-
Notifications
You must be signed in to change notification settings - Fork 4.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add support for delegate GDV and method-based vtable GDV #68703
Add support for delegate GDV and method-based vtable GDV #68703
Conversation
Allow method handle histograms in .mibc files and in the PGO text format. Contributes to dotnet#44610.
…ethod-handle-instrumentation
Tagging subscribers to this area: @JulieLeeMSFT Issue DetailsAdd support for instrumenting delegate calls and vtable calls into method handle histograms. Use these histograms to do GDV for delegate calls and also support method-based GDV for vtable calls. For instrumentation we now support class probes at interface call sites, method probes at delegate call sites and both class probes and method probes at vtable call sites. For vtable calls, when turned on, instrumentation produces both histograms as PGO data so that the JIT can later make the choice about what is the best form of guard to use at that site. For guarding, there are some things to take into account. Delegate calls currently (practically) always point to precode, so this prototype is just guarding on For vtable calls the runtime will backpatch the slots when tiering, so the JIT guards the address retrieved from the vtable against an indirection of the slot, which is slightly more expensive than a class-based guard. Currently the instrumentation is enabled conditionally with Simple microbenchmark: public static void Main()
{
long[] nums = Enumerable.Range(0, 100000).Select(i => (long)i).ToArray();
long finalSum = 0;
Stopwatch timer = Stopwatch.StartNew();
for (int i = 0; i < 10000; i++)
{
finalSum += Sum(nums, i => i + i);
}
Console.WriteLine("{0} in {1}ms", finalSum, timer.ElapsedMilliseconds);
}
[MethodImpl(MethodImplOptions.NoInlining)]
private static long Sum(long[] arr, Func<long, long> t)
{
long sum = 0;
foreach (long l in arr)
{
sum += t(l);
}
return sum;
} With Code diff: ; Assembly listing for method Program:Sum(System.Int64[],System.Func`2[Int64,Int64]):long
; Emitting BLENDED_CODE for X64 CPU with AVX - Windows
; Tier-1 compilation
; optimized code
; optimized using profile data
; rsp based frame
-; partially interruptible
-; with PGO: edge weights are valid, and fgCalledCount is 133
+; fully interruptible
+; with PGO: edge weights are valid, and fgCalledCount is 2
+; 0 inlinees with PGO data; 1 single block inlinees; 0 inlinees without PGO data
; Final local variable assignments
;
-; V00 arg0 [V00,T05] ( 3, 3 ) ref -> rcx class-hnd single-def
-; V01 arg1 [V01,T01] ( 4,200408 ) ref -> rsi class-hnd single-def
-; V02 loc0 [V02,T02] ( 4,200408 ) long -> rdi
-; V03 loc1 [V03,T03] ( 3,100205 ) ref -> rbx class-hnd single-def
-; V04 loc2 [V04,T00] ( 5,400813 ) int -> rbp
-;* V05 loc3 [V05 ] ( 0, 0 ) long -> zero-ref
+; V00 arg0 [V00,T07] ( 3, 3 ) ref -> rcx class-hnd single-def
+; V01 arg1 [V01,T01] ( 6,200332 ) ref -> rsi class-hnd single-def
+; V02 loc0 [V02,T02] ( 4,200332 ) long -> rdi
+; V03 loc1 [V03,T05] ( 3,100167 ) ref -> rbx class-hnd single-def
+; V04 loc2 [V04,T00] ( 5,400661 ) int -> rbp
+; V05 loc3 [V05,T03] ( 3,200330 ) long -> rdx
; V06 OutArgs [V06 ] ( 1, 1 ) lclBlk (32) [rsp+00H] "OutgoingArgSpace"
-; V07 cse0 [V07,T04] ( 3,100205 ) int -> r14 "CSE - aggressive"
+; V07 tmp1 [V07,T04] ( 3,200330 ) long -> rdx "guarded devirt return temp"
+;* V08 tmp2 [V08 ] ( 0, 0 ) ref -> zero-ref class-hnd "guarded devirt this exact temp"
+; V09 cse0 [V09,T06] ( 3,100167 ) int -> r14 "CSE - aggressive"
;
; Lcl frame size = 32
G_M53331_IG01: ;; offset=0000H
4156 push r14
57 push rdi
56 push rsi
55 push rbp
53 push rbx
4883EC20 sub rsp, 32
488BF2 mov rsi, rdx
- ;; size=13 bbWeight=1 PerfScore 5.50
+ ;; size=13 bbWeight=0 PerfScore 0.00
G_M53331_IG02: ;; offset=000DH
33FF xor edi, edi
488BD9 mov rbx, rcx
33ED xor ebp, ebp
448B7308 mov r14d, dword ptr [rbx+8]
4585F6 test r14d, r14d
- 7E18 jle SHORT G_M53331_IG04
+ 7E2A jle SHORT G_M53331_IG05
;; size=16 bbWeight=1 PerfScore 4.00
G_M53331_IG03: ;; offset=001DH
- 8BD5 mov edx, ebp
- 488B54D310 mov rdx, qword ptr [rbx+8*rdx+16]
- 488B4E08 mov rcx, gword ptr [rsi+8]
- FF5618 call [rsi+24]System.Func`2[Int64,Int64][System.Int64,System.Int64]:Invoke(long):long:this
- 4803F8 add rdi, rax
+ 8BC5 mov eax, ebp
+ 488B54C310 mov rdx, qword ptr [rbx+8*rax+16]
+ 48B8E8E23905FC7F0000 mov rax, 0x7FFC0539E2E8
+ 48394618 cmp qword ptr [rsi+24], rax
+ 7521 jne SHORT G_M53331_IG07
+ 488B4608 mov rax, gword ptr [rsi+8]
+ 3800 cmp byte ptr [rax], al
+ 4803D2 add rdx, rdx
+ ;; size=32 bbWeight=100165 PerfScore 1176938.75
+G_M53331_IG04: ;; offset=003DH
+ 4803FA add rdi, rdx
FFC5 inc ebp
443BF5 cmp r14d, ebp
- 7FE8 jg SHORT G_M53331_IG03
- ;; size=24 bbWeight=100203 PerfScore 901827.00
-G_M53331_IG04: ;; offset=0035H
+ 7FD6 jg SHORT G_M53331_IG03
+ ;; size=10 bbWeight=100165 PerfScore 175288.75
+G_M53331_IG05: ;; offset=0047H
488BC7 mov rax, rdi
;; size=3 bbWeight=1 PerfScore 0.25
-G_M53331_IG05: ;; offset=0038H
+G_M53331_IG06: ;; offset=004AH
4883C420 add rsp, 32
5B pop rbx
5D pop rbp
5E pop rsi
5F pop rdi
415E pop r14
C3 ret
;; size=11 bbWeight=1 PerfScore 3.75
+G_M53331_IG07: ;; offset=0055H
+ 488B4E08 mov rcx, gword ptr [rsi+8]
+ FF5618 call [rsi+24]System.Func`2[Int64,Int64][System.Int64,System.Int64]:Invoke(long):long:this
+ 488BD0 mov rdx, rax
+ EBDC jmp SHORT G_M53331_IG04
+ ;; size=12 bbWeight=0 PerfScore 0.00
-; Total bytes of code 67, prolog size 10, PerfScore 901847.20, instruction count 29, allocated bytes for code 67 (MethodHash=59ea2fac) for method Program:Sum(System.Int64[],System.Func`2[Int64,Int64]):long
+; Total bytes of code 97, prolog size 13, PerfScore 1352245.20, instruction count 37, allocated bytes for code 97 (MethodHash=59ea2fac) for method Program:Sum(System.Int64[],System.Func`2[Int64,Int64]):long
; ============================================================ There are some unexpected layout issues due to chained GDV, for example in simple examples TODO:
This is currently based on top of #67919 cc @dotnet/jit-contrib
|
This is amazing. In theory could it hoist the check, clone the loop and then have a loop with a simple add, and a loop with the fallback? |
Yes, we're working on that too. See #65206 which includes a link to a prototype that does this (at least for type tests -- should not be hard to make it work for delegate tests as well). |
Very cool to see this! I just skimmed the changes; will take a closer look -- may not get to it until next week. We should push to get the prereq #67919 merged.
Not sure what kind of messes you're seeing, but GDV even without chaining can cause problems, eg #67318.
Would be interesting to see how many sites are type-polymorphic but method-monomorphic.
We should probably work on streamlining the cost of the class/method probes. Simple thing would be to hoist up the check where we decide to update the table; if we're not going to do a table update we don't need to do all the work to figure out what values we'd record. That should help somewhat, especially for the class/method version where those computations are more complex. |
That optimization makes sense. On another note I'm not sure if the combined profile helper is worth much, since from the call site and class we should be able to get the resolved method anyway. So it might just go away in the final version. |
Azure Pipelines successfully started running 2 pipeline(s). |
Impact for select TechEmpower benchmarks: We don't gain that much, which is not that unexpected since we do not use delegates very much in those benchmarks. Here's the data in terms of how many times a probe at a specific IL offset was hit: The same data for type GDV is https://gist.github.com/jakobbotsch/0903b426eddb69fd830662b228d5ca32 |
This makes us support generating and consuming the method handle histograms in the same way as the type handle histograms by compressing them before they are put in the R2R format. Also finish some SPMI support.
