platform/upstream/coreclr.git
4 years ago[x86/Linux] Fix SIGSEGV during evaluation abort routine. accepted/tizen/5.5/unified/20200608.142150 accepted/tizen/unified/20200608.144754 submit/tizen/20200607.233857 submit/tizen_5.5/20200607.233848
Mikhail Kurinnoi [Thu, 4 Jun 2020 14:48:52 +0000 (17:48 +0300)]
[x86/Linux] Fix SIGSEGV during evaluation abort routine.

In case of evaluation with implicit function call aborted by ```ICorDebugEval::Abort()```, CoreCLR crash with SIGSEGV at line https://github.com/dotnet/runtime/blob/e25517ea27311297c1e3946acb3b4382d5fa7fef/src/coreclr/src/vm/jitinterface.cpp#L14293 since ```m_pJM``` is ```NULL```.
This happens because during ```EECodeInfo::Init()``` call, ```codeAddress``` parameter provide address inside native code region (this address belong to CallDescrWorkerInternal(), libcoreclr.so), but not address inside managed code, so, ```ExecutionManager::FindCodeRange()``` can't find appropriate ```RangeSection```.

During investigation I found, that at line https://github.com/dotnet/runtime/blob/e25517ea27311297c1e3946acb3b4382d5fa7fef/src/coreclr/src/vm/stackwalk.cpp#L2584 current context was not changed properly (we have wrong ```Eip``` register value).
I found, that ```FuncEvalFrame::UpdateRegDisplay()``` code
https://github.com/dotnet/runtime/blob/e25517ea27311297c1e3946acb3b4382d5fa7fef/src/coreclr/src/debug/ee/debugger.inl#L238-L247
don't have x86/Linux support implemented.

I propose changes, that were already made for other ```UpdateRegDisplay()``` implementations in order to provide proper context for x86/Linux.

4 years ago[Tizen] Enable ASan annotation of passing to native code buffers
Andrey Kazmin [Thu, 14 May 2020 13:57:25 +0000 (16:57 +0300)]
[Tizen] Enable ASan annotation of passing to native code buffers

Turn on ASan inteceptors while marshaling managed buffers to native code.
We could not properly annotate already allocated on heap buffers, so
we have to disable pinning of such objects.
Current patch affects only pinning of native arrays.

4 years ago[Tizen] Cleanup confusing symlinks accepted/tizen/5.5/unified/20200601.002556 accepted/tizen/unified/20200601.051718 submit/tizen/20200529.025703 submit/tizen_5.5/20200529.025744
Hyungju Lee [Mon, 25 May 2020 08:19:52 +0000 (17:19 +0900)]
[Tizen] Cleanup confusing symlinks

Change-Id: I032c9ff1e0f43e39d0d05a38eb95dabe2f1b6b25

4 years agoUpdate package version to 3.1.3
Alexander Soldatov/AI Compiler Lab /SRR/Staff Engineer/Samsung Electronics [Fri, 22 May 2020 09:14:11 +0000 (12:14 +0300)]
Update package version to 3.1.3

4 years agoSet vtable offset as contained
JUNG DONG-HEON [Thu, 5 Dec 2019 07:23:39 +0000 (16:23 +0900)]
Set vtable offset as contained

- Can remove a machine instruction which adds vtable offset

4 years ago[Tizen] Change the visibility of symbols for dnetmemoryenumlib accepted/tizen/5.5/unified/20200522.160111 accepted/tizen/unified/20200529.124257 submit/tizen/20200528.065414 submit/tizen_5.5/20200512.212247
Mateusz Moscicki [Fri, 8 May 2020 11:40:06 +0000 (13:40 +0200)]
[Tizen] Change the visibility of symbols for dnetmemoryenumlib

4 years ago[Tizen] Skip tests, which throw System.OutOfMemoryException on TW3 accepted/tizen/5.5/unified/20200423.150450 accepted/tizen/unified/20200423.054553 submit/tizen/20200422.104050 submit/tizen_5.5/20200422.104039
Gleb Balykov [Tue, 24 Mar 2020 15:09:46 +0000 (18:09 +0300)]
[Tizen] Skip tests, which throw System.OutOfMemoryException on TW3

4 years ago[Tizen] Enable Debug builds for i686
Sangwook Kim [Fri, 27 Mar 2020 09:05:25 +0000 (18:05 +0900)]
[Tizen] Enable Debug builds for i686

4 years ago[Tizen] Add a config knob for importing ibc files
Swift Kim [Thu, 1 Aug 2019 06:10:49 +0000 (15:10 +0900)]
[Tizen] Add a config knob for importing ibc files

4 years agoFix PIE options (#26323) accepted/tizen/unified/20200416.080052 submit/tizen/20200415.223728
Jan Vorlicek [Fri, 23 Aug 2019 15:03:01 +0000 (17:03 +0200)]
Fix PIE options (#26323)

* Fix PIE options

We were missing passing the -pie linker option. That means that while we
were compiling our code as position independent, the executables
(not shared libraries) were not marked as position independent and
ASLR was not applied to them. They were always loaded to fixed addresses.

This change adds the missing -pie option and also replaces all the individual
settings of -fPIE / -fPIC on the targets we build by a centralized setting
of CMAKE_POSITION_INDEPENDENT_CODE variable that causes cmake to add the
appropriate compiler options everywhere.

* Fix native parts of coreclr tests build

The native parts of the tests are not built using the root CMakeLists.txt
so I am moving enabling the position independent code to configurecompiler.cmake

Change-Id: Ieafff8984ec23e5fdb00fb0c2fb017e53afbce88

4 years ago[Tizen] Fix build of Runtime_1241 test accepted/tizen/unified/20200402.155517 submit/tizen/20200402.013218
Gleb Balykov [Wed, 25 Mar 2020 15:17:50 +0000 (18:17 +0300)]
[Tizen] Fix build of Runtime_1241 test

4 years ago[Tizen] Add FEATURE_LARGEADDRESS_SUPPORT
Sangwook Kim [Tue, 18 Feb 2020 06:48:15 +0000 (15:48 +0900)]
[Tizen] Add FEATURE_LARGEADDRESS_SUPPORT

Many diagnostic tools are unaware of 32-bit applications which have
large address spaces (> 2GB). Such tools include the TraceEvent library
(required by PerfView and dotnet-trace), and Visual Studio. They assume
the address range 0x80000000 through 0xFFFFFFFF as the system space and
thus often fail to read symbols from event traces generated by CoreCLR.

