Age | Commit message (Collapse) | Author | Files | Lines |
|
cgroup usage is used to trigger oom kills. It includes rss and file cache
of the cgroup.
The implementation was only using the process rss to determine memory load.
This is less than the cgroup usage and leads to oom kills due to GC not
being triggered soon enough.
|
|
|
|
See dotnet/coreclr#18912. This test would always fail once tiered compilation is enabled by default in CoreFX test runs in the CoreCLR repo.
|
|
* Respect Windows script argument to skip package building.
* Bring build.sh logic closer to build.cmd with respect to passing CMAKE flags for building tests
* Make cmake gen script find override file without using script arguments
* build-test.sh can now build native test projects
Remove compileoptions.cmake and push contents into configurecompiler.cmake
which contains compiler configuration from repo root CMakeLists.txt
|
|
Some JitBench fixes
- Fixed `dotnet run` from the `unofficial_dotnet` folder after `Directory.Build.targets` was added to `coreclr/tests/src`. This file is auto-imported and was adding dependencies and properties for targets that were not necessary. Added files to the root JitBench folder to override.
- Updated JitBench to be able to run against netcoreapp3.0. I haven't changed the default yet, could consider doing so once there are no pending perf issues being looked into.
- Added a `RetargetProjects()` function to the setup steps to replace `TargetFramework` and `RuntimeFrameworkVersion` values in the relevant `.csproj` files to run against the expected runtime. This reduces some dependency on other repos containing the actual perf projects such that we don't have to update every repo to support environment variables to set those values.
- Disabled tiering for the `MinOpts` config
|
|
Generalize the bail out pattern to handle cases from spans where we may
see a known array length in a bounds check.
Fixes #19454
|
|
|
|
Fix non-lclVar odd-byte struct passing
|
|
* Make handle checks accurate on Windows
We can actually check the async state of a handle on Windows, so check directly when we have the API available.
Also allow all filetypes through when explicitly using extended syntax `\\?\` .
* Fix nits
* Don't throw in the handle constructor overloads
* Whoops
* Removing the close handle entirely. No adverse effects from closing a bad handle- or leaving it open for that matter.
|
|
(#19505)
This reverts commit 855ddf52c7b05994cf6728b9169fa6cdc694a537. Issue https://github.com/dotnet/coreclr/issues/19124 was fixed by PR https://github.com/dotnet/coreclr/pull/19234, so re-enabling the test in minopts and with tiering.
|
|
Corrects an issue with the conversion table for Gregorian to Japanese lunisolar dates.
See: #19450.
Issue: data discrepancy in year 1962.
Table entry for 1962 has incorrect days per month for months 6 and 7.
Notes
The DaysPerMonth flag uses the binary literal for ease of readability.
|
|
We were previously marking all lcl vars as implicitly referenced in minopts,
but there are some special varargs params that can't ever be marked as
referenced, either implicitly or explicitly.
Special case these when computing or checking ref counts in minopts.
Closes #19349.
|
|
(#19359)" (#19501)
This reverts commit d485659348af3528b688ac5b7e4c58dacba6de58.
|
|
|
|
x86 GcInfo headers
x86 GcSlotTable
x86 GcTransitions
Update expected xml for R2RDumpTests
Add license headers
Allow multiple GcTransitions with same codeOffset
Add index property in GcSlot, make GcSlot a class instead of struct, add missing spaces, update tests
Remove placeholder functions for parsing partially interruptible pointer tables
Implement partially interruptible GcInfo
Example GcInfo output:
CodeLength: 22 bytes
InfoHdr:
PrologSize: 7
EpilogSize: 4
EpilogCount: 1
EpilogAtEnd: yes
Callee-saved regs = EBP
EbpFrame: yes
Fully Interruptible: yes
DoubleAlign: no
Arguments Size: 0 DWORDs
Stack Frame Size: 1 DWORDs
UntrackedCnt: 1
VarPtrTableSize: 0
GenericsContext: 0
GenericsContextIsMethodDesc: 0
ReturnKind: RT_Scalar
RevPInvokeOffset: 0
Epilogs: 18
GcSlots:
-------------------------
[EBP-4]
Flags: GC_SLOT_UNTRACKED
LowBits:
-------------------------
28fc: 55 push ebp
28fd: 8b ec mov ebp, esp
28ff: 50 push eax
2900: 89 4d fc mov dword ptr [ebp - 4], ecx
2903: 8b 4d fc mov ecx, dword ptr [ebp - 4]
2906: ff 15 10 10 01 10 call dword ptr [268505104]
reg ECX becoming live
290c: 90 nop
reg ECX becoming dead
290d: 90 nop
290e: 8b e5 mov esp, ebp
2910: 5d pop ebp
2911: c3 ret
|
|
preview1-26815-04, preview1-26815-04, preview1-26815-01, master-20180815-0120, respectively (#19494)
|
|
ExcludeList for R2RDumpTests in wrong ItemGroup
|
|
|
|
|
|
|
|
preview1-26814-06, preview1-26814-01, preview1-26814-01, master-20180814-0056, respectively (#19377)
|
|
|
|
Unused strong name signing paths and related code
|
|
* Translate [rip +- offset] to absoluate RVA's in R2RDump
The existing logic for displaying rip-relative addressed on X64
make it very hard to calculate the final addresses. I have added
a horrendous hack using textual analysis of the disassembled
instruction to translate this notation to absolute RVA's.
