Chris McKinsey [Fri, 24 Jun 2016 19:05:47 +0000 (12:05 -0700)]
Merge pull request #5973 from cmckinsey/DevDiv_216571
Fix SBCG in value-numbering constant assertprop
Brian Sullivan [Fri, 24 Jun 2016 18:18:07 +0000 (11:18 -0700)]
Merge pull request #5940 from briansull/retbuf-nonstandard
ARM64: Change the fixed return buffer argument to be handled as a non-standard argument
Chris McKinsey [Tue, 21 Jun 2016 17:59:32 +0000 (10:59 -0700)]
Fix SBCG in value-numbering constant assertprop
Value-numbering can assign the same value number to nodes
of different type as long as they have the same bit-representation.
However when substituting a constant into a tree node of different
type, we want the bit-exact representation and not the same value.
Assertion prop should reinterpret int/float value changes rather
than coercing them.
Add an il test that exposes a bug in assertionprop when doing
value-numbering based constant propagation. A cpblk can be used
to copy a scalar value between differing types but for which there
is no change in bit representation.
Gaurav Khanna [Fri, 24 Jun 2016 13:55:32 +0000 (06:55 -0700)]
Merge pull request #5986 from dotnet-bot/UpdateDependencies20160624051958
Updating External dependencies to beta-24224-00
Ivan Baravy [Fri, 24 Jun 2016 12:59:51 +0000 (16:59 +0400)]
Fix typo to apply mac build workaround only on OSX (#5988)
/init-tools.sh:72 tries to compare strings like this:
if [ "$OS"=="OSX" ]; then
What it actually does is different to what it looks like:
$ echo $OS
$ if [ "$OS"=="OSX" ]; then echo EQUAL; fi
EQUAL
$ if [ "$OS" == "OSX" ]; then echo EQUAL; fi
$
This commit fixes the typo and makes OSX workaround run on macs only.
chunseoklee [Fri, 24 Jun 2016 12:56:55 +0000 (21:56 +0900)]
Fix Issue #5542 (#5827)
On arm32, NullReference exception for interface call is not handled
properly. Especially, when callsite is a virtualstub(dispatch/resolve), AV exception
occurs on indirect cell. Since this stub is not a part of
managed or markedjithelper, exception handler considers this is a native
seg fault.
I added :
(1) FaultingExceptionFrame construction routine for virtualstub
(2) Checking and IP-adjusting routine for VirtualStub
MyungJoo Ham [Fri, 24 Jun 2016 12:51:53 +0000 (21:51 +0900)]
Use BIT64 define for 64-bit specific code. (#5987)
Using WIN32/WIN64 for 32/64-bit architecture dependent code confuses
developers especially those who develop coreclr for non-Windows
systems. Therefore, such definitions are modified to BIT64.
Fixes #4737
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
dotnet-bot [Fri, 24 Jun 2016 05:19:58 +0000 (05:19 +0000)]
Updating External dependencies to beta-24224-00
Jarret Shook [Fri, 24 Jun 2016 03:35:02 +0000 (20:35 -0700)]
Merge pull request #5983 from prajwal-aithal/devel/arm-ci-checked-fix
ARM-CI: Prevent creation of job for Checked configuration (rebased over #5961)
Jan Kotas [Fri, 24 Jun 2016 03:18:16 +0000 (20:18 -0700)]
Merge pull request #5980 from dotnet-bot/from-tfs
Merge changes from TFS
Prajwal A N [Thu, 23 Jun 2016 07:14:18 +0000 (16:14 +0900)]
ARM-CI: Prevent creation of job for Checked configuration
Previously we checked if the configuration being built is either of Debug or Release
after the job had already been created. This resulted in a job being created for
Checked configuration with no commands to execute (for this job).
We prevent the unnecessary creation of job for the Checked configuration by
returning control at the very beginning of the <LinuxARMEmulator, arm, Checked> iteration
Signed-off-by: Prajwal A N <an.prajwal@samsung.com>
Pat Gavlin [Fri, 24 Jun 2016 02:39:35 +0000 (19:39 -0700)]
Merge pull request #5960 from pgavlin/CheckDefPython3
Make check-definitions compatible with Python 3.