This workaround is to support such scenarios by simply discarding MSBs
of 32-bit instruction pointer values in the trace output. Only a minimal
set of values required for symbol resolution are affected by this
change. Beware that you will have to manually restore the original
values when you inspect them in lldb or etc.

4 years ago[Tizen] Fix usage of ni.exe in readytorun test
Gleb Balykov [Mon, 16 Mar 2020 16:25:20 +0000 (19:25 +0300)]
[Tizen] Fix usage of ni.exe in readytorun test

4 years ago[Tizen] Add System.Private.CoreLib.pdb to coreclr-devel package
Gleb Balykov [Wed, 11 Mar 2020 13:13:19 +0000 (16:13 +0300)]
[Tizen] Add System.Private.CoreLib.pdb to coreclr-devel package

4 years ago[Tizen] Move the executables(ilasm, ildasm) to use in dotnettool from devel rpm
j-h.choi [Tue, 7 Jan 2020 07:14:33 +0000 (16:14 +0900)]
[Tizen] Move the executables(ilasm, ildasm) to use in dotnettool from devel rpm

4 years agomodify integral tryparse to use memcpy (#2295)
John Salem [Thu, 13 Feb 2020 02:14:46 +0000 (18:14 -0800)]
modify integral tryparse to use memcpy (#2295)

4 years agoFix OverflowException from IntPtr casting (#14381)
Swift Kim [Fri, 31 Jan 2020 14:27:40 +0000 (23:27 +0900)]
Fix OverflowException from IntPtr casting (#14381)

4 years ago[Tizen] Reduce arm_phdr_cb call overhead
JUNG DONG-HEON [Thu, 9 Jan 2020 07:38:00 +0000 (16:38 +0900)]
[Tizen] Reduce arm_phdr_cb call overhead

- Too many calls to arm_phdr_cb even though it get the same data.
- It caches an ARM_CB_DATA for libcoreclr.so, then reuse.

4 years ago[Tizen] RPM version of Tizen will be upgrade to 4.14.1.
Woongsuk Cho [Tue, 31 Dec 2019 04:59:26 +0000 (13:59 +0900)]
[Tizen] RPM version of Tizen will be upgrade to 4.14.1.

4 years agoChange bIsFree check in DacValidateMethodTable (#1086)
Dong-Heon Jung [Mon, 23 Dec 2019 19:11:43 +0000 (04:11 +0900)]
Change bIsFree check in DacValidateMethodTable (#1086)

Some commands of SOS validate a method table in DacValidateMethodTable.
In the function, it checks whether a method table is FreeObjectMethodTable or not with GetClass() value.
However, GetClass() should not be NULL. (There is an assert in GetClass())
In this patch, it compares pMT address with g_pFreeObjectMethodTable address only.

4 years ago[Tizen] add pie linker option to createdump
Woongsuk Cho [Thu, 12 Dec 2019 11:41:39 +0000 (20:41 +0900)]
[Tizen] add pie linker option to createdump

4 years agoHandle glibc sys/sysctl.h deprecation (#27048)
Omair Majid [Wed, 16 Oct 2019 09:25:29 +0000 (05:25 -0400)]
Handle glibc sys/sysctl.h deprecation (#27048)

glibc has deprecated sys/sysctl.h:

    In file included from /coreclr/src/pal/src/misc/sysinfo.cpp:32:
    /usr/include/sys/sysctl.h:21:2: error: "The <sys/sysctl.h> header is deprecated and will be removed." [-Werror,-W#warnings]
    #warning "The <sys/sysctl.h> header is deprecated and will be removed."
     ^
    1 error generated.

Fix that by preferring sysconf and only including sys/sysctl.h if
HAVE_SYSCONF is not true. This mirrors the order of the implementation
code in this file (sysinfo.cpp) which checks for HAVE_SYSCONF
before HAVE_SYSCTL.

Fixes #27008

4 years ago[Tizen] Use -mstackrealign to sync up with Tizen build changes
Mikhail Kashkarov [Fri, 6 Dec 2019 15:57:11 +0000 (18:57 +0300)]
[Tizen] Use -mstackrealign to sync up with Tizen build changes

Tizen i586 now builds with -mstackrealign by default:

    "Realign the stack at entry. On the x86, the -mstackrealign option
    generates an alternate prologue and epilogue that realigns the
    run-time stack if necessary. This supports mixing legacy codes that
    keep 4-byte stack alignment with modern codes that keep 16-byte
    stack alignment for SSE compatibility. ..."

Change-Id: I36afd18998829f897c6b2b48687c4ae1d172b84b
Signed-off-by: Mikhail Kashkarov <m.kashkarov@partner.samsung.com>
4 years agoEnable NGEN for methods marked with AggressiveOptimization (#27259)
Dong-Heon Jung [Fri, 18 Oct 2019 00:23:51 +0000 (09:23 +0900)]
Enable NGEN for methods marked with AggressiveOptimization (#27259)

- Methods marked with AggressiveOptimization are not NGENed at all.
- The methods are compiled during the runtime with high JITC overhead.
- It makes launching time slower over 6% in our embedded systems.

4 years agoMark Relocation Section as NotNeeded (#25715)
Dong-Heon Jung [Fri, 30 Aug 2019 21:05:53 +0000 (06:05 +0900)]
Mark Relocation Section as NotNeeded (#25715)

- After relocation, relocation section in zap image is not necessary.
- Mark the section as NotNeeded by giving advice(madvise with MADV_DONTNEED)
- It reduces 120~150KB PSS in tizen sample apps.

4 years ago[Tizen] Use PTRACE_GETREGSET for any arch when creating dump
Swift Kim [Mon, 18 Nov 2019 07:38:56 +0000 (16:38 +0900)]
[Tizen] Use PTRACE_GETREGSET for any arch when creating dump

Also ignore ptrace NT_FPREGSET failures for arm processes on aarch64
kernels. Fixes #25707.

