ELF executables have a fixed load address (0x804800 on 32-bit x86 Linux binaries, and 0x40000 on 64-bit x86_64 binaries).
I read the SO answers (e.g., this one) about the historical reasons for those specific addresses. What I still don't understand is why to use a fixed load address and not a randomized one (given some range to be randomized within)?
why to use a fixed load address and not a randomized one
Traditionally that's how executables worked. If you want a randomized load address, build a PIE binary (which is really a special case of shared library that has startup code in it) with -fPIE and link with -pie flags.
Building with -fPIE introduces runtime overhead, in some cases as bad as 10% performance degradation, which may not be tolerable if you have a large cluster or you need every last bit of performance.
not sure if I understood your question correct, but saying I did, that's sort-off a "legacy" / historical issue, ELF is the file format used by UNIX derived operating systems, both POSIX (IOS) and Unix-like (Linux).
and the elf format simply states that there must be some resolved and absolute virtual address that the code is loaded into and begins running from...
and simply that's how the file format is, and during to historical reasons that cant be changed... you couldn't just "throw" the executable in any memory address and have it run successfully, back in the 90's when the ELF format was introduced problems such as calling functions with virtual tables we're raised and it was decided that the elf format would have absolute addresses within it.
Also think about it, take a look at the elf format -https://en.wikipedia.org/wiki/Executable_and_Linkable_Format
how would you design an OS executable-loader that would be able to handle an executable load it to ANY desired virtual address and have the code run successfully without actually having to change the binary itself... if you would like to do something like that you'd either need to vastly change the output compilers generate or the format itself, which again isn't possible
As time passed the requirement of position independent executing (PIE/PIC) has raised and shared objects we're introduced in order to allow that and ASLR
(Address Space Layout Randomization) - which means that the code could be thrown in any memory address and still be able to execute, that is simply implemented by making sure that all calls within the code itself are relative to the current address of the executed instruction, AND that when the shared object is loaded the OS loader would have to change some data within the binary given that the data changed is not executable instructions (R E) but actual data (RW, e.g .data segment), which also is implemented by calling functions from some "Jump tables" ( which would be changed at load time ) for example PLT / GOT.... those shared objects allow absolute randomization of the addresses the code is loaded to and if you want to execute some more "secure" code you'd have to compile it as a shared object and and dynamically link it and load time or run time..
( hope I've cleared some things out :) )
Related
For global offset table to work, GOT must be at a fixed location from text segment. Now assume that a program needs a shared library. Assume also that the shared library is already loaded by the OS for some other process. Now for our program, since text section of shared library is already loaded, it just needs to load data segment. The shared library text section is mapped back to the virtual address of our process. But what if there is already some data/text or whatever at the fixed offset from the virtual address of our shared library. How does the dynamic linker resolve that conflict? One approach would be to leave R_386_GOTPC in the text section till load time and let the dynamic linker change it the new offset. Is this how it is done in practice.
On GNU, even the same DSO is mapped at different addresses in different processes. No data at all is shared between them. This means that the GOT is just private data (like .data), and is initialized at load time with the proper addresses (either stubs or the proper function addresses with BIND_NOW).
(This assumes that prelink is not in use, which is somewhat broken anyway.)
I've read this tutorial
I could follow the guide and run the code. but I have questions.
1) Why do we need both load-address and run-time address. As I understand it is because we have put .data at flash too; so why we don't run app there, but need start-up code to copy it into RAM?
http://www.bravegnu.org/gnu-eprog/c-startup.html
2) Why we need linker script and start-up code here. Can I not just build C source as below and run it with qemu?
arm-none-eabi-gcc -nostdlib -o sum_array.elf sum_array.c
Many thanks
Your first question was answered in the guide.
When you load a program on an operating system your .data section, basically non-zero globals, are loaded from the "binary" into the right offset in memory for you, so that when your program starts those memory locations that represent your variables have those values.
unsigned int x=5;
unsigned int y;
As a C programmer you write the above code and you expect x to be 5 when you first start using it yes? Well, if are booting from flash, bare metal, you dont have an operating system to copy that value into ram for you, somebody has to do it. Further all of the .data stuff has to be in flash, that number 5 has to be somewhere in flash so that it can be copied to ram. So you need a flash address for it and a ram address for it. Two addresses for the same thing.
