Rahul Kumar [Wed, 6 Jul 2016 21:00:04 +0000 (14:00 -0700)]
Merge pull request #6130 from rahku/crossgen
ARM64: Fix incremental build problem for cross-components
Rahul Kumar [Tue, 5 Jul 2016 22:04:59 +0000 (15:04 -0700)]
Fix incremental build problem for cross-components
Aditya Mandaleeka [Wed, 6 Jul 2016 19:32:47 +0000 (12:32 -0700)]
Merge pull request #6139 from mkborg/fix_lldb_search_order
Fix LLDB search order
Matt Ellis [Wed, 6 Jul 2016 18:38:36 +0000 (11:38 -0700)]
Merge pull request #6146 from ellismg/fix-sign-targets-syntax-error
Fix syntax error in sign.builds
Matt Ellis [Wed, 6 Jul 2016 16:55:23 +0000 (09:55 -0700)]
Fix syntax error in sign.builds
Koundinya Veluri [Wed, 6 Jul 2016 16:50:12 +0000 (09:50 -0700)]
Merge pull request #6015 from kouvel/NamedMutexMoreFlockCases
Named mutex: Use flock instead of pthread process-shared mutex in som…
Gaurav Khanna [Wed, 6 Jul 2016 14:19:00 +0000 (07:19 -0700)]
Merge pull request #6135 from dotnet-bot/UpdateDependencies20160706051943
Updating External dependencies to beta-24306-00
Mikhail I. Krivtsov [Wed, 6 Jul 2016 11:37:39 +0000 (14:37 +0300)]
Fix LLDB search order
By default 'find_library' without 'NO_DEFAULT_PATH' keyword searches
first in default locations specified by '${CMAKE_SYSTEM_PREFIX_PATH}'
and only then in locations specified via 'PATHS' keyword
('${WITH_LLDB_LIBS}' in our case). As result it always picks-up default
LLDB ignoring the one explicitly specified via '${WITH_LLDB_LIBS}'.
This change modifies search order giving priority to '${WITH_LLDB_LIBS}'
and '${WITH_LLDB_INCLUDES}' over default locations.
dotnet-bot [Wed, 6 Jul 2016 05:19:43 +0000 (05:19 +0000)]
Updating External dependencies to beta-24306-00
Koundinya Veluri [Mon, 27 Jun 2016 17:53:28 +0000 (10:53 -0700)]
Named mutex: Use flock instead of pthread process-shared mutex in some more cases
Workaround for #5456:
- Sometimes, a timed wait operation is not getting released, causing a hang
- Due to the hang, it is not possible to detect this issue with code
- Temporarily disabled the use of pthread process-shared mutexes on ARM/ARM64. File locks will be used instead.
Workaround for #5872:
- On Alpine Linux, a pthread process-shared robust mutex is detecting the case where a process abandons the mutex when it exits while holding the lock, but is putting the mutex into an unrecoverable state (ENOTRECOVERABLE) instead of assigning lock ownership to the next thread that is released from a wait for a lock and notifying of abandonment (EOWNERDEAD).
- Added a test case to detect this issue, to have it use file locks instead
Close #5456
Carol Eidt [Tue, 5 Jul 2016 22:55:48 +0000 (15:55 -0700)]
Merge pull request #6098 from CarolEidt/BlkNodeRefactors
More Blk Op Refactoring
Brian Sullivan [Tue, 5 Jul 2016 22:48:14 +0000 (15:48 -0700)]
Merge pull request #6102 from briansull/inliner-dump
Improve the dump for inline return expressions
Bruce Forstall [Tue, 5 Jul 2016 21:58:31 +0000 (14:58 -0700)]
Merge pull request #6101 from BruceForstall/RyuJITx86EH
RyuJIT/x86: add support for EH
Sivarv [Tue, 5 Jul 2016 21:47:48 +0000 (14:47 -0700)]
Merge pull request #6094 from mikedn/x86-uncontained-long
Allow uncontained GT_LONG nodes
Bruce Forstall [Tue, 5 Jul 2016 20:33:28 +0000 (13:33 -0700)]
Merge pull request #6124 from dotnet-bot/from-tfs
Merge changes from TFS
Bruce Forstall [Tue, 5 Jul 2016 20:29:59 +0000 (13:29 -0700)]
Merge pull request #6100 from BruceForstall/BotrX86EHUpdate
Add information on x86 EH to CLR ABI
Rahul Kumar [Tue, 5 Jul 2016 18:49:29 +0000 (11:49 -0700)]
Merge pull request #6093 from rahku/arm64retbuf
Revert Pinvoke ILStub calli signature for desktop clr
Maoni Stephens [Tue, 5 Jul 2016 18:47:29 +0000 (11:47 -0700)]
only use config on coreclr
[tfs-changeset: 1616092]
Bruce Forstall [Tue, 5 Jul 2016 16:31:21 +0000 (09:31 -0700)]
Merge pull request #6106 from mikedn/modopt-feedback
Address div/mod CR feedback
Carol Eidt [Sat, 2 Jul 2016 00:03:52 +0000 (17:03 -0700)]
More Blk Op Refactoring
Code refactoring in preparation for 1st Class Structs:
Factor out the legacy code generation for CpObj.
