Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add new bswap* instructions #554

Merged
merged 2 commits into from
Dec 29, 2023
Merged

Add new bswap* instructions #554

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2d83ab9
Byteswap fixes
dthaler Dec 28, 2023
7618a14
Add bswap instructions
dthaler Dec 28, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 16 additions & 3 deletions src/asm_marshal.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,11 +72,14 @@ static uint8_t imm(Un::Op op) {
switch (op) {
case Op::NEG: return 0;
case Op::BE16:
case Op::LE16: return 16;
case Op::LE16:
case Op::SWAP16: return 16;
case Op::BE32:
case Op::LE32: return 32;
case Op::LE32:
case Op::SWAP32: return 32;
case Op::BE64:
case Op::LE64: return 64;
case Op::LE64:
case Op::SWAP64: return 64;
}
assert(false);
return {};
Expand Down Expand Up @@ -156,6 +159,16 @@ struct MarshalVisitor {
.offset = 0,
.imm = imm(b.op),
}};
case Un::Op::SWAP16:
case Un::Op::SWAP32:
case Un::Op::SWAP64:
return {ebpf_inst{
.opcode = static_cast<uint8_t>(INST_CLS_ALU64 | (0xd << 4)),
.dst = b.dst.v,
.src = 0,
.offset = 0,
.imm = imm(b.op),
}};
}
assert(false);
return {};
Expand Down
3 changes: 3 additions & 0 deletions src/asm_ostream.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -224,6 +224,9 @@ struct InstructionPrinterVisitor {
case Un::Op::LE16: os_ << "le16 "; break;
case Un::Op::LE32: os_ << "le32 "; break;
case Un::Op::LE64: os_ << "le64 "; break;
case Un::Op::SWAP16: os_ << "swap16 "; break;
case Un::Op::SWAP32: os_ << "swap32 "; break;
case Un::Op::SWAP64: os_ << "swap64 "; break;
case Un::Op::NEG: os_ << "-"; break;
}
os_ << b.dst;
Expand Down
5 changes: 4 additions & 1 deletion src/asm_parse.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,7 @@ using crab::linear_expression_t;
#define OPASSIGN R"_(\s*(\S*)=\s*)_"
#define ASSIGN R"_(\s*=\s*)_"
#define LONGLONG R"_(\s*(ll|)\s*)_"
#define UNOP R"_((-|be16|be32|be64))_"
#define UNOP R"_((-|be16|be32|be64|le16|le32|le64|swap16|swap32|swap64))_"

#define PLUSMINUS R"_((\s*[+-])\s*)_"
#define LPAREN R"_(\s*\(\s*)_"
Expand Down Expand Up @@ -69,6 +69,9 @@ static const std::map<std::string, Un::Op> str_to_unop = {
{"le16", Un::Op::LE16},
{"le32", Un::Op::LE32},
{"le64", Un::Op::LE64},
{"swap16", Un::Op::SWAP16},
{"swap32", Un::Op::SWAP32},
{"swap64", Un::Op::SWAP64},
{"-", Un::Op::NEG},
};

Expand Down
3 changes: 3 additions & 0 deletions src/asm_syntax.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,6 +108,9 @@ struct Un {
LE16, // dst = htole16(dst)
LE32, // dst = htole32(dst)
LE64, // dst = htole64(dst)
SWAP16, // dst = bswap16(dst)
SWAP32, // dst = bswap32(dst)
SWAP64, // dst = bswap64(dst)
NEG, // dst = -dst
};

