maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/ARM/interval-update-remat.ll

163 lines
14 KiB
LLVM
Raw Normal View History

Recommit r265547, and r265610,r265639,r265657 on top of it, plus two fixes with one about error verify-regalloc reported, and another about live range update of phi after rematerialization. r265547: Replace analyzeSiblingValues with new algorithm to fix its compile time issue. The patch is to solve PR17409 and its duplicates. analyzeSiblingValues is a N x N complexity algorithm where N is the number of siblings generated by reg splitting. Although it causes siginificant compile time issue when N is large, it is also important for performance since it removes redundent spills and enables rematerialization. To solve the compile time issue, the patch removes analyzeSiblingValues and replaces it with lower cost alternatives containing two parts. The first part creates a new spill hoisting method in postOptimization of register allocation. It does spill hoisting at once after all the spills are generated instead of inside every instance of selectOrSplit. The second part queries the define expr of the original register for rematerializaiton and keep it always available during register allocation even if it is already dead. It deletes those dead instructions only in postOptimization. With the two parts in the patch, it can remove analyzeSiblingValues without sacrificing performance. Patches on top of r265547: r265610 "Fix the compare-clang diff error introduced by r265547." r265639 "Fix the sanitizer bootstrap error in r265547." r265657 "InlineSpiller.cpp: Escap \@ in r265547. [-Wdocumentation]" Differential Revision: http://reviews.llvm.org/D15302 Differential Revision: http://reviews.llvm.org/D18934 Differential Revision: http://reviews.llvm.org/D18935 Differential Revision: http://reviews.llvm.org/D18936 llvm-svn: 266162
2016-04-13 11:08:27 +08:00
; RUN: llc -verify-regalloc < %s
; PR27275: When enabling remat for vreg defined by PHIs, make sure the update
; of the live range removes dead phi. Otherwise, we may end up with PHIs with
; incorrect operands and that will trigger assertions or verifier failures
; in later passes.
target datalayout = "e-m:o-p:32:32-f64:32:64-v64:32:64-v128:32:128-a:0:32-n32-S32"
target triple = "thumbv7-apple-ios9.0.0"
%class.SOCKSClientSocketPoolTest_AsyncSOCKSConnectError_Test.1.226.276.1301.2326 = type { %class.MockTransportClientSocketPool.0.225.275.1300.2325, i32 }
%class.MockTransportClientSocketPool.0.225.275.1300.2325 = type { i8 }
%class.StaticSocketDataProvider.6.231.281.1306.2331 = type { i8, %struct.MockConnect.5.230.280.1305.2330 }
%struct.MockConnect.5.230.280.1305.2330 = type { %class.IPEndPoint.4.229.279.1304.2329 }
%class.IPEndPoint.4.229.279.1304.2329 = type { %class.IPAddress.3.228.278.1303.2328 }
%class.IPAddress.3.228.278.1303.2328 = type { %"class.(anonymous namespace)::vector.2.227.277.1302.2327" }
%"class.(anonymous namespace)::vector.2.227.277.1302.2327" = type { i8 }
%class.TestCompletionCallback.9.234.284.1309.2334 = type { %class.TestCompletionCallbackTemplate.8.233.283.1308.2333, i32 }
%class.TestCompletionCallbackTemplate.8.233.283.1308.2333 = type { i32 }
%class.AssertionResult.24.249.299.1324.2349 = type { i8, %class.scoped_ptr.23.248.298.1323.2348 }
%class.scoped_ptr.23.248.298.1323.2348 = type { %class.Trans_NS___1_basic_string.18.243.293.1318.2343* }
%class.Trans_NS___1_basic_string.18.243.293.1318.2343 = type { %class.Trans_NS___1___libcpp_compressed_pair_imp.17.242.292.1317.2342 }
%class.Trans_NS___1___libcpp_compressed_pair_imp.17.242.292.1317.2342 = type { %"struct.Trans_NS___1_basic_string<char, int, int>::__rep.16.241.291.1316.2341" }
%"struct.Trans_NS___1_basic_string<char, int, int>::__rep.16.241.291.1316.2341" = type { %"struct.Trans_NS___1_basic_string<char, int, int>::__long.15.240.290.1315.2340" }
%"struct.Trans_NS___1_basic_string<char, int, int>::__long.15.240.290.1315.2340" = type { i64, i32 }
%class.AssertHelper.10.235.285.1310.2335 = type { i8 }
%class.Message.13.238.288.1313.2338 = type { %class.scoped_ptr.0.12.237.287.1312.2337 }
%class.scoped_ptr.0.12.237.287.1312.2337 = type { %"class.(anonymous namespace)::basic_stringstream.11.236.286.1311.2336"* }
%"class.(anonymous namespace)::basic_stringstream.11.236.286.1311.2336" = type { i8 }
%class.scoped_refptr.19.244.294.1319.2344 = type { i8 }