Abstract some functionality to member methods.
Move some block-related method definitions.
Make fgOrderBlockOps legacy-only (it is unused in RyuJIT).
Factor out the code to create a block init value.
Add a new IsLocalExpr() method.
Change some variable names to distinguish between objects and their addresses.
Stephen Toub [Tue, 5 Jul 2016 11:59:03 +0000 (07:59 -0400)]
Merge pull request #6120 from dotnet-bot/UpdateDependencies20160705051942
Updating External dependencies to beta-24305-00
James Ko [Tue, 5 Jul 2016 09:50:52 +0000 (05:50 -0400)]
Better performance for MulticastDelegate.Equals (#6113)
Better performance for MulticastDelegate.Equals
Viacheslav Nikolaev [Tue, 5 Jul 2016 06:33:46 +0000 (09:33 +0300)]
One of the flags MAP_SHARED or MAP_PRIVATE has to be specified. MAP_PRIVATE seems to fit here. (#6089)
dotnet-bot [Tue, 5 Jul 2016 05:19:42 +0000 (05:19 +0000)]
Updating External dependencies to beta-24305-00
Stephen Toub [Mon, 4 Jul 2016 10:09:55 +0000 (06:09 -0400)]
Merge pull request #6109 from dotnet-bot/UpdateDependencies20160704051959
Updating External dependencies to beta-24304-00
James Ko [Mon, 4 Jul 2016 05:39:12 +0000 (01:39 -0400)]
Short-circuit on the count, not the array length, in common Encoding methods (#6108)
* Short-circuit using count instead of array.Length
* Use Array.Empty in default Reset implementation
dotnet-bot [Mon, 4 Jul 2016 05:19:58 +0000 (05:19 +0000)]
Updating External dependencies to beta-24304-00
Maoni Stephens [Sun, 3 Jul 2016 20:06:03 +0000 (13:06 -0700)]
Merge pull request #6104 from Maoni0/config
Added 2 configs for Server GC
Mike Danes [Sun, 3 Jul 2016 16:46:15 +0000 (19:46 +0300)]
Address div/mod CR feedback
Maoni0 [Sun, 3 Jul 2016 04:07:23 +0000 (21:07 -0700)]
Added 2 configs for Server GC
complus var GCNoAffinitize or project.json System.GC.NoAffinitize - specify 1/true to disable hard affinity of Server GC threads to CPUs
complus var GCHeapCount or project.json System.GC.HeapCount - specify the # of Server GC threads/heaps, must be smaller than the # of logical CPUs the process is allowed to run on, ie, if you don't specifically affinitize your process it means the # of total logical CPUs on the machine; otherwise this is the # of logical CPUs you affinitized your process to.
