Sanjam Panda [Thu, 6 Jul 2023 17:21:28 +0000 (22:51 +0530)]
[mono] Add cmake check for HAVE_CLOCK_GETTIME (#88428)
Fixes an issue with missing method names when perf is used together with BenchmarkDotNet using Mono on Linux due to incorrect event order of the fork events and the mmap2 events.
Jakob Botsch Nielsen [Thu, 6 Jul 2023 17:10:15 +0000 (19:10 +0200)]
Disable System.Diagnostics.Tests.EventLogMessagesTests.CanReadAndWriteMessages on Windows 11 (#88472)
Fixes blocking status for #88224
Lakshan Fernando [Thu, 6 Jul 2023 15:11:59 +0000 (08:11 -0700)]
Enable reading process environment variables in EventPipe (#87771)
* Process Environment support
* using free to match malloc
* Add test to CI
* Fix FreeBSD build break
* Fix typo in OSX environ def
* FB
Kunal Pathak [Thu, 6 Jul 2023 14:35:26 +0000 (07:35 -0700)]
Inline TLS field access for linux/osx x64/arm64 (#87082)
* wip
* add __tls_get_addr() code in jitinterface
* working model
* linux rely on __tls_get_addr() value
* Add fields for both max/threadSTaticBlocks, have separate for GC/non-gc
* code cleanup
* code cleanup
* add comments
* jit format
* update guid
* review feedback
* fix the offset
* arm64: wip
* linux arm64 model
* arm64: offsetOfThreadStaticBlock adjustment
* Add mrs and tpid0 register
* arm64: use the new mrs/tpidr0
* fix arm64 build and offset calculation:
* arm64: working
* arm64: move to struct model
* arm64: fixed the struct model
* x64: move to struct model
* code refactoring
* #define for field access
* change mrs -> mrs_tpid0
* fix a bug
* remove unwanted method
* another fix
* Add entries in CorInfoType.cs
* Update the #ifdef
* fix the windows scenario:
* review feedback
* fix the data-type
* add osx-arm64 support
* fix osx-arm64 issues
* fix build error
* fix build error after merge
* add osx/x64 support
* fix errors
* fix the macos/x64
* disable for alpine linux
* Disable for R2R
* review feedback
* fix r2r check
* move windows to struct model
* review feedback
* fix the register clobbering in release bits
* Move the linux/x64 logic to .S file
* Use TargetOS::IsMacOS
* disable optimization for single file
* working for linux/x64
* fix some errors for osx/x64
* fix for osx x64/arm64
* fix for arm64 linux/osx
* try disable for musl/arm64
* rename variable
* Rename variable to tlsIndexObject
* Make offset variables as uint32_t
* change the type of indexObj/ftnAddr to void*
* replace ifdef(msc_ver) with ifdef(windows)
* Revert to JIT_TO_EE_TRANSITION_LEAF
* Move code to asmHelpers.S and rename method
* rename the methods per the platform
* fix osx builds
* fix build break
* fix some errors around osx
* rename some more methods
* review feedback
* review feedback
* delete the comment
* make methods static
* remove macos/x64 check
* fix the check for linux/x64
* detect early for single-file linux/x64
* move the assert
* review feedback
* misc fixup
* use fgMorphArgs()
* remove commented code
Alexander Soldatov [Thu, 6 Jul 2023 14:21:49 +0000 (17:21 +0300)]
[RISC-V] Increase instruction group size (#88468)
Similar to ARM default IG size is not enough for prolog generation.
Brennan [Thu, 6 Jul 2023 13:53:22 +0000 (06:53 -0700)]
Optimize Ascii.Equals when widening (#87141)
* Optimize Ascii.Equals when widening
* add BoundedMemory tests to ensure that boundaries are respected
---------
Co-authored-by: Adam Sitnik <adam.sitnik@gmail.com>
Lakshan Fernando [Thu, 6 Jul 2023 12:29:19 +0000 (05:29 -0700)]
Setup support for managed runtime events (#87785)
* preliminery managed runtime event support
* Linux fixes 1
* full portable threadpool support
* checkpoint before python script
* Additional GC event support
* NativeAOT to use sink threadpool file
* CI check
* Enable GC runtime tests
* Post main merge fixups
* change tests that run in CI
* Support native sinks for threading events
* Enable managed exceptions via eventpipe
* Move Exception event to EventPipe
* Add an empty exception event to disabled EventPipe
* make events do no-op in disabled eventpipe
* Update src/coreclr/nativeaot/Runtime/clretwallmain.h
Co-authored-by: Elinor Fung <elfung@microsoft.com>
* FB first stage
* Test change
* FB
* latest FB
* Pulled PR 88365
* FB
---------
Co-authored-by: Elinor Fung <elfung@microsoft.com>
Steve Pfister [Thu, 6 Jul 2023 11:22:37 +0000 (04:22 -0700)]
[Android] Add NdkToolFinder task & Fix up AOT app build (#88210)
Since we bumped to NDK 23, having the aot compiler itself generate shared libraries stopped working. This is due to NDK 23 moving most of the toolchain into a common bin directory. AS was left over in each of these directories as an alias to bin/<triple>-as.
This change adds a task to collect all of the important NDK toolchain paths. It also fixes up the aot build when AOTWithLibraryFiles is set to true and we want the aot compiler to produce shared libraries.
Dong-Heon Jung [Thu, 6 Jul 2023 04:13:54 +0000 (13:13 +0900)]
[RISC-V] Fix floating point (#88311)
* [RISC-V] Fix: passing struct in regs
* Fix a typo in CallDescrWorkerInternal
* [RISC-V] Update floating points
Stephen Toub [Thu, 6 Jul 2023 02:30:01 +0000 (22:30 -0400)]
Delete formatter test blob for `EqualityComparer<byte>.Default` (#88454)
A change was recently made that deleted the specialized type returned, so delete the dedicated formatting test as well.
Jackson Schuster [Thu, 6 Jul 2023 02:07:14 +0000 (19:07 -0700)]
Fix diagnostic message in ComInterfaceGenerator (#88450)
Fixes #88448
David Wrighton [Thu, 6 Jul 2023 01:04:55 +0000 (18:04 -0700)]
Fix unsafe walking of DacEnumerableHash (#88063)
Under the following set of conditions, it is possible for a lookup in the `DacEnumerableHash` to fail to find already present entries.
Given Thread A which is growing the table, and thread B which is attempting to perform a lookup.
1. Thread B reads an `VolatileEntry*` from a bucket in the hashtable. Let this be EntryB. EntryB will have a next pointer which points to EntryThatWeNeedToFind, but it will not yet read that pointer value from EntryB.
