6 ms·
Why Is My Perfectly Good Shellcode Not Working? Cache Coherency on MIPS and ARM
- giancarlostoro 8y agoI had issues with my own shell script this past week or two. But my issue was colliding hypervisors. Fun times. I'm not savvy enough with Bash as it is, it was frustrating. Edit: I guess my shell code breaking frustration is somehow irrelevant to this issue. I mean running into the issue in the article is a rarity these days, at least on standard / mainstream hardware.
- matthewmacleod 8y agoDifferent kind of “shell” code maybe? :)
- saagarjha 8y agoFor multiple reasons: most people don’t use MIPS, don’t have to write shellcode, and don’t have to deal with cache coherency :)
- taneq 8y agoNote: This is talking about not talking about exploiting buffer overruns, not about BASH scripts. I was very confused for a moment there.
- rocqua 8y agoshellcode vs shell-code. Shellcode traditionally meant binary that would do something like exec("/bin/bash") (i.e. code that launches a shell). When abusing a buffer overflow, you would usually want to somehow get this code to be executed. Later, the term expanded to mean any kind of user-provided bytes an attacker wants to be interperted and executed as machine instructions.
- porpoisely 8y agoFYI: It's called shellcode because the load in the exploit/buffer overrun are actual machine instructions to start a command shell. So essentially it is code to open a shell ( shellcode ). If you want to hack a system, the ideal program you want is a command shell. And if you exploit a program with elevated rights, then you could get a shell with elevated rights which would let you ultimately take control over the system. Edit: And if anyone wants to learn about computers/programming, I suggest you try it yourself. It's an excellent way of learning how the system works at a low level.
- TickleSteve 8y ago"Shellcode" in this sense is the small number of instructions that initiates the exploit, i.e. its contained in the data you overflow the buffer with. The intention is that you tricking the firmware into executing your shellcode to ramp up the exploit. The shellcode is "executable data" and the typical prevention for this type of exploit is execution-prevention on data (and stack) regions (for processors that support it).
- signa11 8y agobut there is always return-to-libc style attacks which defeat executable stack protection...
- saagarjha 8y agoAnd hence we have ASLR to prevent against those attacks…
- DyslexicAtheist 8y agoonly for 64-bit, but disclaimers might apply in (32-bit) IoT land: > Our results suggest that, for current 32-bit architectures, address-space randomization is ineffective against the possibility of generic exploit code for a single flaw; and brute force attacks can be efficient, and hence effective [1] https://web.stanford.edu/~blp/papers/asrandom.pdf https://web.stanford.edu/~blp/papers/asrandom.pdf edit: also ROP
- saagarjha 8y ago32-bit IoT usually has a lot more problems that are easier to exploit than being able to brute force ASLR, but yes: if you are able to run attacks quickly, it’s pretty easy to get lucky and find the offset. There really isn’t much you can do on 32 bit to fix this, unfortunately, unless you start doing fancy things like reordering functions themselves to increase the number of bits of entropy (and even then, you’re not getting much…)
- pm215 8y agoThe post seems to be slightly confused about the Arm DSB and ISB instructions. These don't affect the caches at all -- they are barriers, which affect ordering and pipelining. DSB says "don't go any further until all memory accesses made by instructions before this one have completed", and ISB says "flush the pipeline so the next insn is reread from the cache" (among other things). So self-modifying and other tricky code needs to consider these insn ordering issues as well as cache manipulation, and barrier insns aren't a substitute for doing the cache maintenance.
- ajross 8y agoPipeline control certainly does effect caches. That's where the memory access being controlled goes. I mean, it's tempting to try to understand them as an orthogonal thing, but in fact memory hierarchies and CPU pipelines (and other stuff like store forward buffers) are tied at the hip and it's pointless to try to "understand" them in isolation. At best it's an exercise in specification pedantry. The only reason instruction barriers, fences and serialization instructions seem to make sense in docs is that someone at the architecture sat down and wrote a memory and execution model that can be abstracted by whatever API it is they present to the user. And someone doing performance or exploitation work at the level of the linked article really needs to understand that model, not the API. (Though you're absolutely right that this particular article seems to be a bit mixed up about both)
- pm215 8y agoI meant that it doesn't affect it in the sense that it doesn't change the way the cache behaves or ask it to flush or invalidate any of its contents. You can have a pipelined CPU which needs the barrier insns but which has no cache at all (most likely in embedded). And if the problem you're having is that your data is in the dcache but not the icache then the barrier insns won't help, because they're not architecturally defined to do that. I certainly agree that you need to understand the architectural model to write reliable (and cross-cpu portable code) -- and the architectural model for Arm does distinguish barriers from cache maintenance -- though if you're writing exploit shellcode then "works most of the time and we can't do better given the way the bug we're exploiting to get code execution" is sometimes what you're going to end up with, at which point knowing what the specific hardware is doing can help.
- classichasclass 8y agoThe same principles probably apply to Power ISA (isync, hwsync, icbi, etc.), and for that matter SPARC. Lots of PowerPC chips still turn up in embedded applications.
- vectorEQ 8y agoit's so perfect that it doesn't work. great stuff. instantly lost interest.