[LLVMdev] load bytecode from string for jiting problem (original) (raw)
Vikas Bhargava vikasbhargava at gmail.com
Thu Mar 20 14:32:29 PDT 2014
- Previous message: [LLVMdev] load bytecode from string for jiting problem
- Next message: [LLVMdev] load bytecode from string for jiting problem
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hello Willy, Here is the dump from one of my bitcode files:
0000000 42 43 c0 de 21 0c 00 00 25 05 00 00 0b 82 20 00
As expected, 0x42 (= B), 0x43 (= C), xc0 and 0xde are in correct order. In your case, the first byte is read as 37 (= 0x25). I wonder why? When you check the bytes yourself, you get expected results. When the same bytes are read from Stream object, you get a different result (maybe garbage). I would suggest that you put a watchpoint on mbjit->getBufferStart() and single step your program to make sure it is not freed, over written somewhere.
thx Vikas.
On Thu, Mar 20, 2014 at 7:50 AM, Willy WOLFF <willy.wolff at etu.unistra.fr>wrote:
The stack trace is: (gdb) bt #0 0x00000000004fa8c8 in llvm::BitstreamCursor::Read(unsigned int) () #1 0x00000000004fa1d2 in llvm::BitcodeReader:: ParseBitcodeInto(llvm::Module*) () #2 0x0000000000503ae9 in llvm::getLazyBitcodeModule(llvm::MemoryBuffer*, llvm::LLVMContext&) () #3 0x0000000000503eb6 in llvm::parseBitcodeFile(llvm::MemoryBuffer*, llvm::LLVMContext&) () #4 0x00000000004ec195 in jitter (skeletons=, params=0x7fffffffdf40, phistate=0x11adbc0, lower=0, upper=250, instouter=8, instinner=) at /home/willy/hellostuff/withapollo/simpleloop/runtime.cpp:263 #5 0x00000000004ec8fa in apolloruntimehook (info=, skeletons=0xc8b1f0, skeletonsize=, params=0x7fffffffdf40, phistatesize=) at /home/willy/hellostuff/withapollo/simpleloop/runtime.cpp:438 #6 0x00000000004ee753 in ?? () #7 0x00000000004ecbf1 in main (argc=, argv=) at /home/willy/hellostuff/withapollo/simpleloop/simple scevdynamicarray.c:84
On 03/20/2014 10:10 AM, Willy WOLFF wrote: This segfault occuring only under valgrind, in shell way, and in gdb way i have Invalid bitcode signature simplescevdynamicarray: /home/willy/apollo/llvm/include/llvm/Support/ErrorOr.h:258: storagetype *llvm::ErrorOr<llvm::Module *>::getStorage() [T = llvm::Module *]: Assertion `!HasError && "Cannot get value when an error exists!"' failed. Command terminated by signal 6
this is the code I use: long jitter(void* info, skeletonpair *skeletons, long skeletonsize, paramt params, long phistatesize) { InitializeNativeTarget(); InitializeNativeTargetAsmPrinter(); llvm::StringRef sr (skeletons[idx].jitbytecode, skeletons[idx].jitbytecodelength); if (sr.str()[0] == 'B') std::cout << "B ok\n";_ _if (sr.str()[1] == 'C')_ _std::cout << "C ok\n";_ _if (sr.str()[2] == (char) 0xc0)_ _std::cout << "0xc0 ok\n";_ _if (sr.str()[3] == (char) 0xde)_ _std::cout << "0xde ok\n";_ _llvm::MemoryBuffer* mbjit = llvm::MemoryBuffer::getMemBufferCopy (sr);_ _llvm::ErrorOrllvm::Module* ModuleOrErr = llvm::parseBitcodeFile (mbjit, context); if (llvm::errorcode EC = ModuleOrErr.getError()) { std::cout << ModuleOrErr.getError().message() << "\n";_ _}_ _Module* Mjit = ModuleOrErr.get();_ _std::string eeError;_ _ExecutionEngine* nee =_ _EngineBuilder(Mjit).setEngineKind(EngineKind::JIT).setUseMCJIT(true)._ _setErrorStr(&eeError).create();_ _if (!nee) {_ _fprintf(stderr, "Could not create ExecutionEngine: %s\n",_ _eeError.cstr());_ _assert(false);_ _}_ _Function f = retfct(Mjit); // Function* retfct (Module*); return_ _the function we want to jit._ _uint64t fptr = nee->getFunctionAddress(f->getName()); long (*fjited)(paramt, phistatet, long, long, long, long) = (long (*)(paramt, phistatet, long, long, long, long)) (intptrt)fptr; returnfjited (params, phistate, lower, upper, instouter, instinner); } Thanks, -- Willy WOLFF On 20 Mar 2014, at 00:37, Vikas Bhargava wrote: segmentation fault indicates memory corruption and it's hard to tell without seeing the exact use of the APIs. If possible, please post a complete program and gdb stack trace from the core file. If there are multiple threads using the global variables, please let us know. FWIW, I have some tests to write llvm::Module to bitcode files and read them back into llvm::Module and they work just fine with 3.4 (never tried with tip). thx vikas. ========
On Wed, Mar 19, 2014 at 2:58 PM, Willy WOLFF <willy.wolff at etu.unistra.fr <mailto:willy.wolff at etu.unistra.fr>> wrote: all of: ---- // cout << "lsr: " << lsr << "\n"; llvm::MemoryBuffer* mbjit = llvm::MemoryBuffer::getMemBufferCopy (sr); ------ string lsr = sr.str(); // cout << "lsr: " << lsr << "\n"; llvm::MemoryBuffer* mbjit = llvm::MemoryBuffer::getMemBuffer (lsr); ------- string lsr = sr.str(); // cout << "lsr: " << lsr << "\n"; llvm::MemoryBuffer* mbjit = llvm::MemoryBuffer::getMemBufferCopy (lsr); have same result as invalid bit code. The result of valgrind, effectively, i have invalid reads in the parseBitcodeFile: ==536== Conditional jump or move depends on uninitialised value(s) ==536== at 0x501FE3: llvm::BitstreamCursor::Read(unsigned int) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x501A19: llvm::BitcodeReader::ParseBitcodeInto(llvm::Module*) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x50AEC8: llvm::getLazyBitcodeModule(llvm::MemoryBuffer*, llvm::LLVMContext&) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x50B295: llvm::parseBitcodeFile(llvm::MemoryBuffer*, llvm::LLVMContext&) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x4F1231: blahruntimehook (runtime.cpp:348) ==536== by 0x4F46C2: ??? (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x4F2B60: main (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== ==536== Invalid read of size 8 ==536== at 0x501FE8: llvm::BitstreamCursor::Read(unsigned int) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x501A19: llvm::BitcodeReader::ParseBitcodeInto(llvm::Module*) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x50AEC8: llvm::getLazyBitcodeModule(llvm::MemoryBuffer*, llvm::LLVMContext&) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x50B295: llvm::parseBitcodeFile(llvm::MemoryBuffer*, llvm::LLVMContext&) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x4F1231: blahruntimehook (runtime.cpp:348) ==536== by 0x4F46C2: ??? (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x4F2B60: main (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== Address 0x0 is not stack'd, malloc'd or (recently) free'd ==536== ==536== ==536== Process terminating with default action of signal 11 (SIGSEGV) ==536== Access not within mapped region at address 0x0 ==536== at 0x501FE8: llvm::BitstreamCursor::Read(unsigned int) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x501A19: llvm::BitcodeReader::ParseBitcodeInto(llvm::Module*) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x50AEC8: llvm::getLazyBitcodeModule(llvm::MemoryBuffer*, llvm::LLVMContext&) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x50B295: llvm::parseBitcodeFile(llvm::MemoryBuffer*, llvm::LLVMContext&) (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x4F1231: blahruntimehook (runtime.cpp:348) ==536== by 0x4F46C2: ??? (in /home/willy/blahtestscript/newblah/simplescevdynamicarray) ==536== by 0x4F2B60: main (in /home/willy/blahtestscript/newblah/simplescevdynamicarray)
