Reference function of shared library in executable with relocations but without PIC - c++

I am wondering myself if it is possible to build & link a executable using a shared object so that it is not using PIC (therefore PLT) but load-time relocations.
I think if this is possible, the code section has to be re writeable (which should principal be no problem).
If I try with no additional gcc parameters, it uses PIC (usually, to create a PIC shared lib, I have to add -fPIC).
I know that it is possible with data, for that case a R_386_COPY relocation is executed.
So, is this possible for functions? And if, with which gcc parameters?

is this possible for functions?
Sure.
And if, with which gcc parameters?
No version of ld that I know of will do that (as generally this is considered the wrong thing to do). You'll have to build ld from source, and apply a patch to make it do what you want.
the code section has to be re writeable
Correct.
(which should principal be no problem).
Many environments, such as e.g. SELinux prohibit writeable and executable mappings, as such mappings are exceedingly insecure.
So while your binary with writable code section would run in some environments with no problem, it will not run in many others.

Related

Change Linux shared library (.so file) version after it was compiled

I'm compiling Linux libraries (for Android, using NDK's g++, but I bet my question makes sense for any Linux system). When delivering those libraries to partners, I need to mark them with a version number. I must also be able to access the version number programatically (to show it in an "About" dialog or a GetVersion function for instance).
I first compile the libraries with an unversioned flag (version 0.0) and need to change this version to a real one when I'm done testing just before sending it to the partner. I know it would be easier to modify the source and recompile, but we don't want to do that (because we should then test everything again if we recompile the code, we feel like it would be less error prone, see comments to this post and finally because our development environment works this way: we do this process for Windows binaries: we set a 0.0 resources version string (.rc) and we later change it by using verpatch...we'd like to work with the same kind of process when shipping Linux binaries).
What would be the best strategy here?
To summarize, requirements are:
Compile binaries with "unset" version (0.0 or anything else)
Be able to modify this "unset" version to a specific one without having to recompile the binary (ideally, run a 3rd party tool command, as we do with verpatch under Windows)
Be able to have the library code retrieve it's version information at runtime
If your answer is "rename the .so", then please provide a solution for 3.: how to retrieve version name (i.e.: file name) at runtime.
I was thinking of some solutions but have no idea if they could work and how to achieve them.
Have a version variable (one string or 3 int) in the code and have a way to change it in the binary file later? Using a binary sed...?
Have a version variable within a resource and have a way to change it in the binary file later? (as we do for win32/win64)
Use a field of the .so (like SONAME) dedicated to this and have a tool allowing to change it...and make it accessible from C++ code.
Rename the lib + change SONAME (did not find how this can be achieved)...and find a way to retrieve it from C++ code.
...
Note that we use QtCreator to compile the Android .so files, but they may not rely on Qt. So using Qt resources is not an ideal solution.
I am afraid you started to solve your problem from the end. First of all SONAME is provided at link time as a parameter of linker, so in the beginning you need to find a way to get version from source and pass to the linker. One of the possible solutions - use ident utility and supply a version string in your binary, for example:
const char version[] = "$Revision:1.2$"
this string should appear in binary and ident utility will detect it. Or you can parse source file directly with grep or something alike instead. If there is possibility of conflicts put additional marker, that you can use later to detect this string, for example:
const char version[] = "VERSION_1.2_VERSION"
So you detect version number either from source file or from .o file and just pass it to linker. This should work.
As for debug version to have version 0.0 it is easy - just avoid detection when you build debug and just use 0.0 as version unconditionally.
For 3rd party build system I would recommend to use cmake, but this is just my personal preference. Solution can be easily implemented in standard Makefile as well. I am not sure about qmake though.
Discussion with Slava made me realize that any const char* was actually visible in the binary file and could then be easily patched to anything else.
So here is a nice way to fix my own problem:
Create a library with:
a definition of const char version[] = "VERSIONSTRING:00000.00000.00000.00000"; (we need it long enough as we can later safely modify the binary file content but not extend it...)
a GetVersion function that would clean the version variable above (remove VERSIONSTRING: and useless 0). It would return:
0.0 if version is VERSIONSTRING:00000.00000.00000.00000
2.3 if version is VERSIONSTRING:00002.00003.00000.00000
2.3.40 if version is VERSIONSTRING:00002.00003.00040.00000
...
Compile the library, let's name it mylib.so
Load it from a program, ask its version (call GetVersion), it returns 0.0, no surprise
Create a little program (did it in C++, but could be done in Python or any other languauge) that will:
load a whole binary file content in memory (using std::fstream with std::ios_base::binary)
find VERSIONSTRING:00000.00000.00000.00000 in it
confirms it appears once only (to be sure we don't modify something we did not mean to, that's why I prefix the string with VERSIONSTRING, to make it more unic...)
patch it to VERSIONSTRING:00002.00003.00040.00000 if expected binary number is 2.3.40
save the binary file back from patched content
Patch mylib.so using the above tool (requesting version 2.3 for instance)
Run the same program as step 3., it now reports 2.3!
No recompilation nor linking, you patched the binary version!