%class.BoundNetLog.20.245.295.1320.2345 = type { i32 }
%struct.MockReadWrite.7.232.282.1307.2332 = type { i32 }
%"class.(anonymous namespace)::basic_iostream.22.247.297.1322.2347" = type { i8 }
%class.ClientSocketHandle.14.239.289.1314.2339 = type { i8 }
%"class.(anonymous namespace)::__vector_base.21.246.296.1321.2346" = type { i8 }
@.str = private unnamed_addr constant [1 x i8] zeroinitializer, align 1
define void @_ZN53SOCKSClientSocketPoolTest_AsyncSOCKSConnectError_Test6m_fn10Ev(%class.SOCKSClientSocketPoolTest_AsyncSOCKSConnectError_Test.1.226.276.1301.2326* %this) align 2 {
entry:
%socket_data = alloca %class.StaticSocketDataProvider.6.231.281.1306.2331, align 1
%agg.tmp = alloca %struct.MockConnect.5.230.280.1305.2330, align 1
%callback = alloca %class.TestCompletionCallback.9.234.284.1309.2334, align 4
%gtest_ar = alloca %class.AssertionResult.24.249.299.1324.2349, align 4
%temp.lvalue = alloca %class.AssertHelper.10.235.285.1310.2335, align 1
%agg.tmp10 = alloca %class.Message.13.238.288.1313.2338, align 4
%ref.tmp = alloca %class.Trans_NS___1_basic_string.18.243.293.1318.2343, align 4
%agg.tmp16 = alloca %class.scoped_refptr.19.244.294.1319.2344, align 1
%agg.tmp18 = alloca %class.BoundNetLog.20.245.295.1320.2345, align 4
%call2 = call %class.StaticSocketDataProvider.6.231.281.1306.2331* @_ZN24StaticSocketDataProviderC1EP13MockReadWritejS1_j(%class.StaticSocketDataProvider.6.231.281.1306.2331* nonnull %socket_data, %struct.MockReadWrite.7.232.282.1307.2332* undef, i32 1, %struct.MockReadWrite.7.232.282.1307.2332* null, i32 0)
%call3 = call %struct.MockConnect.5.230.280.1305.2330* @_ZN11MockConnectC1Ev(%struct.MockConnect.5.230.280.1305.2330* nonnull %agg.tmp)
call void @_ZN24StaticSocketDataProvider5m_fn8E11MockConnect(%class.StaticSocketDataProvider.6.231.281.1306.2331* nonnull %socket_data, %struct.MockConnect.5.230.280.1305.2330* nonnull %agg.tmp)
%call5 = call %class.TestCompletionCallback.9.234.284.1309.2334* @_ZN22TestCompletionCallbackC1Ev(%class.TestCompletionCallback.9.234.284.1309.2334* nonnull %callback)
%transport_socket_pool_ = getelementptr inbounds %class.SOCKSClientSocketPoolTest_AsyncSOCKSConnectError_Test.1.226.276.1301.2326, %class.SOCKSClientSocketPoolTest_AsyncSOCKSConnectError_Test.1.226.276.1301.2326* %this, i32 0, i32 0
%call6 = call i32 @_ZN29MockTransportClientSocketPool5m_fn9Ev(%class.MockTransportClientSocketPool.0.225.275.1300.2325* %transport_socket_pool_)
call void @_Z11CmpHelperEQPcS_xx(%class.AssertionResult.24.249.299.1324.2349* nonnull sret %gtest_ar, i8* getelementptr inbounds ([1 x i8], [1 x i8]* @.str, i32 0, i32 0), i8* getelementptr inbounds ([1 x i8], [1 x i8]* @.str, i32 0, i32 0), i64 0, i64 undef)
%tmp = load i8, i8* undef, align 4
%tobool.i = icmp eq i8 %tmp, 0
br i1 %tobool.i, label %if.else, label %if.end
if.else: ; preds = %entry
br i1 undef, label %_ZN15AssertionResult5m_fn6Ev.exit, label %cond.true.i
cond.true.i: ; preds = %if.else
%call4.i = call i8* @_ZN25Trans_NS___1_basic_stringIciiE5m_fn1Ev(%class.Trans_NS___1_basic_string.18.243.293.1318.2343* nonnull undef)
br label %_ZN15AssertionResult5m_fn6Ev.exit
_ZN15AssertionResult5m_fn6Ev.exit: ; preds = %cond.true.i, %if.else
%cond.i = phi i8* [ %call4.i, %cond.true.i ], [ getelementptr inbounds ([1 x i8], [1 x i8]* @.str, i32 0, i32 0), %if.else ]
%call9 = call %class.AssertHelper.10.235.285.1310.2335* @_ZN12AssertHelperC1EPKc(%class.AssertHelper.10.235.285.1310.2335* nonnull %temp.lvalue, i8* %cond.i)
%call11 = call %class.Message.13.238.288.1313.2338* @_ZN7MessageC1Ev(%class.Message.13.238.288.1313.2338* nonnull %agg.tmp10)
call void @_ZN12AssertHelperaSE7Message(%class.AssertHelper.10.235.285.1310.2335* nonnull %temp.lvalue, %class.Message.13.238.288.1313.2338* nonnull %agg.tmp10)
%call.i.i.i.i27 = call zeroext i1 @_Z6IsTruev()
%brmerge = or i1 false, undef
br i1 %brmerge, label %_ZN7MessageD1Ev.exit33, label %delete.notnull.i.i.i.i32
delete.notnull.i.i.i.i32: ; preds = %_ZN15AssertionResult5m_fn6Ev.exit
%call.i.i.i.i.i.i31 = call %"class.(anonymous namespace)::basic_iostream.22.247.297.1322.2347"* @_ZN12_GLOBAL__N_114basic_iostreamD2Ev(%"class.(anonymous namespace)::basic_iostream.22.247.297.1322.2347"* undef)
call void @_ZdlPv(i8* undef)
br label %_ZN7MessageD1Ev.exit33
_ZN7MessageD1Ev.exit33: ; preds = %delete.notnull.i.i.i.i32, %_ZN15AssertionResult5m_fn6Ev.exit