2. Thread A reads the pointer to EntryB from the same bucket in the hashtable.
3. Thread A adds EntryB to the new hashtable
4. Thread A sets the bucket in the old hashtable to point to EntryThatWeNeedToFind
5. Thread A sets the next pointer in EntryB to point to NULL.
6. Thread B reads the next pointer, sees that the next pointer is NULL, and falls back to looking in the "new" hashtable for any possible updates
7. Thread B looks in the appropriate bucket in the "new" hashtable. That bucket does not yet contain EntryThatWeNeedToFind, as it hasn't yet been moved from the old hashtable.
8. Thread B returns NULL __*INCORRECTLY*__, indicating that there are no entries with the matching in them.
This change adjust how these linked lists work, but giving each one a version number which can be checked as the linked list finishes reading. The fix allows finding entries from an earlier set of buckets (as those can be in the new list temporarily, and will not interfere with finding the correct result), but will actively detect walking entries in the new list when attempting to walk the old entries.
In addition, there is a drive by fix for an issue where if there are more than ~14,000,000 entries in the table, it allocates a full new list each time the list is added to.
Fixes #85688
David Wrighton [Thu, 6 Jul 2023 01:03:03 +0000 (18:03 -0700)]
Disable use of certains R2R data structures in the DAC (#88446)
- The ReadyToRun_EnclosingTypeMap, ReadyToRun_TypeGenericInfoMap, and ReadyToRun_MethodIsGenericMap provide optimization data which is already present in metadata
- They are not properly DACized
- Propertly DACizing these methods would not improve debugging, so instead they are disabled in the DAC
Stephen Toub [Thu, 6 Jul 2023 01:01:51 +0000 (21:01 -0400)]
Fix which path is taken in Enumerable.Select for empty partition (#88425)
* Fix which path is taken in Enumerable.Select for empty partition
My recent changes to improve perf of some LINQ iterators by implementing `IList<T>` caused Select on an empty partition to start preferring a path that wasn't optimized for empty. This fixes it. I searched and don't see any other operators that would be similarly negatively affected.
* Use is check
David Wrighton [Thu, 6 Jul 2023 00:55:38 +0000 (17:55 -0700)]
Type equivalence support in Crossgen2 (#87899)
We currently have a small behavior hole on Windows where if a binary uses TypeEquivalence, it is technically possible to get crossgen2 to generate invalid code.
This fixes that by adding support for type equivalence to the managed type system, and adjusting crossgen2 to use it. This is complicated by the detail that this may generate some type references which cannot be referenced from an R2R file, so additional error checking was needed (this additional error checking is what fixes #67855).
However, while basic support for type equivalence has been added to the type system, support for actually performing type equivalent interface resolution has not been, so devirtualization of type equivalent interface calls is disabled. At this time, I do not expect to see a customer need for implementing that fairly complicated feature
Fixes #67855
Eric StJohn [Thu, 6 Jul 2023 00:31:01 +0000 (17:31 -0700)]
Add consultants for some extensions areas (#88430)
* Add consultants for some extensions areas
* Update area-owners.md
---------
Co-authored-by: Dan Moseley <danmose@microsoft.com>
Eric StJohn [Wed, 5 Jul 2023 23:55:38 +0000 (16:55 -0700)]
Add docs for AsyncInterfaces (#88432)
Stephen Toub [Wed, 5 Jul 2023 23:41:24 +0000 (19:41 -0400)]
Dedup Task.WhenAll non-generic and generic implementations (#88154)
The generic implementation was calling the non-generic one and then using an additional continuation to extract the resulting `Task<TResult>` due to lack of covariance on classes. We can instead just factor the shared implementation out into a generic with the type parameter constrained to be a Task. This results in simpler code as well as avoiding an extra continuation in the generic case.
As part of cleaning this up:
- I changed code where we need to make a defensive copy of an input collection to use CopyTo; we were already doing this in some places but not others. This saves on an enumerator allocation when enumerating the source collection, as well as multiple interface calls.
- I augmented WhenAny to also special-case `List<Task>`, as that's a common input and we can handle it a bit more efficiently, especially if the collection ends up containing just two tasks.
- I removed the `GenericDelegateCache<TAntecedentResult, TResult>`. That was from a time before the C# compiler supported caching of generic lambdas. It would have needed to have been updated to handle the stronger type coming out of CommonCWAnyLogic, so I instead just got rid of it. We're better off lazily-creating these rarely used delegates, anyway.
June Rhodes [Wed, 5 Jul 2023 23:22:05 +0000 (09:22 +1000)]
Make DoesConsoleSupportAnsi respect DOTNET_SYSTEM_CONSOLE_ALLOW_ANSI_COLOR_REDIRECTION (#88241)
Co-authored-by: Adam Sitnik <adam.sitnik@gmail.com>
Egor Bogatov [Wed, 5 Jul 2023 23:19:11 +0000 (01:19 +0200)]
Add gtNewGenericCon (#88440)
Jakob Botsch Nielsen [Wed, 5 Jul 2023 22:16:25 +0000 (00:16 +0200)]
JIT: Always look for SIMD fields during promotion (#84122)
Alexander Köplinger [Wed, 5 Jul 2023 21:27:26 +0000 (23:27 +0200)]
[mono] Fix sorting custom attributes in ILStrip (#88435)
The change in https://github.com/dotnet/runtime/pull/87923 was subtly wrong, the problem is that RID on a Cecil metadata token masks out the token type.
We actually have to reconstruct the custom attribute coded-index.
Tomas Weinfurt [Wed, 5 Jul 2023 21:17:11 +0000 (14:17 -0700)]
Add option to decrypt Quic traffic in debug builds (#83001)
* Add option to decrypt Quic traffic in debug builds
* feedback from review
* fixes & feedback
* feedback from review
Michał Petryka [Wed, 5 Jul 2023 19:07:19 +0000 (21:07 +0200)]
Unify Default Comparer logic (#88006)
Unifies (Equality)Comparer.Default logic so that it matches between CoreCLR, NativeAOT, Mono in runtime and intrinsic implementations.
Fixes devirt behaviour around Nullable types.
Also enables devirt for non final types in CoreCLR and NativeAOT.
Required for #87579.
Fixes #87391.
Co-authored-by: Jan Kotas <jkotas@microsoft.com>
Michał Petryka [Wed, 5 Jul 2023 19:01:33 +0000 (21:01 +0200)]
Fix AVX512 moves on X86 (#88421)
Extracted from #85875.