-- Willy WOLFF On 19 Mar 2014, at 22:11, Vikas Bhargava wrote: Hi Willy, If the disassembly of the module works fine, then there is nothing wrong with the module. Stream uses the memorybuffer that you pass in parseBitcodeFile. If what Will is saying is true, there is something wrong with your code in "3:", i.e.: MemoryBuffer* mbjit = MemoryBuffer::getMemBuffer (sr.str()); LLVMContext& context = getGlobalContext(); ErrorOr<Module*> ModuleOrErr = parseBitcodeFile (mbjit, context); if (errorcode EC = ModuleOrErr.getError()) { std::cout << ModuleOrErr.getError(). message() << "\n"; assert(false); } Can you post how you modified it in your second reply? For debugging purpose, you can simply use MemoryBuffer::getMemBufferCopy() and not worry about validity of stringref or null-termination. Also, you can run your program through valgrind and check for any invalid reads. HTH Vikas. =======
On Wed, Mar 19, 2014 at 10:32 AM, Willy WOLFF <willy.wolff at etu.unistra.fr <mailto:willy.wolff at etu.unistra.fr>> wrote: I mad the change, and still have the problem. I investigate more the source code of llvm. First, I change isRawBitcode function to print the content of the parameter like this: original: http://llvm.org/docs/doxygen/_html/ReaderWriter8hsource. html#l00081 <http://llvm.org/docs/doxygen/html/ReaderWriter8hsource._ _html#l00081> inline bool isRawBitcode(const unsigned char *BufPtr, const unsigned char *BufEnd) { // These bytes sort of have a hidden message, but it's not in // little-endian this time, and it's a little redundant. errs()<< "isRawBitcode output:\n";_ _for (int i = 0; i < 4; i++)_ _errs() << BufPtr[i] << "\n";_ _if (BufPtr != BufEnd )_ _errs() << "BP != BE ok\n";_ _if (BufPtr[0] == 'B')_ _errs() << "B ok\n";_ _if (BufPtr[1] == 'C')_ _errs() << "C ok\n";_ _if (BufPtr[2] == 0xc0)_ _errs() << "0xc0 ok\n";_ _if (BufPtr[3] == 0xde)_ _errs() << "0xde ok\n";_ _return BufPtr != BufEnd &&_ _BufPtr[0] == 'B' &&_ _BufPtr[1] == 'C' &&_ _BufPtr[2] == 0xc0 &&_ _BufPtr[3] == 0xde;_ _}_ _Second, I change ParseBitcodeInto as this:_ _original:_ _http://llvm.org/docs/doxygen/html/BitcodeReader8cpp_ _source.html#l01971_ _<http://llvm.org/docs/doxygen/html/BitcodeReader8cpp_ _source.html#l01971> ... errs() << "parsebitcodeinto sniff the signature\n"; uint32t bvar = Stream.Read(8); errs() << "B :" << bvar << "\n"; if (bvar != 'B') { errs() << "B :" << bvar << "\n"; return Error(InvalidBitcodeSignature);_ } if (Stream.Read(8) != 'C') { errs() << "C\n"; return Error(InvalidBitcodeSignature);_ } if ( Stream.Read(8) != 0xc0 ) { errs() << "0xc0\n"; return Error(InvalidBitcodeSignature);_ } if ( Stream.Read(8) != 0xde ) { errs() << "0xde\n"; return Error(InvalidBitcodeSignature);_ } // if (Stream.Read(8) != 'B' || // Stream.Read(8) != 'C' || // Stream.Read(4) != 0x0 || // Stream.Read(4) != 0xC || // Stream.Read(4) != 0xE || // Stream.Read(4) != 0xD // ) { ... The output of the code is : isRawBitcode output: B C BP != BE ok B ok C ok 0xc0 ok 0xde ok parsebitcodeinto sniff the signature B :37 B :37