%call13 = call %class.AssertHelper.10.235.285.1310.2335* @_ZN12AssertHelperD1Ev(%class.AssertHelper.10.235.285.1310.2335* nonnull %temp.lvalue)
br label %if.end
if.end: ; preds = %_ZN7MessageD1Ev.exit33, %entry
%message_.i.i = getelementptr inbounds %class.AssertionResult.24.249.299.1324.2349, %class.AssertionResult.24.249.299.1324.2349* %gtest_ar, i32 0, i32 1
%call.i.i.i = call %class.scoped_ptr.23.248.298.1323.2348* @_ZN10scoped_ptrI25Trans_NS___1_basic_stringIciiEED2Ev(%class.scoped_ptr.23.248.298.1323.2348* %message_.i.i)
Remove alignment argument from memcpy/memmove/memset in favour of alignment attributes (Step 1) Summary: This is a resurrection of work first proposed and discussed in Aug 2015: http://lists.llvm.org/pipermail/llvm-dev/2015-August/089384.html and initially landed (but then backed out) in Nov 2015: http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change is the first in a series that allows source and dest to each have their own alignments by using the alignment attribute on their arguments. In this change we: 1) Remove the alignment argument. 2) Add alignment attributes to the source & dest arguments. We, temporarily, require that the alignments for source & dest be equal. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) Downstream users may have to update their lit tests that check for @llvm.memcpy/memmove/memset call/declaration patterns. The following extended sed script may help with updating the majority of your tests, but it does not catch all possible patterns so some manual checking and updating will be required. s~declare void @llvm\.mem(set|cpy|move)\.p([^(]*)\((.*), i32, i1\)~declare void @llvm.mem\1.p\2(\3, i1)~g s~call void @llvm\.memset\.p([^(]*)i8\(i8([^*]*)\* (.*), i8 (.*), i8 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i8(i8\2* \3, i8 \4, i8 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i16\(i8([^*]*)\* (.*), i8 (.*), i16 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i16(i8\2* \3, i8 \4, i16 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i32\(i8([^*]*)\* (.*), i8 (.*), i32 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i32(i8\2* \3, i8 \4, i32 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i64\(i8([^*]*)\* (.*), i8 (.*), i64 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i64(i8\2* \3, i8 \4, i64 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i128\(i8([^*]*)\* (.*), i8 (.*), i128 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i128(i8\2* \3, i8 \4, i128 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i8\(i8([^*]*)\* (.*), i8 (.*), i8 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i8(i8\2* align \6 \3, i8 \4, i8 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i16\(i8([^*]*)\* (.*), i8 (.*), i16 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i16(i8\2* align \6 \3, i8 \4, i16 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i32\(i8([^*]*)\* (.*), i8 (.*), i32 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i32(i8\2* align \6 \3, i8 \4, i32 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i64\(i8([^*]*)\* (.*), i8 (.*), i64 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i64(i8\2* align \6 \3, i8 \4, i64 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i128\(i8([^*]*)\* (.*), i8 (.*), i128 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i128(i8\2* align \6 \3, i8 \4, i128 \5, i1 \7)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i8\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i8 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i8(i8\3* \4, i8\5* \6, i8 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i16\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i16 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i16(i8\3* \4, i8\5* \6, i16 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i32\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i32 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i32(i8\3* \4, i8\5* \6, i32 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i64\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i64 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i64(i8\3* \4, i8\5* \6, i64 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i128\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i128 