<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Fri, Oct 13, 2017 at 4:06 PM, Peter Collingbourne <span dir="ltr"><<a href="mailto:peter@pcc.me.uk" target="_blank">peter@pcc.me.uk</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Here's a small reproducer.</div><div><br></div><div>struct A {</div><div> virtual void f(...);</div><div>};</div><div><br></div><div>struct B : virtual A {</div><div> virtual void b();</div><div> virtual void f(...);</div><div>};</div><div><br></div><div>void B::f(...) {}</div><div><br></div><div>$ clang++ -fsanitize=function fsan.cpp -ffunction-sections -fdata-sections -c -o /dev/null<br></div><div><div>fatal error: error in backend: Cannot represent a difference across sections</div></div><div><br></div><div>Looking at the IR I see this function definition:</div><div>define void @_ZTv0_n24_N1B1fEz(%struct.B* %this, ...) unnamed_addr #0 align 2 prologue <{ i32, i32 }> <{ i32 846595819, i32 trunc (i64 sub (i64 ptrtoint (i8** @0 to i64), i64 ptrtoint (void (%struct.B*, ...)* @_ZN1B1fEz to i64)) to i32) }> {<br></div><div>which appears to cause the error.</div><div><br></div><div>I get basically the same IR if I use a Darwin target triple, so this isn't a Linux-specific issue. (On Darwin we end up successfully creating an object file, but the embedded offset in the text section will presumably be incorrect.)</div><div><br></div><div>Note that we're emitting prologue data on virtual functions, which is unnecessary because -fsanitize=function only checks indirect calls via function pointers. So I imagine that one way to solve the problem would be to turn off prologue data emission on non-virtual functions.</div></div></blockquote><div><br></div><div>Sorry, I meant "limit prologue data emission to non-virtual functions".</div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Peter</div></div><div class="gmail_extra"><div><div class="h5"><br><div class="gmail_quote">On Fri, Oct 13, 2017 at 3:06 PM, Vedant Kumar <span dir="ltr"><<a href="mailto:vsk@apple.com" target="_blank">vsk@apple.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div style="word-wrap:break-word"><br><div><span><blockquote type="cite"><div>On Oct 13, 2017, at 2:52 PM, Eric Christopher <<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>> wrote:</div><br class="m_-3309266437497712727m_1661471059887043044Apple-interchange-newline"><div><div dir="ltr" style="font-family:Helvetica;font-size:12px;font-style:normal;font-variant-caps:normal;font-weight:normal;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px"><br><br><div class="gmail_quote"><div dir="ltr">On Fri, Oct 13, 2017 at 2:50 PM Vedant Kumar <<a href="mailto:vsk@apple.com" target="_blank">vsk@apple.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div style="word-wrap:break-word"><div><blockquote type="cite"><div>On Oct 13, 2017, at 1:44 PM, Eric Christopher <<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>> wrote:</div><br class="m_-3309266437497712727m_1661471059887043044m_2234658512918948257Apple-interchange-newline"><div><div dir="ltr"><br><br><div class="gmail_quote"><div dir="ltr">On Fri, Oct 13, 2017 at 1:42 PM Vedant Kumar <<a href="mailto:vsk@apple.com" target="_blank">vsk@apple.