Commit Graph

14746 Commits

Author SHA1 Message Date
Chris Lattner 6017d47064 Use explicit std:: qualification to avoid relying on Koenig lookup, which
VC++ does not do properly.  Thanks to Morten Ofstad for the patch!

llvm-svn: 16955
2004-10-13 15:11:23 +00:00
Chris Lattner 36fca705b1 Get rid of template templates that were preventing VC from compiling the
set_intersect template.  Thanks to Morten Ofstad and Jeff Cohen for the
patch!

llvm-svn: 16954
2004-10-13 15:09:21 +00:00
Reid Spencer def9848a6c Changes for automake.
llvm-svn: 16953
2004-10-13 11:56:07 +00:00
Reid Spencer d385a67c44 Set up BUILD_MODE based on the configuration so the default build mode
is established from configuration. Also, don't make the ENABLE_ options
AM_CONDITIONALs.

llvm-svn: 16952
2004-10-13 11:53:12 +00:00
Reid Spencer 26bfe011f8 Updates for changes in Makefile rules.
llvm-svn: 16951
2004-10-13 11:48:50 +00:00
Reid Spencer ace94df71f Update to reflect changes in Makefile rules.
llvm-svn: 16950
2004-10-13 11:46:52 +00:00
Reid Spencer 5b240ad1d9 Get rid of the horrendous FIND_PATH macro, have libs and progs linked into
a common directory for simplicity. Provide an easy way to relink some
objects

llvm-svn: 16949
2004-10-13 11:39:51 +00:00
Chris Lattner b15f04623b Update doco, make doxygen more, use standard dividers.
llvm-svn: 16948
2004-10-13 04:44:53 +00:00
Misha Brukman ac00eb4eb6 Save Reid's notes from bugzilla on the proper procedure to upgrade libtool
llvm-svn: 16947
2004-10-13 01:07:55 +00:00
Reid Spencer 946085840c Resolution to PR451.
llvm-svn: 16946
2004-10-13 01:01:03 +00:00
Reid Spencer 5fcd5c70ca Changes for libtool 1.5.10
llvm-svn: 16945
2004-10-13 00:51:45 +00:00
Reid Spencer 2c1c1c46a9 Update for libtool 1.5.10
llvm-svn: 16944
2004-10-13 00:19:25 +00:00
Reid Spencer d65d4d0b14 Harden the version requirements to the set that actually works after three
days of trying to figure it out. Despite some developer's penchant for
relaxing the tool versions, this just isn't possible. Only certain versions
work with certain other versions.

llvm-svn: 16943
2004-10-13 00:18:58 +00:00
Misha Brukman 89f77bce58 Fix file comment header
llvm-svn: 16942
2004-10-12 21:48:02 +00:00
Misha Brukman bf24de0939 On second thought, OpaqueType is not really a good first-class type.
llvm-svn: 16941
2004-10-12 20:35:04 +00:00
Misha Brukman dfd7ada4a2 No space allowed between \ and end-of-line
llvm-svn: 16940
2004-10-12 19:17:11 +00:00
Misha Brukman 25754a7703 Wrap lines at 80 cols
llvm-svn: 16939
2004-10-12 18:35:57 +00:00
Chris Lattner 92e37e092c Add comment
llvm-svn: 16937
2004-10-12 17:08:44 +00:00
Misha Brukman 0338401147 Opaque types are considered to be first-class.
llvm-svn: 16936
2004-10-12 17:06:17 +00:00
Chris Lattner 76a9758cf1 Minor cleanups:
* fit in 80 lines
 * Eliminate extra namespaces
 * Drop llvm::

llvm-svn: 16935
2004-10-12 16:52:09 +00:00
Chris Lattner 5499551206 Don't emit the method into the llvm namespace, let the #includer decide where it goes
llvm-svn: 16934
2004-10-12 16:21:18 +00:00
Chris Lattner cfa7268e20 Add std:: prefix for compilers without correct koenig lookup implemented.
Patch contributed by Paolo Invernizzi