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i128(i8\3* \4, i8\5* \6, i128 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i8\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i8 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i8(i8\3* align \8 \4, i8\5* align \8 \6, i8 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i16\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i16 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i16(i8\3* align \8 \4, i8\5* align \8 \6, i16 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i32\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i32 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i32(i8\3* align \8 \4, i8\5* align \8 \6, i32 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i64\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i64 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i64(i8\3* align \8 \4, i8\5* align \8 \6, i64 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i128\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i128 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i128(i8\3* align \8 \4, i8\5* align \8 \6, i128 \7, i1 \9)~g The remaining changes in the series will: Step 2) Expand the IRBuilder API to allow creation of memcpy/memmove with differing source and dest alignments. Step 3) Update Clang to use the new IRBuilder API. Step 4) Update Polly to use the new IRBuilder API. Step 5) Update LLVM passes that create memcpy/memmove calls to use the new IRBuilder API, and those that use use MemIntrinsicInst::[get|set]Alignment() to use getDestAlignment() and getSourceAlignment() instead. Step 6) Remove the single-alignment IRBuilder API for memcpy/memmove, and the MemIntrinsicInst::[get|set]Alignment() methods. Reviewers: pete, hfinkel, lhames, reames, bollu Reviewed By: reames Subscribers: niosHD, reames, jholewinski, qcolombet, jfb, sanjoy, arsenm, dschuff, dylanmckay, mehdi_amini, sdardis, nemanjai, david2050, nhaehnle, javed.absar, sbc100, jgravelle-google, eraman, aheejin, kbarton, JDevlieghere, asb, rbar, johnrusso, simoncook, jordy.potman.lists, apazos, sabuasal, llvm-commits Differential Revision: https://reviews.llvm.org/D41675 llvm-svn: 322965
2018-01-20 01:13:12 +08:00
call void @llvm.memset.p0i8.i32(i8* align 4 null, i8 0, i32 12, i1 false)
Recommit r265547, and r265610,r265639,r265657 on top of it, plus two fixes with one about error verify-regalloc reported, and another about live range update of phi after rematerialization. r265547: Replace analyzeSiblingValues with new algorithm to fix its compile time issue. The patch is to solve PR17409 and its duplicates. analyzeSiblingValues is a N x N complexity algorithm where N is the number of siblings generated by reg splitting. Although it causes siginificant compile time issue when N is large, it is also important for performance since it removes redundent spills and enables rematerialization. To solve the compile time issue, the patch removes analyzeSiblingValues and replaces it with lower cost alternatives containing two parts. The first part creates a new spill hoisting method in postOptimization of register allocation. It does spill hoisting at once after all the spills are generated instead of inside every instance of selectOrSplit. The second part queries the define expr of the original register for rematerializaiton and keep it always available during register allocation even if it is already dead. It deletes those dead instructions only in postOptimization. With the two parts in the patch, it can remove analyzeSiblingValues without sacrificing performance. Patches on top of r265547: r265610 "Fix the compare-clang diff error introduced by r265547." r265639 "Fix the sanitizer bootstrap error in r265547." r265657 "InlineSpiller.cpp: Escap \@ in r265547. [-Wdocumentation]" Differential Revision: http://reviews.llvm.org/D15302 Differential Revision: http://reviews.llvm.org/D18934 Differential Revision: http://reviews.llvm.org/D18935 Differential Revision: http://reviews.llvm.org/D18936 llvm-svn: 266162
2016-04-13 11:08:27 +08:00
call void @_ZN25Trans_NS___1_basic_stringIciiE5m_fn2Ev(%class.Trans_NS___1_basic_string.18.243.293.1318.2343* nonnull %ref.tmp)
call void @_Z19CreateSOCKSv5Paramsv(%class.scoped_refptr.19.244.294.1319.2344* nonnull sret %agg.tmp16)
