Is there a more or less reliable way to tell whether data at some location in memory is a beginning of a processor instruction or some other data?
For example, E8 3F BD 6A 00 may be call instruction (E8) with relative offset of 0x6ABD3F, or it might be three bytes of data belonging to some other instruction, followed by push 0 (6A 00).
I know the question sounds silly and there is probably no simple way, but maybe instruction set was designed with this problem in mind and maybe some simple code examining +-100 bytes around the location can give an answer that is very likely correct.
I want to know this because I scan program's code and replace all calls to some function with calls to my replacement. It's working this far but it's not impossible that at some point, as I increase number of functions I'm replacing, some data will look exactly like a function call to that exact address, and will be replaced, and this will cause a program to break in a most unexpected fashion. I want to reduce the probability of that.
If it is your code (or another one which retaining linking and debug info), the best way is to scan symbol/relocation tables in object file. Otherwise there's no reliable way to determine if some byte is inctruction or data.
Possibly the most efficient method to qualify data is recursive disassembling. I. e. disassembling code from enty point and from all jump destinations found. But this is not completely reliable, because it does not traverse jump tables (you can try to use some heuristics for this, but this is not completely reliable too).
Solution for your problem would be patch function being replaced itself: overwrite its beginning with jump inctruction to your function.
Unfortunately, there is no 100% reliable way to distinguish code from data. From the CPU point of view, code is code only when some jump opcode induces the processor into trying to execute the bytes as if they were code. You could try to make a control flow analysis by beginning with the program entry point, and following all possible execution paths, but this may fail in the presence of pointers to function.
For your specific problem: I gather that you want to replace an existing function with a replacement of your own. I suggest that you patch the replaced function itself. I.e., instead of locating all calls to the foo() function and replacing them with a call to bar(), just replace the first bytes of foo() with a jump to bar() (a jmp, not a call: you do not want to mess with the stack). This is less satisfactory because of the double jump, but it is reliable.
It is impossible to distinguish data from instruction in general and this is because of von Neumann architecture . Analyzing the code around is helpful and disassembly tools do this. (This may be helpful. If you can't use IDA Pro /it is commercial/, use another disassembly tool.)
Plain code have a very specific entropy, so it's quite easy to distinglish it from most data. However, it's a probabilistic approach, but a large enough buffer of plain code can be recognized (especially compiler output, when you can also recognize patterns, like beginning of a function).
Also, some opcodes are reserved for future, others are available only from kernel mode. In this case by knowing them and knowing how to compute the instruction lengths (you could try a routine written by Z0mbie for that), you can do it.
Thomas suggests the right idea. To implement it properly, you need to disassemble the first few instructions (the part you would overwrite with the JMP) and generate a simple trampoline function that executes them then jumps to the rest of the original function.
There's libraries that do this for you. A well-known one is Detours but it has somewhat awkward licensing conditions. A nice implementation of the same idea with a more permissive license is Mhook.
Related
Context
The function BN_consttime_swap in OpenSSL is a thing of beauty. In this snippet, condition has been computed as 0 or (BN_ULONG)-1:
#define BN_CONSTTIME_SWAP(ind) \
do { \
t = (a->d[ind] ^ b->d[ind]) & condition; \
a->d[ind] ^= t; \
b->d[ind] ^= t; \
} while (0)
…
BN_CONSTTIME_SWAP(9);
…
BN_CONSTTIME_SWAP(8);
…
BN_CONSTTIME_SWAP(7);
The intention is that so as to ensure that higher-level bignum operations take constant time, this function either swaps two bignums or leaves them in place in constant time. When it leaves them in place, it actually reads each word of each bignum, computes a new word that is identical to the old word, and write that result back to the original location.
The intention is that this will take the same time as if the bignums had effectively been swapped.
In this question, I assume a modern, widespread architecture such as those described by Agner Fog in his optimization manuals. Straightforward translation of the C code to assembly (without the C compiler undoing the efforts of the programmer) is also assumed.
Question
I am trying to understand whether the construct above characterizes as a “best effort” sort of constant-time execution, or as perfect constant-time execution.
