llvm-svn: 73815

llvm-svn: 73813

llvm-svn: 73811

step is to make tblgen generate something more appropriate for MCInst,
and generate calls to operand translation routines where needed.
This includes a bunch of #if 0 code which will slowly be refactored into
something sensible.

llvm-svn: 73810

llvm-svn: 73809

llvm-svn: 73808

SCEVUnknowns with identical Instructions to be equal. This allows
it to analze cases such as the attached testcase, where the front-end
has cloned the loop controlling expression. Along with r73805, this
lets IndVarSimplify eliminate all the sign-extend casts in the
loop in the attached testcase.

llvm-svn: 73807

llvm-svn: 73806

in the loop backedge-taken count computation of the maximum
possible trip count.

llvm-svn: 73805

Add debugging code to test for various locking faux-pas's, when running in single threaded mode.  This should help improve testing coverage for
threading support, without having extensive actually concurrent clients yet.

llvm-svn: 73803

llvm-svn: 73802

Revert r73790, and replace it with a significantly less ugly solution.  Rather than trying to make the global reader-writer lock work,
create separate recursive mutexes for each value map.  The recursive-ness fixes the double-acquiring issue, which having one per ValueMap
lets us continue to maintain some concurrency.

llvm-svn: 73801

on X86.  Not useful yet.

llvm-svn: 73799

llvm-svn: 73798

llvm-svn: 73796

so that it can access the TargetData member (when available) and
use ValueTracking.h information to compute information for
SCEVUnknown Values.

Also add GetMinLeadingZeros and GetMinSignBits functions,
with minimal implementations.

llvm-svn: 73794

hand-crafted testcase which demonstrates the bug that was exposed
in 254.gap.

llvm-svn: 73793

llvm-svn: 73792

llvm-svn: 73791

gets involved, and we end up trying to recursively acquire a writer lock.  The fix for this is slightly horrible,
and involves passing a boolean "locked" parameter around in Constants.cpp, but it's better than having locked and
unlocked versions of most of the code.

llvm-svn: 73790

which it isn't prepared to handle.

llvm-svn: 73787

llvm-svn: 73786

llvm-svn: 73784

llvm-svn: 73783

llvm-svn: 73781

Patch by Иван Сорокин.

llvm-svn: 73779

llvm-svn: 73778

llvm-svn: 73777

expressions, when interesting.

llvm-svn: 73776

llvm-svn: 73775

expression in IVUsers, because in the case of a use of a non-linear
addrec outside of a loop, this causes the addrec to be evaluated as
a linear addrec.

llvm-svn: 73774

llvm-svn: 73773

Patch by Benjamin Kramer.

llvm-svn: 73772

Nicolas, this is probably the cause of the failures you were seeing.

llvm-svn: 73770

as if they were multiple uses of the same instruction.  This interacts
well with the existing loadpre that j-t does to open up many new jump
threads earlier.

llvm-svn: 73768

llvm-svn: 73767

llvm-svn: 73766

place it with -pedantic.  Remove -Wunused since it
is implied by -Wall.  Group -Wno-unused-parameter
with -Wall -W since it is the combination of these
two that turns on -Wunused-parameter.

llvm-svn: 73765

llvm-svn: 73761

handle with an SSE2 instruction.

llvm-svn: 73760