%callback_.i = getelementptr inbounds %class.TestCompletionCallback.9.234.284.1309.2334, %class.TestCompletionCallback.9.234.284.1309.2334* %callback, i32 0, i32 1
%pool_ = getelementptr inbounds %class.SOCKSClientSocketPoolTest_AsyncSOCKSConnectError_Test.1.226.276.1301.2326, %class.SOCKSClientSocketPoolTest_AsyncSOCKSConnectError_Test.1.226.276.1301.2326* %this, i32 0, i32 1
%tmp1 = getelementptr inbounds %class.BoundNetLog.20.245.295.1320.2345, %class.BoundNetLog.20.245.295.1320.2345* %agg.tmp18, i32 0, i32 0
store i32 0, i32* %tmp1, align 4
call void @_ZN18ClientSocketHandle5m_fn3IPiEEvRK25Trans_NS___1_basic_stringIciiE13scoped_refptr15RequestPriorityN16ClientSocketPool13RespectLimitsERiT_11BoundNetLog(%class.ClientSocketHandle.14.239.289.1314.2339* nonnull undef, %class.Trans_NS___1_basic_string.18.243.293.1318.2343* nonnull dereferenceable(12) %ref.tmp, %class.scoped_refptr.19.244.294.1319.2344* nonnull %agg.tmp16, i32 0, i32 1, i32* nonnull dereferenceable(4) %callback_.i, i32* %pool_, %class.BoundNetLog.20.245.295.1320.2345* nonnull %agg.tmp18)
%call19 = call %class.BoundNetLog.20.245.295.1320.2345* @_ZN11BoundNetLogD1Ev(%class.BoundNetLog.20.245.295.1320.2345* nonnull %agg.tmp18)
call void @_Z11CmpHelperEQPcS_xx(%class.AssertionResult.24.249.299.1324.2349* nonnull sret undef, i8* getelementptr inbounds ([1 x i8], [1 x i8]* @.str, i32 0, i32 0), i8* getelementptr inbounds ([1 x i8], [1 x i8]* @.str, i32 0, i32 0), i64 -1, i64 0)
br i1 undef, label %if.then.i.i.i.i, label %_ZN7MessageD1Ev.exit
if.then.i.i.i.i: ; preds = %if.end
%tmp2 = load %"class.(anonymous namespace)::basic_stringstream.11.236.286.1311.2336"*, %"class.(anonymous namespace)::basic_stringstream.11.236.286.1311.2336"** undef, align 4
br label %_ZN7MessageD1Ev.exit
_ZN7MessageD1Ev.exit: ; preds = %if.then.i.i.i.i, %if.end
%connect_.i.i = getelementptr inbounds %class.StaticSocketDataProvider.6.231.281.1306.2331, %class.StaticSocketDataProvider.6.231.281.1306.2331* %socket_data, i32 0, i32 1
%tmp3 = bitcast %struct.MockConnect.5.230.280.1305.2330* %connect_.i.i to %"class.(anonymous namespace)::__vector_base.21.246.296.1321.2346"*
%call.i.i.i.i.i.i.i.i.i.i = call %"class.(anonymous namespace)::__vector_base.21.246.296.1321.2346"* @_ZN12_GLOBAL__N_113__vector_baseD2Ev(%"class.(anonymous namespace)::__vector_base.21.246.296.1321.2346"* %tmp3)
ret void
}
; Function Attrs: argmemonly nounwind
Add address space mangling to lifetime intrinsics In preparation for allowing allocas to have non-0 addrspace. llvm-svn: 299876
2017-04-11 04:18:21 +08:00
declare void @llvm.lifetime.start.p0i8(i64, i8* nocapture) #0
Recommit r265547, and r265610,r265639,r265657 on top of it, plus two fixes with one about error verify-regalloc reported, and another about live range update of phi after rematerialization. r265547: Replace analyzeSiblingValues with new algorithm to fix its compile time issue. The patch is to solve PR17409 and its duplicates. analyzeSiblingValues is a N x N complexity algorithm where N is the number of siblings generated by reg splitting. Although it causes siginificant compile time issue when N is large, it is also important for performance since it removes redundent spills and enables rematerialization. To solve the compile time issue, the patch removes analyzeSiblingValues and replaces it with lower cost alternatives containing two parts. The first part creates a new spill hoisting method in postOptimization of register allocation. It does spill hoisting at once after all the spills are generated instead of inside every instance of selectOrSplit. The second part queries the define expr of the original register for rematerializaiton and keep it always available during register allocation even if it is already dead. It deletes those dead instructions only in postOptimization. With the two parts in the patch, it can remove analyzeSiblingValues without sacrificing performance. Patches on top of r265547: r265610 "Fix the compare-clang diff error introduced by r265547." r265639 "Fix the sanitizer bootstrap error in r265547." r265657 "InlineSpiller.cpp: Escap \@ in r265547. [-Wdocumentation]" Differential Revision: http://reviews.llvm.org/D15302 Differential Revision: http://reviews.llvm.org/D18934 Differential Revision: http://reviews.llvm.org/D18935 Differential Revision: http://reviews.llvm.org/D18936 llvm-svn: 266162
2016-04-13 11:08:27 +08:00