Expand Down
18 changes: 12 additions & 6 deletions src/asm_unmarshal.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,14 +122,20 @@ struct Unmarshaller {
note("arsh32 is not allowed");
return Bin::Op::ARSH;
case INST_ALU_OP_END:
if ((inst.opcode & INST_CLS_MASK) == INST_CLS_ALU64) {
if (inst.opcode & INST_END_BE)
throw InvalidInstruction(pc, "invalid endian immediate");
switch (inst.imm) {
case 16: return Un::Op::SWAP16;
case 32: return Un::Op::SWAP32;
case 64: return Un::Op::SWAP64;
default: throw InvalidInstruction(pc, "invalid endian immediate");
}
}
switch (inst.imm) {
case 16: return (inst.opcode & INST_END_BE) ? Un::Op::BE16 : Un::Op::LE16;
case 32:
if ((inst.opcode & INST_CLS_MASK) == INST_CLS_ALU64)
throw InvalidInstruction(pc, "invalid endian immediate 32 for 64 bit instruction");
return (inst.opcode & INST_END_BE) ? Un::Op::BE32 : Un::Op::LE32;
case 64:
return (inst.opcode & INST_END_BE) ? Un::Op::BE64 : Un::Op::LE64;
case 32: return (inst.opcode & INST_END_BE) ? Un::Op::BE32 : Un::Op::LE32;
case 64: return (inst.opcode & INST_END_BE) ? Un::Op::BE64 : Un::Op::LE64;
default:
throw InvalidInstruction(pc, "invalid endian immediate");
}
Expand Down
12 changes: 12 additions & 0 deletions src/crab/ebpf_domain.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1484,6 +1484,18 @@ void ebpf_domain_t::operator()(const Un& stmt) {
swap_endianness(dst.svalue, int64_t(0), boost::endian::native_to_little<int64_t>);
swap_endianness(dst.svalue, uint64_t(0), boost::endian::native_to_little<uint64_t>);
break;
case Un::Op::SWAP16:
swap_endianness(dst.svalue, uint16_t(0), boost::endian::endian_reverse<uint16_t>);
swap_endianness(dst.uvalue, uint16_t(0), boost::endian::endian_reverse<uint16_t>);
break;
case Un::Op::SWAP32:
swap_endianness(dst.svalue, uint32_t(0), boost::endian::endian_reverse<uint32_t>);
swap_endianness(dst.uvalue, uint32_t(0), boost::endian::endian_reverse<uint32_t>);
break;
case Un::Op::SWAP64:
swap_endianness(dst.svalue, int64_t(0), boost::endian::endian_reverse<int64_t>);
swap_endianness(dst.uvalue, uint64_t(0), boost::endian::endian_reverse<uint64_t>);
break;
case Un::Op::NEG:
neg(dst.svalue, dst.uvalue, stmt.is64 ? 64 : 32);
havoc_offsets(stmt.dst);
Expand Down
6 changes: 5 additions & 1 deletion src/test/test_marshal.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,7 +47,9 @@ static void check_unmarshal_fail(ebpf_inst inst, std::string expected_error_mess
program_info info{.platform = &g_ebpf_platform_linux,
.type = g_ebpf_platform_linux.get_program_type("unspec", "unspec")};
std::vector<ebpf_inst> insns = {inst};
std::string error_message = std::get<std::string>(unmarshal(raw_program{"", "", insns, info}));
auto result = unmarshal(raw_program{"", "", insns, info});
REQUIRE(std::holds_alternative<std::string>(result));
std::string error_message = std::get<std::string>(result);
REQUIRE(error_message == expected_error_message);
}

Expand Down Expand Up @@ -270,4 +272,6 @@ TEST_CASE("fail unmarshal", "[disasm][marshal]") {
check_unmarshal_fail(ebpf_inst{.opcode = INST_CLS_JMP32}, "0: jump out of bounds\n");
check_unmarshal_fail(ebpf_inst{.opcode = 0x90 | INST_CLS_JMP32}, "0: Bad instruction\n");
check_unmarshal_fail(ebpf_inst{.opcode = 0x10 | INST_CLS_JMP32}, "0: jump out of bounds\n");
check_unmarshal_fail(ebpf_inst{.opcode = INST_ALU_OP_END | INST_CLS_ALU, .imm = 0}, "0: invalid endian immediate\n");
check_unmarshal_fail(ebpf_inst{.opcode = INST_ALU_OP_END | INST_CLS_ALU64, .imm = 0}, "0: invalid endian immediate\n");
}
120 changes: 120 additions & 0 deletions test-data/unop.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,3 +40,123 @@ code:
r1 = -r1