Chris McKinsey [Fri, 24 Jun 2016 02:18:52 +0000 (19:18 -0700)]
Merge pull request #5923 from mikedn/x86-store-lclvar
Fix statement insertion point in GT_STORE_LCL_VAR decomposition
Evgeny Pavlov [Fri, 24 Jun 2016 02:00:47 +0000 (06:00 +0400)]
[SOS][Linux] Support of reading local variables from portable PDB (#5897)
* Initial support of reading local variables from portable pdb using System.Diagnostics.Debug.SymbolReader.dll
* Use SysAllocString and SysFreeString for memory management
* Fix coding style after review
Jarret Shook [Fri, 24 Jun 2016 00:46:28 +0000 (17:46 -0700)]
Merge pull request #5961 from jashook/arm32_ci_ubuntu_cross_fix
Seperate the ubuntu-arm build and emulator build
Jan Kotas [Thu, 23 Jun 2016 23:52:36 +0000 (16:52 -0700)]
Undo accidental change
[tfs-changeset: 1614521]
Jan Kotas [Thu, 23 Jun 2016 23:51:31 +0000 (16:51 -0700)]
Add missing contracts
[tfs-changeset: 1614520]
jashoo [Thu, 23 Jun 2016 17:23:09 +0000 (10:23 -0700)]
Seperate the ubuntu-arm build and emulator build
The name for the arm32 ubuntu cross build and the arm32 emulator build are the same. This
also disables the ubuntu cross build per pr.
Russ Keldorph [Thu, 23 Jun 2016 23:39:03 +0000 (16:39 -0700)]
Merge pull request #5968 from RussKeldorph/jitstress
Move R2R JitStress jobs into the jitstress folder with the others.
Pat Gavlin [Thu, 23 Jun 2016 23:32:48 +0000 (16:32 -0700)]
Merge pull request #5959 from pgavlin/AOTPlaceholderFile
Add AOT placeholder files to the JIT package.
Jan Kotas [Thu, 23 Jun 2016 21:44:44 +0000 (14:44 -0700)]
Add braces to avoid operator priority confusion (#5964)
Brian Sullivan [Thu, 23 Jun 2016 21:15:37 +0000 (14:15 -0700)]
Support for updating a nonStandard arg in-place the ArrayStack when it is modified by fgMorphTree
Aditya Mandaleeka [Thu, 23 Jun 2016 20:59:35 +0000 (13:59 -0700)]
Merge pull request #5963 from dotnet-bot/from-tfs
Merge changes from TFS
Pat Gavlin [Thu, 23 Jun 2016 16:41:49 +0000 (09:41 -0700)]
Add AOT placeholder files to the JIT package.
This is intended to prevent accidental inclusion in UWP projects.
Ivan Baravy [Thu, 23 Jun 2016 20:18:06 +0000 (00:18 +0400)]
Fix GC profiling on arm-softfp (#5952)
This commit updates one preprocessor directive inside gc.h to initialize
dwEtwRootKind field of ScanContext when GC_PROFILING is defined while
FEATURE_EVENT_TRACE is not. This happens to platforms like arm-softfp.
Russ Keldorph [Thu, 23 Jun 2016 19:28:07 +0000 (12:28 -0700)]
Move R2R JitStress jobs into the jitstress folder with the others.
Bruce Forstall [Thu, 23 Jun 2016 19:32:21 +0000 (12:32 -0700)]
Merge pull request #5939 from BruceForstall/FixPinvoke
Fix RyuJIT/x86 P/Invoke inlining
Brian Sullivan [Thu, 23 Jun 2016 18:25:53 +0000 (11:25 -0700)]
Merge branch 'master' into retbuf-nonstandard
Jan Kotas [Thu, 23 Jun 2016 18:13:01 +0000 (11:13 -0700)]
Fix build break in internal build
[tfs-changeset: 1614475]
Rahul Kumar [Thu, 23 Jun 2016 18:06:43 +0000 (11:06 -0700)]
Merge pull request #5935 from rahku/test
Fix test runner batch file to provide correct TPA path
Brian Sullivan [Thu, 23 Jun 2016 17:37:53 +0000 (10:37 -0700)]
Merge pull request #5936 from briansull/fullIntArgRegMask
Introduce the method fullIntArgRegMask() to replace many uses of RBM_ARG_REGS
Matt Ellis [Thu, 23 Jun 2016 17:35:38 +0000 (10:35 -0700)]
Merge pull request #5937 from ellismg/pass-build-id-flag-to-linker
Pass --build-id=sha1 to linker explicitly
Jarret Shook [Thu, 23 Jun 2016 17:28:50 +0000 (10:28 -0700)]
Merge pull request #5962 from dotnet/revert-5945-devel/arm-ci-checked-fix
Revert "ARM-CI: Prevent creation of job for Checked configuration"
Jarret Shook [Thu, 23 Jun 2016 17:28:42 +0000 (10:28 -0700)]
Revert "ARM-CI: Prevent creation of job for Checked configuration"
Pat Gavlin [Thu, 23 Jun 2016 17:10:25 +0000 (10:10 -0700)]
Make check-definitions compatible with Python 3.