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div style="word-wrap:break-word"><div><blockquote type="cite"><div>On Oct 13, 2017, at 1:39 PM, Vedant Kumar <<a href="mailto:vsk@apple.com" target="_blank">vsk@apple.com</a>> wrote:</div><br class="m_-3309266437497712727m_1661471059887043044m_2234658512918948257m_-5414524569825149145Apple-interchange-newline"><div><div style="word-wrap:break-word"><div>Hey Eric,</div><div><br></div><div>I'm sorry for the breakage. I made sure to check the run-time tests in compiler-rt but we could have missing coverage there.</div><div><br></div><div>The original version of this patch restricted the prologue data changes to Darwin only. We can switch back to that easily, just let me know.</div></div></div></blockquote><div><br></div></div></div><div style="word-wrap:break-word"><div>Actually I'll go ahead and work a patch up.</div></div><div style="word-wrap:break-word"><div><br></div></div></blockquote><div><br></div><div>Appreciated :)</div><div><br></div><div>Basically we were getting an error of:</div><div><br></div><div><span style="color:rgb(33,33,33);font-size:13px">error: Cannot represent a difference across sections</span><br style="color:rgb(33,33,33);font-size:13px"></div><div><span style="color:rgb(33,33,33);font-size:13px"><br></span></div><div><span style="color:rgb(33,33,33);font-size:13px">trying to compile things with the current code.</span></div></div></div></div></blockquote><div><br></div></div></div><div style="word-wrap:break-word"><div><div>Oh I see.. well, we started using a difference between the address of a function and the address of a global, so the error makes sense.</div><div><br></div><div>I'd be interested in any factors that could narrow the problem down (e.g using a specific linker, using -ffunction-sections, using data-sections, etc). Basically I'm not sure why this would work on some Linux setups but not others.</div><div><br></div></div></div></blockquote><div><br></div><div>Definitely using the latter two options and gold as a linker. I'll see what Han can come up with.</div></div></div></div></blockquote><div><br></div></span>Gotcha. Well, -ffunction-sections appears to be untested in compiler-rt/test/ubsan, at least.</div><div><br></div><div>There's a test somewhere in there called function.cpp -- it would be great if we could cover the *-sections options there. I'm not sure whether that's what caused the failure, but the extra coverage couldn't hurt :). I would do it myself but I don't have a Linux machine to test on.</div><span class="m_-3309266437497712727HOEnZb"><font color="#888888"><div><br></div><div>vedant</div></font></span><div><div class="m_-3309266437497712727h5"><div><br><blockquote type="cite"><div><div dir="ltr" style="font-family:Helvetica;font-size:12px;font-style:normal;font-variant-caps:normal;font-weight:normal;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px"><div class="gmail_quote"><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div style="word-wrap:break-word"><div><div></div><div>While we figure that out here's a patch to limit the impact on non-Darwin platforms:</div><div><a href="https://reviews.llvm.org/D38903" target="_blank">https://reviews.llvm.org/D3890<wbr>3</a></div></div></div></blockquote><div><br></div><div>*goes a looking*</div><div><br></div><div>Thanks!