And that begins to answer your second question, for every line of C code you write you assume things like for example that any function can call any other function. You would like to be able to call functions yes? And you would like to be able to have local variables, and you would like the variable x above to be 5 and you might assume that y will be zero, although, thankfully, compilers are starting to warn about that. The startup code at a minimum for generic C sets up the stack pointer, which allows you to call other functions and have local variables and have functions more than one or two lines of code long, it zeros the .bss so that the y variable above is zero and it copies the value 5 over to ram so that x is ready to go when the code your entry point C function is run.
If you dont have an operating system then you have to have code to do this, and yes, there are many many many sandboxes and toolchains that are setup for various platforms that already have the startup and linker script so that you can just
gcc -O myprog.elf myprog.c
Now that doesnt mean you can make system calls without a...system...printf, fopen, etc. But if you download one of these toolchains it does mean that you dont actually have to write the linker script nor the bootstrap.
But it is still valuable information, note that the startup code and linker script are required for operating system based programs too, it is just that native compilers for your operating system assume you are going to mostly write programs for that operating system, and as a result they provide a linker script and startup code in that toolchain.
1) The .data section contains variables. Variables are, well, variable -- they change at run time. The variables need to be in RAM so that they can be easily changed at run time. Flash, unlike RAM, is not easily changed at run time. The flash contains the initial values of the variables in the .data section. The startup code copies the .data section from flash to RAM to initialize the run-time variables in RAM.
2) Linker-script: The object code created by your compiler has not been located into the microcontroller's memory map. This is the job of the linker and that is why you need a linker script. The linker script is input to the linker and provides some instructions on the location and extent of the system's memory.
Startup code: Your C program that begins at main does not run in a vacuum but makes some assumptions about the environment. For example, it assumes that the initialized variables are already initialized before main executes. The startup code is necessary to put in place all the things that are assumed to be in place when main executes (i.e., the "run-time environment"). The stack pointer is another example of something that gets initialized in the startup code, before main executes. And if you are using C++ then the constructors of static objects are called from the startup code, before main executes.
1) Why do we need both load-address and run-time address.
While it is in most cases possible to run code from memory mapped ROM, often code will execute faster from RAM. In some cases also there may be a much larger RAM that ROM and application code may compressed in ROM, so the executable code may not simply be copied from ROM also decompressed - allowing a much larger application than the available ROM.
In situations where the code is stored on non-memory mapped mass-storage media such as NAND flash, it cannot be executed directly in any case and must be loaded into RAM by some sort of bootloader.
2) Why we need linker script and start-up code here. Can I not just build C source as below and run it with qemu?
The linker script defines the memory layout of you target and application. Since this tutorial is for bare-metal programming, there is no OS to handle that for you. Similarly the start-up code is required to at least set an initial stack-pointer, initialise static data, and jump to main. On an embedded system it is also necessary to initialise various hardware such as the PLL, memory controllers etc.
Low-level details on linking and loading of (PE) programs in Windows.
I'm looking for an answer or tutorial that clarifies how a Windows program are linked and loaded into memory after it has been assembled.
Especially, I'm uncertain about the following points:
After the program is assembled, some instructions may reference memory within the .DATA section. How are these references translated, when the program is loaded into memory starting at some arbitrary address? Does RVA's and relative memory references take care of these issues (BaseOfCode and BaseOfData RVA-fields of the PE-header)?
Is the program always loaded at the address specified in ImageBase header field? What if a loaded (DLL) module specifies the same base?
First I'm going to answer your second question:
No, a module (being an exe or dll) is not allways loaded at the base address. This can happen for two reasons, either there is some other module already loaded and there is no space for loading it at the base address contained in the headers, or because of ASLR (Address Space Layout Randomization) which mean modules are loaded at random slots for exploit mitigation purposes.
To address the first question (it is related to the second one):
The way a memory location is refered to can be relative or absolute. Usually jumps and function calls are relative (though they can be absolute), which say: "go this many bytes from the current instruction pointer". Regardless of where the module is loaded, relative jumps and calls will work.
When it comes to addressing data, they are usually absolute references, that is, "access these 4-byte datum at this address". And a full virtual address is specified, not an RVA but a VA.