Aditya Mandaleeka [Sun, 3 Jul 2016 01:11:31 +0000 (18:11 -0700)]
Merge pull request #6099 from adityamandaleeka/remove_virtual_owned
Remove VIRTUALOwnedRegion
Gaurav Khanna [Sat, 2 Jul 2016 16:12:54 +0000 (09:12 -0700)]
Merge pull request #6086 from dotnet-bot/UpdateDependencies20160701051708
Updating External dependencies to beta-24301-00
Sivarv [Sat, 2 Jul 2016 13:06:44 +0000 (06:06 -0700)]
Merge pull request #6092 from sivarv/lsrawork
Lower, LSRA and Codegen support for reg optional operands
Mike Danes [Fri, 1 Jul 2016 19:54:04 +0000 (22:54 +0300)]
Allow uncontained GT_LONG nodes
When the result of a TYP_LONG operation is not used the GT_LONG node inserted during decomposition remains in the tree and unlike other GT_LONG nodes is not contained and needs special handling.
This usually happens with arithmetic operations that require overflow checks, the result of the operation may be ignored but the operation itself cannot be eliminated because it has a side-effect.
Brian Sullivan [Sat, 2 Jul 2016 01:31:16 +0000 (18:31 -0700)]
Improve the dump for inline return expressions
Also don't do a deep copy of the inline return expression
Bruce Forstall [Sat, 2 Jul 2016 00:57:57 +0000 (17:57 -0700)]
Add information on x86 EH to CLR ABI
Aditya Mandaleeka [Sat, 2 Jul 2016 00:09:47 +0000 (17:09 -0700)]
Remove VIRTUALOwnedRegion.
sivarv [Fri, 1 Jul 2016 18:41:02 +0000 (11:41 -0700)]
Consider spilled lcl var as contained memory operands for codegen purpose.
Bruce Forstall [Fri, 1 Jul 2016 23:55:01 +0000 (16:55 -0700)]
RyuJIT/x86: add support for EH
This change implements the necessary x86 support in RyuJIT for exception handling.
It removes the EH-related NYI. The number of NYI fallbacks in an x86 test run drops
by about 13000. There are 10 new asserts which are now suppressed: 6 appear to be
new and 4 are new instances of existing known asserts.
The various changes are:
1. Implement code generation for BBJ_CALLFINALLY/BBJ_ALWAYS blocks for invoking a finally.
This requires appropriately modifying the ShadowSP slot structure before the call.
2. Implement BBJ_EHFINALLYRET/BBJ_EHFILTERRET control transfer. The related GT_RETFILT
code was already implemented.
3. Implement GT_END_LFIN for handling return-from-finally cleanup, namely, clearing the
appropriate slot in the ShadowSP structure.
Bruce Forstall [Fri, 1 Jul 2016 23:47:51 +0000 (16:47 -0700)]
Merge pull request #6096 from dotnet-bot/from-tfs
Merge changes from TFS
Maoni Stephens [Fri, 1 Jul 2016 21:47:30 +0000 (14:47 -0700)]
Fixed build breaks related to perf counter code on desktop. perf counters are only enabled on desktop.
[tfs-changeset: 1615834]
Rahul Kumar [Fri, 1 Jul 2016 19:20:39 +0000 (12:20 -0700)]
Revert Pinvoke ILStub calli signature for desktop clr
Aditya Mandaleeka [Fri, 1 Jul 2016 20:11:10 +0000 (13:11 -0700)]
Merge pull request #6083 from adityamandaleeka/upload_ci_dumps
Start using dumpling service to upload dumps from CI runs
Pat Gavlin [Fri, 1 Jul 2016 19:33:16 +0000 (12:33 -0700)]
Merge pull request #6031 from mikedn/x86-float-unspill
Remove misplaced call to genUnspillRegIfNeeded
Maoni Stephens [Fri, 1 Jul 2016 17:49:06 +0000 (10:49 -0700)]
Merge pull request #6084 from Maoni0/gc_etw
GC ETW fixes
William Godbe [Fri, 1 Jul 2016 17:22:44 +0000 (10:22 -0700)]
Merge pull request #6091 from wtgodbe/buildfix
do all coreclr copying BEFORE building
wtgodbe [Fri, 1 Jul 2016 17:15:06 +0000 (10:15 -0700)]
do all coreclr copying BEFORE building
chunseoklee [Fri, 1 Jul 2016 08:41:20 +0000 (17:41 +0900)]
FIX #6074 : cannot find exception handler on arm (#6088)
dotnet-bot [Fri, 1 Jul 2016 05:17:08 +0000 (05:17 +0000)]
Updating External dependencies to beta-24301-00
Kyungwoo Lee [Fri, 1 Jul 2016 01:49:05 +0000 (18:49 -0700)]
Merge pull request #6082 from kyulee1/gc3
ARM64: Enable GCStress0x3 Private Job
Andy Ayers [Fri, 1 Jul 2016 00:47:24 +0000 (17:47 -0700)]
Merge pull request #6078 from AndyAyersMS/ModelPolicyAvailability
Inliner: make ModelPolicy available in release builds
Maoni0 [Fri, 24 Jun 2016 22:45:51 +0000 (15:45 -0700)]
GC ETW fixes. The code for getting data for ETW and perf counter was written by different
folks and was very intertwined. For full CLR we always have both defined but for coreclr
perf counters are not enabled so some things for ETW were just 0. Need to make sure when one
is not defined the rest are still getting the data they need.
