[llvm-dev] RFC [ThinLTO]: Promoting more aggressively in order to reduce incremental link time and allow sharing between linkage units

Thu Apr 7 11:59:22 PDT 2016

On Thu, Apr 7, 2016 at 11:26 AM, Mehdi Amini <mehdi.amini at apple.com> wrote:

>
> On Apr 7, 2016, at 10:58 AM, Xinliang David Li <davidxl at google.com> wrote:
>
>
>
> On Wed, Apr 6, 2016 at 9:53 PM, Mehdi Amini <mehdi.amini at apple.com> wrote:
>
>>
>> On Apr 6, 2016, at 9:40 PM, Teresa Johnson <tejohnson at google.com> wrote:
>>
>>
>>
>> On Wed, Apr 6, 2016 at 5:13 PM, Peter Collingbourne <peter at pcc.me.uk>
>> wrote:
>>
>>>
>>>
>>> On Wed, Apr 6, 2016 at 4:53 PM, Mehdi Amini <mehdi.amini at apple.com>
>>> wrote:
>>>
>>>>
>>>> On Apr 6, 2016, at 4:41 PM, Peter Collingbourne <peter at pcc.me.uk>
>>>> wrote:
>>>>
>>>> Hi all,
>>>>
>>>> I'd like to propose changes to how we do promotion of global values in
>>>> ThinLTO. The goal here is to make it possible to pre-compile parts of the
>>>> translation unit to native code at compile time. For example, if we know
>>>> that:
>>>>
>>>> 1) A function is a leaf function, so it will never import any other
>>>> functions, and
>>>>
>>>>
>>>> It still may be imported somewhere else right?
>>>>
>>>> 2) The function's instruction count falls above a threshold specified
>>>> at compile time, so it will never be imported.
>>>>
>>>>
>>>> It won’t be imported, but unless it is a “leaf” it may import and
>>>> inline itself.
>>>>
>>>
>>>> or
>>>> 3) The compile-time threshold is zero, so there is no possibility of
>>>> functions being imported (What's the utility of this? Consider a program
>>>> transformation that requires whole-program information, such as CFI. During
>>>> development, the import threshold may be set to zero in order to minimize
>>>> the incremental link time while still providing the same CFI enforcement
>>>> that would be used in production builds of the application.)
>>>>
>>>> then the function's body will not be affected by link-time decisions,
>>>> and we might as well produce its object code at compile time.
>>>>
>>>>
>>>> Reading this last sentence, it seems exactly the “non-LTO” case?
>>>>
>>>
>>> Yes, basically the point of this proposal is to be able to split the
>>> linkage unit into LTO and non-LTO parts.
>>>
>>>
>>>> This will also allow the object code to be shared between linkage units
>>>> (this should hopefully help solve a major scalability problem for Chromium,
>>>> as that project contains a large number of test binaries based on common
>>>> libraries).
>>>>
>>>> This can be done with a change to the intermediate object file format.
>>>> We can represent object files as native code containing statically compiled
>>>> functions and global data in the .text,. data, .rodata (etc.) sections,
>>>> with an .llvmbc section (or, I suppose, "__LLVM, __bitcode" when targeting
>>>> Mach-O) containing bitcode for functions to be compiled at link time.
>>>>
>>>> In order to make this work, we need to make sure that references from
>>>> link-time compiled functions to statically compiled functions work
>>>> correctly in the case where the statically compiled function has internal
>>>> linkage. We can do this by promoting every global value with internal
>>>> linkage, using a hash of the external names (as I mentioned in [1]).
>>>>
>>>>
>> Mehdi - I know you were keen to reduce the amount of promotion. Is that
>> still an issue for you assuming linker GC (dead stripping)?
>>
>>
>> Yes: we do better optimization on internal function in general.
>>
>
> Inliner is one of the affected optimization -- however this sounds like a
> matter of tuning to teach inliner about promoted static functions.
>
>
> The inliner compute a tradeoff between pseudo runtime cost and binary
> size, the existing bonus for static functions is when there is a single
> call site because it makes the binary increase inexistant (dropping the
> static after inline). We promote function because we think we are likely to
> introduce a reference to it somewhere else, so “lying” to the inliner is
> not necessarily a good idea.
>

It is not lying to the inliner. If a static (before promotion) function is
a candidate to be inlined in the original defining module, it is probably
more likely to inlined in other importing modules where more context is
available. In other words, the inliner can apply the same bonus to
'promoted' static functions as if references in other modules will also
disappear.  Of course, we can not assume it has single callsite.

Comdat functions can be handled similarly.

> That said we (actually Bruno did) prototyped it already with somehow good
> results :)
> I’m not convinced yet that it should be independent of promoted or not
> promoted though.
>

Generally true (see the comdat case).

>
> Assuming we solve the inliner issue, then remain the “optimizations other
> than inliner”. We can probably solve most but I suspect it won’t be
> “trivial” either.
>

Any such optimizations in mind?

David

>
> —
> Mehdi
>
>
>
> David
>
>
>
>> Our benchmarks showed that it can really make some difference, and many
>> cases were ThinLTO didn’t perform as well as FullLTO were because of this
>> promotion.
>> (binary size has never been my concern here)
>>
>
>
>>
>> With this proposal we will need to stick with the current promote
>> everything scheme.
>>
>>
>> I don’t think so: you would need do it only for “internal functions that
>> a leaf and aren’t likely to be imported/inlined”.
>> That said any function that we emit the binary at compile time instead of
>> link time will contribute to inhibit optimizations for LTO/ThinLTO. The
>> gain in compile time has to be really important to make it worth it.
>> (Of course the CFI use-case is a totally different tradeoff).
>>
>> Peter: have you thought about debug info by the way?
>>
>> —
>> Mehdi
>>
>>
>>
>>
>>
>>>
>>>> I imagine that for some linkers, it may not be possible to deal with
>>>> this scheme. For example, I did some investigation last year and discovered
>>>> that I could not use the gold plugin interface to load a native object file
>>>> if we had already claimed it as an IR file. I wouldn't be surprised to
>>>> learn that ld64 has similar problems.
>>>>
>>>>
>>>> I suspect ld64 would need to be update to handle this scheme. Somehow
>>>> it need to be able to extract the object from the section.
>>>>
>>>
>>> Do you mean the bitcode object? There's already support in LLVM for this
>>> (http://llvm-cs.pcc.me.uk/lib/Object/IRObjectFile.cpp#269). I suspect
>>> the tricky part (which I was unsuccessful at doing with gold) will be
>>> convincing the linker that the bitcode object and the regular object are
>>> two separate things.
>>>
>>> Otherwise this should work, but I suspect the applicability (leaving CFI
>>>> aside) may not concern that many functions, so I’m not sure about the
>>>> impact.
>>>>
>>>
>>> I'm also curious about the applicability in the regular ThinLTO case.
>>> One thing that may help with applicability is that I suspect that not only
>>> leaf functions but also functions that only call (directly or indirectly)
>>> functions in the same TU, as well as functions that only make calls via
>>> function pointers or vtables may fall under the criteria for non-importing.
>>>
>>
>> With indirect call profiling the function pointer case should not be a
>> blocker for importing.
>>
>> Teresa
>>
>>
>>>
>>> Peter
>>>
>>
>>
>>
>> --
>> Teresa Johnson |  Software Engineer |  tejohnson at google.com |
>> 408-460-2413
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20160407/db29841e/attachment-0001.html>