Cherry pick fixes from 5.0 to swift-5.1-branch #22821
Conversation
For `Initializer` decl context, type context analyzer should look into its parent context because that might constrain initializer's type. rdar://problem/48208253
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
Please remove my two patches from this PR -- I'm merging them in #22819 |
bob-wilson
force-pushed
the
cherry-picks-for-5.1
branch
from
ed3aeab to
edc2076
Compare
|
OK, let's try again without Slava's changes. |
|
@swift-ci please test |
|
Build failed |
|
Build failed |
bob-wilson
force-pushed
the
cherry-picks-for-5.1
branch
from
edc2076 to
7964a9d
Compare
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
I don't understand these build failures. Trying yet again.... |
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
I reproduced the failure locally. The IRGen/class_resilience_objc.swift test is failing this CHECK line: // CHECK: @"$s21class_resilience_objc27ClassWithEmptyThenResilientC9resilient0I7_struct0H3IntVvpWvd" = hidden global [[INT]] 0, because the output has: hidden global i64 8, align 8 @rjmccall Do you have any idea what is wrong here? There were some differences between 5.0 and master and I tried to get the right adjustments for 5.1 (which is in between those). |
|
@bob-wilson I already pulled that commit to 5.1 in #22818. |
When two access scopes have different access types, it is not safe to merge them into one access. Doing so will introduce false conflicts which manifest as crashes in user code. To do this optimization, we would need additional data flow information about *potential* read conflicts before converting a read to a modify access. Fixes rdar://48239213: Fatal access conflict detected.
bob-wilson
force-pushed
the
cherry-picks-for-5.1
branch
from
7964a9d to
cafdf05
Compare
|
OK, dropping John's change.... Let's try again. |
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
@swift-ci please test linux platform |
|
Build failed |
|
@swift-ci please test linux platform |
The PR includes 2 fixes that were merged to swift-5.0-branch without being pulled in for the 5.1 branch:
rdar://problem/48208253: Rintaro's code completion fix
rdar://problem/48239213: Andy's change to disable merging of read/modify exclusivity access markers