In particular, I am concerned about the scenario where bignum a is already in the L1 data cache when the function BN_consttime_swap is called, and the code just after the function returns start working on the bignum a right away. On a modern processor, enough instructions can be in-flight at the same time for the copy not to be technically finished when the bignum a is used. The mechanism allowing the instructions after the call to BN_consttime_swap to work on a is memory dependence speculation. Let us assume naive memory dependence speculation for the sake of the argument.
What the question seems to boil down to is this:
When the processor finally detects that the code after BN_consttime_swap read from memory that had, contrary to speculation, been written to inside the function, does it cancel the speculative execution as soon as it detects that the address had been written to, or does it allow itself to keep it when it detects that the value that has been written is the same as the value that was already there?
In the first case, BN_consttime_swap looks like it implements perfect constant-time. In the second case, it is only best-effort constant-time: if the bignums were not swapped, execution of the code that comes after the call to BN_consttime_swap will be measurably faster than if they had been swapped.
Even in the second case, this is something that looks like it could be fixed for the foreseeable future (as long as processors remain naive enough) by, for each word of each of the two bignums, writing a value different from the two possible final values before writing either the old value again or the new value. The volatile type qualifier may need to be involved at some point to prevent an ordinary compiler to over-optimize the sequence, but it still sounds possible.
NOTE: I know about store forwarding, but store forwarding is only a shortcut. It does not prevent a read being executed before the write it is supposed to come after. And in some circumstances it fails, although one would not expect it to in this case.
Straightforward translation of the C code to assembly (without the C compiler undoing the efforts of the programmer) is also assumed.
I know it's not the thrust of your question, and I know that you know this, but I need to rant for a minute. This does not even qualify as a "best effort" attempt to provide constant-time execution. A compiler is licensed to check the value of condition, and skip the whole thing if condition is zero. Obfuscating the setting of condition makes this less likely to happen, but is no guarantee.
Purportedly "constant-time" code should not be written in C, full stop. Even if it is constant time today, on the compilers that you test, a smarter compiler will come along and defeat you. One of your users will use this compiler before you do, and they will not be aware of the risk to which you have exposed them. There are exactly three ways to achieve constant time that I am aware of: dedicated hardware, assembly, or a DSL that generates machine code plus a proof of constant-time execution.
Rant aside, on to the actual architecture question at hand: assuming a stupidly naive compiler, this code is constant time on the µarches with which I am familiar enough to evaluate the question, and I expect it to broadly be true for one simple reason: power. I expect that checking in a store queue or cache if a value being stored matches the value already present and conditionally short-circuiting the store or avoiding dirtying the cache line on every store consumes more energy than would be saved in the rare occasion that you get to avoid some work. However, I am not a CPU designer, and do not presume to speak on their behalf, so take this with several tablespoons of salt, and please consult one before assuming this to be true.
This blog post, and the comments made by the author, Henry, on the subject of this question should be considered as authoritative as anyone should allowed to expect. I will reproduce the latter here for archival:
I didn’t think the case of overwriting a memory location with the same value had a practical use. I think the answer is that in current processors, the value of the store is irrelevant, only the address is important.
Out here in academia, I’ve heard of two approaches to doing memory disambiguation: Address-based, or value-based. As far as I know, current processors all do address-based disambiguation.
I think the current microbenchmark has some evidence that the value isn’t relevant. Many of the cases involve repeatedly storing the same value into the same location (particularly those with offset = 0). These were not abnormally fast.
Address-based schemes uses a store queue and a load queue to track outstanding memory operations. Loads check the store queue to for an address match (Should this load do store-to-load forwarding instead of reading from cache?), while stores check the load queue (Did this store clobber the location of a later load I allowed to execute early?). These checks are based entirely on addresses (where a store and load collided). One advantage of this scheme is that it’s a fairly straightforward extension on top of store-to-load forwarding, since the store queue search is also used there.
Value-based schemes get rid of the associative search (i.e., faster, lower power, etc.), but requires a better predictor to do store-to-load forwarding (Now you have to guess whether and where to forward, rather than searching the SQ). These schemes check for ordering violations (and incorrect forwarding) by re-executing loads at commit time and checking whether their values are correct. In these schemes, if you have a conflicting store (or made some other mistake) that still resulted in the correct result value, it would not be detected as an ordering violation.