This should be ready. I am enabling delegate GDV by default in this PR, while keeping the vtable profiling disabled. The impact on size of PGO data and SPC.dll with delegate GDV is:
For TechEmpower impact on PGO, see the comment above. cc @dotnet/jit-contrib PTAL @AndyAyersMS @EgorBo |
${{ if in(parameters.testGroup, 'gcstress-extra', 'r2r-extra', 'clrinterpreter') }}: | ||
${{ if in(parameters.testGroup, 'gcstress-extra', 'r2r-extra', 'clrinterpreter', 'pgo') }}: | ||
timeoutInMinutes: 510 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I am thinking to keep these timeout changes until we get around to splitting up the jobs, it fixes the problem we are seeing with timeouts and I have not seen any issues with making the timeout this high.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
sounds good.
|
||
// Possibly instrument. Note for OSR+PGO we will instrument when | ||
// optimizing and (currently) won't devirtualize. We may want | ||
// to revisit -- if we can devirtualize we should be able to | ||
// suppress the probe. | ||
// | ||
// We strip BBINSTR from inlinees currently, so we'll only | ||
// do this for the root method calls. | ||
// | ||
if (opts.jitFlags->IsSet(JitFlags::JIT_FLAG_BBINSTR)) | ||
{ | ||
assert(opts.OptimizationDisabled() || opts.IsOSR()); | ||
assert(!compIsForInlining()); | ||
|
||
// During importation, optionally flag this block as one that | ||
// contains calls requiring class profiling. Ideally perhaps | ||
// we'd just keep track of the calls themselves, so we don't | ||
// have to search for them later. | ||
// | ||
if ((call->gtCallType != CT_INDIRECT) && opts.jitFlags->IsSet(JitFlags::JIT_FLAG_BBINSTR) && | ||
!opts.jitFlags->IsSet(JitFlags::JIT_FLAG_PREJIT) && (JitConfig.JitClassProfiling() > 0) && | ||
!isLateDevirtualization) | ||
{ | ||
JITDUMP("\n ... marking [%06u] in " FMT_BB " for class profile instrumentation\n", dspTreeID(call), | ||
compCurBB->bbNum); | ||
ClassProfileCandidateInfo* pInfo = new (this, CMK_Inlining) ClassProfileCandidateInfo; | ||
|
||
// Record some info needed for the class profiling probe. | ||
// | ||
pInfo->ilOffset = ilOffset; | ||
pInfo->probeIndex = info.compClassProbeCount++; | ||
call->gtClassProfileCandidateInfo = pInfo; | ||
|
||
// Flag block as needing scrutiny | ||
// | ||
compCurBB->bbFlags |= BBF_HAS_CLASS_PROFILE; | ||
} | ||
return; | ||
} | ||
|
||
// Bail if optimizations are disabled. | ||
if (opts.OptimizationDisabled()) | ||
{ | ||
return; | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This code was extracted into impConsiderCallProbe
which is called in impImportCall
instead. Unlike impDevirtualizeCall
, impConsiderCallProbe
is called for delegates too.
// TODO-GDV: This can be simplified to just use likelyClasses and | ||
// likelyMethods now that we have multiple candidates here. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'll do this in a follow-up change.
uint32_t likelyClassAttribs = 0; | ||
if (likelyClass != NO_CLASS_HANDLE) | ||
{ | ||
likelyClassAttribs = info.compCompHnd->getClassAttribs(likelyClass); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This stuff is easier to review with whitespace ignored.
Test failure is #68376. superpmi failures are expected due to JIT-EE GUID change. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Overall this looks great.
${{ if in(parameters.testGroup, 'gcstress-extra', 'r2r-extra', 'clrinterpreter') }}: | ||
${{ if in(parameters.testGroup, 'gcstress-extra', 'r2r-extra', 'clrinterpreter', 'pgo') }}: | ||
timeoutInMinutes: 510 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
sounds good.
// Reusing the call target for delegates is more | ||
// complicated. Essentially we need to do the | ||
// transformation done in LowerDelegateInvoke by converting | ||
// the call to CT_INDIRECT and reusing the target address. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
At one point I convinced myself we should do this lowering much earlier anyways, so we had a shot at propagating the method of locally created delegate to the call site.
Not sure if that would help here or not.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
From a code-base standpoint I also think that would be nice, we have many places we need to special case IsDelegateInvoke()
since that looks just like a normal managed function call but isn't.
It's possible that would help here, assuming CSE would then handle the shared call target for the check and cold case. In that case we perhaps wouldn't even need the special handling here. Maybe something to experiment with in a future PR.
Also PTAL @davidwrighton for the non-JIT/SPMI changes. I have bumped the R2R minor version again since we now may see One thing I would like to point attention to is that old versions of dotnet-pgo (from before #67919) fail with an "Unknown PGO type" error if they see this new data. So one needs to use a version of dotnet-pgo that matches the version of the runtime used to produce the data. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@AndyAyersMS I believe I addressed your feedback if you can sign off. I hope to merge this tonight to have it included in the weekend jit-pgo runs. |
* Use the signature types when building ABI info (#70635) * Use signature types when building the ABI info This will allow us to let "mismatched" struct types as call arguments, which in turn is expected to simplify and de-pessimize much of the code working around the need to keep precise handles on struct nodes. Credit to @jakobbotsch for being able to make this change. * Work around the "InferOpSizeAlign" problem Handling it generally requires solving a larger problem of "eeGetArgSizeAlignment" not working well on ARM. (See also the use of "eeGetArgSizeAlignment" in "lvaInitUserArgs") * [wasm] Build WasmAppHost with UseAppHost=false (#70606) * [wasm] Build WasmAppHost with UseAppHost=false - WasmAppHost is included in the WebAssembly.Sdk pack, and doesn't have a platform specific package. - Since, this is being built for x64, the build defaults to using the app host, which means that we get a single binary file that can be run directly. - But this also means that the binary included in the WebAssembly.Sdk platform-agnostic package will target the platform where it was built. - Instead, build with UseAppHost=false, and update the wasm host to use the managed assembly instead. And update the RunCommand to use `dotnet exec WasmAppHost.dll`. * fix * JIT: Optimize expansion of indir cell addresses for CFG (#70491) On ARM64 it is expensive to materialize the address of an indirection cell since that requires multiple instructions. This is particular a problem when CFG is enabled where validation + call uses the indir cell twice. This changes the CFG logic to create a local in this case. We also were forcing the indir cell argument node into the register. This is not ideal for this particular case, so remove this logic and let LSRA deal with it. ARM still needs this logic. Fix #65076 * Handle mis-sized structs in XARCH `PUTARG_STK` codegen (#68611) * Handle "mis-sized" structs in XARCH PUTARG_STK codegen Previously, for structs which were not a multiple of TARGET_POINTER_SIZE in size: 1) On x86, we copied them to a local in morph. This was a small CQ issue. 2) On Unix x64, we didn't handle them at all and could read out of bounds in "genStructPutArgUnroll". This change fixes both issues by properly detecting cases where we need to be careful and updating codegen to use correct load sizes. * Add a test * Improve distribution of CoseHeaderLabel hash codes (#70695) * Implement digest one shots for browser implementation * [wasm] Don't use a fixed listening port for the proxy (#70681) Fixes https://github.com/dotnet/runtime/issues/70670 . * Fix SOS tests on 7.0 (#70677) The wcscpy_s in ClrDataAccess::GetRegisterName was failing with an invalid parameter exception because the prefixLen and regLen didn't include the terminating null wchar. * [wasm] Enable some previously failing WBT/AOT tests on windows (#70595) * [wasm] Enable some previously failing WBT/AOT tests on windows Fixes https://github.com/dotnet/runtime/issues/61725 . * [wasm] WBT: Fix a failing satellite assembly test The test fails because it is trying to find the string `got: こんにちは` in the app output, but it sees `?????`. This is due to xharness not setting the console output to UTF8, and then the tests not doing the same. Instead of that, we can move the check to be in the app itself, thus removing the need for any of this. Fixes https://github.com/dotnet/runtime/issues/58709 . * List other architectures and DOTNET_ROOT environment variables in dotnet --info (#70403) * [mono] Remove dead code. (#70671) A variable initialized to false and never modified means we can remove the variable and any code that depends on that variable not being false. This is a consequence of 15ab9f985ed45feaa619df70d288fbd0acd5c45f, where code that assigned the variable was removed. * Avoid throwing Win32Exception from HTTP authentication (#70474) * Avoid throwing Win32Exception from HTTP authentication When server sends malformed NTLM challenge the NT authentication processing would throw an unexpected Win32Exception from HttpClientHandler.Send[Async] calls. This aligns the behavior to WinHTTP handler where the Unauthorized reply with challenge token is returned back to the client. Similarly, failure to validate the last MIC token in Negotiate scheme could result in Win32Exception. Handle it by throwing HttpRequestException instead. * Make the unit test more resilient * Add trace to Negotiate authentication * Dispose connection instead of draining the response * Remove outdated ActiveIssue * Fix usage of GSS_KRB5_CRED_NO_CI_FLAGS_X (#70447) * Fix build detection of GSS_KRB5_CRED_NO_CI_FLAGS_X * Don't use GSS_KRB5_CRED_NO_CI_FLAGS_X