4 years ago[Tizen] Add createdump to build output
Sangwook Kim [Mon, 18 Nov 2019 01:20:57 +0000 (10:20 +0900)]
[Tizen] Add createdump to build output

4 years ago[Tizen] Add coreclr_preload_assembly to CoreCLR host API
Konstantin Baladurin [Fri, 5 Jul 2019 16:36:28 +0000 (19:36 +0300)]
[Tizen] Add coreclr_preload_assembly to CoreCLR host API

4 years ago[Tizen] Enable Tizen ASan runtime support
Vyacheslav Cherkashin [Fri, 19 Jul 2019 12:05:47 +0000 (15:05 +0300)]
[Tizen] Enable Tizen ASan runtime support

Enable libasansi.so support (libasan.so with switchable interceptors).

Change-Id: I9ee9b47b7beab55f036ffc0697ffab2583e9701c
Signed-off-by: Vyacheslav Cherkashin <v.cherkashin@samsung.com>
4 years ago[Tizen] Implement ASan wrapper for Linux AMD64
Vyacheslav Cherkashin [Mon, 29 Jul 2019 16:02:37 +0000 (19:02 +0300)]
[Tizen] Implement ASan wrapper for Linux AMD64

Change-Id: I48446ce7c8771a4c75149512bb7d8a8cb3fae8e5
Signed-off-by: Vyacheslav Cherkashin <v.cherkashin@samsung.com>
4 years ago[Tizen] Implement ASan wrapper for Linux ARM32
Vyacheslav Cherkashin [Mon, 22 Jul 2019 10:40:47 +0000 (13:40 +0300)]
[Tizen] Implement ASan wrapper for Linux ARM32

This commit implements wrappers that allow interception transitions
from managed to external unmanaged code (CIL -> native) and back
(native -> CIL). This allows enable/disable ASan during transitions.
Due to this, we sanitize only external code, which allows us to
achieve acceptable performance.

Change-Id: I53ecdc14d28f7210cd9e7f5bd4db0c8ef5ed81fc
Signed-off-by: Vyacheslav Cherkashin <v.cherkashin@samsung.com>
4 years ago[Tizen] Implement detecting of sanitized libraries
Andrey Drobyshev [Tue, 16 Jul 2019 12:23:18 +0000 (15:23 +0300)]
[Tizen] Implement detecting of sanitized libraries

Parse ".dynamic" section (ELF dynamic array tags) of the module being
added, find ".rel(a).plt" section and search it for presence of
'__asan_init' symbol.

Change-Id: Ie7cc4c818b791b5f00713b42ba15131325b8152c
Signed-off-by: Andrey Drobyshev <a.drobyshev@samsung.com>
4 years agoPrevent freeing of the profiler on process shutdown.
David Mason [Wed, 18 Sep 2019 20:19:36 +0000 (13:19 -0700)]
Prevent freeing of the profiler on process shutdown.

4 years ago[Tizen] skip dotnet specific arguments in corerun
Konstantin Baladurin [Fri, 4 Oct 2019 16:45:44 +0000 (19:45 +0300)]
[Tizen] skip dotnet specific arguments in corerun

Now we use corerun to run corefx tests instead of dotnet, because last
one isn't available for Tizen/armel. So we need to skip dotnet specific
arguments, we patch corerun for it because Microsoft.DotNet.RemoteExecutor
tries to execute binary that it gets from /proc/self/maps, so we need a
binary that will behave like dotnet.

4 years agoBuild error fix on FEATURE_PREJIT=true
Swift Kim [Fri, 12 Jul 2019 06:50:08 +0000 (15:50 +0900)]
Build error fix on FEATURE_PREJIT=true

This fix is to update usages of SetupGcCoverage() under
FEATURE_PREJIT aligned to the signature change in #25261.

4 years ago[Linux/x86] Use ebp from current context during unwinding (#26789)
Konstantin Baladurin [Thu, 26 Sep 2019 17:06:57 +0000 (20:06 +0300)]
[Linux/x86] Use ebp from current context during unwinding (#26789)

pCurrentContextPointers in REGDISPLAY can contain NULLs so we need to use
ebp value from pCurrentContext. This patch contains following changes:

- GetRegdisplayFP returns ebp from pCurrentContext
- GetRegdisplayFP is used instead of *GetEbpLocation()
- Set##reg##Location also updates register value in pCurrentContext

4 years ago[JIT/x86] Fix LinearScan::allocateRegisters (#26649)
Konstantin Baladurin [Fri, 13 Sep 2019 17:28:08 +0000 (20:28 +0300)]
[JIT/x86] Fix LinearScan::allocateRegisters (#26649)

Check for `lvLRACandidate` instead of `!lvDoNotEnregister` when checking whether `this` may be enregistered and has an Interval.

4 years agoLinux/x86: fix build (#26594)
Konstantin Baladurin [Mon, 9 Sep 2019 15:40:14 +0000 (00:40 +0900)]
Linux/x86: fix build (#26594)

4 years ago[Tizen] Pack test libs to coreclr-test
Gleb Balykov [Wed, 11 Sep 2019 13:53:14 +0000 (16:53 +0300)]
[Tizen] Pack test libs to coreclr-test

4 years ago[Tizen] Add unsupported tests
Gleb Balykov [Tue, 10 Mar 2020 17:22:57 +0000 (20:22 +0300)]
[Tizen] Add unsupported tests

4 years ago[Tizen] Replace new runtest.sh, which calls dotnet internally, with old runtest.sh
Gleb Balykov [Wed, 21 Aug 2019 16:47:37 +0000 (19:47 +0300)]
[Tizen] Replace new runtest.sh, which calls dotnet internally, with old runtest.sh

4 years ago[Tizen] Partially revert a6292a6.
Mikhail Kurinnoi [Mon, 19 Aug 2019 12:06:58 +0000 (15:06 +0300)]
[Tizen] Partially revert a6292a6.

NetcoreDBG depends from PAL functions.
Instead of SOS plugin, that use PAL static libs, debugger should be able
to operate with any runtime version and can't be statically linked to PAL.