declare %class.StaticSocketDataProvider.6.231.281.1306.2331* @_ZN24StaticSocketDataProviderC1EP13MockReadWritejS1_j(%class.StaticSocketDataProvider.6.231.281.1306.2331* returned, %struct.MockReadWrite.7.232.282.1307.2332*, i32, %struct.MockReadWrite.7.232.282.1307.2332*, i32) unnamed_addr
declare void @_ZN24StaticSocketDataProvider5m_fn8E11MockConnect(%class.StaticSocketDataProvider.6.231.281.1306.2331*, %struct.MockConnect.5.230.280.1305.2330*)
declare %struct.MockConnect.5.230.280.1305.2330* @_ZN11MockConnectC1Ev(%struct.MockConnect.5.230.280.1305.2330* returned) unnamed_addr
declare %class.TestCompletionCallback.9.234.284.1309.2334* @_ZN22TestCompletionCallbackC1Ev(%class.TestCompletionCallback.9.234.284.1309.2334* returned) unnamed_addr
declare i32 @_ZN29MockTransportClientSocketPool5m_fn9Ev(%class.MockTransportClientSocketPool.0.225.275.1300.2325*)
declare %class.AssertHelper.10.235.285.1310.2335* @_ZN12AssertHelperC1EPKc(%class.AssertHelper.10.235.285.1310.2335* returned, i8*) unnamed_addr
declare void @_ZN12AssertHelperaSE7Message(%class.AssertHelper.10.235.285.1310.2335*, %class.Message.13.238.288.1313.2338*)
declare %class.Message.13.238.288.1313.2338* @_ZN7MessageC1Ev(%class.Message.13.238.288.1313.2338* returned) unnamed_addr
declare %class.AssertHelper.10.235.285.1310.2335* @_ZN12AssertHelperD1Ev(%class.AssertHelper.10.235.285.1310.2335* returned) unnamed_addr
; Function Attrs: argmemonly nounwind
Add address space mangling to lifetime intrinsics In preparation for allowing allocas to have non-0 addrspace. llvm-svn: 299876
2017-04-11 04:18:21 +08:00
declare void @llvm.lifetime.end.p0i8(i64, i8* nocapture) #0
Recommit r265547, and r265610,r265639,r265657 on top of it, plus two fixes with one about error verify-regalloc reported, and another about live range update of phi after rematerialization. r265547: Replace analyzeSiblingValues with new algorithm to fix its compile time issue. The patch is to solve PR17409 and its duplicates. analyzeSiblingValues is a N x N complexity algorithm where N is the number of siblings generated by reg splitting. Although it causes siginificant compile time issue when N is large, it is also important for performance since it removes redundent spills and enables rematerialization. To solve the compile time issue, the patch removes analyzeSiblingValues and replaces it with lower cost alternatives containing two parts. The first part creates a new spill hoisting method in postOptimization of register allocation. It does spill hoisting at once after all the spills are generated instead of inside every instance of selectOrSplit. The second part queries the define expr of the original register for rematerializaiton and keep it always available during register allocation even if it is already dead. It deletes those dead instructions only in postOptimization. With the two parts in the patch, it can remove analyzeSiblingValues without sacrificing performance. Patches on top of r265547: r265610 "Fix the compare-clang diff error introduced by r265547." r265639 "Fix the sanitizer bootstrap error in r265547." r265657 "InlineSpiller.cpp: Escap \@ in r265547. [-Wdocumentation]" Differential Revision: http://reviews.llvm.org/D15302 Differential Revision: http://reviews.llvm.org/D18934 Differential Revision: http://reviews.llvm.org/D18935 Differential Revision: http://reviews.llvm.org/D18936 llvm-svn: 266162
2016-04-13 11:08:27 +08:00
declare void @_ZN18ClientSocketHandle5m_fn3IPiEEvRK25Trans_NS___1_basic_stringIciiE13scoped_refptr15RequestPriorityN16ClientSocketPool13RespectLimitsERiT_11BoundNetLog(%class.ClientSocketHandle.14.239.289.1314.2339*, %class.Trans_NS___1_basic_string.18.243.293.1318.2343* dereferenceable(12), %class.scoped_refptr.19.244.294.1319.2344*, i32, i32, i32* dereferenceable(4), i32*, %class.BoundNetLog.20.245.295.1320.2345*)
declare void @_Z19CreateSOCKSv5Paramsv(%class.scoped_refptr.19.244.294.1319.2344* sret)
; Function Attrs: argmemonly nounwind