If a module is not loaded at its base address, absolute references will all be broken, they are no longer pointing to the correct place the linker assumed they should point to. Let's say the ImageBase is 0x04000000 and you have a variable at RVA 0x000000F4, the VA will be 0x040000F4. Now imagine the module is loaded not at its BaseAddress, but at 0x05000000, everything is moved 0x1000 bytes forward, so the VA of your variable is actually 0x050000F4, but the machine code that accessess the data still has the old address hardcoded, so the program is corrupted. In order to fix this, linkers store in the executable where these absolute references are, so they can be fixed by adding to them how much the executable has been displaced: the delta offset, the difference between where the image is loaded and the image base contained in the headers of the executable file. In this case it's 0x1000. This process is called Base Relocation and is performed at load time by the operating system: before the code starts executing.
Sometimes a module has no relocations, so it can't be loaded anywhere else but at its base address. See How do I determine if an EXE (or DLL) participate in ASLR, i.e. is relocatable?
For more information on ASLR: https://insights.sei.cmu.edu/cert/2014/02/differences-between-aslr-on-windows-and-linux.html
There is another way to move the executable in memory and still have it run correctly. There exists something called Position Independent Code. Code crafted in such a way that it will run anywhere in memory without the need for the loader to perform base relocations.
This is very common in Linux shared libraries and it is done addressing data relatively (access this data item at this distance from the instruction pointer).
To do this, in the x64 architecture there is RIP-relative addressing, in x86 a trick is used to emulate it: get the content of the instruction pointer and then calculate the VA of a variable by adding to it a constant offset.
This is very well explained here:
https://www.technovelty.org/linux/plt-and-got-the-key-to-code-sharing-and-dynamic-libraries.html
I don't think PIC code is common in Windows, more often than not, Windows modules contain base relocations to fix absolute addresses when it is loaded somewhere else than its prefered base address, although I'm not exactly sure of this last paragraph so take it with a grain of salt.
More info:
http://opensecuritytraining.info/LifeOfBinaries.html
How are windows DLL actually shared? (a bit confusing because I didn't explain myself well when asking the question).
https://www.iecc.com/linker/
I hope I've helped :)
Before asking my question, I would like to cover some few technical details I want to make sure I've got correct:
A Position Independent Executable (PIE) is a program that would be able to execute regardless of which memory address it is loaded into, right?
ASLR (Address Space Layout Randomization) pretty much states that in order to keep addresses static, we would randomize them in some manner,
I've read that specifically within Linux and Unix based systems, implementing ASLR is possible regardless of if our code is a PIE, if it is PIE, all jumps, calls and offsets are relative hence we have no problem.
If it's not, code somehow gets modified and addresses are edited regardless of whether the code is an executable or a shared object.
Now this leads me to ask a few questions
If ASLR is possible to implement within codes that aren't PIE and are executables AND NOT SHARED / RELOCATABLE OBJECT (I KNOW HOW RELOCATION WORKS WITHIN RELOCATABLE OBJECTS!!!!), how is it done? ELF format should hold no section that states where within the code sections are functions so the kernel loader could modify it, right? ASLR should be a kernel functionality so how on earth could, for example, an executable containing, for example, these instructions.
pseudo code:
inc_eax:
add eax, 5
ret
main:
mov eax, 5
mov ebx, 6
call ABSOLUTE_ADDRES{inc_eax}
How would the kernel executable loader know how to change the
addresses if they aren't stored in some relocatable table within the ELF
file and aren't relative in order to load the executable into some
random address?
Let's say I'm wrong, and in order to implement ASLR you must have a
PIE executable. All segments are relative. How would one compile a
C++ OOP code and make it work, for example, if I have some instance
of a class using a pointer to a virtual table within its struct,
and that virtual table should hold absolute addresses, hence I
wouldn't be able to compile a pure PIE for C++ programs that have
usage of run time virtual tables, and again ASLR isn't possible....
I doubt that virtual tables would contain relative addresses and
there would be a different virtual table for each call of some
virtual function...
My last and least significant question is regarding ELF and PIE — is there some special way to detect an ELF executable is PIE? I'm familiar with the ELF format so I doubt that there is a way, but I might be wrong. Anyway, if there isn't a way, how does the kernel loader know if our executable is PIE hence it could use ASLR on it.
I've got this all messed up in my head and I'd love it if someone could help me here.
Your question appears to be a mish-mash of confusion and misunderstanding.
A Position Independent Executable (PIE) is a program that would be able to execute regardless of which memory address it is loaded into, right?
Almost. A PIE binary usually can not be loaded into memory at arbitrary address, as its PT_LOAD segments will have some alignment requirements (e.g. 0x400, or 0x10000). But it can be loaded and will run correctly if loaded into memory at address satisfying the alignment requirements.