llvm-svn: 16933
2004-10-12 16:10:39 +00:00
Chris Lattner 00648e1f86 Transform memmove -> memcpy when the source is obviously constant memory.
llvm-svn: 16932
2004-10-12 04:52:52 +00:00
Chris Lattner c14504c92c New testcase for memmove -> memcpy transform
llvm-svn: 16931
2004-10-12 04:51:55 +00:00
Chris Lattner db6ddca8d2 Right, calls have Operand#0 == the called function, don't forget this.
llvm-svn: 16930
2004-10-12 04:45:20 +00:00
Chris Lattner 1ccab843e2 Minor tweaks
llvm-svn: 16929
2004-10-12 04:32:37 +00:00
Chris Lattner 2b33e70e7f New header. Classes can be added as needed.
llvm-svn: 16928
2004-10-12 04:20:46 +00:00
Chris Lattner 89046ca244 Implement a new method
llvm-svn: 16927
2004-10-12 04:20:25 +00:00
Chris Lattner ec901cc6cd This nutty patch has been in my tree since before 1.3 went out, and it needs
to go in.  This patch allows us to compute the trip count of loops controlled
by values loaded from constant arrays.  The cannonnical example of this is
strlen when passed a constant argument:

for (int i = 0; "constantstring"[i]; ++i) ;
return i;

In this case, it will compute that the loop executes 14 times, which means
that the exit value of i is 14.  Because of this, the loop gets DCE'd and
we are happy.  This also applies to anything that does similar things, e.g.
loops like this:

  const float Array[] = { 0.1, 2.1, 3.2, 23.21 };
  for (int i = 0; Array[i] < 20; ++i)

and is actually fairly general.

The problem with this is that it almost never triggers.  The reason is that
we run indvars and the loop optimizer only at compile time, which is before
things like strlen and strcpy have been inlined into the program from libc.
Because of this, it almost never is used (it triggers twice in specint2k).

I'm committing it because it DOES work, may be useful in the future, and
doesn't slow us down at all.  If/when we start running the loop optimizer
at link-time (-O4?) this will be very nice indeed :)

llvm-svn: 16926
2004-10-12 01:49:27 +00:00
Chris Lattner 7cabf6f87a Fix a REALLY obscure bug in my previous checkin, which was splicing the END
marker from one ilist into the middle of another basic block!

llvm-svn: 16925
2004-10-12 01:02:29 +00:00
Chris Lattner 9776f7259b Handle a common case more carefully. In particular, instead of transforming
pointer recurrences into expressions from this:

  %P_addr.0.i.0 = phi sbyte* [ getelementptr ([8 x sbyte]* %.str_1, int 0, int 0), %entry ], [ %inc.0.i, %no_exit.i ]
  %inc.0.i = getelementptr sbyte* %P_addr.0.i.0, int 1            ; <sbyte*> [#uses=2]

into this:

  %inc.0.i = getelementptr sbyte* getelementptr ([8 x sbyte]* %.str_1, int 0, int 0), int %inc.0.i.rec

Actually create something nice, like this:

  %inc.0.i = getelementptr [8 x sbyte]* %.str_1, int 0, int %inc.0.i.rec

llvm-svn: 16924
2004-10-11 23:06:50 +00:00
Chris Lattner 13128ab8fb Allow creation of GEP constantexprs with a vector of value* operands as
well as a vector of constant*'s.  It turns out that this is more efficient
and all of the clients want to do that, so we should cater to them.