on NTLM * Handle GSS_S_UNAVAILABLE error from gss_set_cred_option (NTLM wrapped in Negotiate) * Make the GSSAPI shim work with krb5 1.13 * Preserve the gss_acquire_cred minor status * Use IndexOf in WebUtility (#70700) The IndexOfHtmlDecodingChars method was iterating character by character looking for either a `&` or a surrogate, but then the slow path if one of those is found doesn't special-case surrogates. So, we can just collapse this to a vectorized `IndexOf('&')`, which makes the fast path of detecting whether there's anything to decode much faster if there's any meaningful amount of input prior to a `&`. (I experimented with also using `IndexOf('&')` in the main routine, but it made cases with lots of entities slower, and so I'm not including that here.) * Hoist the invariants out of multi-level nested loops (#68061) * first working version * Skip check of VN hoisting * Account for duplicate blocks * clean up * wip * isCommaTree && hasExcep * revert lsra changes * Update hoisting condition - Only update if node to be hoisted has side-effects and the sibling that is before that throws exception * Change to BasicBlockList * organize preheaders * update hoistedInCurLoop and hoistedInSiblingLoop * Reverse the loop order * cleanup and jit-format * Revert "Minor fix to display IG01 weight correctly" This reverts commit 757120e863b2da188db2593da1b7142fd1ecf191. * simplify code * Remove m_hoistedVNInSiblingLoop * Add back ResetHoistedInCurLoop Fix igWeight * Remove reversal of loop processing order * jit format * Experimental: Also generate PerfScore: * review feedback * fix the superpmi script * Revert superpmi asmdiffs change * Rename method * Add a comment * Delete GTF_ASSERTION_PROP_LONG (#70521) * Convert exception help context parsing to managed (#70251) * Move help link parsing to managed * Cleanup wtoi and nativeIsDigit * Fix metasig declaration * Guard GetHelpContext under FEATURE_COMINTEROP * Remove definition of GetHelpLink and guard more methods under cominterop * DWORD should be uint * Add help context marshaling test * Adjust marshaling test * Fix #ifdef with ILLink * Fix method signature mismatch * Specify string marshaling * Throwing test should not test successful HRESULT * Implement ISupportErrorInfo * Test interface in InterfaceSupportsErrorInfo * Delete tests for .NET Framework Xsltc.exe tool (#70706) * Use regex matching for Xsltc test baseline files * Delete xsltc.exe tests entirely * Reenable C4244 in repo (#70026) * Reenable 4244 warning for Mono. * Enable compiler errors for gcc/clang. * Fix issues enabling -Werror=implicit-int-conversion * Revert turning on implicit-int-conversion as error. This check was triggering failures on parts of the PAL that require additional scrutiny beyond the scope of this work. * Add SlidingWindow and FixedWindow to RateLimitPartition (#68782) * align GC memory load calculation on Linux with Docker and Kubernetes (#64128) * align memory load calculation in GC with Docker and Kubernetes * pass inactive file field name as a parameter * [wasm] Fix Debug configuration builds (#70683) * Fix cmake error ``` Manually-specified variables were not used by the project: CONFIGURATION_WASM_OPT_FLAGS ``` * Build the interpreter with -O1 on Wasm in Debug configs Otherwise `interp_exec_method` and `generate_code` can easily overflow the stack in some browsers with even a few recursive calls (for example during .cctor initializaiton) * [wasm] Make runtime_is_initialized promise callbacks one-shot (#70694) * [wasm] Make runtime_is_initialized promise callbacks one-shot Throw if runtime_is_initialized_resolve or runtime_is_initialized_reject is called more than once * Add a slightly generalized GuardedPromise<T> object Protects against multiple-resolve, multiple-reject, reject after resolve and resolve after reject. Does not protect against the executor throwing. * Do not use ExecutableMemoryAllocator on 64 bit platforms if default base address usage is requested (#70563) * [wasm][debugger] Implement get bytes from loaded_files using debugger protocol. (#69072) * Implement get bytes from loaded_files using debugger protocol. * fix pdb size == nul * Adressing @radical comments. * Fix build. * fix compilation * Addressing @radical comments. Co-authored-by: Ankit Jain <[email protected]> * Add doc on Unix temporary file security practice (#70585) * Add doc on Unix temporary file security practice * Update unix-tmp.md * Update unix-tmp.md * Add example permissions encoding * Update docs/design/security/unix-tmp.md Co-authored-by: Dan Moseley <[email protected]> * Update unix-tmp.md Co-authored-by: Dan Moseley <[email protected]> * ConfigurationBinder.Bind on virtual properties duplicates elements (#70592) * ConfigurationBinder.Bind on virtual properties duplicates elements * simplify GetAllProperties * return array instead of list * Obsolete AssemblyName.CodeBase, AssemblyName.EscapedCodeBase (#70534) * JIT: Enable JitConsumeProfileForCasts by default (#69869) * Fix AttributePresence in composite mode (#70737) According to my comparative perf measurements composite framework spends about 5M more instructions in the method coreclr.dll!CMiniMdTemplate<CMiniMd>::SearchTableForMultipleRows(CMiniColDef sColumn) I tracked this down to the R2R table AttributePresence which is used to speed up attribute queries against the ECMA metadata model. In composite mode we were putting the table in the image header, not component header, and so the runtime was unable to locate it. This change fixes generation of the table in Crossgen2; I have verified that with this change I no longer see the 5M outlier. Thanks Tomas * Handle mis-sized structs in ARM/64 `PUTARG_SPLIT` codegen (#70249) * Add a test * Fix out-of-bounds loads in ARM/64's "genPutArgSplit" * Split the test out * Refactor how implicit byrefs are morphed (#70243) * Refactor implicit by-refs morphing Move it to "fgMorphLocal". Enabled folding in "LocalAddressVisitor". * Add FEATURE_IMPLICIT_BYREFS And use it in all the necessary places. Resolves the TP regression on Unix x64, turning it into a TP improvement. * Add a zero-diff quirk * Do not change `gtRetClsHnd` in `impNormStructVal` (#70699) * Do not change "gtRetClsHdl" in "impNormStructVal" Doing so leads breaking the proper ABI handling for the call. * Add a test * Support `PUTARG_STK(STRUCT LCL_VAR/LCL_FLD)` on ARM/64 (#70256) * Separate out LowerPutArgStk Call it from "LowerArg" instead of "ContainCheckCallOperands", as it does more than just containment analysis. * Support "PUTARG_STK(STRUCT LCL_VAR/LCL_FLD)" on ARM/64 To test this, transform "OBJ(LCL_VAR|FLD_ADDR)" to "LCL_FLD" on lowering. This additionally simplified the codegen code as it doesn't have to support two representations of the same thing. * Support `PUTARG_STK(STRUCT LCL_VAR/LCL_FLD)` on x86/Unix x64 (#70702) * Add GenTree::GetLayout * x86/Unix x64: lowering * x86/Unix x64: "genConsumePutStructArgStk" * x86/Unix x64: "genStructPutArgUnroll" * x86/Unix x64: codegen asserts * x86: "genStructPutArgPush" * Unix x64: "genStructPutArgPartialRepMovs" * Allow `TYP_STRUCT` `LCL_FLD` on the RHS of block copies (#70633) * Delete the unused "GTF_USE_FLAGS" It used to indicate that a branch operation didn't need to materialize its operand and could just "use flags" instead, but that purpose has long been lost now that we have explicit SETCC nodes in lowering. * Make GTF_DONT_EXTEND a shared flag So that it can be used for "LCL_FLD" as well as "GT_IND". No diffs. * Enable TYP_STRUCT on the RHS * fgMorphBlockOperand * Tweak TYP_STRUCT LCL_FLD costs Model it as two load, like OBJ. Note we could be more precise here, by using the register type of the layout. For now, we defer. * Block CSE Preserve previous behavior to avoid diffs. * Revert "Fix usage of GSS_KRB5_CRED_NO_CI_FLAGS_X (#70447)" (#70747) This reverts commit 84f7cad00ad834c365b5cd1297e1166525146b50. * Add note about backward branch constraints to ECMA-335 augments (#70760) * [NativeAOT] Enabling return address hijacking on ARM64 Windows (#70740) * Enabling return address hijacking on ARM64 Windows * pass right flags to the GcInfoDecoder * Fix value numbering of HWI loads (#70621) * Fix numbering of HWI loads Take into account all operands and describe the exception sets precisely. * Add a test * Recast the check * Small MsQuicStream refactorings (#70433) * Inline state transition helpers Fixes #55437 * Add high level comments for HandleEventReceive and ReadAsync * Updating HWIntrinsicInfo::lookupId to not accelerate Vector256 APIs when AVX2 is unsupported (#70686) * Updating HWIntrinsicINfo::lookupId to not accelerate Vector256 APIs when AVX2 is unsupported * Fixing a check in lookupId to properly negate the condition * Ensure the special-cased EA_32BYTE constants only happen when AVX/AVX2 are supported * Fixing a bad assert in morph * Dsiable CertificateValidationRemoteServer.ConnectWithRevocation_WithCallback on Android (#70768) * NativeAOT Unlock assignability comparison in reflection-free (#70726) * NativeAOT Unlock assignability comparison in reflection-free Closes #69960 * Document support for `IsAssignableFrom` * Document support for `IsInstanceOfType` * Document `IsAssignableTo` * Add suggestion from Michal * Update src/coreclr/nativeaot/System.Private.DisabledReflection/src/Internal/Reflection/RuntimeTypeInfo.cs Co-authored-by: Michal Strehovský <[email protected]> * Remove whitespaces Co-authored-by: Michal Strehovský <[email protected]> * [Mono] Fix C4018 round I (#70417) * First round of change of fixing C4018 * Address part of review feedback * Change the type of idx to unsigned for `effective_table_slow` * Add idx range check after the type was changed * Address review feedback for `class-init.c` * Change the return type of `*table_num_rows*` to `guint32`. Deal with the consequence of return