Loading GNU ld script with dlopen

I have an C++14 code that should load an arbitrary shared object file with dlopen. Unfortunately on some systems (e.g. my archlinux, reportedly also applies to some .so on ubuntu and gentoo), these so-files can be "GNU ld scripts" instead of the actual binaries.
For reference, here is the content of my /usr/lib/libm.so:
/* GNU ld script
*/
OUTPUT_FORMAT(elf64-x86-64)
GROUP ( /usr/lib/libm.so.6 AS_NEEDED ( /usr/lib/libmvec.so.1 ) )
I have found a couple of code-pieces that deal with this issue in ghc or ruby. I would like to avoid resorting to manually parsing the text file based parsing the dlerror text and the file. I feel that is terribly evil and I won't be able to implement and maintain corner cases of this format.
Is there a clean way to implement handling this case? Frankly I am puzzled as to why dlopen does not actually handle these tranparaently.
Note: Considering the aforementioned patches I think this is not simply an issue with my system configuration / versions. If this should work out-of-the-box with dlopen (bug instead of missing feature), please let me know.
The linker scripts are intended to be used by the linker, not the run-time linker.
The GNU ld script comment should have been a giveaway: this is for ld, not for ld.so. ;-)
See for instance: http://www.math.utah.edu/docs/info/ld_3.html
So I guess using this with dlopen() would mean mimicking/importing part of ld's magic for this, which would confirm your fears about resorting to manually parsing the text and maintaining terribly evil code.
EDIT: There seems to be one thing that can help you though:
https://www.sourceware.org/ml/libc-alpha/2011-07/msg00152.html
<gnu/lib-names.h> should contain a define LIBM_SO which should point you to the correct file that you can actually dlopen().
That means that normally no evil code would be necessary.

How to determine in which .SO library is given C function?

I have this problem all the time in Linux programming. As long as all the manuals and almost all the source code for Linux are C-centric, all references to some function needs only some include <something.h> line and the function is accessible from the C/C++ code.
But I am programming in assembly language and know almost nothing about C/C++.
In order to be able to call some function, I have to import it from the corresponding .so library.
How to determine the file name of the library? It often differs from the name of the library itself and is not specified in the manuals.
For example, the name of the XLib is actually libX11.so.6. The name of the XShm extension library seems to be libXext.so.6.
Is there easy way to determine the secret real name of the library, using provided C manuals and references?
This is another not-100%-accurate method that may give you some ideas as to how you can narrow things down a bit. It doesn't exactly fit the question because it uses common linux utilities instead of man files, but it may still be helpful.
Use your distribution's package management software.
For example, on Arch Linux, if you were interested in a function in GLFW/glfw3.h, you could find out who owns that file:
$ pacman -Qo /usr/include/GLFW/glfw3.h
/usr/include/GLFW/glfw3.h is owned by glfw 3.1-1
Find out which .so files are in that package:
$ pacman -Ql glfw | grep 'so$'
glfw /usr/lib/libglfw.so
And, if needed, find the actual file that link points to:
$ readlink -f /usr/lib/libglfw.so
/usr/lib/libglfw.so.3.1
This will depend on your distribution. I believe on Ubuntu/Debian you'd use dpkg-query instead.
Edit: DevSolar points out in a comment that you can use apt-file search <header> and apt-file list <package> instead of dpkg-query -S <header> and dpkg-query -L <package>. apt-file appears to work even for packages that aren't installed (though it seems slower?).
I also noticed that (on my Ubuntu VM at least) that, e.g., libglfw-dev contains the libglfw.so symlink, while libglfw2 contains the actual libglfw.so.2 object.
Once you have a set of .so files, you can check them for whatever function you are interested in:
$ nm -D /usr/lib/libglfw.so | grep "glfwCreateWindow"
0000000000007cd0 T glfwCreateWindow
Note that I pulled this last step from a comment on the previous question and don't fully understand it. Maybe you could even skip the earlier steps and rely on nm and grep alone?
This is not a sure fire way, but it can help in many cases.
Basically, you can usually find the library name at the bottom of the man page.
Eg, man XCreateWindow says libX11 on the last line. Then you look for libX11.so and use nm or readelf to see all exported functions.
Another example, man XShm says libXext at the bottom. And so on.
UPDATE
If the function is in section (2) of the man pages, it's a system call (see man man) and is provided by glibc, which would be libc-2.??.so.
Lastly (thanks Basile), if the function does not mention the library, it is also most likely provided by glibc.
DISCLAIMER: Again this is not a 100% accurate method -- but it should help in most cases.
You can ask gcc to tell you which file it would use for linking like so:
gcc --print-file-name=libX11.so
Sample output:
/usr/lib/gcc/x86_64-linux-gnu/4.9/../../../x86_64-linux-gnu/libX11.so
This file will usually be a symlink, so you'll have to pipe it through readlink or realpath to get the actual file. For example:
readlink -f $(gcc --print-file-name=libXext.so)
Sample output:
/usr/lib/x86_64-linux-gnu/libXext.so.6.4.0
As I commented, you could use gcc to link your program, and then it should be able to accept -lX11 ; by using gcc -v instead of gcc you'll find out what is actually linked and how.
However, you have a much more significant issue than finding the lib*.so.*; most C or C++ APIs are described in header files, and these C or C++ header files also contain symbolic constants (like O_RDONLY for open(2)...) or macros (like WIFEXITED in POSIX wait ...) whose value or expansion you should manually find in header files or documentations. (Quite often, such constants are either preprocessor #define-d constants or enum values). Also, some headers -in particular in C++- contains a lot of inline-d functions (or macros)!
A possible way might be to generate some C files to find all these constants, enums, macros, inlined functions..., and/or to customize the GCC compiler (e.g. with MELT ...) to find them.
So my message is that for better or worse, the C language is deeply tied to Linux & POSIX.
You might restrict yourself to use only syscalls(2) from your assembler code. Then you won't use libX11 and you don't need any header or constant (except the ones for syscalls, starting from <asm/unistd.h>).
BTW, in 2015, coding entirely in assembler for performance reasons is a mistake. The compiler is generating better code than you reasonably can (as soon as you have more than a few hundred machine instructions). In practice, you can code in assembler with GCC by using extended asm instructions in your C functions.
Or are you building your own compiler ? Then you should have told so in your question!
Read also the Program Library HowTo & the Linux Assembly HowTo