</div><div><br></div><div>-eric </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div style="word-wrap:break-word"><div><div><br></div><div>vedant</div></div></div><div style="word-wrap:break-word"><div><br><blockquote type="cite"><div><div dir="ltr"><div class="gmail_quote"><div><span style="color:rgb(33,33,33);font-size:13px"><br></span></div><div><span style="color:rgb(33,33,33);font-size:13px">Thanks!</span></div><div><br></div><div><font color="#212121">-eric</font></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div style="word-wrap:break-word"><div></div><div>vedant</div></div><div style="word-wrap:break-word"><div><br><blockquote type="cite"><div><div style="word-wrap:break-word"><div><br></div><div>vedant</div><div><br></div><br><div><blockquote type="cite"><div>On Oct 13, 2017, at 1:33 PM, Eric Christopher <<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>> wrote:</div><br class="m_-3309266437497712727m_1661471059887043044m_2234658512918948257m_-5414524569825149145Apple-interchange-newline"><div><div dir="ltr">Hi Vedant,<div><br></div><div>So this actually broke -fsanitize=function on linux. Han is working up a testcase for it, but letting you know for now that we'll probably need some change here.</div><div><br></div><div>-eric<br><br><div class="gmail_quote"><div dir="ltr">On Tue, Sep 12, 2017 at 5:05 PM Vedant Kumar via cfe-commits <<a href="mailto:cfe-commits@lists.llvm.org" target="_blank">cfe-commits@lists.llvm.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">Author: vedantk<br>Date: Tue Sep 12 17:04:35 2017<br>New Revision: 313096<br><br>URL:<span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project?rev=313096&view=rev" rel="noreferrer" target="_blank">http://llvm.org/viewvc/ll<wbr>vm-project?rev=313096&view=rev</a><br>Log:<br>[ubsan] Function Sanitizer: Don't require writable text segments<br><br>This change will make it possible to use -fsanitize=function on Darwin and<br>possibly on other platforms. It fixes an issue with the way RTTI is stored into<br>function prologue data.<br><br>On Darwin, addresses stored in prologue data can't require run-time fixups and<br>must be PC-relative. Run-time fixups are undesirable because they necessitate<br>writable text segments, which can lead to security issues. And absolute<br>addresses are undesirable because they break PIE mode.<br><br>The fix is to create a private global which points to the RTTI, and then to<br>encode a PC-relative reference to the global into prologue data.<br><br>Differential Revision:<span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span><a href="https://reviews.llvm.org/D37597" rel="noreferrer" target="_blank">https://reviews.llvm<wbr>.org/D37597</a><br><br>Modified:<br> <span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span>cfe/trunk/lib/CodeGen/CGExpr<wbr>.cpp<br> <span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span>cfe/trunk/lib/CodeGen/CodeGe<wbr>nFunction.cpp<br> <span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span>cfe/trunk/lib/CodeGen/CodeGe<wbr>nFunction.h<br> <span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span>cfe/trunk/lib/CodeGen/Target<wbr>Info.cpp<br> <span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span>cfe/trunk/test/CodeGenCXX/ca<wbr>tch-undef-behavior.cpp<br><br>Modified: cfe/trunk/lib/CodeGen/CGExpr.c<wbr>pp<br>URL:<span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGExpr.cpp?rev=313096&r1=313095&r2=313096&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/ll<wbr>vm-project/cfe/trunk/lib/CodeG<wbr>en/CGExpr.cpp?rev=313096&r1=<wbr>313095&r2=313096&view=diff</a><br>==============================<wbr>==============================<wbr>==================<br>--- cfe/trunk/lib/CodeGen/CGExpr.c<wbr>pp (original)<br>+++ cfe/trunk/lib/CodeGen/CGExpr.c<wbr>pp Tue Sep 12 17:04:35 2017<br>@@ -4409,10 +4409,7 @@ RValue CodeGenFunction::EmitCall(Qual<wbr>Typ<br> SanitizerScope SanScope(this);<br> llvm::Constant *FTRTTIConst =<br> CGM.GetAddrOfRTTIDescriptor(Q<wbr>ualType(FnType, 0), /*ForEH=*/true);<br>- llvm::Type *PrefixStructTyElems[] = {<br>- PrefixSig->getType(),<br>- FTRTTIConst->getType()<br>- };<br>+ llvm::Type *PrefixStructTyElems[] = {PrefixSig->getType(), Int32Ty};<br> llvm::StructType *PrefixStructTy = llvm::StructType::get(<br> CGM.getLLVMContext(), PrefixStructTyElems, /*isPacked=*/true);<br><br>@@ -4433,8 +4430,10 @@ RValue CodeGenFunction::EmitCall(Qual<wbr>Typ<br> EmitBlock(TypeCheck);<br> llvm::Value *CalleeRTTIPtr =<br> Builder.CreateConstGEP2_32(Pr<wbr>efixStructTy, CalleePrefixStruct, 0, 1);<br>- llvm::Value *CalleeRTTI =<br>+ llvm::Value *CalleeRTTIEncoded =<br> Builder.CreateAlignedLoad(Cal<wbr>leeRTTIPtr, getPointerAlign());<br>+ llvm::Value *CalleeRTTI =<br>+ DecodeAddrUsedInPrologue(Calle<wbr>ePtr, CalleeRTTIEncoded);<br> llvm::Value *CalleeRTTIMatch =<br> Builder.CreateICmpEQ(CalleeRT<wbr>TI, FTRTTIConst);<br> llvm::Constant *StaticData[] = {<br><br>Modified: cfe/trunk/lib/CodeGen/CodeGenF<wbr>unction.cpp<br>URL:<span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CodeGenFunction.cpp?rev=313096&r1=313095&r2=313096&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/ll<wbr>vm-project/cfe/trunk/lib/CodeG<wbr>en/CodeGenFunction.cpp?rev=<wbr>313096&r1=313095&r2=313096&<wbr>view=diff</a><br>==============================<wbr>==============================<wbr>==================<br>--- cfe/trunk/lib/CodeGen/CodeGenF<wbr>unction.cpp (original)<br>+++ cfe/trunk/lib/CodeGen/CodeGenF<wbr>unction.cpp Tue Sep 12 17:04:35 2017<br>@@ -429,6 +429,43 @@ bool CodeGenFunction::ShouldXRayIns<wbr>trume<br> return CGM.getCodeGenOpts().XRayInstr<wbr>umentFunctions;<br> }<br><br>+llvm::Constant *<br>+CodeGenFunction::EncodeAddrFo<wbr>rUseInPrologue(llvm::Function *F,<br>+ llvm::Constant *Addr) {<br>+ // Addresses stored in prologue data can't require run-time fixups and must<br>+ // be PC-relative. Run-time fixups are undesirable because they necessitate<br>+ // writable text segments, which are unsafe. And absolute addresses are<br>+ // undesirable because they break PIE mode.<br>+<br>+ // Add a layer of indirection through a private global. Taking its address<br>+ // won't result in a run-time fixup, even if Addr has linkonce_odr linkage.<br>+ auto *GV = new llvm::GlobalVariable(CGM.getMo<wbr>dule(), Addr->getType(),<br>+ /*isConstant=*/true,<br>+ llvm::GlobalValue::PrivateLink<wbr>age, Addr);<br>+<br>+ // Create a PC-relative address.