Alexander Köplinger [Wed, 5 Jul 2023 19:00:40 +0000 (21:00 +0200)]
Fix Codespaces prebuild after CMake upgrade and set policy in Mono (#88410)
The prebuild on Codespaces broke after the minimum CMake version was bumped in https://github.com/dotnet/runtime/pull/86530 because the container was still using Ubuntu 20.04 (Focal) which only has CMake 3.16.1.
Upgrade to Ubuntu 22.04 (Jammy) so we have a new enough CMake.
Also fix a missing policy in the Mono CMakeLists.txt that was accidentally removed.
Jakob Botsch Nielsen [Wed, 5 Jul 2023 18:37:09 +0000 (20:37 +0200)]
JIT: Compute BB weights with higher precision, use tolerance in if-conversion (#88385)
In https://github.com/dotnet/runtime/issues/88376#issuecomment-
1620371447
I noticed a spurious diff in the System.Tests.Perf_Int32.ToStringHex
benchmark that looked like this:
```diff
G_M62987_IG13:
and edi, esi
mov ebx, dword ptr [rbp+88H]
- mov ecx, 1
test ebx, ebx
- cmovle ebx, ecx
+ jg SHORT G_M62987_IG15
+ ;; size=12 bbWeight=1.00 PerfScore 2.50
+G_M62987_IG14:
+ mov ebx, 1
+ ;; size=5 bbWeight=0.98 PerfScore 0.24
+G_M62987_IG15:
```
Investigating, it turns out to be caused by a check in if-conversion for
detecting loops, that is using a very sharp boundary. It turns out that
the integral block weights are exactly the same (in both the base and
diff), but the floating point calculation ended up with exactly 1 and
very close to 1 in the base/diff cases respectively.
So two changes to address it:
* Switch getBBWeight to multiply by BB_UNITY_WEIGHT after dividing
* Switch if-conversion to have a 5% tolerance so that losing just a
single count in the wrong place will not cause a decision change. Note
that the check here is used as a cheap "is inside loop" check.
dotnet-maestro[bot] [Wed, 5 Jul 2023 18:24:16 +0000 (13:24 -0500)]
[main] Update dependencies from dnceng/internal/dotnet-optimization (#88132)
* Update dependencies from https://dev.azure.com/dnceng/internal/_git/dotnet-optimization build
20230627.5
optimization.linux-arm64.MIBC.Runtime , optimization.linux-x64.MIBC.Runtime , optimization.windows_nt-arm64.MIBC.Runtime , optimization.windows_nt-x64.MIBC.Runtime , optimization.windows_nt-x86.MIBC.Runtime , optimization.PGO.CoreCLR
From Version 1.0.0-prerelease.23326.3 -> To Version 1.0.0-prerelease.23327.5
* Update dependencies from https://dev.azure.com/dnceng/internal/_git/dotnet-optimization build
20230628.5
optimization.linux-arm64.MIBC.Runtime , optimization.linux-x64.MIBC.Runtime , optimization.windows_nt-arm64.MIBC.Runtime , optimization.windows_nt-x64.MIBC.Runtime , optimization.windows_nt-x86.MIBC.Runtime , optimization.PGO.CoreCLR
From Version 1.0.0-prerelease.23326.3 -> To Version 1.0.0-prerelease.23328.5
* Update dependencies from https://dev.azure.com/dnceng/internal/_git/dotnet-optimization build
20230629.5
optimization.linux-arm64.MIBC.Runtime , optimization.linux-x64.MIBC.Runtime , optimization.windows_nt-arm64.MIBC.Runtime , optimization.windows_nt-x64.MIBC.Runtime , optimization.windows_nt-x86.MIBC.Runtime , optimization.PGO.CoreCLR
From Version 1.0.0-prerelease.23326.3 -> To Version 1.0.0-prerelease.23329.5
* Update dependencies from https://dev.azure.com/dnceng/internal/_git/dotnet-optimization build
20230630.5
optimization.linux-arm64.MIBC.Runtime , optimization.linux-x64.MIBC.Runtime , optimization.windows_nt-arm64.MIBC.Runtime , optimization.windows_nt-x64.MIBC.Runtime , optimization.windows_nt-x86.MIBC.Runtime , optimization.PGO.CoreCLR
From Version 1.0.0-prerelease.23326.3 -> To Version 1.0.0-prerelease.23330.5
* Update dependencies from https://dev.azure.com/dnceng/internal/_git/dotnet-optimization build
20230701.5
optimization.linux-arm64.MIBC.Runtime , optimization.linux-x64.MIBC.Runtime , optimization.windows_nt-arm64.MIBC.Runtime , optimization.windows_nt-x64.MIBC.Runtime , optimization.windows_nt-x86.MIBC.Runtime , optimization.PGO.CoreCLR
From Version 1.0.0-prerelease.23326.3 -> To Version 1.0.0-prerelease.23351.5
* Update dependencies from https://dev.azure.com/dnceng/internal/_git/dotnet-optimization build
20230704.4
optimization.linux-arm64.MIBC.Runtime , optimization.linux-x64.MIBC.Runtime , optimization.windows_nt-arm64.MIBC.Runtime , optimization.windows_nt-x64.MIBC.Runtime , optimization.windows_nt-x86.MIBC.Runtime , optimization.PGO.CoreCLR
From Version 1.0.0-prerelease.23326.3 -> To Version 1.0.0-prerelease.23354.4
---------
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
Ulrich Weigand [Wed, 5 Jul 2023 18:18:25 +0000 (20:18 +0200)]
[mono] Fix InlineArrayAttribute access on big-endian platforms (#88417)
Use "guint32" instead of "gpointer" to access the length field in
InlineArrayAttribute; fixes serious regression after PR #88419.
Kunal Pathak [Wed, 5 Jul 2023 17:05:14 +0000 (10:05 -0700)]
Fix SIMD node DOT retType (#88363)
* Fix SIMD node DOT retType
* no need for lengthsquared
Eric StJohn [Wed, 5 Jul 2023 16:54:49 +0000 (09:54 -0700)]
Suppress System.Drawing.Common dependency from System.Windows.Extensions (#88157)
* Suppress System.Drawing.Common dependency from System.Windows.Extensions
* Add Drawing reference to Windows.Extensions tests
Dan Moseley [Wed, 5 Jul 2023 16:11:54 +0000 (09:11 -0700)]
Inline more helpers in RegexParser (#88392)
* inline RightCharMoveRight()
* inline Textpos()
* inline MoveLeft()
* inline MoveRight()
* Inline CharAt()
* Inline RightChar()
* Inline RightChar()
* Inline CharsRight()
* Inline Textto()
* rename _pattern and _currentPos
* cleanup Category table
* Cleanup ScanBlank()
* reorder <
* reorder - =
* reorder >=
* reorder - >
* more
* bool param name
* comments
* merge
Steve Pfister [Wed, 5 Jul 2023 16:10:17 +0000 (09:10 -0700)]
[iOS] Add temp cindex attribute to build on xcode 15 (#88261)
Without CursorKind 437, the cross build for iOS/tvOS and iOS/tvOSSimulator will fail. This change adds a temporary attribute to avoid that.
