GC Fixes for Upstream Update #143
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
> /runtime/src/coreclr/vm/cgensys.h:107:26: error: stdcall calling convention is not supported on builtin function [-Werror,-Wignored-attributes]
…e are being set (dotnet#83516) * Cleanup some HWIntrinsic logic to ensure the right gtType and simdSize are being set * Apply formatting patch * Add a missing GetLower call * Fix an assert for lowering TYP_SIMD12 where its handled as TYP_SIMD16 * Ensure GetLower is used in Dot for TYP_SIMD32 * Apply formatting patch * Insert after, not before, for the _GetLower to avoid a codegen regression * Put the _GetLower in the right place to avoid the codegen regression * Don't change the simd size of TYP_SIMD8/12 DotProduct unnecessarily
Replaced _divided-by_ with _multiplied-by_ in both IMultiplyOperators.cs and NFloat.cs. Fixes dotnet#80521
… Build ID 2140030 (dotnet#83703)
* xxAny, xAll comparisons in progress. * xxAny, xxAll comparisons, part 2. * [mono][jit] Adding compare all/any operations. Fixed umov,smov macros. * Removed superfluous changes. * Restored newline at the end of HelloWorld. * Fixed unmatched brace. * Indentation. * Normalized boolean values to 0/1. SIMD_EXTR_ constants have friendlier names. Equality/Inequality are now also intrinsics. * Fixed element type for comparisons. * Temporarily disabled intrinsics. Will be permanenty reenabled once all are implemented.
…#83675) * SN_Sum operation on arm64. Fixed dup. Replaced addv, addp, faddp with their generalized variants. Added OP_EXTRACTx opcodes to arm64 codegen. Added horizontal sums. * Fixed smov macro. Added SN_ToScalar. Fixed code style. * Fixed vector sums of nint/nuint. * Temporarily disable intrinsics, until all are implemented.
* Fix. * Regex approach is not necessary. * Fix encoding on Windows. * Fixed build error. * Reverted unnecessary changes. Blocked relink with unicode. * Revet + nit. * Applied @kg's review.
…me (dotnet#83711) * Load AOT module of a container assembly using assembly name * Use mono_image_init to init the image * Implement mono_loader_lock on load_container_amodule * Avoid recursive invocation by setting container_assm_name to NULL
Merge commit 'edb161ab06ddc69d27aee2c0e990a60221ebbe92' into unity-upstream-tracking/staging
Casing on the platform name changed
This is a temporary patch to allow our NULL GC to work with upstream changes. Check the upper bounds in JIT_WriteBarrier_PreGrow64 and JIT_WriteBarrier_WriteWatch_PreGrow64. We must remove this when we return to default coreclr gc.
joncham
reviewed
Mar 23, 2023
| cmp rsi, r8 | ||
|
|
||
| #ifdef FEATURE_MANUALLY_MANAGED_CARD_BUNDLES | ||
| .byte 0x73, 0x4b |
joncham
approved these changes
Mar 23, 2023
bholmes
commented
Mar 23, 2023
| cmp rsi, r10 | ||
|
|
||
| #ifdef FEATURE_MANUALLY_MANAGED_CARD_BUNDLES | ||
| .byte 0x73, 0x3b |
There was a problem hiding this comment.
The offset of 0x3b is probably wrong as there is an extra shr instruction on line 468.
Not sure what I am going to do about it as it may not be used.
There was a problem hiding this comment.
I think I have a better way to do this...
TLDR;
We need to force the bReqUpperBoundsCheck bool to be true here...
https://github.com/dotnet/runtime/blob/main/src/coreclr/vm/amd64/jitinterfaceamd64.cpp#L833
Will swap this out and try again.
|
What I think is going on with the tests...
|
This reverts commit 5522dd8.
With the null gc, we always want to have the upper bounds check This ensures that the tiny bounds we setup will always miss. Remove this when we remove the null GC.
bholmes
force-pushed
the
unity-upstream-tracking/staging
branch
from
759a329 to
a276f28
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Need to implement a simple version of Frozen Segments
Adding "FEATURE_UNITY_NULLGC_WRITEBARRIER_PATCH"
to workaround write barrier problems with our GC. We need to
remove FEATURE_UNITY_NULLGC_WRITEBARRIER_PATCH
completely when we move back to the default GC.