4 years agoadd access(2) call before dlopening files
Yaroslav Yamshchikov [Thu, 18 Jul 2019 10:14:14 +0000 (13:14 +0300)]
add access(2) call before dlopening files

4 years agocorbbtprof: set byte alignment for CORBBTPROF structures (#25816)
Konstantin Baladurin [Tue, 23 Jul 2019 04:01:22 +0000 (07:01 +0300)]
corbbtprof: set byte alignment for CORBBTPROF structures (#25816)

Fix patch fixes SIGBUG that occurs due to unaligned read/write

4 years agoFail to explicitly tail call on x86 unix. (#25032)
Jarret Shook [Fri, 19 Jul 2019 04:07:08 +0000 (21:07 -0700)]
Fail to explicitly tail call on x86 unix. (#25032)

* Fail to explicitly tail call on x86 unix.

* Correctly return 100

* Correct return value

* Add noway assert in morphTailCall to avoid morphing slow tail calls on unix.

* Address feedback

4 years ago[Tizen] Disable jithost arena cache
Gleb Balykov [Fri, 2 Aug 2019 14:28:27 +0000 (17:28 +0300)]
[Tizen] Disable jithost arena cache

4 years ago[Tizen] Disable IBC Logger as a default
DongHeon Jung [Thu, 18 Jul 2019 08:28:55 +0000 (17:28 +0900)]
[Tizen] Disable IBC Logger as a default

- Profile information is collected by ibc logger.
  Hower it is not used and saved into profile file.
- The patch disables IBC logger which is enabled by default.
- It disables IBC logger only with ibclogger.h file.
  IBCLOGGER_ENABLED definition is only used in ibclogger files.

4 years agoFix a build error when IBCLOGGER_ENABLED is not defined (#25691)
Dong-Heon Jung [Wed, 17 Jul 2019 03:09:40 +0000 (12:09 +0900)]
Fix a build error when IBCLOGGER_ENABLED is not defined (#25691)

- Even if DACCESS_COMPILE or CROSSGEN_COMPILE is defined,
  coreclr can be built without IBCLOGGER_ENABLED definition.

4 years ago[Tizen] Enable PGO for Linux/arm
Konstantin Baladurin [Thu, 18 Apr 2019 09:29:56 +0000 (12:29 +0300)]
[Tizen] Enable PGO for Linux/arm

4 years ago[Tizen] seperate PIC and PIE to fix x86_64 build error
Woongsuk Cho [Thu, 9 May 2019 09:02:22 +0000 (18:02 +0900)]
[Tizen] seperate PIC and PIE to fix x86_64 build error

4 years ago[Tizen] Add a library to retrieve memory regions for a coredump
Mateusz Moscicki [Wed, 10 Apr 2019 09:40:33 +0000 (11:40 +0200)]
[Tizen] Add a library to retrieve memory regions for a coredump

This is needed to save a minicoredump of .NET applications to allows
reconstruct managed stack by lldb and SOS plugin.

4 years ago[Tizen] update cscope files to be ignored
Hyungju Lee [Wed, 13 Mar 2019 00:28:54 +0000 (09:28 +0900)]
[Tizen] update cscope files to be ignored

Change-Id: Ia0e1c6aa651c4fd88dc58b901729896a7a27681d

4 years agoFix OOPStackUnwinderX86::Unwind crash when Eip is invalid
Igor Kulaychuk [Wed, 25 Apr 2018 18:31:59 +0000 (21:31 +0300)]
Fix OOPStackUnwinderX86::Unwind crash when Eip is invalid

4 years agoFix EECodeManager::GetAmbientSP on x86/Linux
Igor Kulaychuk [Fri, 20 Apr 2018 20:52:57 +0000 (23:52 +0300)]
Fix EECodeManager::GetAmbientSP on x86/Linux

4 years ago[Tizen] Add -pie to linker option
Hyungju Lee [Mon, 2 Jul 2018 00:28:08 +0000 (09:28 +0900)]
[Tizen] Add -pie to linker option

There have been no -pie linker option.
This patch adds -pie linker option into crossgen(for tizen)

This originates from 0024-Add-pie-to-linker-option.patch

4 years ago[Tizen] Add support for GBS
junghyuk.park [Mon, 25 Jun 2018 09:47:41 +0000 (18:47 +0900)]
[Tizen] Add support for GBS

4 years ago[Tizen] Add prebuilt libicu-57.1 libraries
junghyuk.park [Mon, 25 Jun 2018 07:03:57 +0000 (16:03 +0900)]
[Tizen] Add prebuilt libicu-57.1 libraries

4 years ago[Tizen] Add BuildTools v3.1.3
Gleb Balykov [Tue, 10 Mar 2020 16:22:08 +0000 (19:22 +0300)]
[Tizen] Add BuildTools v3.1.3

4 years ago[3.1] Fail FuncEval if slot backpatching lock is held by any thread (#28006)
Koundinya Veluri [Tue, 18 Feb 2020 21:46:37 +0000 (16:46 -0500)]
[3.1] Fail FuncEval if slot backpatching lock is held by any thread (#28006)

- In many cases cooperative GC mode is entered after acquiring the slot backpatching lock and the thread may block for debugger suspension while holding the lock. A FuncEval may time out on entering the lock if for example it calls a virtual or interface method for the first time. Failing the FuncEval when the lock is held enables the debugger to fall back to other options for expression evaluation.
- Also added polls for debugger suspension before acquiring the slot backpatching lock on background threads that often operate in preemptive GC mode. A common case is when the debugger breaks while the tiering delay timer is active, the timer ticks shortly afterwards (after debugger suspension completes) and if a thread pool thread is already available, the background thread would block while holding the lock. The poll checks for debugger suspension and pulses the GC mode to block before acquiring the lock.

Risks:
- The fix is only a heuristic and lessens the problem when it is detected that the lock is held by some thread. Since the lock is acquired in preemptive GC mode, it is still possible that after the check at the start of a FuncEval, another thread acquires the lock and the FuncEval may time out. The polling makes it less likely for the lock to be taken by background tiering work, for example if a FuncEval starts while rejitting a method.
- The expression evaluation experience may be worse when it is detected that the lock is held, and may still happen from unfortunate timing
- Low risk for the change itself

Port of https://github.com/dotnet/runtime/pull/2380
Fix for https://github.com/dotnet/runtime/issues/1537

4 years agoPort dotnet/runtime#31946 to release/3.1 branch (#28014)
Levi Broderick [Tue, 18 Feb 2020 21:32:57 +0000 (13:32 -0800)]
Port dotnet/runtime#31946 to release/3.1 branch (#28014)

When string.Replace is given a target string with zero collation weight, it would enter an infinite loop. It is now changed so that the call to Replace terminates when such a condition is encountered.