As part of this effort I have also encapsulated the CorDisTools
helper in a new class Disassembler that also contains customizable
provisions for handling special assembly cases on the individual
architectures.
Thanks
Tomas
* Temporarily block out disassembly to make tests pass
In my initial commit I removed the line blocking out disassembly
however this ends up failing several lab tests so I'm putting the
line back.
Thanks
Tomas
|
|
Fixed SOS plugin for core dumps.
|
|
* Update URLs and exclusion list for staging
* Update Failing tests
* Fix IO test namespace
* Add 2 failing Ubuntu tests
* Exclude test under Ubuntu
* Add regressed test
* Roll-back CI URLs
* Add issue number
|
|
Disable 3 tests for Windows/arm32 HeapVerify runs
|
|
in stress modes. The issue is #18988
|
|
Disable VectorDot_ro.cmd_9267 for Windows arm64 R2R
|
|
Tracked by #18991
|
|
Disable test b163200 under GCStress
|
|
Tracked by #19464
|
|
They all time out.
Fixes #19462
|
|
Exclude 437044.cmd_10304 from GCStress runs
|
|
This test times out on ARM Windows R2R GCStress15 runs.
|
|
This option is the opposite of -msbuildonunsupportedplatform, and an
managed version of of -skipnative.
This is useful for bootstrapping on new Linux/x64 distributions, where
dotnet may not work (yet) but coreclr will try and use it anyway,
failing the bootstrap scripts.
See also https://github.com/dotnet/source-build/issues/663 for
additional background information.
|
|
Disable System.Linq.Expressions.Tests against #19457
|
|
|
|
`int i` appears to be unused so i removed it.
Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
|
|
Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
|
|
* Fix a couple of apartment state issues
Fix for https://github.com/dotnet/coreclr/issues/17822
- The apartment state now defaults to MTA for the main thread along with a CoInitialize
- Calling `Thread.SetApartmentState` with STA now fails as expected (different behavior from previous netcore, same behavior as netfx)
Fix for https://github.com/dotnet/coreclr/issues/17787
- `WaitHandle.WaitAll` for multiple handles is not supported on an STA thread due to issues described in https://github.com/dotnet/coreclr/issues/17787#issuecomment-385117537
- It now throws `NotSupportedException` as expected (different behavior from previous netcore, same behavior as netfx)
Fix for https://github.com/dotnet/coreclr/issues/19225
* Temporarily exclude invalid CoreFX test
|
|
* Create Microsoft.Internal.IO
Includes all of Path, Directory, *Info, etc. from System.IO with targets of NetFX 4.6.1 and NetCore 2.1.
* Tweaks / minor fixes
* Remove netstandard build.
* Add package
Update to build for 4.7.2 only
* Rename folder to prepare for project rename
* Change the name to Microsoft.IO.Redist
* Namespace tweaks
* Use open key, tweak references
* More tweaks on feedback
Also grab latest build tools for 4.7.x RIDs
Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
|
|
Disable type loader assert that fires when resolving a type from a WinMD. Instead, compilation should fail gracefully, or at runtime a type load exception will be thrown.
|
|
|
|
(#19427)
- Sealed Runtime makes `is RuntimeType` and similar checks faster. These checks are fairly common in reflection.
- Delete support for introspection only loads from the runtime. We do not plan to use in .NET Core. The support for introspection loads inherited from RuntimeType and thus it is incompatible with sealed RuntimeType.
|
|
|
|
* m removed from names, spaces after \\ and braces added
* Moved to shared
|
|
* Don't early terminate on null for 64bit NR HashCode
* Improved GetNonRandomizedHashCode
* Update message on GetNonRandomizedHashCode
* Consume null terminator rather than special case odd lengths
|
|
Disable R2RDumpTest for JIT stress modes
|
|
|