Fix bounds check in bridged ASCII String comparison

[Catfish-Man]

Fixes rdar://problem/50407214

'count' is the UTF8 count of self, and the only String that we've proven is ASCII at this point is 'other', so 'count' can be greater than 'length'.

[Catfish-Man]

@swift-ci please test

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - de23660888e21c6ba0f46c7eb4aa13a6aed6b2cf

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - de23660888e21c6ba0f46c7eb4aa13a6aed6b2cf

[airspeedswift]

@swift-ci please benchmark

[swift-ci]

Performance: -O

TEST	OLD	NEW	DELTA	RATIO
Regression
Calculator	135	153	+13.3%	0.88x (?)
OpenClose	61	67	+9.8%	0.91x (?)
Chars2	3100	3400	+9.7%	0.91x (?)

Performance: -Osize

TEST	OLD	NEW	DELTA	RATIO
Regression
Calculator	140	160	+14.3%	0.88x
PrefixCountableRangeLazy	15	17	+13.3%	0.88x (?)
OpenClose	65	71	+9.2%	0.92x (?)
Chars2	3150	3400	+7.9%	0.93x (?)

Performance: -Onone

TEST	OLD	NEW	DELTA	RATIO
Regression
DataAppendDataLargeToLarge	17800	24200	+36.0%	0.74x (?)

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 mini
  Model Identifier: Macmini8,1
  Processor Name: Intel Core i7
  Processor Speed: 3.2 GHz
  Number of Processors: 1
  Total Number of Cores: 6
  L2 Cache (per Core): 256 KB
  L3 Cache: 12 MB
  Memory: 64 GB

[Catfish-Man]

Test failures are due to forcing ASAN on for this test. Asking some folks what to do about that

[aroben]

Do you think those performance regressions are real? I wonder why this change would hurt performance; aren't we passing a smaller byte count to memcmp now?

[Catfish-Man]

The perf regressions aren't real, none of those tests use the code in question

[Catfish-Man]

@swift-ci please test

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - de23660888e21c6ba0f46c7eb4aa13a6aed6b2cf

[aroben]

Yikes, they are very large-looking unreal regressions.

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - de23660888e21c6ba0f46c7eb4aa13a6aed6b2cf

[Catfish-Man]

@milseman pointed out an improved approach that makes this clearer and is likely a perf win at the same time :)

[Catfish-Man]

@swift-ci please benchmark

[Catfish-Man]

@swift-ci please test

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - e67267549ce4836aa63fb1ce37c35487062b493d

[Catfish-Man]

@swift-ci please benchmark

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - e67267549ce4836aa63fb1ce37c35487062b493d

[swift-ci]

Performance: -O

TEST	OLD	NEW	DELTA	RATIO
Regression
ObjectiveCBridgeStringIsEqualAllSwift	29	32	+10.3%	0.91x (?)

Performance: -Osize

TEST	OLD	NEW	DELTA	RATIO
Regression
ObjectiveCBridgeStringIsEqualAllSwift	29	32	+10.3%	0.91x (?)

Performance: -Onone

TEST	OLD	NEW	DELTA	RATIO
Improvement
ObjectiveCBridgeStubNSDataAppend	1960	1800	-8.2%	1.09x (?)

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 mini
  Model Identifier: Macmini8,1
  Processor Name: Intel Core i7
  Processor Speed: 3.2 GHz
  Number of Processors: 1
  Total Number of Cores: 6
  L2 Cache (per Core): 256 KB
  L3 Cache: 12 MB
  Memory: 64 GB

[Catfish-Man]

That's very suspicious. The case that should have improved is maybe not covered by the benchmarks, and the case that claims to have maybe regressed is a) a very very low workload test, and b) takes a branch just before all the changed code.

I'm still voting no actual change there.

[Catfish-Man]

(the thing that will improve is when the Swift string is non-ascii and the NSString is ascii, we won't do all the work to build breadcrumbs)

[milseman]

@Catfish-Man did this get cherry-picked to 5.1?