Could future processors move to value-based schemes? I suspect they might. They were proposed in the mid-2000s(?) to reduce the complexity of the memory execution hardware.
The idea behind constant-time implementation is not to actually perform everything in constant time. That will never happen on an out-of-order architecture.
The requirement is that no secret information can be revealed by timing analysis.
To prevent this there are basically two requirements:
a) Do not use anything secret as a stop condition for a loop, or as a predicate to a branch. Failing to do so will open you to a branch prediction attack https://eprint.iacr.org/2006/351.pdf
b) Do not use anything secret as an index to memory access. This leads to cache timing attacks http://www.daemonology.net/papers/htt.pdf
As for your code: assuming that your secret is "condition" and possibly the contents of a and b the code is perfectly constant time in the sense that its execution does not depend on the actual contents of a, b and condition. Of course the locality of a and b in memory will affect the execution time of the loop, but not the CONTENTS which are secret.
That is assuming of course condition was computed in a constant time manner.
As for C optimizations: the compiler can only optimize code based on information it knows. If "condition" is truly secret the compiler should not be able to discern it contents and optimize. If it can be deducted from your code then the compiler will most likely make optimization for the 0 case.
So, I've been trying to write an emulator, or at least understand how stuff works. I have a decent grasp of assembly, particularly z80 and x86, but I've never really understood how an object file (or in my case, a .gb ROM file) indicates the start and end of an instruction.
I'm trying to parse out the opcode for each instruction, but it occurred to me that it's not like there's a line break after every instruction. So how does this happen? To me, it just looks like a bunch of bytes, with no way to tell the difference between an opcode and its operands.
For most CPUs - and I believe Z80 falls in this category - the length of an instruction is implicit.
That is, you must decode the instruction in order to figure out how long it is.
If you're writing an emulator you don't really ever need to be able to obtain a full disassembly. You know what the program counter is now, you know whether you're expecting a fresh opcode, an address, a CB page opcode or whatever and you just deal with it. What people end up writing, in effect, is usually a per-opcode recursive descent parser.
To get to a full disassembler, most people impute some mild simulation, recursively tracking flow. Instructions are found, data is then left by deduction.
Not so much on the GB where storage was plentiful (by comparison) and piracy had a physical barrier, but on other platforms it was reasonably common to save space or to effect disassembly-proof code by writing code where a branch into the middle of an opcode would create a multiplexed second stream of operations, or where the same thing might be achieved by suddenly reusing valid data as valid code. One of Orlando's 6502 efforts even re-used some of the loader text — regular ASCII — as decrypting code. That sort of stuff is very hard to crack because there's no simple assembly for it and a disassembler therefore usually won't be able to figure out what to do heuristically. Conversely, on a suitably accurate emulator such code should just work exactly as it did originally.
So the question is:
How to optimize function entry & exit code in a portable way for speed using GCC, plain C?
I am interested in relevant options etc. My goal is writing a CPU emulator where the instruction set is decoded using call tables. I already eliminated any function call I could reasonably eliminate, but due to the structure of the instruction set, doing 2-3 such calls per emulated instruction is necessary (so I can neither eliminate any more branch mispredictions here, either).
Based on analysing the assembly (x86, 32bits) output the option -fomit-frame-pointer seems worthwhile (once I don't care for the lost debug-ability here). Otherwise in general if I look over the complete emulator it seems like it could be better with better overall register and stack management (don't saving every single thing on every entry), my impression of the generated assembly is that it tampers more with the stack than how much useful work it does.
So the situation is basically that there is a whole lot of little functions which are called many-many times, and which can not be eliminated from the code.
I don't want to switch over from the interpreting emulation since this should be the most portable approach to do this thing (more portable anyway than any solution which would recompile).
On x86-32, the ABIs for common operating systems have standard calling conventions that use the stack to pass arguments to functions, because there are few general-purpose registers. One way to improve function calls that take only a few arguments (and relatively simple arguments) would be to use a different calling convention (like fastcall) that use registers to pass the arguments. If moving to x86-64 is an option (and it should be, it's been around for ages...), the ABIs are much better for fast function calls, because the number of general-purpose registers doubled.