<br>+ auto *GOTAsInt = llvm::ConstantExpr::getPtrToIn<wbr>t(GV, IntPtrTy);<br>+ auto *FuncAsInt = llvm::ConstantExpr::getPtrToIn<wbr>t(F, IntPtrTy);<br>+ auto *PCRelAsInt = llvm::ConstantExpr::getSub(GOT<wbr>AsInt, FuncAsInt);<br>+ return (IntPtrTy == Int32Ty)<br>+ ? PCRelAsInt<br>+ : llvm::ConstantExpr::getTrunc(P<wbr>CRelAsInt, Int32Ty);<br>+}<br>+<br>+llvm::Value *<br>+CodeGenFunction::DecodeAddrUs<wbr>edInPrologue(llvm::Value *F,<br>+ llvm::Value *EncodedAddr) {<br>+ // Reconstruct the address of the global.<br>+ auto *PCRelAsInt = Builder.CreateSExt(EncodedAddr<wbr>, IntPtrTy);<br>+ auto *FuncAsInt = Builder.CreatePtrToInt(F, IntPtrTy, "<a href="http://func_addr.int/" rel="noreferrer" target="_blank">func_addr.int</a>");<br>+ auto *GOTAsInt = Builder.CreateAdd(PCRelAsInt, FuncAsInt, "<a href="http://global_addr.int/" rel="noreferrer" target="_blank">global_addr.int</a>");<br>+ auto *GOTAddr = Builder.CreateIntToPtr(GOTAsIn<wbr>t, Int8PtrPtrTy, "global_addr");<br>+<br>+ // Load the original pointer through the global.<br>+ return Builder.CreateLoad(Address(GOT<wbr>Addr, getPointerAlign()),<br>+ "decoded_addr");<br>+}<br>+<br> /// EmitFunctionInstrumentation - Emit LLVM code to call the specified<br> /// instrumentation function with the current function and the call site, if<br> /// function instrumentation is enabled.<br>@@ -856,7 +893,10 @@ void CodeGenFunction::StartFunction<wbr>(Glob<br> CGM.getTargetCodeGenInfo().ge<wbr>tUBSanFunctionSignature(CGM)) {<br> llvm::Constant *FTRTTIConst =<br> CGM.GetAddrOfRTTIDescriptor(F<wbr>D->getType(), /*ForEH=*/true);<br>- llvm::Constant *PrologueStructElems[] = { PrologueSig, FTRTTIConst };<br>+ llvm::Constant *FTRTTIConstEncoded =<br>+ EncodeAddrForUseInPrologue(Fn, FTRTTIConst);<br>+ llvm::Constant *PrologueStructElems[] = {PrologueSig,<br>+ FTRTTIConstEncoded};<br> llvm::Constant *PrologueStructConst =<br> llvm::ConstantStruct::getAnon<wbr>(PrologueStructElems, /*Packed=*/true);<br> Fn->setPrologueData(PrologueS<wbr>tructConst);<br><br>Modified: cfe/trunk/lib/CodeGen/CodeGenF<wbr>unction.h<br>URL:<span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CodeGenFunction.h?rev=313096&r1=313095&r2=313096&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/ll<wbr>vm-project/cfe/trunk/lib/CodeG<wbr>en/CodeGenFunction.h?rev=31309<wbr>6&r1=313095&r2=313096&view=<wbr>diff</a><br>==============================<wbr>==============================<wbr>==================<br>--- cfe/trunk/lib/CodeGen/CodeGenF<wbr>unction.h (original)<br>+++ cfe/trunk/lib/CodeGen/CodeGenF<wbr>unction.h Tue Sep 12 17:04:35 2017<br>@@ -1776,6 +1776,15 @@ public:<br> /// EmitMCountInstrumentation - Emit call to .mcount.<br> void EmitMCountInstrumentation();<br><br>+ /// Encode an address into a form suitable for use in a function prologue.<br>+ llvm::Constant *EncodeAddrForUseInPrologue(ll<wbr>vm::Function *F,<br>+ llvm::Constant *Addr);<br>+<br>+ /// Decode an address used in a function prologue, encoded by \c<br>+ /// EncodeAddrForUseInPrologue.<br>+ llvm::Value *DecodeAddrUsedInPrologue(llvm<wbr>::Value *F,<br>+ llvm::Value *EncodedAddr);<br>+<br> /// EmitFunctionProlog - Emit the target specific LLVM code to load the<br> /// arguments for the given function. This is also responsible for naming the<br> /// LLVM function arguments.