llvm-svn: 17934

* Eliminate the releaseMemory method, this is not an analysis
* Change the fixed, active, and inactive lists of intervals to maintain an
  iterator for the current position in the interval.  This allows us to do
  constant time increments of the iterator instead of having to do a binary
  search to find our liverange in our liveinterval all of the time, which
  substantially speeds up cases where LiveIntervals have many LiveRanges
  - which is very common for physical registers.  On targets with many
  physregs, this can make a noticable difference.

  With a release build of LLC for PPC, this halves the time in
  processInactiveIntervals and processActiveIntervals, from 1.5s to .75s.

  This also lays the ground for more to come.

llvm-svn: 17933

llvm-svn: 17932

llvm-svn: 17931

iterator/begin/end members.

llvm-svn: 17930

llvm-svn: 17929

llvm-svn: 17927

llvm-svn: 17926

llvm-svn: 17925

ReturnInst constructed is the same as if NULL was passed instead of
the void value.

llvm-svn: 17923

llvm-svn: 17921

* Don't include Makefile.rules in set of preconditions, it never has to be
  copied to objdir.
* Enable the "update makefile first before executing targets" feature in
  gnu make by *not* using a full path to the Makefile in the rule.

llvm-svn: 17920

* ensure trailing spaces are eliminated so they don't factor into the
  length of a member's name.
* make sure all the bytes of a name are written even if the name ends in
  multiple null characters (bug in OSX ar)
* make sure we provide the full member name when searching for symbols so
  the module name is not accidentally duplicated.

llvm-svn: 17918

name of the module: "Archive.a(object.o)"

llvm-svn: 17917

llvm-svn: 17916

llvm-svn: 17915

llvm-svn: 17914

by saying what we mean

llvm-svn: 17913

llvm-svn: 17912

llvm-svn: 17911

llvm-svn: 17910

*do* include the length of the long file in the length of the member and
they are *not* null terminated.

llvm-svn: 17909

Added my own defMap. Only saving values that are not loop invariant. Fixed a couple of assertions that were triggered due to registers not being allocated. These both had to do with PHINodes.

llvm-svn: 17907

to 34.7s with a profile build.

llvm-svn: 17906

llvm-svn: 17905

llvm-svn: 17904

llvm-svn: 17903

llvm-svn: 17902

* Don't use variable length arrays (replaced with alloca)

llvm-svn: 17901

being linked do not need to be remapped any longer.

llvm-svn: 17900

llvm-svn: 17899

out this recent regression

llvm-svn: 17898

llvm-svn: 17897

llvm-svn: 17896

llvm-svn: 17895

You can't have implicit defs that overlap explicit defs, or implicit
defs that alias one another.

llvm-svn: 17894

by splicing function bodies from the src module to the destination module.

This speeds up linking quite a bit, e.g. gccld time on 176.gcc from 26s -> 20s
when forming the .rbc file, with a profile build.  One of the really strange
but cool effects of this patch is that it speeds up the optimizers as well,
from 12s -> 10.7s, presumably because of better locality???

In any case, this is just a first step.  We can trivially get rid of the
LocalMap now and do other simplifications.

llvm-svn: 17893

llvm-svn: 17892

Patch contributed by Jeff Cohen.

llvm-svn: 17891

Patch contributed by Jeff Cohen.

llvm-svn: 17890