Pat Gavlin [Thu, 30 Jun 2016 23:36:19 +0000 (16:36 -0700)]
Merge pull request #6077 from pgavlin/RemoveGtfDead
Remove GTF_DEAD.
Mike McLaughlin [Thu, 30 Jun 2016 22:49:46 +0000 (15:49 -0700)]
Fix the hang in the debugger tests repo on OS X. (#6079)
The transport pipe closes during shudown where blocked/deadlocked with the pending read on the transport thread.
Aditya Mandaleeka [Thu, 30 Jun 2016 22:20:22 +0000 (15:20 -0700)]
Upload CI dumps to dumpling.
Kyungwoo Lee [Thu, 30 Jun 2016 22:00:07 +0000 (15:00 -0700)]
ARM64: Enable GCStress0x3 Private Job
William Godbe [Thu, 30 Jun 2016 21:40:53 +0000 (14:40 -0700)]
Merge pull request #6080 from wtgodbe/artifacts
Use HTML publisher for code coverage reports in CI
wtgodbe [Thu, 30 Jun 2016 21:32:44 +0000 (14:32 -0700)]
Use HTML publisher for code coverage reports in CI
Andy Ayers [Tue, 28 Jun 2016 22:07:01 +0000 (15:07 -0700)]
Inliner: make ModelPolicy available in release builds
Make the ModelPolicy available in release builds of the jit. Enable this
policy by setting:
COMPLUS_JitInlinePolicyModel=1
This works for either jitting or prejitting.
Update the ModelPolicy so callee never inline decisions are always
propagated back to the runtime.
Pat Gavlin [Thu, 30 Jun 2016 17:26:05 +0000 (10:26 -0700)]
Remove GTF_DEAD.
This flag was only used by CSE to mark dead trees s.t. these trees
were not considered during CSE replacement. The same information,
however, can be provided by the gtCSEnum field: a tree with this
field set to NO_CSE is treated identically to a tree marked with
GTF_DEAD. The GTF_DEAD flag is therefore redundant and can be
eliminated.
This is part of #5999.
Mike Danes [Tue, 28 Jun 2016 18:28:51 +0000 (21:28 +0300)]
Remove misplaced call to genUnspillRegIfNeeded
Evgeny Pavlov [Thu, 30 Jun 2016 17:40:25 +0000 (21:40 +0400)]
[SOS] [Linux] Initial support of reading source and line number by native offset (#6010)
* Initial support of reading source and line number by native offset
using portable PDB reader
Jarret Shook [Thu, 30 Jun 2016 15:41:41 +0000 (08:41 -0700)]
Merge pull request #5958 from jashook/disable_timeouts_under_gcstress
Disable tests that timeout under gcstress
Mike McLaughlin [Thu, 30 Jun 2016 05:06:06 +0000 (22:06 -0700)]
Fix undefined locals in desktop test builds. (#6073)
When for desktop out of the ProjectK branch, the previous change left unassigned member variables. Moved the pragma to include them.
