• Sun Oct 18 2009 Jakub Jelinek jakub@redhat.com 4.4.2-4
• update from gcc-4_4-branch
• PRs c++/37204, c++/37766, c++/37875, c++/38798, c++/40092,
  libstdc++/40654, libstdc++/40826

• fix VTA ICE on invalid pointer arithmetics (#529512)

• Sat Oct 17 2009 Jakub Jelinek jakub@redhat.com 4.4.2-3

• fix VTA handling in the scheduler (PR debug/41535)

• fix up %%check section to be able to find ecj1

• Fri Oct 16 2009 Jakub Jelinek jakub@redhat.com 4.4.2-2

• update from gcc-4_4-branch
• PR target/40913
• VTA backports
• PR debug/41717

• fix Ada .eh_frame generation (PR debug/40521)

• Thu Oct 15 2009 Jakub Jelinek jakub@redhat.com 4.4.2-1

• update from gcc-4_4-branch
• GCC 4.4.2 release
• PRs middle-end/22072, target/41665
• don't emit -Wpadded warnings for builtin structures

• don't generate .eh_frame, but generate .debug_frame when -g and none of
  -fasynchronous-unwind-tables/-fexceptions/-funwind-tables is used
  (PR debug/40521)

• Wed Oct 14 2009 Jakub Jelinek jakub@redhat.com 4.4.1-22

• update from gcc-4_4-branch
• PRs target/26515, target/38948
• fix s390{,x} BLKmode symbol handling
• fix i?86 testqi splitter (#528206, PR target/41680)
• VTA backports
• introduce debug temps (PRs debug/41264, debug/41338, debug/41343,
  debug/41447, target/41693)
• build debug stmts on updates (PR debug/41616)
• fix another with/without -save-temps debug info difference
  (#526841, PR preprocessor/41543)
• fix invalid ranges in .debug_loc section (PR debug/41695)
  %if 0%{?rhel} >= 6

• if -mcpu= isn't specified, default to -mcpu=power4 (#463549)
  %endif

• Sat Oct 10 2009 Jakub Jelinek jakub@redhat.com 4.4.1-21

• update from gcc-4_4-branch
• fix s390{,x} prefetch for pre-z10 CPUs (#524552)
• VTA backports
• fix debug info differences with/without -save-temps
  (PR preprocessor/41445)
• fix ICE with small BLKmode returning call (#516028,
  PR rtl-optimization/41646)

These packages fix a lot of, mainly debug info related, issues. Can you please first, for a short moment, tag it into dist-f12-override so that I can also build updated libtool which depends on the exact same gcc %{version} and then when that is build, tag both gcc and libtool into dist-f12 at the same time?

Besides the standard %check testing each gcc build performs it has been also tested by building kernel (without -fno-var-tracking-assignments hack) and glibc (again, without that temporary workaround).

BTW, on the other side, now that binutils-*-33.fc12 is tagged into dist-f12, it should be untagged from dist-f12-override.