The initialized linked failures are the failures which block further development for me at least. PR 25886 is a huge failure as we are now over the 255 limit for tree codes for some languages (well Objective-C++ only but it gives us a hint of what is to come soon). PR 25860 is an ICE due to SSA PRE/ load PRE, there is a fix which just needs to be applied. PR 25569 fails on powerpc64-linux, though it is only with Fortran code (no C example has been found yet). PR 25905 is a bootstrap failure on m68k due to an ICE in combine.c PR 25890 is a testsuite failure on x86_64-linux-gnu (and a couple others) due to an ICE in combine.c I might add others later today. There is a tree vectorizer bug which I might add.
PR 25636: reload is causing a write to read only memory, exposed by PRE changes.
PR 25259: libgomp depends on stdint.h.
PR 25910: libgomp fails to compile on i686-linux-gnu.
PR 25865: libgomp dects TLS as working when it does not. Causes a build failure on *-netbsd*.
PR 25884: perl is now required to build GCC (libgomp).
ICE in c-common.c due to PARM_DECL and ADDR_EXPR and not having visibility.
(In reply to comment #6) > ICE in c-common.c due to PARM_DECL and ADDR_EXPR and not having visibility. That was PR 25861.
25911: -O2 -ffast-math -ftree-vectorize ICEs on reasonable code.
PR 25931: build failure for powerpc-darwin target (maybe others too).
PR 25908: vtables and classes no longer work on powerpc-darwin
PR 25942: libgomp needs correct version of autoconf/automake installed to build.
PR 25959: C++ building on powerpc-* is broken.
PR 25703: wrong code on x86-linux, there is a C example from the duplicated bug. PR 25961: Bootstrap failure on ia64-* after Zack's patches.
Closing as works for me as Mark's email works for me.