Kyungwoo Lee [Thu, 30 Jun 2016 03:20:53 +0000 (20:20 -0700)]
Merge pull request #6075 from kyulee1/settime
ARM64: Set timeout per scenario
Kyungwoo Lee [Thu, 30 Jun 2016 01:26:41 +0000 (18:26 -0700)]
ARM64: Set timeout per scenario
William Godbe [Wed, 29 Jun 2016 21:44:42 +0000 (14:44 -0700)]
Merge pull request #6072 from wtgodbe/archive2
Archive the right artifacts in coverage build
wtgodbe [Wed, 29 Jun 2016 21:36:01 +0000 (14:36 -0700)]
Archive the right artifacts in coverage build
Brian Sullivan [Wed, 29 Jun 2016 20:46:55 +0000 (13:46 -0700)]
Merge pull request #6044 from briansull/split-lvIsMultiRegArgOrRet
Split the lvIsMultiiRegArgOrRet into two bits
Aditya Mandaleeka [Wed, 29 Jun 2016 19:50:06 +0000 (12:50 -0700)]
Merge pull request #6041 from adityamandaleeka/fix_strip_logic
Fix platform logic in symbol stripping code
William Godbe [Wed, 29 Jun 2016 19:20:16 +0000 (12:20 -0700)]
Merge pull request #6070 from wtgodbe/sign
Only sign non-arm64 binaries
Andy Ayers [Wed, 29 Jun 2016 19:12:07 +0000 (12:12 -0700)]
Merge pull request #6069 from dotnet-bot/from-tfs
Merge changes from TFS
wtgodbe [Wed, 29 Jun 2016 18:14:53 +0000 (11:14 -0700)]
Only sign non-arm64 binaries
Brian Sullivan [Wed, 29 Jun 2016 00:32:33 +0000 (17:32 -0700)]
Split the lvIsMultiRegArgOrRet into two bits
Currently for all targets that have MultiReg return LclVars we require that they not be struct promoted.
But on ARM64 we do allow for MultiReg #argument LclVars to be struct promoted.
Thus we have to disentagle this value into two distinct bool values to implement MultiReg returns for Arm64.
jashoo [Wed, 29 Jun 2016 17:32:41 +0000 (10:32 -0700)]
Disable tests that timeout under gcstress
After increasing the timeout time for gcstress, these tests continue to fail
due to timeouts. To get the gcstress legs green and keep them within a manageable
amount of time they will be disabled.
Andy Ayers [Wed, 29 Jun 2016 17:22:43 +0000 (10:22 -0700)]
Have the jit capture the SuperPMI method index for use in correlating outputs from different runs.
Useful for matching SPMI outputs in cases where jit changes cause differential replay failures so lining up results side-by-side doesn't work.
If the jit is not hosted under SPMI, the index will report as zero.
[tfs-changeset: 1615278]
Mike McLaughlin [Wed, 29 Jun 2016 15:52:57 +0000 (08:52 -0700)]
Fix building on desktop bug #234633. (#6045)
ifdef the references to System.Diagnostic.StackFrame.
Kyungwoo Lee [Wed, 29 Jun 2016 15:51:07 +0000 (08:51 -0700)]
Merge pull request #6054 from kyulee1/gcjob
ARM64: Enable a gcstress private job
Kyungwoo Lee [Wed, 29 Jun 2016 15:30:27 +0000 (08:30 -0700)]
Merge pull request #6040 from rahku/arm64retbuf
ARM64: update tests.lst
Gaurav Khanna [Wed, 29 Jun 2016 15:25:26 +0000 (08:25 -0700)]
Merge pull request #6000 from gkhanna79/Arm64Nuget
Generate nuget packages for Arm64 build
Kyungwoo Lee [Wed, 29 Jun 2016 15:20:10 +0000 (08:20 -0700)]
ARM64: Enable a gcstress private job
Gaurav Khanna [Wed, 29 Jun 2016 15:16:48 +0000 (08:16 -0700)]
Merge pull request #6049 from dotnet-bot/UpdateDependencies20160629051946
Updating External dependencies to beta-24229-00
Kyungwoo Lee [Wed, 29 Jun 2016 14:52:37 +0000 (07:52 -0700)]
Merge pull request #6053 from kyulee1/ilrtrel
Relase build only for ilasm round trip tests in CI
Kyungwoo Lee [Wed, 29 Jun 2016 14:41:41 +0000 (07:41 -0700)]
Relase build only for ilasm round trip tests in CI
This eliminates spurious job display in Jenkins web which is unnecessary.