llvm-svn: 16923
2004-10-11 22:52:25 +00:00
Chris Lattner 02a71e7748 Implement remove/eraseFromParent methods
llvm-svn: 16922
2004-10-11 22:21:39 +00:00
Chris Lattner 273207f32f Add long overdue methods
llvm-svn: 16921
2004-10-11 22:21:13 +00:00
Chris Lattner be29306939 Make this build
llvm-svn: 16919
2004-10-11 20:53:28 +00:00
Chris Lattner a92af96c56 Reenable the transform, turning X/-10 < 1 into X > -10
llvm-svn: 16918
2004-10-11 19:40:04 +00:00
Chris Lattner 58043a1473 Fix a warning that is emitted on the suns
llvm-svn: 16917
2004-10-11 15:50:40 +00:00
Chris Lattner 004e250cd2 This patch implements two things (sorry).
First, it allows SRA of globals that have embedded arrays, implementing
GlobalOpt/globalsra-partial.llx.  This comes up infrequently, but does allow,
for example, deleting several stores to dead parts of globals in dhrystone.

Second, this implements GlobalOpt/malloc-promote-*.llx, which is the
following nifty transformation:

Basically if a global pointer is initialized with malloc, and we can tell
that the program won't notice, we transform this:

struct foo *FooPtr;
...
  FooPtr = malloc(sizeof(struct foo));
...
  FooPtr->A   FooPtr->B

Into:

struct foo FooPtrBody;
...
   FooPtrBody.A  FooPtrBody.B

This comes up occasionally, for example, the 'disp' global in 183.equake (where
the xform speeds the CBE version of the program up from 56.16s to 52.40s (7%)
on apoc), and the 'desired_accept', 'fixLRBT', 'macroArray', & 'key_queue'
globals in 300.twolf (speeding it up from 22.29s to 21.55s (3.4%)).

The nice thing about this xform is that it exposes the resulting global to
global variable optimization and makes alias analysis easier in addition to
eliminating a few loads.

llvm-svn: 16916
2004-10-11 05:54:41 +00:00
Chris Lattner 615bd303d1 This testcase ensures that we can SRA a global even if part of the global
cannot be SRA'd

llvm-svn: 16915
2004-10-11 05:00:30 +00:00
Chris Lattner 214956dd9d These testcases check to see if we can promote malloc's to globals in some
cases.

llvm-svn: 16914
2004-10-11 05:00:12 +00:00
Chris Lattner 155cfd237b Don't miss global optimizations because we run before the inliner
llvm-svn: 16913
2004-10-11 04:47:18 +00:00
Chris Lattner 6faf3949f6 Fix SingleSource/Benchmarks/McGill/chomp
llvm-svn: 16912
2004-10-11 04:07:27 +00:00
Chris Lattner b2b7f90caf If we are trying to create a ConstantExpr cast that is really a GEP to the
first element of an array, return a GEP instead of a cast.  This allows us
to transparently fold this:

int* getelementptr (int* cast ([100 x int]* %Gbody to int*), int 40)

into this:

int* getelementptr ([100 x int]* %Gbody, int 0, int 40)

llvm-svn: 16911
2004-10-11 03:57:30 +00:00
Misha Brukman 8c8cbb8b71 Fix grammar
llvm-svn: 16910
2004-10-11 02:06:40 +00:00
Misha Brukman 8da107a68f Remove exhuberant-style comments which remained from copying file from `sample'
llvm-svn: 16909
2004-10-11 00:56:41 +00:00
Misha Brukman 6e200dedd7 Fix file header comment; remove sample project (extra-verbose) comments
llvm-svn: 16908
2004-10-11 00:55:47 +00:00
Misha Brukman e67286641e Fix file header comment
llvm-svn: 16907
2004-10-11 00:54:59 +00:00
Misha Brukman 1df01504c6 ModuloScheduling has moved to lib/Target/SparcV9
llvm-svn: 16906
2004-10-10 23:37:40 +00:00
Misha Brukman b4cb9fc7c9 Add ModuloScheduling to the recursive build tree
llvm-svn: 16905
2004-10-10 23:36:09 +00:00
Misha Brukman 9da1134bfb Adjust header file inclusion due to move
llvm-svn: 16904
2004-10-10 23:34:50 +00:00