Yesterday I did a mistake in a preparation of the patch for PR65710. Here is the patch correcting the mistake. The patch was bootstrapped and tested on x86/x86-64, ppc64, and aarch64. Committed as rev. 221983. I am committing the patch also to gcc-4.9-branch a bit later. 2015-04-10 Vladimir Makarov PR target/65710 * lra-assigns.c (spill_for): Update smallest_bad_spills_num. Print bad_spills_num and insn_pseudos_num.