4 years agoFix AppDomain.SetPrincipalPolicy bug for new threads (#32104) (#28019)
Eirik Tsarpalis [Tue, 18 Feb 2020 21:31:53 +0000 (21:31 +0000)]
Fix AppDomain.SetPrincipalPolicy bug for new threads (#32104) (#28019)

* fix principal policy for new threads

Fixes #31717

Co-authored-by: Marco Rossignoli <marco.rossignoli@gmail.com>
4 years agoFix build on systems with glibc >= 2.30 (#28012)
Omair Majid [Tue, 18 Feb 2020 18:27:24 +0000 (13:27 -0500)]
Fix build on systems with glibc >= 2.30 (#28012)

On newer systems with glibc 2.30, the compiler emits a warning:

    In file included from coreclr/src/pal/src/misc/sysinfo.cpp:32:
    /usr/include/sys/sysctl.h:21:2: error: "The <sys/sysctl.h> header is deprecated and will be removed." [-Werror,-W#warnings]
    #warning "The <sys/sysctl.h> header is deprecated and will be removed."
     ^

The glibc 2.30 release notes cover this at
https://sourceware.org/ml/libc-alpha/2019-08/msg00029.html:

* The Linux-specific <sys/sysctl.h> header and the sysctl function have been
  deprecated and will be removed from a future version of glibc.
  Application should directly access /proc instead.  For obtaining random
  bits, the getentropy function can be used.

To keep coreclr release/3.1 building, disable treating the #warning as an
error. Clang and GCC have separate flags to turn this error off.

4 years agoPort dotnet/runtime#31904 to release/3.1 (#28013)
Levi Broderick [Tue, 18 Feb 2020 18:06:38 +0000 (10:06 -0800)]
Port dotnet/runtime#31904 to release/3.1 (#28013)

Remove BMI2 from ASCII and UTF-16 processing hot paths, as not all processors have optimized implementations of pext/pdep

4 years agoUpdate dependencies from https://github.com/dotnet/arcade build 20200213.5 (#28018)
dotnet-maestro[bot] [Fri, 14 Feb 2020 01:45:08 +0000 (01:45 +0000)]
Update dependencies from https://github.com/dotnet/arcade build 20200213.5 (#28018)

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.20113.5
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.20113.5
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.20113.5
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.20113.5

4 years agoPort of dotnet/runtime#1059 to 3.1 branch (#27986)
Carol Eidt [Thu, 13 Feb 2020 22:27:49 +0000 (14:27 -0800)]
Port of dotnet/runtime#1059 to 3.1 branch (#27986)

This is the fix for #27924. This is a GC hole bug that was found externally, #27590.
The cause is that the JIT was using the target type of the subtract when it needed
to make a copy of the source, but it needs to use the source type.

## Customer Impact
Corruption of state that is non-deterministic and hard to track down.

## Regression?
Not a recent regression, but exposed by Unsafe.ByteOffset.

## Testing
The fix has been verified in the runtime repo.

## Risk
Low: The fix is straightfoward and only impacts 3 lines of code.

4 years agoPort fix for #1241 to 3.1 (#27983)
Carol Eidt [Thu, 13 Feb 2020 22:27:34 +0000 (14:27 -0800)]
Port fix for #1241 to 3.1 (#27983)

4 years agoPort PR #258 to 3.1 (#27984)
Carol Eidt [Thu, 13 Feb 2020 22:27:22 +0000 (14:27 -0800)]
Port PR #258 to 3.1 (#27984)

* Port PR #258 to 3.1

* Fix test proj file

4 years agoFix GC heap corruption on ARM. (#27985)
Anton Lapounov [Thu, 13 Feb 2020 22:27:12 +0000 (14:27 -0800)]
Fix GC heap corruption on ARM. (#27985)

Port of dotnet/runtime#1389.

4 years agoPort the 5.0 fix for issue #1104 (#28003)
Brian Sullivan [Thu, 13 Feb 2020 22:26:51 +0000 (14:26 -0800)]
Port the 5.0 fix for issue #1104 (#28003)

* Port the 5.0 fix for issue #1104 3.1

- Pull Request Runtime\#1734
- This change corrects a cut-and paste typo with a previous commit.
- Includes test case Runtime_1104.cs

* Updated the csproj to use 3.1 syntax

4 years agoupdate branding for 3.1.3 (#28016)
Anirudh Agnihotry [Thu, 13 Feb 2020 22:24:18 +0000 (14:24 -0800)]
update branding for 3.1.3 (#28016)

4 years ago[automated] Merge branch 'release/3.0' => 'release/3.1' (#28002)
dotnet-maestro-bot [Fri, 17 Jan 2020 02:10:56 +0000 (18:10 -0800)]
[automated] Merge branch 'release/3.0' => 'release/3.1' (#28002)

* Fix Segfault in PerfInfo Image Logging (#26910)

* update branding for 3.0.2 (#27896)

* update branding

* remove stable properties from coreclr as it never stablises

* [release/3.0] Fix use of ilasm during test builds when using a 3.0 SDK (#27820)

* Add fix for restore of ilasm

The restore logic in buildtools tried to use a netcoreapp2.1 TFM for Linux-musl.
This ended up restoring Linux-x64 binaries which then broke the test build.
This change works around this by saving a copy of the depproj that BuildTools used into
the tree and restore it as a 3.0 app manually in init-tools on our side of the build.

