew, did you have a test case (I don't have a direct access to the Linaro archives) ?
Thanks a lot, Christian On 10/01/2014 12:43 PM, Ramana Radhakrishnan wrote: > On Wed, Oct 1, 2014 at 10:03 AM, Christian Bruel <christian.br...@st.com> > wrote: >> Hi Ramana, >> >> Your patch https://gcc.gnu.org/ml/gcc-patches/2012-02/msg01492.html >> seems to have not been applied for 4.10. Are there any stoppers or is it >> an omission ? > Short answer, no, not an omission. It could not be made to work > properly for a few reasons. When I continued with it the problem I was > hitting was the assumption that you can branch anywhere. IIRC using an > indirect branch wasn't possible because you couldn't find enough > registers because of pass ordering issues. > > Matthew Gretton-Dann and Christophe Lyon at Linaro worked on it for > sometime and they hit other problems. There are probably enough mails > in the archive to document this history. > > regards > Ramana
