[PATCH] Support to lazily load Metadata

[Rafael Espíndola]

This is awesome. Sorry for taking so long to review it.

+  // Delay parsing Metadata if WillMaterializeAll is false and
+  // ShouldLazyLoadMetadata is true.
+  if (std::error_code EC = R->ParseBitcodeInto(
+          M, !WillMaterializeAll && ShouldLazyLoadMetadata))

Why the check for WillMaterializeAll? The nice thing about omitting it
is that the lazy path will look a bit more like the non-lazy path.

It looks like a materializeMetadata will be needed for the gold
plugin, but I can add that as a second patch. LGTM!

Thanks,
Rafael



On 27 February 2015 at 14:11, Manman Ren <mren at apple.com> wrote:
> Hi All,
>
> This patch defers loading of Metadata until we materialize a module, a function or explicitly the Metadata.
>
> In this patch, we only enable lazy loading of Metadata when creating LTOModule in local contexts.
> These can only really be used for parsing symbols, so it's unnecessary to ever load the metadata blocks.
>
> Implementation —————
> We add a parameter called ShouldLazyLoadMetadata to a few functions.
>
> We only defer loading metadata inside ParseModule when ShouldLazyLoadMetadata
> is true and we have not loaded any Metadata block yet.
>
> This commit implements all-or-nothing loading of Metadata. If there is a
> request to load any metadata block, we will load all deferred metadata blocks.
>
> We make sure the deferred metadata blocks are loaded before we materialize any
> function or a module.
>
> The default value of the added parameter ShouldLazyLoadMetadata for
> getLazyBitcodeModule is false, so the default behavior stays the same.
>
> Next Step —————
> If we are going to enable lazy-loading of Metadata for other usages of
> getLazyBitcodeModule, where deferred metadata blocks need to be loaded, we can
> expose BitcodeReader::materializeMetadata to Module, similar to
> Module::materialize.
>
> Thanks for reviewing!
> Manman
>