Import and use the Python 3-compatible print_function from __future__.
Bruce Forstall [Thu, 12 May 2016 14:31:13 +0000 (07:31 -0700)]
Enable RyuJIT/x86 PInvoke lowering
Fixes #4181 "NYI_X86: Implement PInvoke frame init inlining for x86"
The main work here is to handle the custom calling convention for the
x86 CORINFO_HELP_INIT_PINVOKE_FRAME helper call: it takes EDI as an argument,
trashes only EAX, and returns the TCB in ESI.
The code changes are as follows:
1. Lowering::InsertPInvokeMethodProlog(): don't pass the "secret stub param" for x86.
Also, don't store the InlinedCallFrame.m_pCallSiteSP in the prolog: for x86 this is done
at the call site, due to the floating stack pointer.
2. LinearScan::getKillSetForNode(): for helper calls, call compHelperCallKillSet() to get the killMask,
to account for non-standard kill sets.
3. Morph.cpp::fgMorphArgs(): set non-standard arguments for CORINFO_HELP_INIT_PINVOKE_FRAME.
4. compHelperCallKillSet(): set the correct kill set for CORINFO_HELP_INIT_PINVOKE_FRAME.
5. codegenxarch.cpp::genCallInstruction(): set the ABI return register for CORINFO_HELP_INIT_PINVOKE_FRAME.
6. lowerxarch.cpp::TreeNodeInfoInit(): set the GT_CALL dstCandidates for CORINFO_HELP_INIT_PINVOKE_FRAME.
5 & 6 are both needed to avoid a copy.
With this change, the #1 NYI with 18415 hits over the tests is gone.
The total number of NYI is now 29516.
kvochko [Thu, 23 Jun 2016 16:23:20 +0000 (20:23 +0400)]
Enable out-of-proc unwinding on ARM (#5946)
Fix #5855
Michelle McDaniel [Thu, 23 Jun 2016 15:14:49 +0000 (08:14 -0700)]
Merge pull request #5890 from adiaaida/x86LongCallFinal
Enable GT_CALL with long ret types for x86 RyuJIT
Jarret Shook [Thu, 23 Jun 2016 15:05:21 +0000 (08:05 -0700)]
Merge pull request #5945 from prajwal-aithal/devel/arm-ci-checked-fix
ARM-CI: Prevent creation of job for Checked configuration
Gaurav Khanna [Thu, 23 Jun 2016 14:17:56 +0000 (07:17 -0700)]
Merge pull request #5944 from dotnet-bot/UpdateDependencies20160623052003
Updating External dependencies to beta-24223-00
MyungJoo Ham [Thu, 23 Jun 2016 11:13:46 +0000 (20:13 +0900)]
Scripts: verify compiler definitions of native and managed (#4675)
* Scripts: find out compiler definitions of CMake
In order to find mismatch between native and managed,
we need to know the list of definitions of native.
The copmiler definitions are stored at cmake.definitions
This addresses the complaints of #4674
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
* Scripts: add check-definitions.py
scripts/check-definitions.py checks the consistency between
the native-build (CMake) compiler definitions and
the managed-build (MSBuild/mscorlib) compiler definitions
at build-time and prints out potentially dangerous
inconsistencies.
In order to get the proper results, managed build should
be executed after the native build (build.sh will do so
if no options such as skipnative or skipmanaged are given.)
