<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/154462>154462</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            `parseSourceString` runs over input `StringRef`
        </td>
    </tr>

    <tr>
      <th>Labels</th>
      <td>
            new issue
      </td>
    </tr>

    <tr>
      <th>Assignees</th>
      <td>
      </td>
    </tr>

    <tr>
      <th>Reporter</th>
      <td>
          SchrodingerZhu
      </td>
    </tr>
</table>

<pre>
    Consider the following parsing code:
```c++
// MLIR
#include <llvm/Support/MemoryBuffer.h>
#include <llvm/Support/SourceMgr.h>
#include <mlir/Dialect/Arith/IR/Arith.h>
#include <mlir/Dialect/ControlFlow/IR/ControlFlow.h>
#include <mlir/Dialect/DLTI/DLTI.h>
#include <mlir/Dialect/Func/IR/FuncOps.h>
#include <mlir/Dialect/LLVMIR/LLVMDialect.h>
#include <mlir/Dialect/MemRef/IR/MemRef.h>
#include <mlir/Dialect/SCF/IR/SCF.h>
#include <mlir/Dialect/UB/IR/UBOps.h>
#include <mlir/IR/AsmState.h>
#include <mlir/IR/BuiltinAttributes.h>
#include <mlir/IR/BuiltinOps.h>
#include <mlir/IR/BuiltinTypes.h>
#include <mlir/IR/Diagnostics.h>
#include <mlir/IR/DialectRegistry.h>
#include <mlir/InitAllDialects.h>
#include <mlir/Interfaces/DataLayoutInterfaces.h>
#include <mlir/Parser/Parser.h>
#include <mlir/Support/LLVM.h>
#include <mlir/Support/LogicalResult.h>
#include <mlir/Target/LLVMIR/Import.h>

using namespace mlir;
const char *data = "module {} sadfkhjasdf";
int main() {

  std::string_view sliced(data, 9);

  // 1) Build a registry and MLIR context with required dialects.
 DialectRegistry registry;
  mlir::registerAllDialects(registry);
 MLIRContext context(registry);
  context.loadAllAvailableDialects();

// 2) Parse the incoming MLIR module from string.
#ifdef DO_COPY
 llvm::SourceMgr sourceMgr;
  auto buffer = llvm::MemoryBuffer::getMemBufferCopy(sliced);
 sourceMgr.AddNewSourceBuffer(std::move(buffer), llvm::SMLoc());
 OwningOpRef<ModuleOp> module = parseSourceFile<ModuleOp>(sourceMgr, &context);
#else
  OwningOpRef<ModuleOp> module = parseSourceString<ModuleOp>(sliced, &context);
#endif
}
```

When compiled with `LLVM 20` with 
```
clang++ main.cpp -lLLVM -lMLIR
```

The MLIR Asm Parser runs over the slice:

```
loc("-":1:11): error: custom op 'sadfkhjasdf' is unknown (tried 'builtin.sadfkhjasdf' as well)
```

with `DO_COPY`, the problem solves.
</pre>
<img width="1" height="1" alt="" src="http://email.email.llvm.org/o/eJyUVt1u4ygUfhpygxpRHOfnwhdOMpEqNdtR09nV7s0Iw4nNDAYv4GT79iuwnSYzO61XSmQM5zu_nw-HOSdLDZChdI3S7YS1vjI2O_DKGiF1Cfavqp0URrxmG6OdFGCxrwAfjVLmLHWJG2ZdeHIjACU5Ijmak-7HEV2HH8kR3SG6w_vHh-f4lkjNVSsAo2Sj1KlGdHdom8ZYj-huD7Wxr-v2eAQ7rVDy6WPIwbSWw778hXytpEV0t5VMAQ_yuZW-QnT38Dy8jARujPbWqJ0y5wF-tTVSyfbx5aF_jETsWs0He2H91LiRyMfH3_cRFhb99kjoHupnOA5mu7eR0MNmN-AOm91I0Jf1gPmy_ijArnCuPnjm4WPJdSuVlzr33sqi9TBCeQ8Z5Ugv-_LajNG8lazUxnnJxwmH9DxDKZ23r-8DtPS5Uj3kA-Xagz0yDi4YYZ49slfT-rftd9GfmXXwtnhX9u0rDRQcK2pKyZl6Bteq9_n6wmwJN0x_qIOSC4jkbWxQmtXgGsYBR2QS-hI32nnMK2YxorlgnmGUbDGitDaiVYDRYo0WW-yYOH6vvjEnjojSDiu1xzWTGtEloqsoGa1h7LwInTDJnbdSl19PEs7YKclBILoMVhDd4BWiq05TBPUt8j7oCnwSmGHbVx0zLWLzxNxoD_94fJa-whb-bqUFgcVQ8aDpB8ZclHS2cB99cK87AXvFGUSXF_nBvWh50xvuHfhPueFwqgwTuVL5iUnFCgVX2q-D7kOmIeRIpHi1SM1NHQoWA-7rcLSmxl06pz0TjgKOePv0dfP0-c9gPV4KMa7LbYDdsBo8ZK03uIhXS6z0G-j61ul2SvB7qLudjWleEV0ORRwivuif5kL8BufOcq-ELi9EqM0JEF0W_cEq1P_K3_2j4X1uLqqfzlrq8qkJLTjZ7GManhqUfBpSEtwPly90RndSwa1gcOASP91gROeX6g1FoAkoBzE1_9fgIVbjZ5N9in5pTwt5DMvF9npW6BjxRwUac1M3UoHoSI7mJHzZmAapfusWxxXTZTdqxA9yypsG36mIulPDzHFr6KWCjl-5qzvuWWxb7bA59RNOjKMfaG7hqi8WvYu9IL8P__sYY47BWhP4g3nrvKmxaTCii5v-scDS4VZ_1-asMaJLbyWIIFV018j0B2nm8BmUCgZ-imNI0fAdzEnIfPC_saZQUGNn1AncdCKyRKySFZtAdr9I0_n8nlIyqTKxghklsxUp5glLCUuB8KIgK1okC0EJmciMEpqSJSUkmRFKpsksXaRsloh0xsVsxtGMQM2kmgZGT40tJ9K5FrL7dDab04liBSgXB0xKNZxxPA2pS7cTmwXQXdGWDs2Iks67NzVeegUZmpOfWTcnV-WSuml9yEJ3GAg8J5PWqqzyvnGhiLHTlNJXbTHlJgyP_QwZzTfWfOtmkOhbuBJ7508Z_TcAAP__PUx0tQ">