<br><br>Modified: cfe/trunk/lib/CodeGen/TargetIn<wbr>fo.cpp<br>URL:<span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/TargetInfo.cpp?rev=313096&r1=313095&r2=313096&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/ll<wbr>vm-project/cfe/trunk/lib/CodeG<wbr>en/TargetInfo.cpp?rev=313096&<wbr>r1=313095&r2=313096&view=diff</a><br>==============================<wbr>==============================<wbr>==================<br>--- cfe/trunk/lib/CodeGen/TargetIn<wbr>fo.cpp (original)<br>+++ cfe/trunk/lib/CodeGen/TargetIn<wbr>fo.cpp Tue Sep 12 17:04:35 2017<br>@@ -1086,8 +1086,8 @@ public:<br> getUBSanFunctionSignature(Cod<wbr>eGen::CodeGenModule &CGM) const override {<br> unsigned Sig = (0xeb << 0) | // jmp rel8<br> <span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span>(0x06 << 8) | // .+0x08<br>- ('F' << 16) |<br>- ('T' << 24);<br>+ ('v' << 16) |<br>+ ('2' << 24);<br> return llvm::ConstantInt::get(CGM.Int<wbr>32Ty, Sig);<br> }<br><br>@@ -2277,17 +2277,10 @@ public:<br><br> llvm::Constant *<br> getUBSanFunctionSignature(Cod<wbr>eGen::CodeGenModule &CGM) const override {<br>- unsigned Sig;<br>- if (getABIInfo().has64BitPointers<wbr>())<br>- Sig = (0xeb << 0) | // jmp rel8<br>- (0x0a << 8) | // .+0x0c<br>- ('F' << 16) |<br>- ('T' << 24);<br>- else<br>- Sig = (0xeb << 0) | // jmp rel8<br>- (0x06 << 8) | // .+0x08<br>- ('F' << 16) |<br>- ('T' << 24);<br>+ unsigned Sig = (0xeb << 0) | // jmp rel8<br>+ (0x06 << 8) | // .+0x08<br>+ ('v' << 16) |<br>+ ('2' << 24);<br> return llvm::ConstantInt::get(CGM.Int<wbr>32Ty, Sig);<br> }<br><br><br>Modified: cfe/trunk/test/CodeGenCXX/catc<wbr>h-undef-behavior.cpp<br>URL:<span class="m_-3309266437497712727m_1661471059887043044Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGenCXX/catch-undef-behavior.cpp?rev=313096&r1=313095&r2=313096&view=diff" rel="noreferrer" target="_blank">http://llvm.org/viewvc/ll<wbr>vm-project/cfe/trunk/test/Code<wbr>GenCXX/catch-undef-behavior.<wbr>cpp?rev=313096&r1=313095&r2=<wbr>313096&view=diff</a><br>==============================<wbr>==============================<wbr>==================<br>--- cfe/trunk/test/CodeGenCXX/catc<wbr>h-undef-behavior.cpp (original)<br>+++ cfe/trunk/test/CodeGenCXX/catc<wbr>h-undef-behavior.cpp Tue Sep 12 17:04:35 2017<br>@@ -16,6 +16,10 @@ struct S {<br> // Check that type mismatch handler is not modified by ASan.<br> // CHECK-ASAN: private unnamed_addr global { { [{{.*}} x i8]*, i32, i32 }, { i16, i16, [4 x i8] }*, i8*, i8 } { {{.*}}, { i16, i16, [4 x i8] }* [[TYPE_DESCR]], {{.*}} }<br><br>+// CHECK: [[IndirectRTTI_ZTIFvPFviEE:@.+<wbr>]] = private constant i8* bitcast ({ i8*, i8* }* @_ZTIFvPFviEE to i8*)<br>+// CHECK-X86: [[IndirectRTTI_ZTIFvPFviEE:@.+<wbr>]] = private constant i8* bitcast ({ i8*, i8* }* @_ZTIFvPFviEE to i8*)<br>+// CHECK-X32: [[IndirectRTTI_ZTIFvPFviEE:@.+<wbr>]] = private constant i8* bitcast ({ i8*, i8* }* @_ZTIFvPFviEE to i8*)<br>+<br> struct T : S {};<br><br> // CHECK-LABEL: @_Z17reference_binding<br>@@ -395,23 +399,30 @@ void downcast_reference(B &b) {<br> // CHECK-NEXT: br i1 [[AND]]<br> }<br><br>-// CHECK-LABEL: @_Z22indirect_function_callPFv<wbr>iE({{.*}} prologue <{ i32, i8* }> <{ i32 1413876459, i8* bitcast ({ i8*, i8* }* @_ZTIFvPFviEE to i8*) }><br>-// CHECK-X32: @_Z22indirect_function_callPFv<wbr>iE({{.*}} prologue <{ i32, i8* }> <{ i32 1413875435, i8* bitcast ({ i8*, i8* }* @_ZTIFvPFviEE to i8*) }><br>-// CHECK-X86: @_Z22indirect_function_callPFv<wbr>iE({{.