This question already has answers here:
Building an assembler
(4 answers)
How Do You Make An Assembler? [closed]
(4 answers)
Closed 9 years ago.
I've recently been trying to immerse myself in the world of assembly programming with the eventual goal of creating my own programming language. I want my first real project to be a simple assembler written in C that will be able to assemble a very small portion of the x86 machine language and create a Windows executable. No macros, no linkers. Just assembly.
On paper, it seems simple enough. Assembly code comes in, machine code comes out.
But as soon as I thinking about all the details, it suddenly becomes very daunting. What conventions does the operating system demand? How do I align data and calculate jumps? What does the inside of an executable even look like?
I'm feeling lost. There aren't any tutorials on this that I could find and looking at the source code of popular assemblers was not inspiring (I'm willing to try again, though).
Where do I go from here? How would you have done it? Are there any good tutorials or literature on this topic?
I have written a few myself (assemblers and disassemblers) and I would not start with x86. If you know x86 or any other instruction set you can pick up and learn the syntax for another instruction set in short order (an evening/afternoon), at least the lions share of it. The act of writing an assembler (or disassembler) will definitely teach you an instruction set, fast, and you will know that instruction set better than many seasoned assembly programmers for that instruction set who have not examined the microcode at that level. msp430, pdp11, and thumb (not thumb2 extensions) (or mips or openrisc) are all good places to start, not a lot of instructions, not overly complicated, etc.
I recommend a disassembler first, and with that a fixed length instruction set like arm or thumb or mips or openrisc, etc. If not then at least use a disassembler (definitely choose an instruction set for which you already have an assembler, linker, and disassembler) and with pencil and paper understand the relationship between the machine code and the assembly, in particular the branches, they usually have one or more quirks like the program counter is an instruction or two ahead when the offset is added, to gain another bit they sometimes measure in whole instructions not bytes.
It is pretty easy to brute force parse the text with a C program to read the instructions. A harder task but perhaps as educational, would be to use bison/flex and learn that programming language to allow those tools to create (an even more extreme brute force) parser which then interfaces to your code to tell you what was found where.
The assembler itself is pretty straight forward, just read the ascii and set the bits in the machine code. Branches and other pc relative instructions are a little more painful as they can take multiple passes through the source/tables to completely resolve.
mov r0,r1
mov r2 ,#1
the assembler begins parsing the text for a line (being defined as the bytes that follow a carriage return 0xD or line feed 0xA), discard the white space (spaces and tabs) until you get to something non white space, then strncmp that with the known mnemonics. if you hit one then parse the possible combinations of that instruction, in the simple case above after the mov skip over the white space to non-white space, perhaps the first thing you find must be a register, then optional white space, then a comma. remove the whitespace and comma and compare that against a table of strings or just parse through it. Once that register is done then go past where the comma is found and lets say it is either another register or an immediate. If immediate lets say it has to have a # sign, if register lets say it has to start with a lower or upper case 'r'. after parsing that register or immediate, then make sure there is nothing else on the line that shouldnt be on the line. build the machine code for this instruciton or at least as much as you can, and move on to the next line. It may be tedious but it is not difficult to parse ascii...
at a minimum you will want a table/array that accumulates the machine code/data as it is created, plus some method for marking instructions as being incomplete, the pc-relative instructions to be completed on a future pass. you will also want a table/array that collects the labels you find and the address/offset in the machine code table where found. As well as the labels used in the instruction as a destination/source and the offset in the table/array holding the partially complete instruction they go with. after the first pass, then go back through these tables until you have matched up all the label definitions with the labels used as a source or destination, using the label definition address/offset to compute the distance to the instruction in question and then finish creating the machine code for that instruction. (some disassembly may be required and/or use some other method for remembering what kind of encoding it was when you come back to it later to finish building the machine code).
