Commit Graph

980 Commits

Author SHA1 Message Date
Jim Laskey 5c3bdbad9b No longer used, and confuses maintainers
llvm-svn: 30248
2006-09-11 16:02:58 +00:00
Evan Cheng 1ec61ba00a 1) With X86 lowering change, the following can no longer happen since
the branch's chain is also produced by cmp.
          [ch, r : ld]
             ^ ^
             | |
      [XX]--/   \- [flag : cmp]
       ^             ^
       |             |
       \---[br flag]-

Remove an isel check which prevents loads from being folded into cmp / test
instructions.

2) Whenever possible, delete a selected node to allow more load folding
opportunities. Note not all nodes can be deleted after it has been
selected. Some may have simply morphed; some have not changed at all (e.g.
EntryToken).

llvm-svn: 30242
2006-09-11 02:24:43 +00:00
Evan Cheng 5a798b0648 Generated isel should favors explicit constant operand (+2) over an operand with a predicate (+1).
llvm-svn: 30180
2006-09-08 07:26:39 +00:00
Chris Lattner a2d7d1c4fa Fix JIT encoding of two-addr instructions.
llvm-svn: 30111
2006-09-05 03:01:52 +00:00
Chris Lattner 13a5dcddce Fix a long-standing wart in the code generator: two-address instruction lowering
actually *removes* one of the operands, instead of just assigning both operands
the same register.  This make reasoning about instructions unnecessarily complex,
because you need to know if you are before or after register allocation to match
up operand #'s with the target description file.

Changing this also gets rid of a bunch of hacky code in various places.

This patch also includes changes to fold loads into cmp/test instructions in
the X86 backend, along with a significant simplification to the X86 spill
folding code.

llvm-svn: 30108
2006-09-05 02:12:02 +00:00
Chris Lattner 70fd01a1b8 Make sure to clear CurDefmPrefix = 0, otherwise stuff after a defm won't
parse right.  This fixes TableGen/MultiClass.td

llvm-svn: 30037
2006-09-01 22:07:27 +00:00
Chris Lattner c05fac8b83 fix an assertion with multidefs. Def inside of multiclasses don't need to
be complete.

llvm-svn: 30034
2006-09-01 21:59:03 +00:00
Chris Lattner aec1af2239 regenerate
llvm-svn: 30029
2006-09-01 21:14:42 +00:00
Chris Lattner 66b7235877 Add often-requested support for defining "multiclasses" which can be instantiated.
This allows you to define multiple definitions with one line, encouraging
more .td file factoring.

llvm-svn: 30027
2006-09-01 21:13:49 +00:00
Evan Cheng 9b675bd769 Issue cvs co with -P option to prune empty directories. Remove the unnecessary
update stage.

llvm-svn: 29955
2006-08-29 18:01:18 +00:00
Reid Spencer 97c598083e Add virtual methods to all subclasses so they are not overloaded.
llvm-svn: 29930
2006-08-28 00:12:25 +00:00
Chris Lattner 251468e9d0 Use compiler.h instead of hand rolling our own macro
llvm-svn: 29913
2006-08-27 13:16:24 +00:00
Evan Cheng 44ce57d5c2 Do not emit getTargetNode() and SelectNodeTo() which takes more than 3
SDOperand arguments. Use the variants which take an array and number instead.

llvm-svn: 29906
2006-08-27 08:11:28 +00:00
Evan Cheng 34b70eea5c SelectNodeTo now returns a SDNode*.
llvm-svn: 29901
2006-08-26 08:00:10 +00:00
Evan Cheng b7b5ba5aaf Minor getCopyToReg() call change.
llvm-svn: 29900
2006-08-26 07:39:28 +00:00
Evan Cheng 61413a3d72 Select() no longer require Result operand by reference.
llvm-svn: 29898
2006-08-26 05:34:46 +00:00
Evan Cheng a8b2b6c7ff Remove dead code.
llvm-svn: 29897
2006-08-26 01:34:15 +00:00
Evan Cheng eaf9fba2ff A bit more clean up.
llvm-svn: 29893
2006-08-26 01:02:19 +00:00
Evan Cheng 548bd0dc95 - Clean up tablegen dag isel generator code.
- Clean up the code generated by tablegen:
  * AddToISelQueue now takes one argument.
  * ComplexPattern matching condition can now be shared.
  * Eliminate passing unnecessary arguments to emit routines.
  * Eliminate some unneeded SDOperand declarations in select routines.
  * Other minor clean ups.
- This reduces foot print slightly: X86ISelDAGToDAG.o is reduced from 971k
  to 823k.