post: ["r1.type=number", "r1.svalue=[-5, 5]"]
---
test-case: be16 singleton

pre: ["r1.type=number", "r1.svalue=6636321", "r1.uvalue=6636321"]

code:
<start>: |
r1 = be16 r1 ; 0x654321 -> 0x2143

post: ["r1.type=number", "r1.svalue=8515", "r1.uvalue=8515"]
---
test-case: be32 singleton

pre: ["r1.type=number", "r1.svalue=40926266145", "r1.uvalue=40926266145"]

code:
<start>: |
r1 = be32 r1 ; 0x0987654321 -> 0x21436587

post: ["r1.type=number", "r1.svalue=558065031", "r1.uvalue=558065031"]
---
test-case: be64 singleton

pre: ["r1.type=number", "r1.svalue=40926266145", "r1.uvalue=40926266145"]

code:
<start>: |
r1 = be64 r1 ; 0x0987654321 -> 0x2143658709000000

post: ["r1.type=number", "r1.svalue=2396871057337221120", "r1.uvalue=2396871057337221120"]
---
test-case: be16 range

pre: ["r1.type=number", "r1.svalue=[0, 2]", "r1.uvalue=[0, 2]", "r1.svalue=r1.uvalue"]

code:
<start>: |
r1 = be16 r1 ; [0x0000, 0x0002] -> [0x0000, 0x2000] but currently we just lose the range

post: ["r1.type=number"]
---
test-case: le16 singleton

pre: ["r1.type=number", "r1.svalue=6636321", "r1.uvalue=6636321"]

code:
<start>: |
r1 = le16 r1 ; 0x654321 -> 0x4321

post: ["r1.type=number", "r1.svalue=17185", "r1.uvalue=17185"]
---
test-case: le32 singleton

pre: ["r1.type=number", "r1.svalue=40926266145", "r1.uvalue=40926266145"]

code:
<start>: |
r1 = le32 r1 ; 0x0987654321 -> 0x87654321

post: ["r1.type=number", "r1.svalue=2271560481", "r1.uvalue=2271560481"]
---
test-case: le64 singleton

pre: ["r1.type=number", "r1.svalue=40926266145", "r1.uvalue=40926266145"]

code:
<start>: |
r1 = le64 r1 ; 0x0987654321 -> 0x2143658709000000

post: ["r1.type=number", "r1.svalue=40926266145", "r1.uvalue=40926266145"]
---
test-case: le16 range

pre: ["r1.type=number", "r1.svalue=[0, 2]", "r1.uvalue=[0, 2]", "r1.svalue=r1.uvalue"]

code:
<start>: |
r1 = le16 r1 ; this could preserve the range but we don't support that yet

post: ["r1.type=number"]
---
test-case: swap16 singleton

pre: ["r1.type=number", "r1.svalue=6636321", "r1.uvalue=6636321"]

code:
<start>: |
r1 = swap16 r1 ; 0x654321 -> 0x2143

post: ["r1.type=number", "r1.svalue=8515", "r1.uvalue=8515"]
---
test-case: swap32 singleton

pre: ["r1.type=number", "r1.svalue=40926266145", "r1.uvalue=40926266145"]

code:
<start>: |
r1 = swap32 r1 ; 0x0987654321 -> 0x21436587

post: ["r1.type=number", "r1.svalue=558065031", "r1.uvalue=558065031"]
---
test-case: swap64 singleton

pre: ["r1.type=number", "r1.svalue=40926266145", "r1.uvalue=40926266145"]

code:
<start>: |
r1 = swap64 r1 ; 0x0987654321 -> 0x2143658709000000

post: ["r1.type=number", "r1.svalue=2396871057337221120", "r1.uvalue=2396871057337221120"]
---
test-case: swap16 range

pre: ["r1.type=number", "r1.svalue=[0, 2]", "r1.uvalue=[0, 2]", "r1.svalue=r1.uvalue"]

code:
<start>: |
r1 = swap16 r1 ; [0x0000, 0x0002] -> [0x0000, 0x2000] but currently we just lose the range

post: ["r1.type=number"]