Fix #4674
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
* Scripts: allow check-definitions py to ignore the harmless
The third argument of check-definitions.py specifies harmless keywords
to be suppressed from emitting warning messages.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
* Scripts: add ignored cdefine keywords for warning
As an example of how to declare compiler definition keywords
that are harmless to be inconsistent between the native and the
managed, we have added FEATURE_IMPLICIT_TLS and FEATURE_HIJACK.
Developers may add more keywords in System.Private.CoreLib.csproj
if the keywords are verified to be harmless; i.e., although
the keywords exist in both cmake and clr.coreclr.props,
the keywords are NEVER used in either side of the sources
or the keywords only happen to have the same name while they
denote the completely different semantics and may be disjoint.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Prajwal A N [Thu, 23 Jun 2016 07:14:18 +0000 (16:14 +0900)]
ARM-CI: Prevent creation of job for Checked configuration
Previously we checked if the configuration being built is either of Debug or Release
after the job had already been created. This resulted in a job being created for
Checked configuration with no commands to execute (for this job).
We prevent the unnecessary creation of job for the Checked configuration by
returning control at the very beginning of the <LinuxARMEmulator, arm, Checked> iteration
Signed-off-by: Prajwal A N <an.prajwal@samsung.com>
Maoni Stephens [Thu, 23 Jun 2016 07:07:49 +0000 (00:07 -0700)]
Merge pull request #5934 from Maoni0/gcwhere
enable GCWhere on PAL
Sean Gillespie [Thu, 23 Jun 2016 06:40:41 +0000 (23:40 -0700)]
Build the GC as part of its own CMake target (#5932)
dotnet-bot [Thu, 23 Jun 2016 05:20:03 +0000 (05:20 +0000)]
Updating External dependencies to beta-24223-00
Jarret Shook [Thu, 23 Jun 2016 03:46:08 +0000 (20:46 -0700)]
Merge pull request #5211 from prajwal-aithal/devel/automate-arm-ci
ARM-CI: Make the Linux ARM emulator check automatic
Prajwal A N [Wed, 25 May 2016 10:48:38 +0000 (19:48 +0900)]
ARM-CI: Make the Linux ARM emulator check automatic
* Previously the Linux ARM emulator build had to be triggered manually
* Making this check automatic based on the stability of emulator build
Signed-off-by: Prajwal A N <an.prajwal@samsung.com>
Jarret Shook [Thu, 23 Jun 2016 02:49:09 +0000 (19:49 -0700)]
Merge pull request #5942 from prajwal-aithal/devel/arm-ci-results-fix
ARM-CI: Resolving typo in result path
Prajwal A N [Thu, 23 Jun 2016 02:14:10 +0000 (11:14 +0900)]
ARM-CI: Resolving typo in result path
Signed-off-by: Prajwal A N <an.prajwal@samsung.com>
Jarret Shook [Thu, 23 Jun 2016 01:54:49 +0000 (18:54 -0700)]
Merge pull request #5941 from prajwal-aithal/devel/arm-ci-results-fix
ARM-CI: Change result archiving from emulated to cross build
Prajwal A N [Thu, 23 Jun 2016 01:16:34 +0000 (10:16 +0900)]
ARM-CI: Change result archiving from emulated to cross build
Previous PR assumed emulated build and thus archived the binaries inside the emulator.
But since cross build produces binaries inside the source folder, change the result path
Signed-off-by: Prajwal A N <an.prajwal@samsung.com>
Brian Sullivan [Thu, 23 Jun 2016 00:40:39 +0000 (17:40 -0700)]
Change the Arm64 fixed return buffer argument to be handled as a non-standard argument
We already have support for the handling of non-standard arguments and
it is simpler and better to handle the fixed return buffer argument as a non-standard argument.
Matt Ellis [Thu, 23 Jun 2016 00:31:15 +0000 (17:31 -0700)]
Pass --build-id=sha1 to linker explicitly
On some platforms, a build-id was not being added to native artifacts
and we would like it to be present. Explicitly pass --build-id=sha1 to
the linker.
