[MISched] Compare right next cluster node

[wangpc-pp]

We support bottom-up and bidirectonal postra scheduling now, but we
only compare successive next cluster node as if we are doing topdown
scheduling. This makes load/store clustering and macro fusions wrong.

This patch makes sure that we can get the right cluster node by the
scheduling direction.

[wangpc-pp]

Ping.

This fixes wrong macro fusion result when enabling bottom-up/bi-directional post-ra scheduling.

[wangpc-pp]

Ping.

[michaelmaitland]

Is there any test case that would should this fixing a regression?

[wangpc-pp]

Is there any test case that would should this fixing a regression?

Yeah, like what I have written in the description, macro fusion and load/store clustering will be wrong if using bidirectional or bottom up postra scheduling. I found this issue when trying to enable postra scheduling.

Please see https://github.com/llvm/llvm-project/pull/115864/commits, without this fix, you will see the wrong result.

[michaelmaitland]

Is there any test case that would should this fixing a regression?

Yeah, like what I have written in the description, macro fusion and load/store clustering will be wrong if using bidirectional or bottom up postra scheduling. I found this issue when trying to enable postra scheduling.

Please see https://github.com/llvm/llvm-project/pull/115864/commits, without this fix, you will see the wrong result.

Can we include a test in this PR that shows how this fixes the behavior that is incorrect prior to this patch?

[wangpc-pp]

Is there any test case that would should this fixing a regression?

Yeah, like what I have written in the description, macro fusion and load/store clustering will be wrong if using bidirectional or bottom up postra scheduling. I found this issue when trying to enable postra scheduling.
Please see https://github.com/llvm/llvm-project/pull/115864/commits, without this fix, you will see the wrong result.

Can we include a test in this PR that shows how this fixes the behavior that is incorrect prior to this patch?

Sure, I can add some RUN lines to existing macro fusion tests.

[wangpc-pp]

Is there any test case that would should this fixing a regression?

Yeah, like what I have written in the description, macro fusion and load/store clustering will be wrong if using bidirectional or bottom up postra scheduling. I found this issue when trying to enable postra scheduling.
Please see https://github.com/llvm/llvm-project/pull/115864/commits, without this fix, you will see the wrong result.

Can we include a test in this PR that shows how this fixes the behavior that is incorrect prior to this patch?

Sure, I can add some RUN lines to existing macro fusion tests.

Done. You can see the second commit.

[michaelmaitland]

LGTM