Remove alignment argument from memcpy/memmove/memset in favour of alignment attributes (Step 1) Summary: This is a resurrection of work first proposed and discussed in Aug 2015: http://lists.llvm.org/pipermail/llvm-dev/2015-August/089384.html and initially landed (but then backed out) in Nov 2015: http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change is the first in a series that allows source and dest to each have their own alignments by using the alignment attribute on their arguments. In this change we: 1) Remove the alignment argument. 2) Add alignment attributes to the source & dest arguments. We, temporarily, require that the alignments for source & dest be equal. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) Downstream users may have to update their lit tests that check for @llvm.memcpy/memmove/memset call/declaration patterns. The following extended sed script may help with updating the majority of your tests, but it does not catch all possible patterns so some manual checking and updating will be required. s~declare void @llvm\.mem(set|cpy|move)\.p([^(]*)\((.*), i32, i1\)~declare void @llvm.mem\1.p\2(\3, i1)~g s~call void @llvm\.memset\.p([^(]*)i8\(i8([^*]*)\* (.*), i8 (.*), i8 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i8(i8\2* \3, i8 \4, i8 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i16\(i8([^*]*)\* (.*), i8 (.*), i16 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i16(i8\2* \3, i8 \4, i16 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i32\(i8([^*]*)\* (.*), i8 (.*), i32 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i32(i8\2* \3, i8 \4, i32 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i64\(i8([^*]*)\* (.*), i8 (.*), i64 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i64(i8\2* \3, i8 \4, i64 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i128\(i8([^*]*)\* (.*), i8 (.*), i128 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i128(i8\2* \3, i8 \4, i128 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i8\(i8([^*]*)\* (.*), i8 (.*), i8 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i8(i8\2* align \6 \3, i8 \4, i8 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i16\(i8([^*]*)\* (.*), i8 (.*), i16 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i16(i8\2* align \6 \3, i8 \4, i16 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i32\(i8([^*]*)\* (.*), i8 (.*), i32 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i32(i8\2* align \6 \3, i8 \4, i32 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i64\(i8([^*]*)\* (.*), i8 (.*), i64 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i64(i8\2* align \6 \3, i8 \4, i64 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i128\(i8([^*]*)\* (.*), i8 (.*), i128 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i128(i8\2* align \6 \3, i8 \4, i128 \5, i1 \7)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i8\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i8 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i8(i8\3* \4, i8\5* \6, i8 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i16\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i16 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i16(i8\3* \4, i8\5* \6, i16 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i32\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i32 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i32(i8\3* \4, i8\5* \6, i32 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i64\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i64 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i64(i8\3* \4, i8\5* \6, i64 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i128\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i128 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i128(i8\3* \4, i8\5* \6, i128 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i8\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i8 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i8(i8\3* align \8 \4, i8\5* align \8 \6, i8 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i16\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i16 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i16(i8\3* align \8 \4, i8\5* align \8 \6, i16 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i32\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i32 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i32(i8\3* align \8 \4, i8\5* align \8 \6, i32 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i64\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i64 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i64(i8\3* align \8 \4, i8\5* align \8 \6, i64 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i128\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i128 