Fixes #5796
Brian Sullivan [Wed, 22 Jun 2016 21:36:45 +0000 (14:36 -0700)]
Introduce the method fullIntArgRegMask()
This method returns the full integer argument register mask
that also include the fixed return buffer argument register.
Brian Sullivan [Thu, 23 Jun 2016 00:27:32 +0000 (17:27 -0700)]
Merge pull request #5931 from briansull/fix-crossgen-arm64
Fix for Arm64 crossgen break
Maoni0 [Wed, 22 Jun 2016 19:23:11 +0000 (12:23 -0700)]
enable GCWhere on PAL and fixed a bug in 859307b where it's missing a function in Linux
Rahul Kumar [Thu, 23 Jun 2016 00:14:38 +0000 (17:14 -0700)]
Fix test runner batch file to provide correct TPA path
Aditya Mandaleeka [Thu, 23 Jun 2016 00:11:12 +0000 (17:11 -0700)]
Merge pull request #3484 from Dmitry-Me/fixBrokenCondition
Fix impossible condition
Jarret Shook [Wed, 22 Jun 2016 23:49:49 +0000 (16:49 -0700)]
Merge pull request #5933 from prajwal-aithal/devel/arm-ci-script-fix
ARM-CI: Use CI cross build script in emulator
Prajwal A N [Wed, 22 Jun 2016 23:21:54 +0000 (08:21 +0900)]
ARM-CI: Use CI cross build script in emulator
Previous PRs added the build commands to the groovy script.
But this did not work for various reasons.
The current patch makes the CI system execute the bash script
that does the cross build. The bash script is already supplied with the emulator
Signed-off-by: Prajwal A N <an.prajwal@samsung.com>
Bruce Forstall [Wed, 22 Jun 2016 23:15:41 +0000 (16:15 -0700)]
Merge pull request #5929 from BruceForstall/CleanUpPinvokeLowering
Clean up PInvoke lowering
Brian Sullivan [Wed, 22 Jun 2016 22:48:45 +0000 (15:48 -0700)]
Fix for Arm64 crossgen break
Remorphing a call with an x8 Return buffer arg asserts
Bruce Forstall [Wed, 22 Jun 2016 21:37:31 +0000 (14:37 -0700)]
Clean up PInvoke lowering
Mostly, add lots of comments.
Michelle McDaniel [Tue, 31 May 2016 15:40:16 +0000 (08:40 -0700)]
Enable GT_CALL with long ret types for x86 RyuJIT
1) Enables genMultiRegCallStoreToLocal for x86 RyuJIT. Forces long return
types to be stored to the stack after returning from a call.
2) Update lvaPromoteLongVars to not promote a long if it is a multi reg
arg or ret.
3) Adds NYI for call arguments that are longs and contain adds or
subtracts. We are currently producing the wrong order of operations so
that we can push the argument immediately after performing the arithmetic.
We will do the hi adc/sbb before the carry bit has been set by doing the
lo operation.
4) Adds an NYI for morphing a node into a call node if the call will have
a long return type.
5) Moves the logic for forcing var = call() for calls with long return
types to lower::DecomposeNode().
Mike Danes [Wed, 22 Jun 2016 20:59:17 +0000 (23:59 +0300)]
Add regression test
Kyungwoo Lee [Wed, 22 Jun 2016 20:43:29 +0000 (13:43 -0700)]
Merge pull request #5926 from kyulee1/cmpxchg
ARM64: Skip CompareExchangeTClass Test
Bruce Forstall [Wed, 22 Jun 2016 20:41:25 +0000 (13:41 -0700)]
Merge pull request #5915 from BruceForstall/FgMorphArgsCleanup
Clean up fgMorphArgs
Rahul Kumar [Wed, 22 Jun 2016 20:35:13 +0000 (13:35 -0700)]
Merge pull request #5891 from rahku/arm64retbuf
Arm64:return buffer arg should be passed in x8
Carol Eidt [Wed, 22 Jun 2016 20:09:55 +0000 (13:09 -0700)]
Merge pull request #5389 from parjong/fix/issue_5357
Run fgMorphCopyBlock for newly inserted assignment
Carol Eidt [Wed, 22 Jun 2016 20:02:32 +0000 (13:02 -0700)]
Merge pull request #5871 from mikedn/modopt
Improve div/mod by const power of 2
Aditya Mandaleeka [Wed, 22 Jun 2016 19:49:30 +0000 (12:49 -0700)]
Merge pull request #5925 from RussKeldorph/j32_issues
Fix syntax error in x86_jit32_issues.targets
Pat Gavlin [Wed, 22 Jun 2016 18:27:09 +0000 (11:27 -0700)]
Merge pull request #5904 from pgavlin/WalkTreePostExecutionOrder
Obey GTF_REVERSE_OPS in fgWalkTreePost.