ASLR (Address Space Layout Randomization) pretty much states that in order to keep addresses static we would randomize them in some manner,
I can't parse the above statement in any meaningful way.
ASLR is a technique for randomizing various parts of address space, in order to make "known address" attacks more difficult.
Note that ASLR predates PIE binaries, and does not in any way require PIE. When ASLR was introduced, it randomized placement of stack, heap, and shared libraries. The placement of (non-PIE) main executable could not be randomized.
ASLR has been considered a success, and therefore extended to also support PIE main binary, which is really a specially crafted shared library (and has ET_DYN file type).
call ABSOLUTE_ADDRES{inc_eax}
how would the kernel executable loader know how to change the addresses if > they aren't stored in some relocatable table
Simple: on x86, there is no instruction to call ABSOLUTE_ADDRESS -- all calls are relative.
2 ... I wouldn't be able to compile a pure PIE for C++ programs that have usage of run time virtual tables, and again ASLR isn't possible..
PIE binary requires relocation, just like a shared library. Virtual tables in PIE binaries work exactly the same way they work in shared libraries: ld-linux.so.2 updates GOT (global offset table) before transferring control to the PIE binary.
3 ... is there some special way to detect an ELF executable is PIE
Simple: a PIE binary has ELF file type set to ET_DYN (a non-PIE binary will have type ET_EXEC). If you run file a.out on a PIE executable, you'll see that it's a "shared library".
I've been looking into Window's PE format lately and I have noticed that in most examples,
people tend to set the ImageBase offset value in the optional header to something unreasonably high like 0x400000.
What could make it unfavorable not to map an image at offset 0x0?
First off, that's not a default of Windows or the PE file format, it is the default for the linker's /BASE option when you use it to link an EXE. The default for a DLL is 0x10000000.
Selecting /BASE:0 would be bad choice, no program can ever run at that base address. The first 64 KB of the address space is reserved and can never be mapped. Primarily to catch null pointer dereference bugs. And expanded to 64KB to catch pointer bugs in programs that started life in 16-bits and got recompiled to 32-bits.
Why 0x40000 and not 0x10000 is the default is a historical accident as well and goes back to at least Windows 95. Which reserved the first 4 megabytes of the address space for the "16-bit/MS-DOS Compatibility Arena". I don't remember much about it, Windows 9x had a very different 16-bit VM implementation from NT. You can read some more about it in this ancient KB article. It certainly isn't relevant anymore these days, a 64-bit OS will readily allocate heap memory in the space between 0x010000 and 0x400000.
There isn't any point in changing the /BASE option for an EXE. However, there's lots of point changing it for a DLL. They are much more effective if they don't overlap and thus don't have to be relocated, they won't take any space in the paging file and can be shared between processes. There's even an SDK tool for it so you can change it after building, rebase.exe
Practically, the impact of setting /BASE to 0 depends on the Address Space Layout Randomization (ASLR) setting of your image (which is also put by the Linker - /DYNAMICBASE:NO).
Should your image have /BASE:0 and ASLR is on (/DYNAMICBASE:YES), then your image will start and run because the loader will automatically load it at a "valid" address.
Should your image have /BASE:0 and ASLR is off (/DYNAMICBASE:NO), then your image will NOT start because the loader will NOT load it at the desired based address (which is, as explained above, unvalid/reserved).
If you map it to address 0 then that means the code expects to be running starting at address zero.
For the OS, address zero is NULL, which is an invalid address.
(Not "fundamentally", but for modern-day OSes, it is.)
Also, in general you don't want anything in the lower 16 MiB of memory (even virtual), for numerous reasons.
But what's the alternative? It has to be mapped somewhere, so they chose 0x400000... no particular reason for that particular address, probably. It was probably just handy.
Microsoft chose that address as the default starting address specified by the linker at which the PE file will be memory mapped. The linker assumes this address and and can optimize the executable with that assumption. When the file is memory mapped at that address the code can be run without needing to modify any internal offsets.
If for some reason the file cannot be loaded to that location (another exe/dll already loaded there) relocations will need to occur before the executable can run which will increase load times.
Lower memory addresses are usually assumed to contain low level system routines and are generally left alone. The only real requirement for the ImageBase address is that it is a multiple of 0x10000.
Recommended reading:
http://msdn.microsoft.com/en-us/library/ms809762.aspx