[X86] haddsub-2.ll - remove duplicate scalar tests (#207183)
These vectorisation tests exist in PhaseOrdering/X86/hadd.ll and
PhaseOrdering/X86/hsub.ll and the result shuffle tests exist in
CodeGen/X86/haddsub.ll and CodeGen/X86/phaddsub.ll
Help towards #143000
[X86] phaddsub.ll - add SSE2/SSE3 test coverage (#207179)
Ensure integer HADD/SUB instructions don't get generated without SSSE3 and that the codegen isn't entirely terrible
[SCCP] Avoid markConstant() in more places (#206494)
This extends the fix from 991455e69e93c0ce88e927eddd28a9ab34d1f8b2 to
two more places. In both of these, we can run into the same situation
where we first mark as constant with a vector that has poison values,
and then again without poison elements, resulting in an assertion
failure.
(An alternative would be to change markConstant() to allow calls with
different constant if the new constant is a vector that only refines
some previously poison elements?)
[AArch64][InstCombine] Combine add with [su]adalp intrinsics (#204796)
It can happen that the [su]adalp uses a zero accumulator and its result
is used as an operand into an ADD instruction. We can fold those.
add(adalp(%pred, zeroinitializer, %in), %acc)
-->
adalp(%pred, %acc, %in)
[Dexter] Switch to using script-mode by default
This patch changes the default mode of Dexter from heuristic-mode to
script-mode. The --use-script argument is replaced with --use-heuristic,
some comments/docs/error messages are updated accordingly, and tests have
their flags switched accordingly.
[Dexter] Update lldb-based dexter-tests to use script-mode
This patch replaces uses of heuristic-mode Dexter in the dexter-tests suite
with uses of the script-mode, for tests that use DAP (via lldb-dap). The
updates are largely straightforward but occasionally non-trivial, and in
some cases some slight modifications have been made to keep the "spirit" of
the test intact.