Jakob Botsch Nielsen [Wed, 5 Jul 2023 15:16:33 +0000 (17:16 +0200)]
JIT: Propagate multi-reg-index for upper-vector-restore ref positions (#88380)
`LinearScan::resolveRegisters` will write the register back to the IR
node for an upper-vector-restore RP, so without this propagation we
would overwrite the register assignment for an unrelated field.
For example, for IR like
```scala
N001 ( 9, 6) [000090] m------N--- t90 = LCL_VAR struct<JIT.HardwareIntrinsics.Arm._AdvSimd.SimpleTernaryOpTest__AbsoluteDifferenceWideningUpperAndAdd_Vector128_UInt32+TestStruct, 48>(P) V00 loc0
▌ simd16 field V00._fld1 (fldOffset=0x0) -> V14 tmp10 (last use)
▌ simd16 field V00._fld2 (fldOffset=0x10) -> V15 tmp11 (last use)
▌ simd16 field V00._fld3 (fldOffset=0x20) -> V16 tmp12 (last use) $540
┌──▌ t90 struct
N002 ( 10, 7) [000091] ----------- ▌ RETURN struct $VN.Void
```
we could build ref positions such as
```
[000091] 449.#639 U14 UVRs UVRes NA │ │ │ │ │ │ │ │ │ │ │ │ │V16a│ │ │ │
449.#640 d0 Fixd Keep d0 │ │ │ │ │ │ │ │ │ │ │ │ │V16a│ │ │ │
449.#641 V14 Use * ReLod d0 │ │ │ │ │ │ │ │ │ │ │V14a│ │V16a│ │ │ │
Keep d0 │ │ │ │ │ │ │ │ │ │ │V14i│ │V16a│ │ │ │
449.#642 U15 UVRs UVRes NA │ │ │ │ │ │ │ │ │ │ │ │ │V16a│ │ │ │
449.#643 d1 Fixd Keep d1 │ │ │ │ │ │ │ │ │ │ │ │ │V16a│ │ │ │
449.#644 V15 Use * ReLod d1 │ │ │ │ │ │ │ │ │ │ │ │V15a│V16a│ │ │ │
Keep d1 │ │ │ │ │ │ │ │ │ │ │ │V15i│V16a│ │ │ │
449.#645 d2 Fixd Keep d2 │ │ │ │ │ │ │ │ │ │ │ │ │V16a│ │ │ │
449.#646 V16 Use * Keep d2 │ │ │ │ │ │ │ │ │ │ │ │ │V16i│ │ │ │
```
When writing back register assignments the upper-vector-restore at #642
ended up overwriting the assignment for the first field on [000090],
resulting in
```scala
N447 ( 9, 6) [000090] m------N--z t90 = LCL_VAR struct<JIT.HardwareIntrinsics.Arm._AdvSimd.SimpleTernaryOpTest__AbsoluteDifferenceWideningUpperAndAdd_Vector128_UInt32+TestStruct, 48>(P) V00 loc0 NA
▌ simd16 field V00._fld1 (fldOffset=0x0) -> V14 tmp10 (last use)
▌ simd16 field V00._fld2 (fldOffset=0x10) -> V15 tmp11 (last use)
▌ simd16 field V00._fld3 (fldOffset=0x20) -> V16 tmp12 d2 (last use) REG NA,d1,d2 $540
┌──▌ t90 struct
N449 ( 10, 7) [000091] ----------- ▌ RETURN struct REG NA $VN.Void
```
(note the REG NA instead of REG d0).
Tom Deseyn [Wed, 5 Jul 2023 14:44:44 +0000 (16:44 +0200)]
Tar: set directory modification times while extracting. (#88231)
* Tar: set directory modification times while extracting.
* Extend test and change entry write order.
* Get rid of some enumeration allocations.
* test: check DateTime using InRange.
* Extend comment explaining the order of entries.
* Apply suggestions from code review
Co-authored-by: Dan Moseley <danmose@microsoft.com>
* Try including AltDirectorySeparatorChar.
* Revert "Try including AltDirectorySeparatorChar."
This reverts commit
086aac43a897826394156692bd82f79feb3eca9d.
* Try using Directory.SetLastWriteTime instead of File's.
* Tie up some ends.
---------
Co-authored-by: Dan Moseley <danmose@microsoft.com>
dotnet bot [Wed, 5 Jul 2023 13:16:16 +0000 (15:16 +0200)]
Localized file check-in by OneLocBuild Task: Build definition ID 679: Build ID 2214454 (#88399)
Eirik Tsarpalis [Wed, 5 Jul 2023 12:55:21 +0000 (13:55 +0100)]
Implement JsonTypeInfoResolver.WithAddedModifier (#88255)
Vlad Brezae [Wed, 5 Jul 2023 12:36:20 +0000 (15:36 +0300)]
[mono][interp] Add vectorization for Vector4 (#87822)
* [mono][interp] Reuse vectorization for Vector4
Which is pretty much Vector<float>
* [mono][interp] Implement Vector.One for float gparam
* [mono][interp] Extract code to be reused later
* [mono][interp] Intrinsify Vector<T> ctor and Vector4 ctor
The constructors receive each element as an argument. This is the same as `Vector128.Create` so we reuse that code. This ends up generating a MINT_SIMD_V128_I*_CREATE opcode which receives each element value as an argument.
* [mono][interp] Track r4 var value during interpreter optimizations
Since it is common to instantiate Vector4 with constant values, add also support for tracking R4 values during optimizations. If all elements are constant, we generate instead a single MINT_SIMD_V128_LDC, which has the entire vector value embedded in the instruction stream.
* [mono][interp] Fix Vector ctor instrinsic when called explicitly
* [mono][interp] Replace ldloca + stobj.vt pair with mov.vt
* [mono][interp] Return early if vector_klass is not simd type
It is not marked as simd type if the generic param is not valid (ex not primitive type). Future checks are not expecting this scenario and can potentially crash.
