Refactor FixedWidthInteger init

[meg-gupta]

Without jump threading enabled in ossa, the FixedWidthInteger's init function which is an inline always function causes large cfgs due to the truncation case. This PR separates the truncation case into a different function with no always inlining.

[meg-gupta]

@swift-ci benchmark

[swift-ci]

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	4594	6879	+49.7%	0.67x (?)
Improvement	OLD	NEW	DELTA	RATIO
DictionaryKeysContainsNative	30	28	-6.7%	1.07x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
DictionaryKeysContainsNative	26	29	+11.5%	0.90x (?)
Improvement	OLD	NEW	DELTA	RATIO
UTF8Decode_InitFromData_ascii_as_ascii	801	745	-7.0%	1.08x (?)

Code size: -Osize

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
NSError	885	994	+12.3%	0.89x (?)
Improvement	OLD	NEW	DELTA	RATIO
DataCreateMedium	667900	357100	-46.5%	1.87x
Data.init.Sequence.64kB.Count0	4573	2461	-46.2%	1.86x
Data.init.Sequence.64kB.Count0.I	4543	2467	-45.7%	1.84x
Data.init.Sequence.2047B.Count0.I	7253	3995	-44.9%	1.82x
Data.init.Sequence.2049B.Count0.I	7244	4001	-44.8%	1.81x
Data.init.Sequence.809B.Count0.I	5845	3265	-44.1%	1.79x
Data.append.Sequence.64kB.Count0.I	4477	2518	-43.8%	1.78x
Data.append.Sequence.64kB.Count0	4479	2522	-43.7%	1.78x
Data.init.Sequence.513B.Count0.I	5659	3198	-43.5%	1.77x
Data.init.Sequence.511B.Count0.I	5637	3187	-43.5%	1.77x
Data.init.Sequence.809B.Count0	5834	3311	-43.2%	1.76x
Data.append.Sequence.809B.Count0.I	5686	3255	-42.8%	1.75x
Data.append.Sequence.809B.Count0	5668	3246	-42.7%	1.75x
Data.init.Sequence.64kB.Count	4481	2584	-42.3%	1.73x
Data.init.Sequence.64kB.Count.I	4475	2582	-42.3%	1.73x
Data.append.Sequence.64kB.Count	4479	2585	-42.3%	1.73x
Data.append.Sequence.64kB.Count.I	4476	2584	-42.3%	1.73x
Data.init.Sequence.2049B.Count.I	7035	4065	-42.2%	1.73x
Data.init.Sequence.2047B.Count.I	7018	4062	-42.1%	1.73x
Data.init.Sequence.809B.Count.I	5596	3255	-41.8%	1.72x
DataCreateSmall	78970	46020	-41.7%	1.72x
Data.init.Sequence.513B.Count.I	5360	3151	-41.2%	1.70x
Data.init.Sequence.511B.Count.I	5334	3139	-41.2%	1.70x
Data.init.Sequence.809B.Count	5592	3291	-41.1%	1.70x
Data.append.Sequence.809B.Count.I	5666	3336	-41.1%	1.70x
Data.append.Sequence.809B.Count	5662	3335	-41.1%	1.70x

Code size: -swiftlibs

How to read the data

The tables contain differences in performance which are larger than 8% and differences in code size which are larger than 1%.

If you see any unexpected regressions, you should consider fixing the
regressions before you merge the PR.

Noise: Sometimes the performance results (not code size!) contain false
alarms. Unexpected regressions which are marked with '(?)' are probably noise.
If you see regressions which you cannot explain you can try to run the
benchmarks again. If regressions still show up, please consult with the
performance team (@eeckstein).

Hardware Overview

  Model Name: Mac Pro
  Model Identifier: MacPro6,1
  Processor Name: 12-Core Intel Xeon E5
  Processor Speed: 2.7 GHz
  Number of Processors: 1
  Total Number of Cores: 12
  L2 Cache (per Core): 256 KB
  L3 Cache: 30 MB
  Memory: 64 GB

[meg-gupta]

@swift-ci test

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - 0e5d73d

[meg-gupta]

Clarify if @_alwaysEmitIntoClient is fine to use here

[lorentey]

It looks fine to me from an ABI compatibility viewpoint! 👍

[meg-gupta]

@swift-ci test Linux platform

[atrick]

LGTM

Someone from the stdlib, @lorentey ? could do a quick sanity check.

I don't think this is a temporary workaround at all. I think it fixes an underlying compile-time and code size problem that jump-threading was somewhat hiding from us. Running jump-threading doesn't change the fact that the compiler is generating a giant CFG to begin with.

[meg-gupta]

@lorentey can you please take a look ?

[meg-gupta]

@swift-ci test Linux platform