| Age | Commit message (Collapse) | Author |
|---|
|
- If an unknown intrinsic appears in the bitcode file,
it is reported but execution can proceed.
- If an unknown intrinsic is encountered during execution of some path,
- the intrinsic is reported
- this path is treated as an error
- execution of other paths can proceed
To be more precise, there is a list of "known unknown intrinsics".
Intrinsics not on this list will prevent execution.
|
|
Changes:
- IntrinsicCleaner accepts fshr/fshl as accepted intrinsics
- Executor::executeCall converts fshr/fshl to urem/zext/concat/shift/extract
- Klee/main suppresses warnings about externals that are LLVM reserved
(i.e., begin with "llvm.")
- New test exercises 32 and 7 bit versions including oversize shift values
Test values are based on LLVM's test for fshl/fshr
- Changes that depend on existence of fshr/fshl are guarded by
#if LLVM_VERSION_CODE >= LLVM_VERSION(7, 0)
or
; REQUIRES: geq-llvm-7.0
|
|
|
|
llvm.objectsize is used in several optimisation during compile time. Lowering
these intrinsics took a conservative approach returning always the value for
unknown. Instead, lower to the object's real size, if possible. Otherwise,
a conservative value is used.
Since LLVM 4.0, the function `llvm::lowerObjectSizeCall()` does exactly
this. Use this function or preserve the old behaviour for older LLVM versions.
|
|
|
|
|
|
|
|
|
|
|
|
function and updated some .ll tests to use --optimize=false instead of --disable-opt
|
|
@llvm.objectsize has now three aguments, so fix the tests accordingly.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
|
|
Clone some tests to have their 3.7 version. 'call's, 'load's and
'getelementptr's match the new specification in them.
@andreamattavelli: Fixed test cases: BitCastAlias test cases included
modification to alias specifications that require LLVM 3.8
[v2] added comments what was changed and why
[v3] the new tests are without suffix, the old ones have ".leq36".
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
|
|
|
|
introduced in LLVM 2.7. Previously KLEE would emit the following error
message when ``IntrinsicLowering::LowerIntrinsicCall()`` was called on
the intrinsic
```
LLVM ERROR: Code generator does not support intrinsic function 'llvm.objectsize.i64.p0i8'!
```
The ``IntrinsicCleaner`` pass now lowers this intrinsic to a constant
integer depending on the second argument to the intrinsic. This
corresponds to the case where the size of the object pointed to by the
first argument is unknown.
An alternative design would be to handle this intrinsic in the Executor
where is actually possible to know the size of objects during execution.
However that would be much more complicated because if the pointer is
symbolic we would have to fork for every object that could be pointed
to.
The implementation is similar to #260 but we handle the second argument
to the intrinsic correctly and also have a simple test case.
Unfortunately we have to have a different version of the test case
for LLVM 2.9 because the expected suffix for the intrinsic is different
in LLVM 2.9.
|
