[go: up one dir, main page]
Skip to content
This repository has been archived by the owner on Aug 23, 2022. It is now read-only.

Unable to Uplift x86 PE: Check failed 'seg_type' Must be non NULL #740

Open
SaifRushdHadad opened this issue Mar 4, 2021 · 0 comments
Open

Unable to Uplift x86 PE: Check failed 'seg_type' Must be non NULL #740

SaifRushdHadad opened this issue Mar 4, 2021 · 0 comments

Comments

@SaifRushdHadad
Copy link

I am getting the error Check failed: 'seg_type' Must be non NULL when trying to uplift a x86 PE skeleton helloworld file. I am not sure if it is my install, based on the documentation McSema should fully support x86 PEs, I was wondering if there is a test suite corpus that is available that I could test my installation against?

Tool versioning:

  • McSema v3.0.23 LLVM 10 - Ubuntu 18.04
  • IDA Pro 7.1

Steps to reproduce:

echo 'void main() {};' > hello.c && \
i686-w64-mingw32-gcc hello.c -fPIC -pie -o hello.exe
"${MCSEMAPATH}/mcsema/bin/mcsema-disass-3" \
>     --disassembler "${IDAPATH}/idat" \
>     --arch x86 \
>     --os windows \
>     --entrypoint main\
>     --binary "${WORKDIR}/${TARGET}.exe" \
>     --output "${WORKDIR}/${TARGET}.cfg" \
>     --log_file "${WORKDIR}/${TARGET}.log"
"${MCSEMAPATH}/mcsema/bin/mcsema-lift-10.0" \
>     --arch x86 \
>     --os windows \
>     --cfg "${WORKDIR}/${TARGET}.cfg" \
>     --output "${WORKDIR}/${TARGET}.bc"\
>     --explicit_args \
>     --merge_segments \
>     --name_lifted_sections \
>     -semantics_search_paths "${MCSEMAPATH}/remill/share/remill/10.0/semantics/"
E20210304 14:17:08.631868 19859 Arch.cpp:379] Unable to decode instruction at 4026c0 with bytes ff ff ff ff b0 26 40 00 00 00 00 00 ff ff ff and error: GENERAL_ERROR
E20210304 14:17:08.632352 19859 Arch.cpp:957] DecodeXED() could not decode the following opcodes: (X86 4026c0 (BYTES ff ff ff ff b0 26 40 00 00 00 00 00 ff ff ff) !NO-FUNCTION!)
E20210304 14:17:08.632378 19859 Function.cpp:1644] Could not decode instruction at 4026c0 reachable from instruction 4026bf in function _register_frame_ctor at 4026b0: (X86 4026c0 (BYTES ff ff ff ff b0 26 40 00 00 00 00 00 ff ff ff) !NO-FUNCTION!)
E20210304 14:17:09.014456 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 400000 in block inst_401010 in function sub_401010__pre_c_init
E20210304 14:17:09.014518 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 400000 in block inst_4022ca in function sub_4022b0___FindPESectionByName
E20210304 14:17:09.014537 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 400000 in block inst_402340 in function sub_402340____mingw_GetSectionForAddress
E20210304 14:17:09.014551 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 400000 in block inst_4023b0 in function sub_4023b0____mingw_GetSectionCount
E20210304 14:17:09.014565 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 400000 in block inst_4023e0 in function sub_4023e0___FindPESectionExec
E20210304 14:17:09.014585 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 400000 in block inst_402450 in function sub_402450___GetPEImageBase
E20210304 14:17:09.014602 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 400000 in block inst_402480 in function sub_402480___IsNonwritableInCurrentImage
E20210304 14:17:09.014619 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 400000 in block inst_402500 in function sub_402500____mingw_enum_import_library_names
E20210304 14:17:09.014639 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 40003c in block inst_4010b0 in function sub_401010__pre_c_init
E20210304 14:17:09.014963 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 40003c in block inst_4022ee in function sub_4022b0___FindPESectionByName
E20210304 14:17:09.015137 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 40003c in block inst_402360 in function sub_402340____mingw_GetSectionForAddress
E20210304 14:17:09.015158 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 40003c in block inst_4023ce in function sub_4023b0____mingw_GetSectionCount
E20210304 14:17:09.015174 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 40003c in block inst_40240e in function sub_4023e0___FindPESectionExec
E20210304 14:17:09.015190 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 40003c in block inst_4024a0 in function sub_402480___IsNonwritableInCurrentImage
E20210304 14:17:09.015213 19859 Optimize.cpp:1252] Missed cross-reference to absolute address 40003c in block inst_40252e in function sub_402500____mingw_enum_import_library_names
F20210304 14:17:09.156836 19859 Segment.cpp:484] Check failed: 'seg_type' Must be non NULL
*** Check failure stack trace: ***
    @           0x85c08c  google::LogMessageFatal::~LogMessageFatal()
    @           0x481292  mcsema::DefineDataSegments()
    @           0x471f31  mcsema::LiftCodeIntoModule()
    @           0x487856  main
    @     0x7fd79dcd8bf7  __libc_start_main
    @           0x43897a  _start
[1]    19859 abort (core dumped)  "${MCSEMAPATH}/mcsema/bin/mcsema-lift-10.0" --arch x86 --os windows --cfg

Relevant code:

mcsema/mcsema/BC/Segment.cpp

Lines 479 to 484 in b0f7790

auto seg_type = llvm::dyn_cast<llvm::StructType>(remill::GetValueType(seg));
// This might be null if there are two lifted variables with same name and
// one of them is exported and the exported variable is having xrefs or
// notnull.
CHECK_NOTNULL(seg_type); 

  auto seg_type = llvm::dyn_cast<llvm::StructType>(remill::GetValueType(seg));

  // This might be null if there are two lifted variables with same name and
  // one of them is exported and the exported variable is having xrefs or
  // notnull.
  CHECK_NOTNULL(seg_type);
  seg->setInitializer(FillDataSegment(cfg_module, cfg_seg, seg_type));
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@SaifRushdHadad