Kyungwoo Lee [Wed, 22 Jun 2016 18:19:24 +0000 (11:19 -0700)]
ARM64: Skip CompareExchangeTClass Test
This test seems to fail often in CI. Disable it for now.
Sejong Oh [Wed, 22 Jun 2016 18:12:38 +0000 (11:12 -0700)]
Merge pull request #5905 from sejongoh/group_jitstress_jobs
Move all JIT stress jobs into a sub directory on CI
Russ Keldorph [Wed, 22 Jun 2016 18:09:12 +0000 (11:09 -0700)]
Fix syntax error in x86_jit32_issues.targets
William Godbe [Wed, 22 Jun 2016 17:46:46 +0000 (10:46 -0700)]
Merge pull request #5924 from wtgodbe/groovalicious
Run all tests under server GC during code coverage job
wtgodbe [Wed, 22 Jun 2016 17:43:55 +0000 (10:43 -0700)]
Run all tests under server GC
John Chen [Wed, 22 Jun 2016 17:43:23 +0000 (10:43 -0700)]
Merge pull request #5921 from JohnChen0/master
Disable a test case till issue #5501 is fixed
Evgeny Pavlov [Wed, 22 Jun 2016 16:56:51 +0000 (20:56 +0400)]
[SOS] [Linux] Fix incorrect symbol loading from portable pdb (#5896)
* Fix incorrect symbol loading from portable pdb
* Change return value to S_OK
Mike McLaughlin [Wed, 22 Jun 2016 16:55:33 +0000 (09:55 -0700)]
Generate symbols packages for Windows CoreCLR binaries. (#5902)
* Generate symbols packages for Windows CoreCLR binaries.
Issue #5832.
* Add dll/exe back to symbol package.
Mike Danes [Wed, 22 Jun 2016 16:44:34 +0000 (19:44 +0300)]
Fix statement insertion point in GT_STORE_LCL_VAR decomposition
In some cases a newly created statement is inserted between the current statement and its embedded statements instead of being inserted after the current statement and its embedded statements.
Gaurav Khanna [Wed, 22 Jun 2016 16:31:25 +0000 (09:31 -0700)]
Merge pull request #5919 from gkhanna79/FixCOC
Add Code of Conduct details to README.md
John Chen [Wed, 22 Jun 2016 16:12:44 +0000 (09:12 -0700)]
Disable a test case till issue #5501 is fixed
Mike Danes [Mon, 9 May 2016 22:01:12 +0000 (01:01 +0300)]
Optimize integer div/mod by const power of 2 in lowering
Optimizing GT_DIV/GT_UDIV/GT_MOD/GT_UMOD by power of 2 in codegen is problematic because the xarch DIV instruction has special register requirements. By the time codegen decides to perform the optimization the rax and rdx registers have been already allocated by LSRA even though they're not always needed (as it happens in the case of unsigned division where CDQ isn't used).
Since the JIT can't represent a CDQ instruction in its IR an arithmetic shift (GT_RSH) has been instead to extract the dividend sign. xarch's SAR is larger than CDQ but it has the advantage that it doesn't require specific registers. Also, arithmetic shifts are available on architectures other than xarch.
Example: method "static int foo(int x) => x / 8;" is now compiled to
mov eax, ecx
mov edx, eax
sar edx, 31
and edx, 7
add edx, eax
mov eax, edx
sar eax, 3
instead of
mov eax, ecx
cdq
and edx, 7
add eax, edx
sar eax, 3
As a side-effect of this change the optimization now also works when the divisor is too large to be contained. Previously this wasn't possible because the divisor constant needed to be modified during codegen but the constant was already loaded into a register.