type change of `table_info_get_rows`. Correct the type of a few local variables which store the return of `mono_metadata_token_index`. * Update return type * Address review feedbacks of metadata.c * Fix native crash * Make counter private to for-loop * Address review feedbacks * Address review feedbacks * Fix race condition in LoaderAllocator::CompareExchangeValueInHandle (#70765) The method was not actually using interlocked operation. It made it possible for multiple different values to win. Fixes #70742 * Enable redirection on arm64 (#70769) * Handle HW exceptions on Windows without redirection (#70428) This change modifies the way hardware exceptions are propagated from the vectored exception handler. Until now, runtime was returning from the vectored exception handler with instruction pointer in the context set to an asm helper. That redirected the thread to that helper and we have raised an exception from its call chain. While working on CET support, it was found that propagating exceptions from the vectored exception handler directly works just fine. So this change makes it work that way for all Windows targets. * Predicate for valid Tracker Reference is inverted (#70709) * Predicate for valid Tracker Reference is inverted * Fix test to not inline wrapper creation function. * JIT: Enable addressing modes for gc types (#70749) * JIT: Faster long->int overflow checks (#69865) * Update .net version for asp.net spmi collection script (#70778) * Enable IlcTrimMetadata by default (#70201) Enables more aggressive metadata stripping by default. In this mode, the AOT compiler only generates reflection metadata for artifacts that were deemed reflection-visible by either dataflow analysis, or user inputs. Previously we would consider everything that got generated reflection-visible. * Change the defaults within the compiler (making the conservative mode opt-in). * Add tests for scenarios in https://github.com/dotnet/runtimelab/issues/656 (resolves dotnet/runtimelab#656). * Update tests * Do more marking where it was necessary * Remove GSS_KRB5_CRED_NO_CI_FLAGS_X code (#70772) The support for building with GSS_KRB5_CRED_NO_CI_FLAGS_X was broken for quite some time. Attempts to reenable it failed due to bug in the krb5 GSSAPI implementation resulting in invalid memory accesses. * Make ConnectWithRevocation_StapledOcsp more resilient to being slow * Upgrade SDK to Preview 5 (#70117) Upgrade to released Preview 5 build * Find&replace FastInterlock* with Interlocked* (#70784) FastInterlock* operations were #defined as aliases to regular Interlock* operations. Deleted the FastInterlock* aliases and replaced usage with Interlocked* operations. * JIT: rework optCanonicalizeLoop (#70792) Rework loop canonicalization in anticipation of extending it to handle non-loop edges like those seen in #70802 (details in #70569). The code to fix up flow out of `h` was not needed and has been removed. We now assert upstream that no such fix up will be needed. * Avoid infinite loop in case of heap corruption (#70759) Co-authored-by: Juan Hoyos <[email protected]> Co-authored-by: Noah Falk <[email protected]> * [wasm][debugger] Remove workaround to get pauseOnExceptions setting (#70748) * Add SubstitutionParser that uses XPath Navigator into NativeAOT (#64671) * Add end-to-end test for NTLM/Negotiate authentication against fake server (#70630) * Add end-to-end test for NTLM/Negotiate authentication against fake server * Simplify the test since HttpClientHandler is always SocketsHttpHandler for the test environment * Fix test condition * Remove extra comment * Enable ThreadAbort with CET enabled via the QueueUserAPC2 (#70803) This change makes ThreadAbort work when CET is enabled. Instead of thread redirection, it uses the new user mode APC mechanism to get a thread to a handler and then throws the ThreadAbortException from there if the thread was interrupted at a safe location. I have verified it works using mdbg tests and also by manual testing in the Visual Studio 2022 using a test app that creates an instance of classes with properties containing infinite loop, wait on a lock, wait on a handle, infinite loop inside of a lock and infinite loop in finally. For the testing, I've enabled this separately from the CET so that the missing support for CET in the debugger code doesn't cause troubles. So we could enable this without CET enabled too, but I'd prefer doing that separately. * Test additional NativeAOT Lib testing (#70212) * Test additional NativeAOT Lib testing * more libraries to un nativeaot rolling build * FB * FB2 * only write results file if specified in args * excluding failing tests * oops, missed pull before a forced push * excluding some recently added tests that fail * fix typo with end element * FB * Fix emitDispJumpList for Arm64 (#70542) * fix jitdump * Fix arm build * Another format * Fix regression in runtime-jit-experimental (#70797) The newly-introduced `emitRemoveJumpToNextInst` optimization caused a regression when hot/cold-splitting, where jumps from the last hot instruction to the first cold instruction were erroneously removed. This is fixed by disabling the `isRemovableJmpCandidate` flag for branches between hot/cold sections. On an unrelated note, a JIT dump message has been added to indicate stress-splitting is occurring. * JIT: relax fwd sub restriction on changing class handle (#70587) For HW SIMD types, at least. Fixes #64879. * Use LastIndexOf in ZipArchiveEntry.GetFileName_Unix (#70701) There's not particularly good reason to open-code the loop here. * Update dogfooding instructions (#70833) - Delete note about multilevel lookup. It is not relevant anymore. - Fix nightly feed url to net7 - Replace .NET Core with just .NET * Add icon to dotnet.exe on Windows (#70578) * Add two new stages to prepare for our new custom type marshalling design. (#70598) * Simplify `Environment.IsWindows8OrAbove`. (#70818) * Simplify Environment.IsWindows8OrAbove. Since .NET 5, the regular Windows version detecton code always returns the correct version. * Remove two unused interop files. * Delete redundant aliases for primitive types (#70805) * Add limited support for LocalCertificateSelectionCallback for QUIC (#70716) * Inline state transition helpers Fixes #55437 * Add high level comments for HandleEventReceive and ReadAsync * [QUIC] Call `LocalCertificateSelectionCallback` to get client certificate * Code review feedback * JIT: Optimize range checks for X >> CNS and for short/byte indices (#67141) Co-authored-by: SingleAccretion <[email protected]> * Tighten checks in `areArgumentsContiguous` (#70829) * Tighten checks in "areArgumentsContiguous" * Add a test * Narrow cast fix (#70518) * Added genActualTypeSize. Remove narrow cast if the actual type sizes are the same * Trying to fix build * Fixing build * Removed genActualTypeSize. Using genActualType instead * Added helper function * Comments * Moving back to morph * Removing part of the test * Added fgOptimizeCastOnAssignment * Fixing build * Removing extra bits * Removed fgIsSafeToRemoveIntToIntCastOnAssignment, inlining the implementation, adding extra logic when actual types are not the same * [wasm] Disable WBT test failing on windows (#70808) Issue: https://github.com/dotnet/runtime/issues/70675 * Enable IDE1005 (Delegate invocation can be simplified) (#70522) * Disable long running test (#70843) * [wasm] Enabled WriteToReadOnly tests - emscripten bug got fixed (#70814) Fixes #53021. Enabled tests from #53021 - they are passing now because issue reported by @radekdoulik got fixed: https://github.com/emscripten-core/emscripten/issues/14299. * Use IndexOf{Any} in a few more places (#70176) * Use IndexOf{Any} in a few more places * Address PR feedback * Fix corerun for assemblies with mismatched filename (#70800) corerun gracefully handled filename not matching the assembly name. Handling of this case regressed in #68186. This change is fixing the regression. Fixes #68455 * Fix a few Stream-related issues in System.Net.Http (#70731) * Fix a few Stream-related issues in System.Net.Http * Put back WriteTimeout * Fix tests * Catch (more) mismatched args in `fgMorphArgs` (#70846) * Catch (more) mismatched args in "fgMorphArgs" * Add a test * [coop] fix coop suspend timeout cue card (#70828) the BLOCKING_SUSPEND_REQUESTED state is treated as suspend in full coop mode (the thread keeps running, but by definition it's not allowed to access managed resources and it will self-suspend if it tries to enter GC Unsafe mode by calling a runtime API or managed code). It is bad in hybrid suspend mode (the thread should be preemptively suspended, but we timed out before the signal handler had a chance to run). The corresponding suspension logic in the code is: https://github.com/dotnet/runtime/blob/3fc61ebb562afc327a8fc6de5c82d76e86bf6f5d/src/mono/mono/utils/mono-threads.c#L1149-L1158 * Fixing the handling of Positive NaN in Math.Min for float/double (#70795) * Adding tests validating Positive NaN for Max, MaxMagnitude, Min, and MinMagnitude * Fixing the handling of Positive NaN in Math.Min for float/double * Fixing the Max/Min code comments to use greater and lesser * Adding a code comment clarifying the sign toggling behavior * JIT: break loop canonicalization into two stages (#70809) For a given loop, we need to separate out the true backedge, any non-loop backedges, and any inner loop backedges so that they all target distinct blocks. Otherwise, we may violate assumptions that the loop entry dominates all blocks in the loop and that all backedges that reach top come from within the loop. This seems simplest to do with two rounds of canonicalization, one that moves the non-loop edges, and another that moves the true backedge. Fixes #70802. * Use `Array.Empty()` in `System.Reflection.Metadata`. (#70862) It is now available in