Egor Bogatov [Wed, 5 Jul 2023 12:26:24 +0000 (14:26 +0200)]
Improve non-SIMD block copy/init codegen (#88393)
Jo Shields [Wed, 5 Jul 2023 10:47:48 +0000 (06:47 -0400)]
Add clr.tools subset to runtime-community (#88160)
Co-authored-by: Adeel Mujahid <3840695+am11@users.noreply.github.com>
Kevin Jones [Wed, 5 Jul 2023 09:52:59 +0000 (05:52 -0400)]
Permit trailing data after PKCS12 (#88340)
Adam Sitnik [Wed, 5 Jul 2023 07:45:54 +0000 (09:45 +0200)]
Faster optimized frozen dictionary creation (6/6) (#88093)
* don't use custom ToArray for small frozen collections, up to 50% gain for creation time for collections with <= 4 items
* for these types GetHashCode returns their value casted to int, so when we receive a Dictionary/HashSet where there are key we know that all hash codes are unique and we can avoid some work later
10-15% CPU time gain and 15-20% allocation reduction for FrozenDictionary and FrozenHashSet where TKey is uint, short, ushort, byte, sbyte
* move Length Buckets code to a dedicated helper type to reduce code duplication and decrease code size
* add tests for Frozen Dictionaries with key being uint, short, ushort, byte, sbyte, nint and nuint
* fix discovered bug: IntPtr started implementing IComparable<IntPtr> in .NET 5
Jo Shields [Wed, 5 Jul 2023 03:41:44 +0000 (23:41 -0400)]
Bump to libc++ without libatomic linkage (#88402)
Closes: https://github.com/dotnet/runtime/issues/88361
Tarek Mahmoud Sayed [Wed, 5 Jul 2023 00:45:20 +0000 (17:45 -0700)]
Move IMeterFactory to System.Diagnostics.DiagnosticSource (#88302)
Zoltan Varga [Wed, 5 Jul 2023 00:26:05 +0000 (20:26 -0400)]
[mono][llvm] Fix another crash when using opaque pointers. (#88397)
Zoltan Varga [Wed, 5 Jul 2023 00:25:43 +0000 (20:25 -0400)]
[mono] Respect KeepNativeSymbols when building cross compilers. (#88398)
Dan Moseley [Wed, 5 Jul 2023 00:11:43 +0000 (17:11 -0700)]
change regex options tests to extension method (#88366)
* extension method for regexoptions
* Revert "extension method for regexoptions"
This reverts commit
95b05489f0245d23d6d265a2cad2fca2ae0975a0.
* Inline UseOption
Vlad Brezae [Tue, 4 Jul 2023 17:42:46 +0000 (20:42 +0300)]
[mono][interp] Extract simd create opcodes in separate methods (#88381)
Newer versions of clang allocate the local buffer repeatedly by bumping the stack. This behavior seems dubious, leading to stack overflows, but extracting the code outside the interp main loop is an improvement anyway for the interpreter.
Qiao Pengcheng [Tue, 4 Jul 2023 17:02:35 +0000 (01:02 +0800)]
[LoongArch64] Fix the NotImplementedException within R2RDump. (#88379)
Jakob Botsch Nielsen [Tue, 4 Jul 2023 16:02:38 +0000 (18:02 +0200)]
JIT: Handle primitive-sized remainders overlapping padding/promotions in physical promotion (#88109)
The remainder may be separated by a bit of padding or other promoted fields but
still fit into a primitive; in this case it is still beneficial to copy it all
as a primitive, instead of falling back to a full block copy.
Example:
```csharp
private S _s;
void Foo()
{
S s = new();
s.A = 10;
s.D = 20;
s.F = 30; // A, D, F gets promoted
_s = s;
}
private struct S
{
public byte A;
public byte B;
public byte C;
public byte D;
public byte E;
public byte F;
}
```
```diff
Processing block operation [000018] that involves replacements
dst+003 <- V04 (V01.[003..004)) (last use)
dst+005 <- V05 (V01.[005..006)) (last use)
Block op remainder: [001..003) [004..005)
- => Remainder strategy: retain a full block op
+ => Remainder strategy: int at +001
```
```diff
; V00 this [V00,T01] ( 3, 3 ) ref -> rcx this class-hnd single-def
;* V01 loc0 [V01 ] ( 0, 0 ) struct ( 8) zero-ref do-not-enreg[SF] ld-addr-op
;# V02 OutArgs [V02 ] ( 1, 1 ) struct ( 0) [rsp+00H] do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
;* V03 tmp1 [V03 ] ( 0, 0 ) ubyte -> zero-ref "V01.[000..001)"
;* V04 tmp2 [V04 ] ( 0, 0 ) ubyte -> zero-ref "V01.[003..004)"
;* V05 tmp3 [V05 ] ( 0, 0 ) ubyte -> zero-ref "V01.[005..006)"
; V06 tmp4 [V06,T00] ( 5, 10 ) byref -> rcx single-def "Spilling address for field-by-field copy"
;
; Lcl frame size = 0
G_M52879_IG01: ;; offset=0000H
;; size=0 bbWeight=1 PerfScore 0.00
G_M52879_IG02: ;; offset=0000H
add rcx, 8
xor eax, eax
- mov dword ptr [rcx], eax
- mov dword ptr [rcx+02H], eax
+ mov dword ptr [rcx+01H], eax
mov byte ptr [rcx], 10
mov byte ptr [rcx+03H], 20
mov byte ptr [rcx+05H], 30
- ;; size=22 bbWeight=1 PerfScore 5.50
-G_M52879_IG03: ;; offset=0016H
+ ;; size=20 bbWeight=1 PerfScore 4.50
+G_M52879_IG03: ;; offset=0014H
ret
;; size=1 bbWeight=1 PerfScore 1.00
-; Total bytes of code 23, prolog size 0, PerfScore 8.80, instruction count 8, allocated bytes for code 23 (MethodHash=
1da13170) for method Program:Foo():this (FullOpts)
+; Total bytes of code 21, prolog size 0, PerfScore 7.60, instruction count 7, allocated bytes for code 21 (MethodHash=
1da13170) for method Program:Foo():this (FullOpts)
; ============================================================
```
We have to be careful, however, since the covering segment can now contain
promoted fields. If this happens we need to make sure we write the promoted
field _after_ the remainder.
Jakob Botsch Nielsen [Tue, 4 Jul 2023 13:18:39 +0000 (15:18 +0200)]
JIT: Quirk a quirk in fgExpandVirtualVtableCallTarget (#88374)
Fix #88349
Ivan Povazan [Tue, 4 Jul 2023 13:02:54 +0000 (15:02 +0200)]
[mono] Reenable disabled tests as mono now respects non-public custom attributes in dynamic assemblies (#88322)
* Reenable disabled tests
* Disable Roslyn analyzer tests in the browser as Monitor.Wait / Monitor.Pulse are not supported on single-threaded WASM
Contributes to: https://github.com/dotnet/runtime/issues/60650
feiyun0112 [Tue, 4 Jul 2023 10:26:27 +0000 (18:26 +0800)]
Email addresses in same domain not equal (#87721)
Different email addresses in same domain should not be equal.