* Change to a 3.0 SDK for servicing

* Bump test versions of ilasm and runtime package to match the RC

* Move off of the old arm32 queue (#27905)

* Add alpine3.10 testing to 3.0 (#27214)

* Fix GetSequencePoints when profiler provides mapping via SetILInstrumentedCodeMap (#27843)

Port #25802 to 3.0.2

* [release/3.0] Update dependencies from dotnet/core-setup (#27910)

* Update dependencies from https://github.com/dotnet/core-setup build 20191016.12

- Microsoft.NETCore.App - 3.0.1-servicing-19516-12

* Update dependencies from https://github.com/dotnet/core-setup build 20191121.02

- Microsoft.NETCore.App - 3.0.2-servicing-19571-02

* update brandint to 3.0.3 (#27993)

* Update branding to 3.0.3 (#27997)

eng/Versions.props doesn't seem to the the source of truth for branding

* Fix infrastructure issues in release 3.0 (#28000)

* Restore Helix SDK using Tools.props to work around MSBuild race condition
* Disable formatting jobs
* Disable ARM windows jobs
* Fix tracing in RHEL6 build

* [release/3.0] Update dependencies from dotnet/arcade (#26889)

* Update dependencies from https://github.com/dotnet/arcade build 20190924.3

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19474.3
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.19474.3
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.19474.3
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.19474.3

* Update NuGet.config

* Update runtime versions

* Update dependencies from https://github.com/dotnet/arcade build 20191119.2

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19569.2
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.19569.2
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.19569.2
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.19569.2

* Update dependencies from https://github.com/dotnet/arcade build 20191122.3

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19572.3
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.19572.3
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.19572.3
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.19572.3

* Update dependencies from https://github.com/dotnet/arcade build 20191127.5

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19577.5
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.19577.5
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.19577.5
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.19577.5

* Add coreclr static feed to NuGet.config

Co-authored-by: Matt Mitchell <mmitche@microsoft.com>
Co-authored-by: Juan Hoyos <juan.hoyos@microsoft.com>
Co-authored-by: Brian Robbins <brianrob@microsoft.com>
Co-authored-by: Anirudh Agnihotry <anirudhagnihotry098@gmail.com>
Co-authored-by: Juan Hoyos <juan.hoyos@microsoft.com>
Co-authored-by: Jarret Shook <jashoo@microsoft.com>
Co-authored-by: David Mason <davmason@microsoft.com>
Co-authored-by: dotnet-maestro[bot] <42748379+dotnet-maestro[bot]@users.noreply.github.com>
Co-authored-by: Matt Mitchell <mmitche@microsoft.com>
4 years ago[release/3.1] Port fix to revert EncoderNLS and DecoderNLS Convert changes (#27996)
Levi Broderick [Wed, 15 Jan 2020 21:04:09 +0000 (13:04 -0800)]
[release/3.1] Port fix to revert EncoderNLS and DecoderNLS Convert changes (#27996)

4 years agoUpdate branding to 3.1.2 (#27998)
Matt Mitchell [Wed, 15 Jan 2020 19:51:12 +0000 (11:51 -0800)]
Update branding to 3.1.2 (#27998)

eng/Versions.props doesn't seem to be the source of truth for branding

4 years agoRevert CPU clipping in the presence of CPU quota to 3.0, 2.x behavior.
Vladimir Sadov [Wed, 15 Jan 2020 03:31:01 +0000 (19:31 -0800)]
Revert CPU clipping in the presence of CPU quota to 3.0, 2.x behavior.

Basically reverting https://github.com/dotnet/coreclr/pull/26806

4 years ago[release/3.1] Port fix for JIT silent bad code (#27972)
Andy Ayers [Tue, 14 Jan 2020 18:36:18 +0000 (10:36 -0800)]
[release/3.1] Port fix for JIT silent bad code (#27972)

* [release/3.1] Port fix for JIT silent bad code

Release/3.1 port of https://github.com/dotnet/runtime/pull/797.
Fixes https://github.com/dotnet/runtime/issues/764

The jit might incorrectly order a read from a struct field with an operation
that modifies the field, so that the read returns the wrong value.

Silent bad code; program behaves incorrectly.

Yes, introduced in the 3.0 cycle.

Verified the user's test case now passes; no diffs seen in any existing framework
or test code.

**Low**: the jit is now spilling the eval stack entries to temps in cases where it
did not before; this should be conservatively safe.

cc: @brucefo

____

If we're appending an assignment whose LHS is is a location within a local
struct, we need to spill all references to that struct from the eval stack.

Update the existing logic for this to handle the case where the LHS is a field
of a local struct, and the field is updated by unusual means (here, `initobj`).

Fixes dotnet/runtime#764.

* Fix test

4 years agoFix CancellationTokenRegistration.Unregister race condition (#27949)
Stephen Toub [Tue, 14 Jan 2020 18:36:07 +0000 (13:36 -0500)]
Fix CancellationTokenRegistration.Unregister race condition (#27949)

4 years agoRelease/3.1 port of dotnet/runtime#239 (#27973)
Andy Ayers [Tue, 14 Jan 2020 18:35:50 +0000 (10:35 -0800)]
Release/3.1 port of dotnet/runtime#239 (#27973)

Fix for #27923

The jit might fail to locate a class handle for a ref class, leading to an
unexpected crash while jitting.

## Customer Impact
Unexpected and hard to diagnose crash/exception

## Regression?
Yes, introduced during the development 3.0 cycle. 2.x behaves correctly.

## Testing
Verified the user's test case now passes; no diffs seen in any existing
framework or test code.

## Risk
**Low**: the jit will now fall back to using the handle for System.Object if no
better option can be found.

cc @BruceForstall

____

In some cases we may end up in lvaSetClass without a valid ref class handle
from either the IR or the stack. Use the handle for object as a conservative
fallback.

4 years agoFix EventSource to stop ignoring EventCommand.SendManifest (#27979)
Sung Yoon Whang [Tue, 14 Jan 2020 18:35:31 +0000 (10:35 -0800)]
Fix EventSource to stop ignoring EventCommand.SendManifest (#27979)

4 years agoStop throwing exception in TimeZoneInfo POSIX parsing (#27969)
Eric Erhardt [Tue, 14 Jan 2020 18:35:16 +0000 (12:35 -0600)]
Stop throwing exception in TimeZoneInfo POSIX parsing (#27969)

IsDaylightSavingTime_CasablancaMultiYearDaylightSavings fails on rhel.8

When parsing the tzdata POSIX string that contains an 'n' Julian date, we are currently throwing an exception, and then falling back to a TimeZoneInfo without DST enabled. However, this is a mistake because there are other DST transitions that were read from the tzdata file that are valid and usable. We shouldn't be throwing that information away.