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i128(i8\3* align \8 \4, i8\5* align \8 \6, i128 \7, i1 \9)~g The remaining changes in the series will: Step 2) Expand the IRBuilder API to allow creation of memcpy/memmove with differing source and dest alignments. Step 3) Update Clang to use the new IRBuilder API. Step 4) Update Polly to use the new IRBuilder API. Step 5) Update LLVM passes that create memcpy/memmove calls to use the new IRBuilder API, and those that use use MemIntrinsicInst::[get|set]Alignment() to use getDestAlignment() and getSourceAlignment() instead. Step 6) Remove the single-alignment IRBuilder API for memcpy/memmove, and the MemIntrinsicInst::[get|set]Alignment() methods. Reviewers: pete, hfinkel, lhames, reames, bollu Reviewed By: reames Subscribers: niosHD, reames, jholewinski, qcolombet, jfb, sanjoy, arsenm, dschuff, dylanmckay, mehdi_amini, sdardis, nemanjai, david2050, nhaehnle, javed.absar, sbc100, jgravelle-google, eraman, aheejin, kbarton, JDevlieghere, asb, rbar, johnrusso, simoncook, jordy.potman.lists, apazos, sabuasal, llvm-commits Differential Revision: https://reviews.llvm.org/D41675 llvm-svn: 322965
2018-01-20 01:13:12 +08:00
declare void @llvm.memset.p0i8.i32(i8* nocapture, i8, i32, i1) #0
Recommit r265547, and r265610,r265639,r265657 on top of it, plus two fixes with one about error verify-regalloc reported, and another about live range update of phi after rematerialization. r265547: Replace analyzeSiblingValues with new algorithm to fix its compile time issue. The patch is to solve PR17409 and its duplicates. analyzeSiblingValues is a N x N complexity algorithm where N is the number of siblings generated by reg splitting. Although it causes siginificant compile time issue when N is large, it is also important for performance since it removes redundent spills and enables rematerialization. To solve the compile time issue, the patch removes analyzeSiblingValues and replaces it with lower cost alternatives containing two parts. The first part creates a new spill hoisting method in postOptimization of register allocation. It does spill hoisting at once after all the spills are generated instead of inside every instance of selectOrSplit. The second part queries the define expr of the original register for rematerializaiton and keep it always available during register allocation even if it is already dead. It deletes those dead instructions only in postOptimization. With the two parts in the patch, it can remove analyzeSiblingValues without sacrificing performance. Patches on top of r265547: r265610 "Fix the compare-clang diff error introduced by r265547." r265639 "Fix the sanitizer bootstrap error in r265547." r265657 "InlineSpiller.cpp: Escap \@ in r265547. [-Wdocumentation]" Differential Revision: http://reviews.llvm.org/D15302 Differential Revision: http://reviews.llvm.org/D18934 Differential Revision: http://reviews.llvm.org/D18935 Differential Revision: http://reviews.llvm.org/D18936 llvm-svn: 266162
2016-04-13 11:08:27 +08:00
declare %class.BoundNetLog.20.245.295.1320.2345* @_ZN11BoundNetLogD1Ev(%class.BoundNetLog.20.245.295.1320.2345* returned) unnamed_addr
declare %class.scoped_refptr.19.244.294.1319.2344* @_ZN13scoped_refptrD1Ev(%class.scoped_refptr.19.244.294.1319.2344* returned) unnamed_addr
declare %"class.(anonymous namespace)::__vector_base.21.246.296.1321.2346"* @_ZN12_GLOBAL__N_113__vector_baseD2Ev(%"class.(anonymous namespace)::__vector_base.21.246.296.1321.2346"* returned) unnamed_addr
declare i8* @_ZN25Trans_NS___1_basic_stringIciiE5m_fn1Ev(%class.Trans_NS___1_basic_string.18.243.293.1318.2343*)
declare zeroext i1 @_Z6IsTruev()
declare void @_ZdlPv(i8*)
declare %"class.(anonymous namespace)::basic_iostream.22.247.297.1322.2347"* @_ZN12_GLOBAL__N_114basic_iostreamD2Ev(%"class.(anonymous namespace)::basic_iostream.22.247.297.1322.2347"* returned) unnamed_addr
declare %class.scoped_ptr.23.248.298.1323.2348* @_ZN10scoped_ptrI25Trans_NS___1_basic_stringIciiEED2Ev(%class.scoped_ptr.23.248.298.1323.2348* readonly returned) unnamed_addr align 2
declare void @_Z11CmpHelperEQPcS_xx(%class.AssertionResult.24.249.299.1324.2349* sret, i8*, i8*, i64, i64)
declare void @_ZN25Trans_NS___1_basic_stringIciiE5m_fn2Ev(%class.Trans_NS___1_basic_string.18.243.293.1318.2343*)
attributes #0 = { argmemonly nounwind }