llvm-svn: 29892
2006-08-26 00:59:04 +00:00
Patrick Jenkins 06725191ab Corrected an error that was introduced yesterday that caused the GCC
version to be left blank if the build failed.

Also corrected a problem where if the build failed the nightly tester
would still try to read the results of the Dejagnutests log even
though it didnt exist.

llvm-svn: 29829
2006-08-22 18:11:19 +00:00
Patrick Jenkins 235abe4ce8 The new nightly tester should stop outputing a message stating the
testresults directory is not present unless the script is run as verbose.

llvm-svn: 29797
2006-08-21 20:45:57 +00:00
Patrick Jenkins deb77ee8eb These changes reflect the changes in the database for how tests are stored
and bring the handing of dejagnu tests into compliance with this new scheme.

llvm-svn: 29772
2006-08-18 18:00:21 +00:00
Patrick Jenkins d515016982 Changing the format dejagnu tests are submitted. We used to only submit multisource information for pass fail. Now we submit all pass/fail test information. This should fix the issue where some results were not being reported.
llvm-svn: 29757
2006-08-17 22:11:03 +00:00
Patrick Jenkins 0cead44fb1 My last commit added some experimental features I have been working on. This commit rolls those back.
llvm-svn: 29738
2006-08-16 22:32:20 +00:00
Patrick Jenkins 2bc1934fb1 Changed the call to countloc.sh so that the correct topdirectory of the llvm build is set.
llvm-svn: 29736
2006-08-16 22:18:41 +00:00
Evan Cheng 63d178f473 SelectNodeTo() may return a SDOperand that is different from the input.
llvm-svn: 29726
2006-08-16 07:30:09 +00:00
Chris Lattner d16696d115 The generated runloop shouldn't require Select to update 'Result'. In fact
we'd like to remove result entirely.

llvm-svn: 29713
2006-08-15 23:42:26 +00:00
Chris Lattner 72608b5387 allow Select to return the node being selected if RAU isn't needed.
llvm-svn: 29712
2006-08-15 23:27:50 +00:00
Reid Spencer 7c63770b0e Simplify this a little bit.
llvm-svn: 29695
2006-08-15 03:50:10 +00:00
Reid Spencer e2bcaaa250 This script takes arguments so remove the check for number of arguments,
just check the first arg to see if its -topdir.

llvm-svn: 29694
2006-08-15 03:48:22 +00:00
Reid Spencer 235c1555c3 Make all tools that use llvmdo have a -topdir option that allows the top
source dir for LLVM to be specified explicitly. This removes the dependency
on the llvm-config script. If the option is not given, then the scripts use
llvm-config which should be both built and in the PATH. This arrangement
provides a useful default for most developers but also allows the nightly
tester to execute countloc.sh before llvm-config is built and without
altering the PATH.

llvm-svn: 29663
2006-08-14 18:49:05 +00:00
Patrick Jenkins d531b509cc Added some comments to clarify code around sub TestDirectory
llvm-svn: 29658
2006-08-14 16:07:14 +00:00
Reid Spencer f82ce13179 1. fix bug by ensuring we start at the llvm source root.
2. make use of the -code-only option on llvmdo
3. Add some debug print statements to find #1

llvm-svn: 29655
2006-08-13 19:03:06 +00:00
Reid Spencer 0cb9ded022 Count the documentation. Otherwise we take a 30,000 hit in the LOC on the
nightly test graph. And, documentation is important.