It's possible that Stream object is not correctly initialized? On 03/13/2014 06:37 PM, Will Dietz wrote: On Thu, Mar 13, 2014 at 9:02 AM, Willy WOLFF <willy.wolff at etu.unistra.fr_ _<mailto:willy.wolff at etu.unistra.fr>> wrote: Hello, I having a weird problem while writing a bytecode module to a string, and after read/parse it for unsing on a jit. I write a pass to export function to module, and put this module inside a global variable. I use WriteBitcodeToFile for this. For debuging, after this write, I try to load the exported module with parseBitcodeFile. This two step works. After, while the compiled program is running, I try to read and parse this global variable for jiting the function. 1) I read the global variable with StringRef sr (gv, gvlength); 2) I manually test this bytecode by (inspired by inline bool isRawBitcode(const unsigned char *BufPtr, const unsigned char *BufEnd) at http://llvm.org/docs/doxygen/ _html/ReaderWriter8hsource.html#l00067 <http://llvm.org/docs/doxygen/_ _html/ReaderWriter8hsource.html#l00067>) if (sr.str()[0] == 'B') std::cout << "B ok\n";_ _if (sr.str()[1] == 'C')_ _std::cout << "C ok\n";_ _if (sr.str()[2] == (char) 0xc0)_ _std::cout << "0xc0 ok\n";_ _if (sr.str()[3] == (char) 0xde)_ _std::cout << "0xde ok\n";_ _3) I try to parse the gv by_ _MemoryBuffer* mbjit = MemoryBuffer::getMemBuffer_ _(sr.str());_ _Not sure if this is your issue, but should be fixed anyway:_ _The std::string created by "sr.str()" ends its lifetime_ _in this_ _statement, and MemoryBuffer for efficiency reasons_ _avoids copying data it doesn't have to (like StringRef)_ _so will be_ _referencing the freed memory._ _To resolve this:_ _* Pass MemoryBuffer your StringRef directly_ _* Use getMemBufferCopy()_ _* Preserve the result of sr.str() into a stack variable_ _and pass that_ _to getMemoryBuffer() instead._ _As a final note, check if your bitcode buffer "string" is_ _null-terminated or not. If not, be sure to be careful and_ _do things like informing MemoryBuffer that this is the case._ _Hope this helps,_ _~Will_ _LLVMContext& context = getGlobalContext();_ _ErrorOr<Module*> ModuleOrErr = parseBitcodeFile (mbjit, context); if (errorcode EC = ModuleOrErr.getError()) { _std::cout << ModuleOrErr.getError().message() << "\n"; assert(false); } This is the execution result: B ok C ok 0xc0 ok 0xde ok Invalid bitcode signature Ok is not working :/ But why ??? For debuging, between 2) and 3), I export the readed module and write to a file on my hard drive, and try llvm-dis, and the dissasembly of the module works. Wath's wrong? Any idea for solve this problem? Thanks you very much. Regards, Willy
LLVM Developers mailing list LLVMdev at cs.uiuc.edu <mailto:LLVMdev at cs.uiuc.edu> http://llvm.cs.uiuc.edu <http://llvm.cs.uiuc.edu/> http://lists.cs.uiuc.edu/_mailman/listinfo/llvmdev <http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev>
LLVM Developers mailing list LLVMdev at cs.uiuc.edu <mailto:LLVMdev at cs.uiuc.edu> http://llvm.cs.uiuc.edu <http://llvm.cs.uiuc.edu/> http://lists.cs.uiuc.edu/_mailman/listinfo/llvmdev <http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev>
LLVM Developers mailing list LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20140320/272d8139/attachment.html>
- Previous message: [LLVMdev] load bytecode from string for jiting problem
- Next message: [LLVMdev] load bytecode from string for jiting problem
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]