So instead, we now skip the POSIX string if we detect an unsupported 'n' Julian date, and just use the last transition as the AdjustmentRule for all the DateTimes in the future. This way we can still make DST determinations correctly for some DateTimes.

Fix https://github.com/dotnet/corefx/issues/42192

4 years agoThis is a point fix for not allowing fast tail calls on windows arm64 for vararg...
Jarret Shook [Tue, 14 Jan 2020 18:35:01 +0000 (10:35 -0800)]
This is a point fix for not allowing fast tail calls on windows arm64 for vararg methods (#27963)

* This is a point fix for not allowing fast tail calls on windows arm64

This only affects windows arm and arm64.

* Fix build break

4 years agoPort to 3.1 - Fix out of range access in GetRecycleMemoryInfo (#27959)
Jan Vorlicek [Tue, 14 Jan 2020 18:34:51 +0000 (19:34 +0100)]
Port to 3.1 - Fix out of range access in GetRecycleMemoryInfo (#27959)

Ports change #26873 to release 3.1 branch.

On OpenVZ virtualized linux, GetCurrentProcessorNumber which uses sched_getcpu()
can return a value greater than the number of processors reported by
sched_getaffinity with CPU_COUNT or sysconf(_SC_NPROCESSORS_ONLN).
For example, taskset -c 2,3 ./MyApp will make CPU_COUNT be 2 but
sched_getcpu() can return 2 or 3, and OpenVZ kernel can make
sysconf(_SC_NPROCESSORS_ONLN) return a limited cpu count but
sched_getcpu() still report the real processor number.

Example of affinity vs current CPU id on OpenVZ:
nproc: 8
nprocOnline: 1
affinity: 1, 0, 0, 0, 0, 0, 0, 0, cpuid: 2
affinity: 1, 0, 0, 0, 0, 0, 0, 0, cpuid: 2
affinity: 1, 0, 0, 0, 0, 0, 0, 0, cpuid: 2
affinity: 1, 0, 0, 0, 0, 0, 0, 0, cpuid: 2
affinity: 1, 0, 0, 0, 0, 0, 0, 0, cpuid: 2
affinity: 1, 0, 0, 0, 0, 0, 0, 0, cpuid: 5
affinity: 1, 0, 0, 0, 0, 0, 0, 0, cpuid: 5

4 years agoPort to 3.1 - Fix VirtualMemoryLogging::logRecords overflow (#27958)
Jan Vorlicek [Tue, 14 Jan 2020 18:34:38 +0000 (19:34 +0100)]
Port to 3.1 - Fix VirtualMemoryLogging::logRecords overflow (#27958)

when VirtualMemoryLogging::recordNumber increments from LONG_MAX,
it became negative number, and the result of i % MaxRecords became
a number from -127 to 0.

When that happens we will ovewrite CRITICAL_SECTION virtual_critsec
which are stored in bss right before logRecords with garbage data.
Then most likely the process will have a GC hang with one or more
GC threads stuck trying to enter or leave critical section.

The fix is to ensure ULONG value are passed to modulo operation.

4 years agoPort to 3.1 - Fix getting affinity set on MUSL on Jetson TX2 (#27957)
Jan Vorlicek [Tue, 14 Jan 2020 18:33:49 +0000 (19:33 +0100)]
Port to 3.1 - Fix getting affinity set on MUSL on Jetson TX2 (#27957)

Ports https://github.com/dotnet/runtime/pull/206 to release/3.1.

The code in PAL_GetCurrentThreadAffinitySet relied on the fact that the
number of processors reported as configured in the system is always
larger than the maximum CPU index. However, it turns out that it is not
true on some devices / distros. The Jetson TX2 reports CPUs 0, 3, 4 and
5 in the affinity mask and the 1 and 2 are never reported. GLIBC reports
6 as the number of configured CPUs, however MUSL reports just 4. The
PAL_GetCurrentThreadAffinitySet was using the number of CPUs reported as
configured as the upper bound for scanning affinity set, so on Jetson
TX2, the affinity mask returned had just two bits set while there were
4 CPUs. That triggered an assert in the GCToOSInterface::Initialize.

This change fixes that by reading the maximum CPU index from the
/proc/cpuinfo. It falls back to using the number of processors
configured when the /proc/cpuinfo is not available (on macOS, FreeBSD, ...)

Fixes https://github.com/dotnet/runtime/issues/170

4 years agoupdate branding to 3.1.2 (#27981)
Anirudh Agnihotry [Tue, 14 Jan 2020 18:16:15 +0000 (10:16 -0800)]
update branding to 3.1.2 (#27981)

4 years agoUse 3.1.100 SDK in 3.1 build (#27965)
Matt Mitchell [Thu, 5 Dec 2019 23:18:05 +0000 (15:18 -0800)]
Use 3.1.100 SDK in 3.1 build (#27965)

To ensure that the 3.1 stack uses a consistent SDK.  The upper portions of the stack require the 3.1 SDK because for TFM purposes.

4 years ago[release/3.1] Update dependencies from dotnet/arcade (#27927)
dotnet-maestro[bot] [Thu, 5 Dec 2019 18:47:54 +0000 (10:47 -0800)]
[release/3.1] Update dependencies from dotnet/arcade (#27927)

* Update dependencies from https://github.com/dotnet/arcade build 20191122.3

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19572.3
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.19572.3
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.19572.3
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.19572.3

* Update dependencies from https://github.com/dotnet/arcade build 20191127.5

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19577.5
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.19577.5
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.19577.5
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.19577.5

4 years ago[automated] Merge branch 'release/3.0' => 'release/3.1' (#27922)
dotnet-maestro-bot [Mon, 2 Dec 2019 19:25:15 +0000 (11:25 -0800)]
[automated] Merge branch 'release/3.0' => 'release/3.1' (#27922)

* Fix Segfault in PerfInfo Image Logging (#26910)

* update branding for 3.0.2 (#27896)

* update branding

* remove stable properties from coreclr as it never stablises

* [release/3.0] Fix use of ilasm during test builds when using a 3.0 SDK (#27820)

* Add fix for restore of ilasm

The restore logic in buildtools tried to use a netcoreapp2.1 TFM for Linux-musl.
This ended up restoring Linux-x64 binaries which then broke the test build.
This change works around this by saving a copy of the depproj that BuildTools used into
the tree and restore it as a 3.0 app manually in init-tools on our side of the build.