llvm-svn: 29654
2006-08-13 18:59:40 +00:00
Reid Spencer 4e6115ea3a Use the new -code-only option to llvmdo so that we only count things that
are considered to be code. This will give a drop in the LOC count on the
nightly testers, but it is more accurate than previous estimates.

llvm-svn: 29653
2006-08-13 18:29:21 +00:00
Reid Spencer 1be8d4c3c3 Add a -code-only option which restricts llvmdo to visiting just those
files that are considered to be code. Documentation is excluded.

llvm-svn: 29652
2006-08-13 18:28:27 +00:00
Reid Spencer 558269c8d6 Changes for LLVM coding standard compliance:
1. Remove tabs
  2. Wrap to 80 columns
  3. Put spaces between "if" or "elsif" and expression
  4. Put { on same line as if statement.
  5. Make indentation be 2 spaces.
Functionality change:
  Run utils/countloc.sh after the build is done. This is necessary because
  that script now uses llvm-config to find the top src directory.

llvm-svn: 29651
2006-08-13 09:53:02 +00:00
Reid Spencer 52aeae0666 Lexer.cpp is a generated file both in lib/AsmParser and projects/Stacker
so we remove it from being counted.

llvm-svn: 29641
2006-08-11 23:58:58 +00:00
Reid Spencer 854b5c7c40 Weed out some cruft and add in some missing extensions.
llvm-svn: 29638
2006-08-11 23:51:05 +00:00
Reid Spencer 8ca43d72aa Weed out cruft that should not be counted and add in extensions we
previously missed.

llvm-svn: 29637
2006-08-11 23:50:27 +00:00
Patrick Jenkins e97e60734a Added two flags (-submit-server and -submit-script) which allow the user to specify where the test results are submitted too. If these are not specified they default to the LLVM.org server.
llvm-svn: 29633
2006-08-11 23:02:09 +00:00
Reid Spencer db3752ab6b Make this script not care where it is run from by getting the llvm src
root from the llvm-config script.

llvm-svn: 29632
2006-08-11 21:55:31 +00:00
Reid Spencer 719c383594 Make llvmdo and llvmgrep invulnerable to where they are run from by getting
the llvm source root from the llvm-config command. The dependency now is
that the correct llvm-config command is in the path. For most developers
this is the case.

llvm-svn: 29630
2006-08-11 21:53:27 +00:00
Reid Spencer 1f0b9a1158 Clean up the HTML output. Add a -filedetails option to report the number of
lines contributed by each user to each file.

llvm-svn: 29624
2006-08-11 20:44:17 +00:00
Reid Spencer c647e96ea0 Remove blank line at end of file, removing test user "foo" from CVS history.
llvm-svn: 29623
2006-08-11 19:42:11 +00:00
Reid Spencer 092ae90eab Drop the -details option, it was broken. If you want to know about a
particular directory, just pass that directory to the script. Also, clean
up the HTML output, put list of skipped dirs at the end, and make a few
other minor improvements.

llvm-svn: 29620
2006-08-11 18:36:55 +00:00
Evan Cheng 962752dcea Use SmallVector instead.
llvm-svn: 29619
2006-08-11 18:33:41 +00:00
Evan Cheng 0d3a754ae8 - Prevent some functions from being inlined to eliminate the code size bloat
introduced by previous commit.
- SelectCode now returns a SDNode*. If it is not null, the selected node
  produces the same number of results as the input node. The seletion loop
  is responsible for calling ReplaceAllUsesWith() to replace the input node
  with the output target node. For other cases, e.g. when load is folded,
  the selection code is responsible for calling ReplaceAllUsesOfValueWith()
  and SelectCode returns NULL.
- Other clean ups.

llvm-svn: 29602
2006-08-11 08:59:35 +00:00
Reid Spencer 6ea83217ad Fix a comment.
llvm-svn: 29590
2006-08-09 19:40:13 +00:00
Chris Lattner f9cab8c965 Revert previous patch
llvm-svn: 29585
2006-08-09 16:44:44 +00:00