*}} prologue <{ i32, i8* }> <{ i32 1413875435, i8* bitcast ({ i8*, i8* }* @_ZTIFvPFviEE to i8*) }><br>+//<br>+// CHECK-LABEL: @_Z22indirect_function_callPFv<wbr>iE({{.*}} prologue <{ i32, i32 }> <{ i32 846595819, i32 trunc (i64 sub (i64 ptrtoint (i8** {{.*}} to i64), i64 ptrtoint (void (void (i32)*)* @_Z22indirect_function_callPFv<wbr>iE to i64)) to i32) }><br>+// CHECK-X32: @_Z22indirect_function_callPFv<wbr>iE({{.*}} prologue <{ i32, i32 }> <{ i32 846595819, i32 sub (i32 ptrtoint (i8** [[IndirectRTTI_ZTIFvPFviEE]] to i32), i32 ptrtoint (void (void (i32)*)* @_Z22indirect_function_callPFv<wbr>iE to i32)) }><br>+// CHECK-X86: @_Z22indirect_function_callPFv<wbr>iE({{.*}} prologue <{ i32, i32 }> <{ i32 846595819, i32 sub (i32 ptrtoint (i8** [[IndirectRTTI_ZTIFvPFviEE]] to i32), i32 ptrtoint (void (void (i32)*)* @_Z22indirect_function_callPFv<wbr>iE to i32)) }><br> void indirect_function_call(void (*p)(int)) {<br>- // CHECK: [[PTR:%.+]] = bitcast void (i32)* {{.*}} to <{ i32, i8* }>*<br>+ // CHECK: [[PTR:%.+]] = bitcast void (i32)* {{.*}} to <{ i32, i32 }>*<br><br> // Signature check<br>- // CHECK-NEXT: [[SIGPTR:%.+]] = getelementptr <{ i32, i8* }>, <{ i32, i8* }>* [[PTR]], i32 0, i32 0<br>+ // CHECK-NEXT: [[SIGPTR:%.+]] = getelementptr <{ i32, i32 }>, <{ i32, i32 }>* [[PTR]], i32 0, i32 0<br> // CHECK-NEXT: [[SIG:%.+]] = load i32, i32* [[SIGPTR]]<br>- // CHECK-NEXT: [[SIGCMP:%.+]] = icmp eq i32 [[SIG]], 1413876459<br>+ // CHECK-NEXT: [[SIGCMP:%.+]] = icmp eq i32 [[SIG]], 846595819<br> // CHECK-NEXT: br i1 [[SIGCMP]]<br><br> // RTTI pointer check<br>- // CHECK: [[RTTIPTR:%.+]] = getelementptr <{ i32, i8* }>, <{ i32, i8* }>* [[PTR]], i32 0, i32 1<br>- // CHECK-NEXT: [[RTTI:%.+]] = load i8*, i8** [[RTTIPTR]]<br>+ // CHECK: [[RTTIPTR:%.+]] = getelementptr <{ i32, i32 }>, <{ i32, i32 }>* [[PTR]], i32 0, i32 1<br>+ // CHECK-NEXT: [[RTTIEncIntTrunc:%.+]] = load i32, i32* [[RTTIPTR]]<br>+ // CHECK-NEXT: [[RTTIEncInt:%.+]] = sext i32 [[RTTIEncIntTrunc]] to i64<br>+ // CHECK-NEXT: [[FuncAddrInt:%.+]] = ptrtoint void (i32)* {{.*}} to i64<br>+ // CHECK-NEXT: [[IndirectGVInt:%.+]] = add i64 [[RTTIEncInt]], [[FuncAddrInt]]<br>+ // CHECK-NEXT: [[IndirectGV:%.+]] = inttoptr i64 [[IndirectGVInt]] to i8**<br>+ // CHECK-NEXT: [[RTTI:%.+]] = load i8*, i8** [[IndirectGV]], align 8<br> // CHECK-NEXT: [[RTTICMP:%.+]] = icmp eq i8* [[RTTI]], bitcast ({ i8*, i8* }* @_ZTIFviE to i8*)<br> // CHECK-NEXT: br i1 [[RTTICMP]]<br>+<br> p(42);<br> }<br><br><br><br>______________________________<wbr>_________________<br>cfe-commits mailing list<br><a href="mailto:cfe-commits@lists.llvm.org" target="_blank">cfe-commits@lists.llvm.org</a><br><a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/<wbr>mailman/listinfo/cfe-commits</a></blockquote></div></div></div></div></blockquote></div></div></div></blockquote></div></div></blockquote></div></div></div></blockquote></div></div></blockquote></div></div></div></blockquote></div><br></div></div></div></blockquote></div><br><br clear="all"><div><br></div></div></div><span class="HOEnZb"><font color="#888888">-- <br><div class="m_-3309266437497712727gmail_signature" data-smartmail="gmail_signature"><div dir="ltr">-- <div>Peter</div></div></div>
</font></span></div>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr">-- <div>Peter</div></div></div>
</div></div>