all frameworks it targets. * Delete StaticallyLinked property from Microsoft.NETCore.Native.Unix.props (#70854) This option has many issues. Anybody trying to experiment with static linking can add `<LinkerArg Include="-static" />` into the local file. * Restore ArchivingUtils.AttemptSetLastWriteTime (#70617) * Fix GC stress failure in LoaderAllocator::CompareExchangeValueInHandle (#70832) InterlockedCompareExchangeT has to be called on a raw Object* to make the GC stress infrastructure happy. * Small performance cleanups in S.S.Cryptography * ARM64 - Optimize `i % 2` (#70599) * Assign proper VNs to "shared constant" CSE defs (#70852) * Assign proper VNs to shared CSE defs They must be those of the original expression, not the "base" constant CSE creates. * Add a test * Enable IDE0054 (Use compound assignment) (#70564) * Enable IDE0054 (Use compound assignment) * Update src/libraries/System.Data.Common/src/System/Data/Common/StringStorage.cs Co-authored-by: Tanner Gooding <[email protected]> Co-authored-by: Tanner Gooding <[email protected]> * Use new byte[] span optimization in a few more places (#70665) * Use new byte[] span optimization in a few more places Separated out of larger change to use CreateSpan (these don't rely on that). * Address PR feedback * Improve TLS1.3 detection in registry for QUIC (#70730) * Improve TLS1.3 detection in registry for QUIC * Split client and server detection * Code review feedback * Generate method bodies for delegate Invoke methods (#70883) There is a dataflow warning suppression in System.Linq.Expressions that assumes we'll always have an invocable method body for delegate Invoke method. We need one in IL. We don't in native code. Emulate what IL Linker does and generate a method body. This is a size regression. Suppression: https://github.com/dotnet/runtime/blob/3b2883b097a773715ca84056885e0ca1488da36e/src/libraries/System.Linq.Expressions/src/System/Dynamic/Utils/TypeUtils.cs#L906-L912 Fixes #70880. * Avoid crashing on unresolved dependencies (#70871) Fixes #70815. * Enable IDE0020 (Use pattern matching) (#70523) * Enable IDE0020 (Use pattern matching) * Update src/libraries/System.Private.Xml/src/System/Xml/Serialization/XmlSerializationWriter.cs Co-authored-by: Buyaa Namnan <[email protected]> * Update variable naming Co-authored-by: Buyaa Namnan <[email protected]> * Fix failure building two libraries tests (#70881) Microsoft.CSharp was hitting an issue due to a vararg constructor. We don't support varargs. They shouldn't make it into the dependency graph. We are already checking in many places. We were erroneously thinking a vararg constructor with no mandatory arguments is a default constructor. Runtime.InteropServices were crashing because we got a MulticastDelegate type into a codepath that expects a MulticastDelegate descendant. * Use hash/hmac one shots in NTLM (#70857) Co-authored-by: Stephen Toub <[email protected]> * Unpin locals, attempt 2 (#70655) Contributes to #40553 * Replace a few instances of PtrToStructure with more efficient marshalling (#70866) * Fix for timer scheduling happening on the incorrect thread when wasm threading is enabled (#70863) * Add event to capture min/max threads (#70063) Event ThreadPoolMinMaxThreads added. Parameters are: ushort MinWorkerThreads ushort MaxWorkerThreads ushort MinIOCompletionThreads ushort MaxIOCompletionThreads ushort ClrInstanceID It is fired in the ThreadPool constructor and in the SetMinThreads/SetMaxThreads functions. * [wasm] Add support for per-project customization of helix work items (#70461) * [wasm] Add support for per-project customization of helix work items Currently, for sending tests to helix: 1. the project binaries are zipped, as part of the project build 2. then separate helix targets files build that adds helix items for those zip files. - for wasm, we test with multiple 'scenarios' - like v8, browser, and nodejs. - so, 3 helix work items are submitted per zip file - If a test project needs to have any customizations, for example, for testing crypto with a managed implementation, and subtlecrypto, which would require: - passing different arguments to xharness, in case of managed, and subtlecrypto - and this should be done only for the browser case - Currently, this would need this would need to be done in `sendtohelix-wasm.targets` special casing for the test project, and scenario. - We add support for importing `$(ProjectName).helix.targets`, and calling a special target in that to add helix items, as needed. - This targets file can be set in the test project like: ```xml <HelixTargetsFile Condition="'$(TargetOS)' == 'Browser'">wasm.helix.targets</HelixTargetsFile> ``` - it will get deployed next to the zip file, and picked up automatically ```xml <Project> <PropertyGroup> <_CryptoProjectName>System.Security.Cryptography.Tests</_CryptoProjectName> <System_Security_Cryptography_Tests_TargetName Condition="'$(Scenario))' == 'WasmTestOnBrowser'">System_Security_Cryptography_Tests_Targ et</System_Security_Cryptography_Tests_TargetName> </PropertyGroup> <Target Name="System_Security_Cryptography_Tests_Target"> <ItemGroup> <HelixWorkItem Include="$(Scenario)-managed-$(_CryptoProjectName)"> <PayloadArchive>$(TestArchiveTestsDir)$(_CryptoProjectName).zip</PayloadArchive> <Command>$(HelixCommand)</Command> <Timeout>$(_workItemTimeout)</Timeout> </HelixWorkItem> <HelixWorkItem Include="$(Scenario)-subtlecrypto-System.Security.Cryptography.Tests"> <PayloadArchive>$(TestArchiveTestsDir)$(_CryptoProjectName).zip</PayloadArchive> <Command>$(HelixCommand)</Command> <Timeout>$(_workItemTimeout)</Timeout> <PreCommands Condition="'$(OS)' == 'Windows_NT'">set "WasmXHarnessArgs=--web-server-use-cors"</PreCommands> <PreCommands Condition="'$(OS)' != 'Windows_NT'">export "WasmXHarnessArgs=--web-server-use-cors"</PreCommands> </HelixWorkItem> </ItemGroup> </Target> ``` - The targets file *must* have these: - a property named like `System_Security_Cryptography_Tests_TargetName`, for a test project named `System.Security.Cryptography.Tests` - if this property is not set, then the default behavior of adding work items will run - The target should add any items it needs to to `@(HelixWorkItem)`. - Examples of adding these can be seen in `sendtohelix*` project files - Remember that all these properties, and targets get imported into the msbuild *global* namespace, so make sure to use unique names to avoid conflicts with other test projects. Future work: this commit only enables it for wasm/library tests, but it should be easy to extract it out, but needs some testing. * [wasm] Helix: test with, and without subtle crypto * disable non-wasm builds * address review feedback * Address review feedback * Fix typo * Revert "disable non-wasm builds" This reverts commit 7ef99e81f82200189dd3f61eeaf00d6ca4ced6d4. * Update src/libraries/System.Security.Cryptography/tests/wasm.helix.targets Co-authored-by: Eric Erhardt <[email protected]> * Address review feedback * remove debug spew * Change the way helix extension targets are discovered. The new approach: To run a custom project specific target for adding/editing @(HelixWorkItem): - In the project add: `<HelixTargetsFile Condition="'$(TargetOS)' == 'Browser'">wasm.helix.targets</HelixTargetsFile>` - This file gets copied next to the test archive as $(MSBuildProjectName).helix.targets - In this `wasm.helix.targets` file, add to $(HelixExtensionTargets) to run your custom target ```xml <PropertyGroup Condition="'$(IsRunningLibraryTests)' == 'true' and '$(Scenario)' == 'WasmTestOnBrowser'"> <HelixExtensionTargets>$(HelixExtensionTargets);_AddHelixCrypoItems</HelixExtensionTargets> ``` - The extension target will be called after the default items are added to `@(HelixWorkItem)`. - Useful properties to condition on: $(Scenario), $(IsRunningLibraryTests) - And add to, change, or remove from @(HelixWorkItem) Example: ```xml <Target Name="_AddHelixCrypoItems"> <ItemGroup> <!-- remove the existing item --> <HelixWorkItem Remove="@(HelixWorkItem)" Condition="'%(OriginalFileName)' == '$(_CryptoProjectName)'" /> <!-- add two new ones - managed, and subtylecrypto --> <HelixWorkItem Include="$(WorkItemPrefix)managed-$(_CryptoProjectName)"> <PayloadArchive>$(TestArchiveTestsDir)$(_CryptoProjectName).zip</PayloadArchive> <Command>$(HelixCommand)</Command> <Timeout>$(_workItemTimeout)</Timeout> <OriginalFileName>$(_CryptoProjectName)</OriginalFileName> </HelixWorkItem> <HelixWorkItem Include="$(WorkItemPrefix)subtlecrypto-$(_CryptoProjectName)"> <PayloadArchive>$(TestArchiveTestsDir)$(_CryptoProjectName).zip</PayloadArchive> <Command>$(HelixCommand)</Command> <Timeout>$(_workItemTimeout)</Timeout> <OriginalFileName>$(_CryptoProjectName)</OriginalFileName> <PreCommands Condition="'$(OS)' == 'Windows_NT'">set "WasmXHarnessArgs=%WasmXHarnessArgs% --web-server-use-cop"</PreCommands> <PreCommands Condition="'$(OS)' != 'Windows_NT'">export "WasmXHarnessArgs=$WasmXHarnessArgs --web-server-use-cop"</PreCommands> </HelixWorkItem> <_CryptoHelixItem Include="@(HelixWorkItem)" Condition="$([System.String]::new('%(HelixWorkItem.Identity)').EndsWith('-$(_CryptoProjectName)'))" /> </ItemGroup> <Error Text="Something went wrong. Expected to have only two work items for $(_CryptoProjectName). But got @(_CryptoHelixItem)" Condition="@(_CryptoHelixItem->Count()) != 2" /> </Target> ``` * cleanup * Move WBT specific stuff into the target too. This will make it simpler to move it off into a targets file later * fix build * fix libtests * fix wbt * [wasm] Bump helix timeout to 90mins for debugger tests on windows Co-authored-by: Eric Erhardt <[email protected]> * Fix bug in Tar preventing extraction of hardlinks or entries starting with `.