The next step is allowing for multiple source files, if that is something you want to allow. Now you have to have labels that dont get resolved by the assembler so you have to leave placeholders in the output and make some flavor of the longest jump/branch instruction because you dont know how far away the destination will be, expect the worse. Then there is the output file format you choose to create/use, then there is the linker which is mostly simple, but you have to remember to fill in the machine code for the final pc relative instructions, no harder than it was in the assembler itself.
Note, writing an assembler is not necessarily related to creating a programming language and then writing a compiler for it, separate thing, different problems. Actually if you want to make a new programming language just use an existing assembler for an existing instruction set. Not required of course, but most teachings and tutorials are going to use the bison/flex approach for programming languages, and there are many college course lecture notes/resources out there for beginning compiler classes that you can just use to get you started then modify the script to add the features of your language. The middle and back ends are the bigger challenge than the front end. there are many books on this topic and many online resources as well. As mentioned in another answer llvm is not a bad place to create a new programming language the middle and backends are done for you, you only need to focus on the programming language itself, the front end.
You should look at LLVM, llvm is a modular compiler back end, the most popular front end is Clang for compiling C/C++/Objective-C. The good thing about LLVM is that you can pick the part of the compiler chain that you are interested in and just focus on that, ignoring all of the others. You want to create your own language, write a parser that generates the LLVM internal representation code, and for free you get all of the middle layer target independent optimisations and compiling to many different targets. Interesting in a compiler for some exotic CPU, write a compiler backend that takes the LLVM intermediated code and generates your assemble. Have some ideas about optimisation technics, automatic threading perhaps, write a middle layer which processes LLVM intermediate code. LLVM is a collection of libraries not a standalone binary like GCC, and so it is very easy to use in you own projects.
What you're looking for is not a tutorial or source code, it's a specification. See http://msdn.microsoft.com/en-us/library/windows/hardware/gg463119.aspx
Once you understand the specification of an executable, write a program to generate one. The executable you build should be as simple as possible. Once you have mastered that, then you can write a simple line-oriented parser that reads instruction names and numeric arguments to generate a block of code to plug into the exe. Later you can add symbols, branches, sections, whatever you want, and that's where something like http://www.davidsalomon.name/assem.advertis/asl.pdf will come in.
P.S. Carl Norum has a good point in the comment above. If your goal is create your own programming language, learning to write an assembler is irrelevant and is very much not the right way to start (unless the language you want to create is an assembly language). There are already assemblers that produce executables from assembler source, so your compiler could produce assembler source and you could avoid the work of recreating the assembler ... and you should. Or you could use something like LLVM, which will solve many other daunting problems of compiler construction. The odds are very small that you will ever actually produce your own programming language, but they're much smaller if you start from scratch and there's no need to. Decide what your goal is and use the best tools available to achieve it.
I need to make a hash table that can eventually be used to write a full assembler.
Basically I will have something like:
foo 100,
and I will need to hash foo and then store the 100 (the address of the command). I was thinking I should just use a 2d array. The second dimension of the array would only be accessed when recording the address (just an int) or when returning the address. There would be no searching done in the second dimension.
If I implement the hash table this way, would it be inefficient? If it is very inefficient, what would be a better way to implement the table?
Edit: I haven't written any code yet. In fact I don't even know what language I'm going to use yet. I want to write it in C so it will be more of a challenge, but I might write it in Java if I feel pressured for time.
If you have every other int in the array unused then in addition to memory waste you're going to use the cache poorly as the cache lines will be underused.
But normally I wouldn't worry about such things when writing an assembler as it's not something very performance demanding as say graphics or heavy computations. At least, I wouldn't rush into optimizing too early.
It is, however, important to keep in mind that once you start assembling large pieces of code (~100,000 lines of assembly) generated automatically (say, from C/C++ code by a compiler), performance will become more and more important as the user experience (wait times) degrades. At that point there will be many candidates for optimization: I/O, parsing, symbol look up, generation of as short as possible jump instructions if they can have multiple encodings for shorter and longer jumps. Expressions and macros will contribute too. You may even consider minimizing white space and comments in the input assembly code in the first place.
Without being able to see any code, there is no reason that this would have to be inefficient. The only reason that it could be is if you pre allocated a bunch of memory that you did not end up using, however without seeing your algorithm you had in mind it is impossible to tell.