This method is linear time in the size of the basic block, which is very
bad for large basic blocks. On the Assembler/2004-09-29-VerifierIsReallySlow.llx
testcase, the verifier changes from taking 50s to 0.23s with this patch.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16593 91177308-0d34-0410-b5e6-96231b3b80d8
from ModulePass. Instead of implementing Pass::run, then should implement
ModulePass::runOnModule.
Also, fix some undefined behavior, expecting | on booleans to evaluate
left-to-right.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16435 91177308-0d34-0410-b5e6-96231b3b80d8
error. We'll strategize on this when we have multiple front ends to deal
with. For now llvm-ld just runs a standard set of transforms.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16333 91177308-0d34-0410-b5e6-96231b3b80d8
libraries list is merged into the "destination" module's list. Also, if the
source module is one of the dependent libraries, it is removed from the
list.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16282 91177308-0d34-0410-b5e6-96231b3b80d8
Move include/Config and include/Support into include/llvm/Config,
include/llvm/ADT and include/llvm/Support. From here on out, all LLVM
public header files must be under include/llvm/.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16137 91177308-0d34-0410-b5e6-96231b3b80d8
error message, print out:
llvm-link: error linking in 'g.2.rbc': Global Variable Collision on
' %struct.D_Reduction* %d_reduction_10_gram' - External linkage globals have
different initializers
That's a bit more concise, huh?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@15958 91177308-0d34-0410-b5e6-96231b3b80d8
Ouch! Changes in the lazy initialization code caused each incorporated
function to reprocess the entire function on every lookup of a value's
slot number. This caused a horrible slowdown in all functions. This
fix made llvm-dis go from "longer than I care to wait" (minutes) on a large
test case to 0.53 seconds.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@15818 91177308-0d34-0410-b5e6-96231b3b80d8
gccld -disable-opt on 252.eon.
This patch deletes it and replaces it with a map. The proper fix for this
is to fix PR411, but this will do in the short term.
gccld on eon now takes 5.51s, which is 50% faster than before this patch. :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@15480 91177308-0d34-0410-b5e6-96231b3b80d8
Do not call FindGlobalNamed when we know we will ignore the result (because
we are not going to link a static symbol anyway). This speeds up
gccld -disable-opt on 252.eon from 8.63s to 8.39s.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@15465 91177308-0d34-0410-b5e6-96231b3b80d8
lookups it does. This shaves another 5% off of bcreading 252.eon. Note that
the proper solution to this problem is to fix PR411, but that will have to
wait until later.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@15455 91177308-0d34-0410-b5e6-96231b3b80d8
(e.g., LICM) into FunctionPassManagers. The problem is that we were
using a C-style cast to cast required analysis passes to PassClass*, but
if it's a FunctionPassManager, and the required analysis pass is an
ImmutablePass, the types aren't really compatible, so the C-style cast
causes a crash.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@15140 91177308-0d34-0410-b5e6-96231b3b80d8