\` (#70853) * Add PlatformDetection.SupportsHardLinkCreation property. * Fix how paths are combined/joined and sanitized on extraction, to ensure paths with redundant segments get properly handled. * Add tests that verify archives with entries whose paths start with .\, including the root folder itself. * Re-enable the hardlink test, condition it to not run if platform does not support extraction of hardlinks. * Remove unnecessary test - This same code is already being tested by TarReader_ExtractToFile_Tests.ExtractEntriesWithSlashDotPrefix * Reuse test code that retrieves memory stream. * Bump test data package version * Add missing typeof(PlatformDetection) in ConditionalFact Co-authored-by: carlossanlop <[email protected]> * Remove #define Sleep (#70868) * Remove #define Sleep * Fix ICorDebugFunction2::GetVersionNumber for arm32. (#69901) * Replace the remaining uses of Marshal.PtrToStructure in core networking (#70900) * Rename ICMP interop to match Windows SDK names * Add support for delegate GDV and method-based vtable GDV (#68703) Add support for instrumenting delegate calls and vtable calls into method handle histograms. Use these histograms to do GDV for delegate calls and also support method-based GDV for vtable calls. For instrumentation we now support class probes at interface call sites, method probes at delegate call sites and both class probes and method probes at vtable call sites. For vtable calls, when turned on, instrumentation produces both histograms as PGO data so that the JIT can later make the choice about what is the best form of guard to use at that site. For guarding, there are some things to take into account. Delegate calls currently (practically) always point to precode, so this PR is just guarding on getFunctionFixedEntryPoint which returns the precode address, and this is generally quite cheap (same cost as class-based GDV). That's the case for delegates pointing to instance methods anyway, this PR does not support static methods yet -- those will be more expensive. For vtable calls the runtime will backpatch the slots when tiering, so the JIT guards the address retrieved from the vtable against an indirection of the slot, which is slightly more expensive than a class-based guard. Currently the instrumentation is enabled conditionally with COMPlus_JitDelegateProfiling=1 (for delegates) and COMPlus_JitVTableProfiling=1 (for vtable calls). Currently delegate profiling is turned on by default while vtable profiling is off by default. * Use ArgumentNullException.ThrowIfNull in a few more places (#70897) * Append the function pointer type (#70923) The issue here was the recursion into PrettyPrintSigWorkerInternal that would reset the buffer. Now, we pass in a new buffer and append it when we return. * [main] Update dependencies from dotnet/runtime dotnet/icu dotnet/xharness dotnet/runtime-assets dotnet/emsdk dotnet/roslyn-analyzers (#70476) * Update dependencies from https://github.com/dotnet/runtime-assets build 20220608.1 Microsoft.DotNet.CilStrip.Sources , System.ComponentModel.TypeConverter.TestData , System.Drawing.Common.TestData , System.Formats.Tar.TestData , System.IO.Compression.TestData , System.IO.Packaging.TestData , System.Net.TestData , System.Private.Runtime.UnicodeData , System.Runtime.Numerics.TestData , System.Runtime.TimeZoneData , System.Security.Cryptography.X509Certificates.TestData , System.Text.RegularExpressions.TestData , System.Windows.Extensions.TestData From Version 7.0.0-beta.22281.1 -> To Version 7.0.0-beta.22308.1 * Update dependencies from https://github.com/dotnet/emsdk build 20220608.2 Microsoft.NET.Workload.Emscripten.Manifest-7.0.100 From Version 7.0.0-preview.6.22281.1 -> To Version 7.0.0-preview.6.22308.2 * Update dependencies from https://github.com/dotnet/xharness build 20220610.1 Microsoft.DotNet.XHarness.CLI , Microsoft.DotNet.XHarness.TestRunners.Common , Microsoft.DotNet.XHarness.TestRunners.Xunit From Version 1.0.0-prerelease.22305.1 -> To Version 1.0.0-prerelease.22310.1 * Revert "Update dependencies from https://github.com/dotnet/emsdk build 20220608.2" This reverts commit bbb4e156ddbad8a2cb7b604246214272085b6622. * Update dependencies from https://github.com/dotnet/icu build 20220609.1 Microsoft.NETCore.Runtime.ICU.Transport From Version 7.0.0-preview.6.22306.1 -> To Version 7.0.0-preview.6.22309.1 * Update dependencies from https://github.com/dotnet/emsdk build 20220608.2 Microsoft.NET.Workload.Emscripten.Manifest-7.0.100 From Version 7.0.0-preview.6.22281.1 -> To Version 7.0.0-preview.6.22308.2 * Update dependencies from https://github.com/dotnet/runtime-assets build 20220610.1 Microsoft.DotNet.CilStrip.Sources , System.ComponentModel.TypeConverter.TestData , System.Drawing.Common.TestData , System.Formats.Tar.TestData , System.IO.Compression.TestData , System.IO.Packaging.TestData , System.Net.TestData , System.Private.Runtime.UnicodeData , System.Runtime.Numerics.TestData , System.Runtime.TimeZoneData , System.Security.Cryptography.X509Certificates.TestData , System.Text.RegularExpressions.TestData , System.Windows.Extensions.TestData From Version 7.0.0-beta.22281.1 -> To Version 7.0.0-beta.22310.1 * Update dependencies from https://github.com/dotnet/roslyn-analyzers build 20220610.1 Microsoft.CodeAnalysis.NetAnalyzers From Version 7.0.0-preview1.22302.1 -> To Version 7.0.0-preview1.22310.1 * Update dependencies from https://github.com/dotnet/runtime build 20220612.5 Microsoft.NET.Sdk.IL , Microsoft.NETCore.App.Runtime.win-x64 , Microsoft.NETCore.DotNetHost , Microsoft.NETCore.DotNetHostPolicy , Microsoft.NETCore.ILAsm , runtime.native.System.IO.Ports , System.Text.Json From Version 7.0.0-preview.6.22305.4 -> To Version 7.0.0-preview.6.22312.5 * Update dependencies from https://github.com/dotnet/icu build 20220613.1 Microsoft.NETCore.Runtime.ICU.Transport From Version 7.0.0-preview.6.22306.1 -> To Version 7.0.0-preview.6.22313.1 * Update dependencies from https://github.com/dotnet/xharness build 20220613.1 Microsoft.DotNet.XHarness.CLI , Microsoft.DotNet.XHarness.TestRunners.Common , Microsoft.DotNet.XHarness.TestRunners.Xunit From Version 1.0.0-prerelease.22310.1 -> To Version 1.0.0-prerelease.22313.1 * Update dependencies from https://github.com/dotnet/runtime-assets build 20220613.1 Microsoft.DotNet.CilStrip.Sources , System.ComponentModel.TypeConverter.TestData , System.Drawing.Common.TestData , System.Formats.Tar.TestData , System.IO.Compression.TestData , System.IO.Packaging.TestData , System.Net.TestData , System.Private.Runtime.UnicodeData , System.Runtime.Numerics.TestData , System.Runtime.TimeZoneData , System.Security.Cryptography.X509Certificates.TestData , System.Text.RegularExpressions.TestData , System.Windows.Extensions.TestData From Version 7.0.0-beta.22281.1 -> To Version 7.0.0-beta.22313.1 * Update dependencies from https://github.com/dotnet/emsdk build 20220613.1 Microsoft.NET.Workload.Emscripten.Manifest-7.0.100 From Version 7.0.0-preview.6.22281.1 -> To Version 7.0.0-preview.6.22313.1 * [wasm] Wasm.Build.Tests: Disable strict version checks for emcc Due to the way we have to update `dotnet/emsdk`, and get the update PRs in `runtime`, the emsdk version can be mismatched. For example, if `dotnet/emsdk` has already updated to `3.1.12`, but `dotnet/runtime` is still on `3.1.7`. This is an expected scenario while working on updating to a newer emscripten. The version mismatch will still show up, but as a warning now. * Update dependencies from https://github.com/dotnet/xharness build 20220614.1 Microsoft.DotNet.XHarness.CLI , Microsoft.DotNet.XHarness.TestRunners.Common , Microsoft.DotNet.XHarness.TestRunners.Xunit From Version 1.0.0-prerelease.22310.1 -> To Version 1.0.0-prerelease.22314.1 * [wasm] Wasm.Build.Tests: Update Skiasharp reference The skiasharp dependent tests fail with: `error : undefined symbol: _ZNKSt3__220__vector_base_commonILb1EE20__throw_length_errorEv`. Update the reference to the latest package which added a binary compiled for `3.1.7`. * [wasm] Workaround a test failure in WBT Currently, there is a mismatch between emsdk versions with the workload emscripten packs using `3.1.12`, and the runtime being built with `3.1.7`. And that is causing `Wasm.Build.Tests.NativeLibraryTests.ProjectUsingSkiaSharp` to fail with: ``` EXEC : error : undefined symbol: _ZNKSt3__220__vector_base_commonILb1EE20__throw_length_errorEv (referenced by top-level compiled C/C++ code) [/datadisks/disk1/work/B1420981/w/ACC90933/e/blz_nativeref_aot_Debug/blz_nativeref_aot_Debug.csproj] EXEC : warning : Link with `-sLLD_REPORT_UNDEFINED` to get more information on undefined symbols [/datadisks/disk1/work/B1420981/w/ACC90933/e/blz_nativeref_aot_Debug/blz_nativeref_aot_Debug.csproj] EXEC : warning : To disable errors for undefined symbols use `-sERROR_ON_UNDEFINED_SYMBOLS=0` [/datadisks/disk1/work/B1420981/w/ACC90933/e/blz_nativeref_aot_Debug/blz_nativeref_aot_Debug.csproj] EXEC : warning : __ZNKSt3__220__vector_base_commonILb1EE20__throw_length_errorEv may need to be added to EXPORTED_FUNCTIONS if it arrives from a system library [/datadisks/disk1/work/B1420981/w/ACC90933/e/blz_nativeref_aot_Debug/blz_nativeref_aot_Debug.csproj] EXEC : error : undefined symbol: _ZNKSt3__221__basic_string_commonILb1EE20__throw_length_errorEv (referenced by top-level compiled C/C++ code) [/datadisks/disk1/work/B1420981/w/ACC90933/e/blz_nativeref_aot_Debug/blz_nativeref_aot_Debug.csproj] EXEC : warning : __ZNKSt3__221__basic_string_commonILb1EE20__throw_length_errorEv may need to be added to EXPORTED_FUNCTIONS if it arrives from a system library [/datadisks/disk1/work/B1420981/w/ACC90933/e/blz_nativeref_aot_Debug/blz_nativeref_aot_Debug.csproj] EXEC : error : Aborting compilation due to previous errors [/datadisks/disk1/work/B1420981/w/ACC90933/e/blz_nativeref_aot_Debug/blz_nativeref_aot_Debug.csproj] emcc : error : '/datadisks/disk1/work/B1420981/w/ACC90933/e/dotnet-workload/packs/Microsoft.NET.Runtime.Emscripten.3.1.12.Node.linux-x64/7.0.0-preview.6.22313.1/tools/bin/node /datadisks/disk1/work/B1420981/w/ACC90933/e/dotnet-workload/packs/Microsoft.NET.Runtime.Emscripten.3.1.12.Sdk.linux-x64/7.0.0-preview.6.22313.1/tools/emscripten/src/compiler.js /datadisks/disk1/work/B1420981/t/tmp44tn7y2d.json' failed (returned 1) [/datadisks/disk1/work/B1420981/w/ACC90933/e/blz_nativeref_aot_Debug/blz_nativeref_aot_Debug.csproj] ``` This commit adds a temporary workaround to ignore the undefined symbols for this specific test. And this should be removed when runtime is updated to `3.1.12` . Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com> Co-authored-by: Alexander Köplinger <[email protected]> Co-authored-by: Larry Ewing <[email protected]> Co-authored-by: Ankit Jain <[email protected]> * Harden for null byrefs (#70317) * Remove enum_flag_Unrestored usage in boxing stubs. * Add AND instruction (21h) to x86 decoder. * Update mono for null ref in interpreter paths. * Disable test on llvmfullaot and wasm * Handle null destination for intrinsics. * Use u8 in a few more places (#70894) * Use u8 in a few more places * A few more u8s * Enable lib tests that used to fail with GC issues (#70831) * enable lib tests that used to fail with GC issues * excluding linq parallel test since that doesn't build * Threading.Thread tests have issues * Excluding new test failures * Remove some dead code / branches (#70146) * Remove some dead code / branches Some signal amidst the noise in an lgtm.com report. * Address PR feedback * superpmi.py: Add `-jitoption` for asmdiffs (#70939) The new `-jitoption` option passes the argument options to both baseline and diff compilers. This is a convenience option: there already is `-base_jit_option` and `-diff_jit_option` to specify passing options to either baseline or diff. The name of the option is the same as that used for `replay`. * EventLogException is missing the original win32 error code (#70629) * Enable IDE0071 (Simplify interpolation) (#70918) * Enable IDE0100 (Remove redundant equality) (#70896) * Enable IDE0065 (Misplaced using directive) (#70919) * Optimization for full range checks (#70145) (#70222) * Update C# compiler (#70947) Brings in support for required members * Allow ValueListBuilder to work with empty scratch spans (#70917) * Enable IDE0030 (Use coalesce expression) (#70948) * Do not reuse implicit byrefs for by-value args (#70955) * Do not reuse implicit byrefs for by-value args * Add a test * Enable `TYP_STRUCT` `LCL_VAR/LCL_FLD` call args on Windows x64 (#70777) * Fix forward sub * Enable folding in local morph * Morph: TYP_STRUCT LCL_FLD * Make SuperPMI more explicit about JIT in error messages (#70938) This makes it easier to figure out which JIT failed in asmdiffs scenarios. e.g., ``` [10:40:50] ERROR: Method 3673 of size 3107 failed to load and compile correctly by JIT2 (C:\gh\runtime\artifacts\tests\coreclr\windows.x64.Checked\Tests\Core_Root\clrjit_universal_arm64_x64.dll). [10:40:50] ERROR: Method 3673 of size 3107 failed to load and compile correctly by JIT1 (C:\gh\runtime2\artifacts\tests\coreclr\windows.x64.Checked\Tests\Core_Root\clrjit_universal_arm64_x64.dll). ``` * Fix illumos-x64 build (#70956) * Disable W^X on macOS under Rosetta emulation (#70912) * Disable W^X on macOS under Rosetta emulation Apple has informed us that double mapping doesn't work properly on Rosetta emulation. This change disables W^X if Rosetta is detected. * Reflect PR feedback * Add support for randomized guard stack cookies (#70806) The compiler was generating GS cookies, but the cookie was always an ASCII encoding of "Hi, mom!". Make the cookie properly per-process randomized. The randomization algorithm is same as in the CoreCLR VM. Fixes #70071. * Fixes #64159 - JsonSerializerContext source generation fails with keyword identifiers (#66876) * Fixes #64159 - initial implementation * Tidy up, and add test for ignored reserved-keyword-named property * PR feedback - use same approach as logging generator * PR feedback * Update src/libraries/System.Text.Json/gen/JsonSourceGenerator.Parser.cs Co-authored-by: Adeel Mujahid <[email protected]> * PR feedback - rename ExactNameSpecifiedInSourceCode * PR feedback - use extracted (and renamed) local variable * Remove commented code * Added `IsVerbatimName` as extension method * Support fields with verbatim names (@) * Use alternative method for checking for verbatim names * Uses `SyntaxFacts` to determine if escaping is needed * Remove extension method * Modified source generator test to include a verbatim field * Minor typo Co-authored-by: Adeel Mujahid <[email protected]> * [main] Update dependencies from dotnet/arcade (#70662) Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com> * Enable IDE0031 (Use null propagation) (#70965) * Fix nullable annotations on MailAddress.TryCreate (#70943) * [QUIC] Removes abstract providers and mock from S.N.Quic (#70421) * Removed abstract providers and mock from S.N.Quic * Removed provider and replaced with statics on Listener and Connection * Added assert * Stop wrapping SIMD nodes in OBJs (#70888) * Upstream coreclr and pal changes to support powerpc (ppc64le) architecture (#69105) * coreclr pal layer chanegs. * Updated the system arch for power architecture. * Fixed the failing PAL SXS exception_handling test case * replaced gint with gint64 to avoid overflow and this has fixed the segmentation fault issue. * coreclr pal layer chanegs. * Fixed the failing PAL SXS exception_handling test case * coreclr pal layer chanegs. * Updated the system arch for power architecture. * Fixed the failing PAL SXS exception_handling test case * replaced gint with gint64 to avoid overflow and this has fixed the segmentation fault issue. * coreclr pal layer chanegs. * Fixed the failing PAL SXS exception_handling test case * Removing intsafe.h file as main does not have this file now.(Already removed in commit 27195f670937c7e21ab68a806396f9d17c57231a) Co-authored-by: Alhad Deshpande <[email protected]> * Collections contain stale object IDs (#70924) It looks like the collections contain stale ObjectID values. These collection should be purged between GCs. * Minor refactoring for `UpgradeToRegexGeneratorAnalyzer` (#70767) * Fix arm64 funclet frame type 5 (#70922) * Fix arm64 funclet frame type 5 * Add code to stress arm64 funclet frame type 5 This can be forced using `COMPlus_JitSaveFpLrWithCalleeSavedRegisters=3`, or will be enabled under stress. * Add a regression test * Fix funclet frame type 5 alignment calculation * Fix a couple bugs; improve documentation 1. Improve the frame shape pictures, especially for arm64 funclet frames. Include the MonitorAcquired element and OSR pad. 2. Fix bug for `SP_to_PSP_slot_delta` for funclet frame type 4 for OSR: don't include osrPad. 3. Fix bug with funclet frame type 5 for `SP_to_FPLR_save_delta` using wrong aligned func size; would be incorrect if one and two SP adjustment full frame aligned sizes are different (e.g., if two SP adjustment required two alignment slots and one SP adjustment thus required none). * Fix default log color behavior when console is redirected (#70504) * JIT: Model string literal objects as invariant and non-GLOB_REF (#70986) * Model string literals as invariant/non GLOB_REF * String literals are never null * Add support for cross module inlining and cross module generic compilation to Crossgen2 (#68919) * Add support for cross module inlining and cross module generic compilation to Crossgen2 - Refactor Module into ModuleBase and Module - The goal is to have allow a subset version of Module which can only hold refs, this is to be used by the manifest module in an R2R image to allow for version resilient cross module references. - Update handling of ModuleBase so that its used everywhere that tokens are parsed from R2R - Remove ENCODE_MODULE_ID_FOR_STATICS and ENCODE_ACTIVE_DEPENDENCY - These were only used for NGEN, and conflict with easy impelmentation for the ModuleBase concept - Remove locking in ENCODE_STRING_HANDLE processing, and tweak comments. Comments applied to the removed ngen based code, and the lock was only necessary for the old ngen thing. - Adjust ComputeLoaderModuleWorker for locating loader module - Follow comment more accurately, to avoid putting every generic into its definition module. This will make R2R function lookup able to find compiled instantiations in some cases. This may be what we want long term, it may not. - Remove MemberRefToDesc map and replace with LookupMap like the other token types. We no longer make use of the hot table, so this is more efficient - Also reduces complexity of implementation of ModuleBase - Build fixup to describe a single method as a standalone blob of data - There are parallel implementations in Crossgen2 and in the runtime - They produce binary identical output - Basic R2RDump support for new fixup - Adjust module indices used within the R2R format to support a module index which refers to the R2R manifest metadata. This requires bumping the R2R version to 6.2 - Add a module index between the set of assembly refs in the index 0 module and the set of assembly refs in the R2R manifest metadata - Adjust compilation dependency rules to include a few critical AsyncStateMachineBox methods - Remove PEImage handling of native metadata which was duplicative - Do not enable any more devirtualization than was already in use, even in the cross module compilation scenario. In particular, do not enable devirtualization of methods where the decl method isn't within the version bubble, even if the decl method could be represented with a cross-module reference token. (This could be fixed, but is out of scope for this initial investigation) Make the compilation deterministic in this new model, even though we are generating new tokens on demand - Implement this by