Example: method "static ulong foo(ulong x) => x /
4294967296;" is now compiled to
mov rax, rcx
shr rax, 32
whereas before a DIV instruction was used.
This change also fixes an issue in fgShouldUseMagicNumberDivide. The optimization that is done in lower can handle negative power of 2 divisors but fgShouldUseMagicNumberDivide handled those cases because it didn't check the absolute value of the divisor.
Example: method "static int foo(int x) => return x / -2;" is now compiled to
mov eax, ecx
mov edx, eax
shr edx, 31
add edx, eax
sar edx, 1
mov eax, edx
neg eax
instead of
mov eax, 0x7FFFFFFF
imul edx:eax, ecx
mov eax, edx
sub eax, ecx
mov edx, eax
shr edx, 31
add eax, edx
Mike Danes [Sat, 14 Nov 2015 08:46:20 +0000 (10:46 +0200)]
Add tests for integer div/mod by const
Bruce Forstall [Wed, 22 Jun 2016 06:25:52 +0000 (23:25 -0700)]
Clean up fgMorphArgs
Simplify a lot of the coding, especially related to non-standard arg and SysV struct arg handling.
Gaurav Khanna [Wed, 22 Jun 2016 15:43:14 +0000 (08:43 -0700)]
Add Code of Conduct details to README.md
chunseoklee [Wed, 22 Jun 2016 15:12:58 +0000 (00:12 +0900)]
fix build break on ARM caused by f1889 (#5914)
Gaurav Khanna [Wed, 22 Jun 2016 13:58:08 +0000 (06:58 -0700)]
Merge pull request #5913 from dotnet-bot/UpdateDependencies20160622051611
Updating External dependencies to beta-24222-00
Jan Vorlicek [Wed, 22 Jun 2016 12:26:17 +0000 (14:26 +0200)]
Cleanup hardware exception handling (#5916)
Currently, we were checking for the exception address being in a marked jit
helper and unwinding to the caller context if it was there at two places.
Once in the HandleHardwareException to get context for the faulting exception
frame and once in DispatchManagedException to get the first managed frame.
This change modifies the code in HandleHardwareException to update the context
record in the exception itself so that we can leave out the same change in
the DispatchManagedException.
I have also added updating of the exception address in the exception record
so that the exception address is set to be in the managed caller of the helper.
dotnet-bot [Wed, 22 Jun 2016 05:16:11 +0000 (05:16 +0000)]
Updating External dependencies to beta-24222-00
Brian Sullivan [Wed, 22 Jun 2016 04:12:32 +0000 (21:12 -0700)]
Merge pull request #5908 from briansull/Fix-Remorph-Call
Fix for Remorph of a call with nonstandard args
Brian Sullivan [Wed, 22 Jun 2016 04:11:15 +0000 (21:11 -0700)]
Merge pull request #5909 from dotnet-bot/from-tfs
Merge changes from TFS
Brian Sullivan [Tue, 21 Jun 2016 23:19:11 +0000 (16:19 -0700)]
Fix for Remorph of a call with nonstandard args
Fadi Hanna [Tue, 21 Jun 2016 23:34:40 +0000 (16:34 -0700)]
Fixing build break on x86 and arm.
[tfs-changeset: 1614031]
Aditya Mandaleeka [Tue, 21 Jun 2016 23:06:05 +0000 (16:06 -0700)]
Merge pull request #3097 from vors/docx
Convert SOSAndICorDebug.docx to SOSAndICorDebug.md
Jarret Shook [Tue, 21 Jun 2016 22:49:33 +0000 (15:49 -0700)]
Merge pull request #5900 from jashook/gc_stress_incompat
Set b14364 as gcstress incompat
Jarret Shook [Tue, 21 Jun 2016 22:49:18 +0000 (15:49 -0700)]
Merge pull request #5898 from jashook/gcstress_timeout_update
Up the default timeout for GCStress
Aditya Mandaleeka [Tue, 21 Jun 2016 22:40:25 +0000 (15:40 -0700)]
Merge pull request #5907 from Maoni0/stress_txt
updated stress run description
Maoni0 [Tue, 21 Jun 2016 22:36:05 +0000 (15:36 -0700)]
added a note about the reason why we ignore failures from tests in the stress run infrastructure