UriComponents.UserInfo will compare also UserInfo from the Uris.
Pavel Savara [Tue, 4 Jul 2023 09:23:46 +0000 (11:23 +0200)]
only log errors of Task/Promise marshaling (#88326)
Marie Píchová [Tue, 4 Jul 2023 09:17:51 +0000 (11:17 +0200)]
Unified source file formatting (#88330)
Meri Khamoyan [Tue, 4 Jul 2023 06:20:43 +0000 (10:20 +0400)]
[OSX] HybridGlobalization Workaround for insufficiently sized destination buffer (#88184)
Workaround for insufficiently sized dest buffer
Michal Strehovský [Tue, 4 Jul 2023 06:20:06 +0000 (15:20 +0900)]
Remove conservative generic scanning logic (#88360)
@EgorBo's #88025 made it possible to undo conservative scanning logic added in dotnet/corert#7618.
Saves 0.4% in size for Stage1 and Stage2 apps.
Michal Strehovský [Tue, 4 Jul 2023 05:48:39 +0000 (14:48 +0900)]
Allow inlining array methods (#88367)
`CanInline` would return false to methods that implement generic interfaces on arrays because we didn't consider the magic `Array<T>` types constructed.
For:
```csharp
static int Main()
{
IReadOnlyCollection<int> l = new int[] { 1, 2, 3 };
return l.Count;
}
```
Before:
```
sub rsp, 40
;; size=4 bbWeight=1 PerfScore 0.25
G_M24375_IG02: ;; offset=0004H
lea rcx, [(reloc 0x4000000000420a70)] ; int[]
mov edx, 3
call CORINFO_HELP_NEWARR_1_VC
lea rcx, [(reloc 0x4000000000420aa8)] ; const ptr
mov rdx, qword ptr [rcx]
mov qword ptr [rax+10H], rdx
mov rdx, qword ptr [rcx+04H]
mov qword ptr [rax+14H], rdx
mov rcx, rax
call System.Array`1[int]:get_Count():int:this
nop
;; size=48 bbWeight=1 PerfScore 9.75
G_M24375_IG03: ;; offset=0034H
add rsp, 40
ret
```
After:
```
sub rsp, 40
;; size=4 bbWeight=1 PerfScore 0.25
G_M24375_IG02: ;; offset=0004H
lea rcx, [(reloc 0x4000000000420a70)] ; int[]
mov edx, 3
call CORINFO_HELP_NEWARR_1_VC
lea rcx, [(reloc 0x4000000000420aa8)] ; const ptr
mov rdx, qword ptr [rcx]
mov qword ptr [rax+10H], rdx
mov rdx, qword ptr [rcx+04H]
mov qword ptr [rax+14H], rdx
mov eax, 3
;; size=44 bbWeight=1 PerfScore 8.50
G_M24375_IG03: ;; offset=0030H
add rsp, 40
ret
```
Dan Moseley [Tue, 4 Jul 2023 05:22:09 +0000 (22:22 -0700)]
Remove empty files (#88362)
Bruce Forstall [Tue, 4 Jul 2023 04:18:09 +0000 (21:18 -0700)]
Fix libraries Windows test runner script (#88354)
Having HAS_TEST_RESULTS unset leads to a Windows CMD script
syntax error.
Clean up a few other things while there:
1. Explicitly set ENABLEEXTENSIONS
2. Set a variable to the test exit code. There's an unsafe assumption
currently that `%ERRORLEVEL%` doesn't change between the test run
and all of the subsequent uses.
3. Remove unnecessary quotes around setting `HAS_TEST_RESULTS=1`
Fixes #88353
Ivan Diaz Sanchez [Tue, 4 Jul 2023 02:03:12 +0000 (19:03 -0700)]
Up Minimum CMake Version to 3.20 (#86530)
* Let's see what happens :)
* CMake 3.20 works on my machine???
* Upped the llibraries CMake version to 3.20
* Upped native libs CMake version to 3.20, and updated the requirements docs.
* Add CMake Policy to 3.17.0 for native libs because of wasm.
* Downgraded native libs CMake to 3.17.0 temporarily, while we figure out the funny troublemaker wasm.
* Added a couple notes regarding CMake on Linux and Windows, and upped the corehost and libunwind CMake versions to 3.20
* Reverted external native components CMake versions, and upped tests to 3.20. Only Mono remains now.
* Added mono to the 3.20 gang. Also, changed the condition for wasm.
* Had to condition mono's CMake version as well because of wasm.
* DEBUG-ONLY COMMIT: Added some CMake logging to help me figure out what is set for wasm to think of the appropriate conditional.
* DEBUG-ONLY COMMIT: I think I have the answer for wasm now/
* Restored the DEBUG-ONLY changes and set 3.20 as the universal CMake, as the missing machines have now been updated.
* Updated requirements doc with notes for Debian 11 and lower.