Kyungwoo Lee [Wed, 29 Jun 2016 13:43:15 +0000 (06:43 -0700)]
Merge pull request #5950 from wateret/fix-5949
[ARM/Linux] Disable unittest `_il_reltypeEqualOp`
Hanjoung Lee [Wed, 29 Jun 2016 05:55:37 +0000 (14:55 +0900)]
Remove unittests that contain undefined behaviour
Following tests contain undefined behaviour.
(Casting double to an integer type for a value that cannot be represent as the integer type.)
- JIT.Methodical.casts.ilseq._il_reltypeEqualOp._il_reltypeEqualOp
- JIT.Regression.CLR-x86-JIT.V1-M10.b08107
dotnet-bot [Wed, 29 Jun 2016 05:19:46 +0000 (05:19 +0000)]
Updating External dependencies to beta-24229-00
Bruce Forstall [Wed, 29 Jun 2016 03:24:53 +0000 (20:24 -0700)]
Merge pull request #6043 from BruceForstall/DisplayPlatformInJitDump
Display the build platform and architecture in the JitDump
Bruce Forstall [Tue, 28 Jun 2016 23:55:25 +0000 (16:55 -0700)]
Display the build platform and architecture in the JitDump
Bruce Forstall [Tue, 28 Jun 2016 23:54:08 +0000 (16:54 -0700)]
Merge pull request #6038 from BruceForstall/PinvokeFeedbackCleanup
Address P/Invoke inlining code review feedback and cleanup
William Godbe [Tue, 28 Jun 2016 23:34:19 +0000 (16:34 -0700)]
Merge pull request #6042 from wtgodbe/archive
Archive coverage reports from the right place
wtgodbe [Tue, 28 Jun 2016 23:30:28 +0000 (16:30 -0700)]
Archive coverage reports from the right place
Brian Sullivan [Tue, 28 Jun 2016 22:51:05 +0000 (15:51 -0700)]
Merge pull request #6035 from briansull/fix-5642
Fix issue #5642
Brian Sullivan [Tue, 28 Jun 2016 22:41:47 +0000 (15:41 -0700)]
Merge pull request #6036 from briansull/dump-fix-inlining
Do not enable the verbose Jit Dump during the inlining decision phase.
Aditya Mandaleeka [Tue, 28 Jun 2016 22:37:25 +0000 (15:37 -0700)]
Fix path to BOTR GC section in gc.cpp (#6039)
Aditya Mandaleeka [Tue, 28 Jun 2016 22:25:11 +0000 (15:25 -0700)]
Fix platform logic in symbol stripping code.
Rahul Kumar [Tue, 28 Jun 2016 22:14:31 +0000 (15:14 -0700)]
update arm64 tests.lst
Bruce Forstall [Tue, 28 Jun 2016 21:56:50 +0000 (14:56 -0700)]
Address P/Invoke inlining code review feedback and cleanup
Address some code review feedback from #5939. Also, do a little cleanup.
Specifically:
1. Make many GenTreeCall accessors 'const'.
2. HasNonStandardArgs() and GetNonStandardArgCount() are used for a specific purpose in the fast
tailcall implementation. However, the fgMorphArgs() "non-standard args" mechanism is now used for
other purposes, so the names of these didn't match their intent. Also, the logic had decayed compared
to the fgMorphArgs logic it was mimicing. I renamed them to add the word "Added", and wrote some
comments to be more explicit about their use, as well as indicate in fgMorphArgs() that they need
to be kept in sync.
3. Added several GenTreeCall::IsHelperCall() accessors, and replace some code with them.
4. Added RBM_INIT_PINVOKE_FRAME_TRASH define to remove an `#ifdef`.
5. Removed an assert loop in LowerFastTailCall() since it's no longer accurate given the new users of `isNonStandard`.
Rahul Kumar [Tue, 28 Jun 2016 22:04:33 +0000 (15:04 -0700)]
Merge pull request #5978 from rahku/arm64retbuf
Fix Pinvoke IL Stubs to emit correct return type in signature of call…
Rahul Kumar [Wed, 22 Jun 2016 21:48:37 +0000 (14:48 -0700)]
Fix Pinvoke IL Stubs to emit correct return type in signature of calli instruction