Machine level IfConversion for ARM

Hi All,

How can I run the IfConversion pass in JIT to optimize my code for ARM as a
Target Architecture?

I'm not sure what you mean. The 32-bit ARM target already runs the ifconverter by default.

Evan

I need to do some analysis, in which I have to selectively convert IFs. I am
trying to find a way to call this optimization pass from the JIT code, thus
I can decide which branches to convert and which to keep. Apparently I
cannot do this using a pass manager.

I was wondering if there are plans to support the stackmap/patchmap
intrinsics for Linux/ELF? I saw that there were corresponding patches
posted earlier but as far as I can tell they haven't made it in yet.

Robin

I don’t believe the changes have made it in yet. In addition to the previous patch set, we’ve got an independent implementation we’ve been using for a while:
We really should get one or the other merged in tree. Could you point me to the previous patch set so I can ping it and review? Philip

I've been trying this patch, which seems to work fine from a quick
look:

    http://comments.gmane.org/gmane.comp.compilers.llvm.cvs/171038

I also found this one, which is similar:

    https://github.com/dropbox/pyston/blob/master/llvm_patches/0001-Support-emitting-stackmap-sections-for-ELF-object-fi.patch

And I believe I saw a 3rd version somewhere, which I don't find
anymore right now though.

These are all just small extensions for the existing intrinsics.

Thanks,

Robin

FYI, this just landed as 214538. If you could test and report back, it would be appreciated.

Philip

I've cherry-picked it into release_35 and it seems to be working fine
(I can access and parse the section, haven't further used the content
yet).

Thanks!

Robin