James Newton-King [Tue, 4 Jul 2023 01:42:39 +0000 (09:42 +0800)]
Improve LoggerFactory and Logger debugging (#88313)
Jo Shields [Mon, 3 Jul 2023 21:53:34 +0000 (17:53 -0400)]
Try adding cross-compilers for Musl (#87841)
* Add cross-compilers for Musl
* Fix LLVM linkage on Musl
Egor Bogatov [Mon, 3 Jul 2023 21:36:59 +0000 (23:36 +0200)]
Pass exact context to getMethodInfo (#88025)
* Modify JIT-EE API
Co-authored-by: Jan Kotas <jkotas@microsoft.com>
Adeel Mujahid [Mon, 3 Jul 2023 21:19:50 +0000 (00:19 +0300)]
Shrink column width in `ilc --help` (#88356)
Jan Kotas [Mon, 3 Jul 2023 20:58:35 +0000 (13:58 -0700)]
Fix condition for ENSURE_PRIMARY_STACK_SIZE (#88342)
Fixes #88339
Alexander Radchenko [Mon, 3 Jul 2023 20:52:51 +0000 (02:52 +0600)]
TcpListener should implement IDisposable (#88043)
* TcpListener should implement IDisposable
* lint
* Code review
* Update src/libraries/System.Net.Sockets/src/System/Net/Sockets/TCPListener.cs
Co-authored-by: Anton Firszov <antonfir@gmail.com>
* Update TCPListener.cs [EditorBrowsable(EditorBrowsableState.Never)]
* Code review
* lint
* Code review
---------
Co-authored-by: Anton Firszov <antonfir@gmail.com>
Filip Navara [Mon, 3 Jul 2023 20:44:14 +0000 (22:44 +0200)]
Fix incorrect sequence number used in NTLM signature calculation (#87998)
Alexander Köplinger [Mon, 3 Jul 2023 20:14:43 +0000 (22:14 +0200)]
[mono] Remove mono_loader_save_bundled_library() (#88331)
SwapnilGaikwad [Mon, 3 Jul 2023 19:43:36 +0000 (20:43 +0100)]
Remove TODO from a comment in System.Guid.cs (#88337)
dotnet bot [Mon, 3 Jul 2023 15:46:39 +0000 (17:46 +0200)]
Localized file check-in by OneLocBuild Task: Build definition ID 679: Build ID 2213468 (#88301)
Adam Sitnik [Mon, 3 Jul 2023 14:14:56 +0000 (16:14 +0200)]
[Crossgen] Don't display list of enum values in help (#88324)
Jo Shields [Mon, 3 Jul 2023 13:46:29 +0000 (09:46 -0400)]
Upgrade to LLVM 16 & C++17, and bundle libc++ on Linux (#87190)
* Upgrade to LLVM 16, and stop focusing on C++11 ABi stuff from CentOS 7
Co-authored-by: Zoltan Varga <vargaz@gmail.com>
Thefrank [Mon, 3 Jul 2023 13:31:49 +0000 (06:31 -0700)]
[mono] fix memfuncs for FreeBSD (#88300)
* fix memfuncs for FreeBSD
* match format
James Newton-King [Mon, 3 Jul 2023 12:43:08 +0000 (20:43 +0800)]
Improve Hosting debugging (#88308)
* Improve Hosting debugging
* Update
* Remove Logger
* PR feedback
Michal Strehovský [Mon, 3 Jul 2023 04:08:33 +0000 (13:08 +0900)]
Expose stack trace metadata stripping as a supported option (#88235)
Nikita Balabaev [Sun, 2 Jul 2023 18:37:07 +0000 (20:37 +0200)]
Fix OptionsValidator source-gen to skip static and const members (#88254)
Co-authored-by: Nikita Balabaev <nbalabaev@microsoft.com>
Stephen Toub [Sat, 1 Jul 2023 22:14:52 +0000 (18:14 -0400)]
Implement IList<T> on some LINQ iterators (#88249)
`ICollection<T>` provides both a Count and a CopyTo, and `IList<T>` an indexer, all of which can make various consumption mechanisms more efficient. We only implement the interfaces when the underlying collection has a fixed size and all of the interface implementations are side-effect free (in particular, while appealing to do so, we don't implement them on various Select iterators).
Some of the serialization tests need to be fixed as a result. The state of Queue's array is a bit different based on how its initialized, and such private details show up in BinaryFormatter output. Rather than special-casing the output per framework and core, I've just changed the test itself to ensure Queue can't see the size of the input collection.
Katelyn Gadd [Sat, 1 Jul 2023 20:08:33 +0000 (13:08 -0700)]
[wasm] Implement complete PackedSimd feature set in interpreter (#87903)
Implement all PackedSIMD APIs in wasm builds of the interpreter based on a table and auto-generated code
Implement most PackedSIMD APIs in jiterpreter
Detect constant simd arguments in jiterpreter and verify them
Simplify unused/temporary dregs in interp
Andrew Au [Sat, 1 Jul 2023 20:04:28 +0000 (13:04 -0700)]
Avoid reporting frozen segments in GetGenerationBounds (#85727)
Co-authored-by: Noah Falk <noahfalk@users.noreply.github.com>
Jakob Botsch Nielsen [Sat, 1 Jul 2023 19:35:12 +0000 (21:35 +0200)]
Unify JIT interface getTypeLayout between VM and crossgen2; refine what is considered "significant padding" (#88238)
* Stop considering auto layout types to have significant padding in the VM. This was already the behavior in crossgen2. This fixes one of the points of #71711.
* Remove special case where we never considered structs with GC pointers to have significant padding. After the above change this has no additional diffs.
* Fix the inline array layout expansion; the numFields of the inline array node was computed incorrectly, and the parent indices of descendants were not updated correctly
Example C#:
```csharp
private (long, int) _tuple;
[MethodImpl(MethodImplOptions.NoInlining)]
private void Foo()
{
_tuple = (5, 10);
}
```
Before:
```asm
; V00 this [V00,T00] ( 3, 3 ) ref -> rcx this class-hnd single-def
;# V01 OutArgs [V01 ] ( 1, 1 ) struct ( 0) [rsp+00H] do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
; V02 tmp1 [V02 ] ( 4, 8 ) struct (16) [rsp+08H] do-not-enreg[SB] ld-addr-op "NewObj constructor temp"
; V03 tmp2 [V03,T01] ( 3, 5 ) long -> [rsp+08H] do-not-enreg[] "field V02.Item1 (fldOffset=0x0)" P-DEP
; V04 tmp3 [V04,T02] ( 3, 5 ) int -> [rsp+10H] do-not-enreg[] "field V02.Item2 (fldOffset=0x8)" P-DEP
;
; Lcl frame size = 24
G_M52879_IG01: ;; offset=0000H
sub rsp, 24
vzeroupper
;; size=7 bbWeight=1 PerfScore 1.25
G_M52879_IG02: ;; offset=0007H
vxorps xmm0, xmm0, xmm0
vmovups xmmword ptr [rsp+08H], xmm0
mov qword ptr [rsp+08H], 5
mov dword ptr [rsp+10H], 10
vmovups xmm0, xmmword ptr [rsp+08H]
vmovups xmmword ptr [rcx+08H], xmm0
;; size=38 bbWeight=1 PerfScore 8.33
G_M52879_IG03: ;; offset=002DH
add rsp, 24
ret
;; size=5 bbWeight=1 PerfScore 1.25
; Total bytes of code 50, prolog size 7, PerfScore 15.83, instruction count 10, allocated bytes for code 50 (MethodHash=
1da13170) for method Program:Foo():this (FullOpts)
```
After:
```asm
; V00 this [V00,T00] ( 4, 4 ) ref -> rcx this class-hnd single-def
;# V01 OutArgs [V01 ] ( 1, 1 ) struct ( 0) [rsp+00H] do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
;* V02 tmp1 [V02 ] ( 0, 0 ) struct (16) zero-ref ld-addr-op "NewObj constructor temp"
;* V03 tmp2 [V03,T01] ( 0, 0 ) long -> zero-ref "field V02.Item1 (fldOffset=0x0)" P-INDEP
;* V04 tmp3 [V04,T02] ( 0, 0 ) int -> zero-ref "field V02.Item2 (fldOffset=0x8)" P-INDEP
;
; Lcl frame size = 0
G_M52879_IG01: ;; offset=0000H
;; size=0 bbWeight=1 PerfScore 0.00
G_M52879_IG02: ;; offset=0000H
mov qword ptr [rcx+08H], 5
mov dword ptr [rcx+10H], 10
;; size=15 bbWeight=1 PerfScore 2.00
G_M52879_IG03: ;; offset=000FH
ret
;; size=1 bbWeight=1 PerfScore 1.00
; Total bytes of code 16, prolog size 0, PerfScore 4.60, instruction count 3, allocated bytes for code 16 (MethodHash=
1da13170) for method Program:Foo():this (FullOpts)
```
Adeel Mujahid [Sat, 1 Jul 2023 19:02:19 +0000 (22:02 +0300)]
Fix make repro cli command (#88289)
Jakob Botsch Nielsen [Sat, 1 Jul 2023 18:36:20 +0000 (20:36 +0200)]
JIT: Remove a quirk in regular promotion (#88130)
Also fix a bug in LSRA: if we undo promotion of one field of a multi-reg
struct and as a result undo for the other fields as well, we must also
clear out the largeVectorVars bitset to indicate these are no longer
candidates.