* Change to a 3.0 SDK for servicing

* Bump test versions of ilasm and runtime package to match the RC

* Move off of the old arm32 queue (#27905)

* Add alpine3.10 testing to 3.0 (#27214)

* Fix GetSequencePoints when profiler provides mapping via SetILInstrumentedCodeMap (#27843)

Port #25802 to 3.0.2

* [release/3.0] Update dependencies from dotnet/core-setup (#27910)

* Update dependencies from https://github.com/dotnet/core-setup build 20191016.12

- Microsoft.NETCore.App - 3.0.1-servicing-19516-12

* Update dependencies from https://github.com/dotnet/core-setup build 20191121.02

- Microsoft.NETCore.App - 3.0.2-servicing-19571-02

4 years agoChange non_virtual_calls_to_instance_methods test to non-il-sdk format (#27952)
Juan Hoyos [Mon, 2 Dec 2019 15:53:14 +0000 (07:53 -0800)]
Change non_virtual_calls_to_instance_methods test to non-il-sdk format (#27952)

* Change non_virtual_calls_to_instance_methods test to non-il-sdk format

* Work around maxpath hit by method name

* Fix compile item to short file name

4 years agoDisable formatting jobs in release/3.1 (#27954)
Jarret Shook [Thu, 28 Nov 2019 15:39:52 +0000 (07:39 -0800)]
Disable formatting jobs in release/3.1 (#27954)

4 years agoupdate branding for 3.1 sevicing (#27921)
Anirudh Agnihotry [Tue, 26 Nov 2019 21:25:03 +0000 (13:25 -0800)]
update branding for 3.1 sevicing (#27921)

* update branding for 3.1 sevicing

* added internal branch

4 years agoNon virtual calls to instance methods non-virtual dispatch on instance interface...
David Wrighton [Fri, 22 Nov 2019 21:26:57 +0000 (13:26 -0800)]
Non virtual calls to instance methods non-virtual dispatch on instance interface methods (#27756) (#27868)

* est for non_virtual_calls_to_instance_methods
* Fix handling of callvirt to instance methods on interface types that are not virtual
- Use call type to indicate if its non-virtual or not, instead of opcode
* Use ilproj to protect against future C# compiler changes
- This test needs to test the use of specific opcodes, and so an IL proj is required

4 years agoMove off of the old arm32 queue (#27906)
Jarret Shook [Thu, 21 Nov 2019 17:31:01 +0000 (09:31 -0800)]
Move off of the old arm32 queue (#27906)

4 years ago[release/3.1] Update dependencies from dotnet/arcade (#26886)
dotnet-maestro[bot] [Thu, 21 Nov 2019 02:56:36 +0000 (18:56 -0800)]
[release/3.1] Update dependencies from dotnet/arcade (#26886)

* Update dependencies from https://github.com/dotnet/arcade build 20190924.3

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19474.3
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.19474.3
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.19474.3
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.19474.3

* Add lots o' sources

* Update dependencies from https://github.com/dotnet/arcade build 20191119.2

- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.19569.2
- Microsoft.DotNet.Build.Tasks.Feed - 2.2.0-beta.19569.2
- Microsoft.DotNet.Build.Tasks.Packaging - 1.0.0-beta.19569.2
- Microsoft.DotNet.Helix.Sdk - 2.0.0-beta.19569.2

4 years agoUpdate PreReleaseVersionLabel to RTM (#27885)
William Godbe [Fri, 15 Nov 2019 00:21:02 +0000 (16:21 -0800)]
Update PreReleaseVersionLabel to RTM (#27885)

4 years ago[release/3.1] Fix use of ilasm during test builds when using a 3.0 SDK (#27818)
Juan Hoyos [Mon, 11 Nov 2019 23:40:59 +0000 (15:40 -0800)]
[release/3.1] Fix use of ilasm during test builds when using a 3.0 SDK (#27818)

* Add fix for restore of ilasm

The restore logic in buildtools tried to use a netcoreapp2.1 TFM for Linux-musl.
This ended up restoring Linux-x64 binaries which then broke the test build.
This change works around this by saving a copy of the depproj that BuildTools used into
the tree and restore it as a 3.0 app manually in init-tools on our side of the build.

* Change to a 3.0 SDK for servicing

4 years agoRevert "stablixe package (#27740)" (#27816)
William Godbe [Mon, 11 Nov 2019 19:15:16 +0000 (11:15 -0800)]
Revert "stablixe package (#27740)" (#27816)

This reverts commit 2bc3c7c6feede9a40deb7474ecdafa8d0809d8af.

4 years agostablixe package (#27740)
Anirudh Agnihotry [Mon, 11 Nov 2019 17:31:24 +0000 (09:31 -0800)]
stablixe package (#27740)

4 years agoMerge pull request #27742 from dotnet/darc-release/3.1-d5e14434-9c47-489d-80e5-c9ee4d...
William Godbe [Mon, 11 Nov 2019 17:31:07 +0000 (09:31 -0800)]
Merge pull request #27742 from dotnet/darc-release/3.1-d5e14434-9c47-489d-80e5-c9ee4d41b969

[release/3.1] Update dependencies from dotnet/core-setup and dotnet/corefx

4 years ago[release/3.1] Reword comments for policheck (#27670)
Jeremy Koritzinsky [Mon, 11 Nov 2019 17:30:52 +0000 (09:30 -0800)]
[release/3.1] Reword comments for policheck (#27670)

* Comment rewordings required by policheck.

* Use AppContainer instead of Windows Store.

4 years agoFixup PrivateAssets->ExcludeAssets
wtgodbe [Fri, 8 Nov 2019 22:29:02 +0000 (14:29 -0800)]
Fixup PrivateAssets->ExcludeAssets