detecting when we need new tokens during a compile, and recompiling with new tokens when necessary - This may result in compiling the same code as much as twice Compile the right set of methods with cross module inlining enabled - Add support for compiling the called virtual methods on generic types - This catches the List<T> and Dictionary<TKey,TValue> scenarios - Add command line switches to enable/disable the new behavior - By default the new behavior is not enabled * Implement Vector128 version of System.Buffers.Text.Base64 DecodeFromUtf8 and EncodeToUtf8 (#70654) * Implement Vector128 version of System.Buffers.Text.Base64.DecodeFromUtf8 Rework the SS3 into a Vector128 version, and add Arm64 support. * SSE3 improvements * Remove superfluous bitwise And * Add comment to SimdShuffle * Inline SimdShuffle * Implement Vector128 version of System.Buffers.Text.Base64.EncodeToUtf8 * Ensure masking on SSE3 Change-Id: I319f94cfc51d0542ae4eb11a8d48b3eb8180553f CustomizedGitHooks: yes * Restore asserts and move zero inside the loop * Neater C# code Change-Id: I2cbe14f4228f8035e7d213b5b58815c4eee35563 CustomizedGitHooks: yes * Make SimdShuffle consistent across X64 and Arm64 * Better looking multiply * Use `gtEffectiveVal` for `GT_ADD` op1 in `optCreateAssertion` (#70228) * Initial work for comma hoisting in cse * Formatting * Adding more ops to comma hoisting * Set regnum * Update optcse.cpp * Using effectiveval instead * Convert fallback path of GetCommandLineArgs to managed (#70608) * P/Invoke definition * Use P/Invoke in managed code * Update managed call site * Add test using private reflection * Native command line should be superset of managed Co-authored-by: Jan Kotas <[email protected]> * [wasm] Enabled WasmTestOnBrowser run on Windows for Browser. (#70890) * Enabled tests - no timeout with new ems. Keeping disabled tests failing from other reasons. * Added @radical's suggestion. * Test CI- enable test that are throwing in Debug. * [wasm][debugger] Fix side effect on Firefox of getting bytes from loaded_files using debugger protocol (#70990) * Fix 70983 * fix typo * Fix compilation * Add AddSystemd() and AddWindowsService() IServiceCollection extension methods (#68580) * Add AddSystemd() IServiceCollection extension method * Add AddWindowsService() IServiceCollection extension method * Don't default to CWD if in C:\Windows\system32 - instead, when CWD is C:\Windows\system32 Hosting will use AppContext.BaseDirectory. This way Windows apps and services that are launched will work by default. HostApplicationBuilder.ContentRootPath can't be changed after construction, so setting it to a workable default for Windows apps. Co-authored-by: Eric Erhardt <[email protected]> Co-authored-by: Martin Costello <[email protected]> * Use RemoteExecutor * Update src/libraries/Microsoft.Extensions.Hosting/tests/UnitTests/HostTests.cs * Skip test on Windows nano server * Respond to PR feedback Co-authored-by: Eric Erhardt <[email protected]> Co-authored-by: Martin Costello <[email protected]> * [main] Update dependencies from dotnet/runtime dotnet/xharness dotnet/icu dotnet/emsdk (#70991) * Update dependencies from https://github.com/dotnet/runtime build 20220619.5 Microsoft.NET.Sdk.IL , Microsoft.NETCore.App.Runtime.win-x64 , Microsoft.NETCore.DotNetHost , Microsoft.NETCore.DotNetHostPolicy , Microsoft.NETCore.ILAsm , runtime.native.System.IO.Ports , System.Text.Json From Version 7.0.0-preview.6.22312.5 -> To Version 7.0.0-preview.6.22319.5 * Update dependencies from https://github.com/dotnet/xharness build 20220620.1 Microsoft.DotNet.XHarness.CLI , Microsoft.DotNet.XHarness.TestRunners.Common , Microsoft.DotNet.XHarness.TestRunners.Xunit From Version 1.0.0-prerelease.22314.1 -> To Version 1.0.0-prerelease.22320.1 * Update dependencies from https://github.com/dotnet/icu build 20220620.2 Microsoft.NETCore.Runtime.ICU.Transport From Version 7.0.0-preview.6.22313.1 -> To Version 7.0.0-preview.6.22320.2 * Update dependencies from https://github.com/dotnet/emsdk build 20220620.1 Microsoft.NET.Workload.Emscripten.Manifest-7.0.100 From Version 7.0.0-preview.6.22313.1 -> To Version 7.0.0-preview.6.22320.1 Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com> * Windows/Arm64: Use 8.1 atomic instructions if they are available (#70921) * Use Fast compareexchange, acquire/release * working windows version * remove printf * some more #ifdef * fix some #ifdef * optimize EnterObjMonitorHelperSpin * Remove FastInterlockedCompareExchange64 for now * Add ZipArchiveEntry.IsEncrypted for password-protection detection (#70036) * Add ZipArchiveEntry.IsEncrypted for password-protection detection * Apply suggestions from code review Co-authored-by: David Cantú <[email protected]> * Run GenerateReferenceAssemblySource for System.IO.Compression * Revert WriteAsync default parameter value removal * Update DeflateStream.WriteAsync to include the default param value * Revert unrelated change from GenerateReferenceAssemblySource * Revert unrelated change after syncing with GenerateReferenceAssemblySource Co-authored-by: David Cantú <[email protected]> * Exposing IRootFunctions.Hypot and IRootFunctions.Root (#71010) * Exposing IRootFunctions.Hypot and IRootFunctions.Root * Adding tests for IRootFunctions.Hypot and IRootFunctions.Root * Add TarEntry conversion constructors (#70325) * ref: Conversion constructors * src: Conversion constructors * tests: Conversion constructors * [wasm] Wasm.Build.Tests: disable tests failing on CI (#71020) * [wasm] Wasm.Build.Tests: disable tests failing on CI .. due to being oomkill'ed. * [wasm] Don't fail the job when firefox tests fail, as they are known to be unstable * Remove allocation in ImmutableArray Builder Sort (#70850) * Remove allocation in ImmutableArray Builder Sort `ImmutableArray<T>.Builder.Sort(Comparison<T> comparison)` allocates a new instance of `Comparer<T>` using `comparison`. Then `ArraySort` allocates a new delegate from `comparer.Compare`. Both allocations can be eliminated by calling `Array.Sort<T>(T[] array, Comparison<T> comparison)` directly. * Sort only valid range of array * Use MemoryExtensions.Sort where available. * Update src/libraries/System.Collections.Immutable/src/System/Collections/Immutable/ImmutableArray_1.Builder.cs Fix spelling mistake. Co-authored-by: Eirik Tsarpalis <[email protected]> Co-authored-by: Eirik Tsarpalis <[email protected]> * [Android] Make sure AdditionalTimeZoneChecks trait is added to xunit-excludes (#70974) The trait is supposed to be included in System.Runtime tests by default, but wasn't because the _withoutCategories property was in tests.props. As a result, the property was resolved before the .csproj properties. This fix moves the property definition to tests.targets. Fixes #70482 * Use LastIndexOfAny in GetFileName_Windows (#71032) * Avoid unnecessary allocations in SanitizeEntryFilePath (#71034) * Fix Ordinal Ignore Case string compare (#71022) * Move allocation of command line argument array to C# (#71021) Also, avoid redundant allocation of managed string for each command line argument. * Add markdown readme for System.Reflection.MetadataLoadContext (#70610) * Create README.md * set red zone size 512 for powerpc64. (#70885) * Ensure consistent reflectability for generic methods (#70977) If we have a method body for `SomeMethod<Foo>` and `SomeMethod<T>` was decided to be reflection-visible, ensure `SomeMethod<Foo>` is also reflection-visible. * Implement NegotiateAuthentication API (#70720) * WIP: Add implementation of NegotiateAuthentication Switch System.Net.Http to use NegotiateAuthentication Fix IsCompleted in managed NTLM implementation * WIP: Update error code mapping * Spanify input of GetOutgoingBlob * Update comments * Move NegotiateStreamPal.Encrypt/Decrypt to shared sources. Unix implementation already had them and they get trimmed anyway. * Revert accidental change * Build fixes. * Fix error handling condition * Update error mapping based on HttpListener usage. * WIP: HttpListener test * Move workaround from HttpListener to low-level SSPI code * Fix build * Clean up * Revert "WIP: HttpListener test" This…
I missed this change. This is a big deal 😀 ! |
Add support for instrumenting delegate calls and vtable calls into method handle histograms. Use these histograms to do GDV for delegate calls and also support method-based GDV for vtable calls.
For instrumentation we now support class probes at interface call sites, method probes at delegate call sites and both class probes and method probes at vtable call sites. For vtable calls, when turned on, instrumentation produces both histograms as PGO data so that the JIT can later make the choice about what is the best form of guard to use at that site.
For guarding, there are some things to take into account. Delegate calls currently (practically) always point to precode, so this PR is just guarding on
getFunctionFixedEntryPoint
which returns the precode address, and this is generally quite cheap (same cost as class-based GDV). That's the case for delegates pointing to instance methods anyway, this PR does not support static methods yet -- those will be more expensive.For vtable calls the runtime will backpatch the slots when tiering, so the JIT guards the address retrieved from the vtable against an indirection of the slot, which is slightly more expensive than a class-based guard.
Currently the instrumentation is enabled conditionally with
COMPlus_JitDelegateProfiling=1
(for delegates) andCOMPlus_JitVTableProfiling=1
(for vtable calls). Currently delegate profiling is turned on by default while vtable profiling is off by default.Simple microbenchmark:
TODO:
Support delegates pointing to value type instance methods(Follow-up)Support delegates pointing to static functions(Follow-up)R2R support(Follow-up)Contributes to #44610
cc @dotnet/jit-contrib