Eirik Tsarpalis [Sat, 1 Jul 2023 08:07:46 +0000 (09:07 +0100)]
Apply a number of improvements and bugfixes to JsonNode (#88194)
* Avoid JsonElement torn read in a few JsonArray/JsonDocument and unify code patterns.
* Simplify and clean up JsonValue implementation -- fix DeepEquality bugs.
* Update src/libraries/System.Text.Json/src/System/Text/Json/Nodes/JsonValueOfTPrimitive.cs
* Avoid delegate allocation in the JsonObject setter.
Pavel Savara [Sat, 1 Jul 2023 07:50:21 +0000 (09:50 +0200)]
[wasm] Introduce <InvariantTimezone> build flag (#87284)
Andy Ayers [Sat, 1 Jul 2023 01:53:48 +0000 (18:53 -0700)]
JIT: Don't set patchpoints in methods with CORINFO_DEBUG_CODE (#88227)
In #88199 the debugger is overriding some jit flags, but has left other flags set
that confuse the jit: both `TIER0` and `DEBUG_CODE` end uip set so one part of the jit
used logic appropriate for `TIER0` and another part did not.
The JIT ended up creating a patchpoint for OSR in a method with localloc and this
combination is not supported by OSR and lead to a crash.
Harden the jit so if the runtime asks for debuggable code, the jit will never
generate patchpoints.
Tarek Mahmoud Sayed [Sat, 1 Jul 2023 00:52:21 +0000 (17:52 -0700)]
Fix Out of memory with source gen tests (#88266)
Egor Bogatov [Fri, 30 Jun 2023 23:17:52 +0000 (01:17 +0200)]
Clean up Guid.DecodeByte (#88267)
Jeremy Koritzinsky [Fri, 30 Jun 2023 21:34:20 +0000 (14:34 -0700)]
Fix build failures when declaring a COM-style enumerable type (#88263)
Ilona Tomkowicz [Fri, 30 Jun 2023 21:15:28 +0000 (23:15 +0200)]
[browser] Fix missing braces when resolving application culture (#88264)
Regression from https://github.com/dotnet/runtime/pull/86255.
Steve Harter [Fri, 30 Jun 2023 21:04:00 +0000 (16:04 -0500)]
Add IHostedLifecycleService (#87335)
Stephen Toub [Fri, 30 Jun 2023 20:48:31 +0000 (16:48 -0400)]
Delete stale SkipOnTargetFramework attributes (and a few tests) (#88251)
These projects don't run tests on .NET Framework.
Stephen Toub [Fri, 30 Jun 2023 20:48:22 +0000 (16:48 -0400)]
Simplify LINQ Shuffler test helper (#88252)
The seed support is dead code, and the whole thing can be simplified down to a one-line helper.
Egor Bogatov [Fri, 30 Jun 2023 20:20:09 +0000 (22:20 +0200)]
Rename JITAllowOptionalRelocs to JitEnableOptionalRelocs (#88243)
Co-authored-by: Michał Petryka <35800402+MichalPetryka@users.noreply.github.com>
Elinor Fung [Fri, 30 Jun 2023 20:08:27 +0000 (16:08 -0400)]
Remove code in nativeaot event trace files under !FEATURE_NATIVEAOT or FEATURE_DTRACE (#88258)
Jeremy Koritzinsky [Fri, 30 Jun 2023 18:30:22 +0000 (11:30 -0700)]
Add NFloat as an interop intrinsic for the source generator (#88257)
Pent Ploompuu [Fri, 30 Jun 2023 16:39:46 +0000 (19:39 +0300)]
Reduce allocations in Microsoft.Extensions.Configuration (#88211)
Co-authored-by: Günther Foidl <gue@korporal.at>
Trung Nguyen [Fri, 30 Jun 2023 16:05:40 +0000 (02:05 +1000)]
[libs] Fix poll events conversion (#86843)
- Use bitwise and to check for all flags instead of using a switch
statement for poll events flags conversion.
- Convert poll events before comparison in assert statement.
Tanner Gooding [Fri, 30 Jun 2023 15:43:01 +0000 (08:43 -0700)]
Tweak the register order for xarch to better account for callee saved (#88151)
* Tweak the register order for xarch to better account for callee saved
* Have a different register order for leaf methods
* Preference R12 after R13 for 2x the savings
* Allow float/vector constants to use EVEX registers
* Fix Runtime_34937 to expect the right codegen on Unix
* Don't add VAR_ORDER_LEAF and add a comment explaining why we prefer callee trash then callee save
* Simplifying the REG_VAR_ORDER_FLT_EVEX_CALLEE_* defines
* Ensure x86 matches the general x64 ordering
* Ensure that AllBitsSet vector constants correctly handle high simd registers
* Ensure that Runtime_34937 expects the right codegen for Windows x64
* Improve the register ordering more
* Ensure scalar AllBitsSet handles EVEX registers
* Don't change register allocation order for x86
* Minimize the diffs
James Newton-King [Fri, 30 Jun 2023 13:19:14 +0000 (21:19 +0800)]
Improve Microsoft.Extensions.DependencyInjection debugging (#88082)