C++ binary identification (manifest)

We have a large set of C++ projects (GCC, Linux, mostly static libraries) with many dependencies between them. Then we compile an executable using these libraries and deploy the binary on the front-end. It would be extremely useful to be able to identify that binary. Ideally what we would like to have is a small script that would retrieve the following information directly from the binary:
$ident binary
$binary : Product=PRODUCT_NAME;Version=0.0.1;Build=xxx;User=xxx...
$ dependency: Product=PRODUCT_NAME1;Version=0.1.1;Build=xxx;User=xxx...
$ dependency: Product=PRODUCT_NAME2;Version=1.0.1;Build=xxx;User=xxx...
So it should display all the information for the binary itself and for all of its dependencies.
Currently our approach is:
During compilation for each product we generate Manifest.h and Manifest.cpp and then inject Manifest.o into binary
ident script parses target binary, finds generated stuff there and prints this information
However this approach is not always reliable for different versions of gcc..
I would like to ask SO community - is there better approach to solve this problem?
Thanks for any advice
One of the catches with storing data in source code (your Manifest.h and .cpp) is the size limit for literal data, which is dependent on the compiler.
My suggestion is to use ld. It allows you to store arbitrary binary data in your ELF file (so does objcopy). If you prefer to write your own solution, have a look at libbfd.
Let us say we have a hello.cpp containing the usual C++ "Hello world" example. Now we have the following make file (GNUmakefile):
hello: hello.o hello.om
$(LINK.cpp) $^ $(LOADLIBES) $(LDLIBS) -o $#
%.om: %.manifest
ld -b binary -o $# $<
%.manifest:
echo "$#" > $#
What I'm doing here is to separate out the linking stage, because I want the manifest (after conversion to ELF object format) linked into the binary as well. Since I am using suffix rules this is one way to go, others are certainly possible, including a better naming scheme for the manifests where they also end up as .o files and GNU make can figure out how to create those. Here I'm being explicit about the recipe. So we have .om files, which are the manifests (arbitrary binary data), created from .manifest files. The recipe states to convert the binary input into an ELF object. The recipe for creating the .manifest itself simply pipes a string into the file.
Obviously the tricky part in your case isn't storing the manifest data, but rather generating it. And frankly I know too little about your build system to even attempt to suggest a recipe for the .manifest generation.
Whatever you throw into your .manifest file should probably be some structured text that can be interpreted by the script you mention or that can even be output by the binary itself if you implement a command line switch (and disregard .so files and .so files hacked into behaving like ordinary executables when run from the shell).
The above make file doesn't take into account the dependencies - or rather it doesn't help you create the dependency list in any way. You can probably coerce GNU make into helping you with that if you express your dependencies clearly for each goal (i.e. the static libraries etc). But it may not be worth it to take that route ...
Also look at:
C/C++ with GCC: Statically add resource files to executable/library and
Is there a Linux equivalent of Windows' "resource files"?
If you want particular names for the symbols generated from the data (in your case the manifest), you need to use a slightly different route and use the method described by John Ripley here.
How to access the symbols? Easy. Declare them as external (C linkage!) data and then use them:
#include <cstdio>
extern "C" char _binary_hello_manifest_start;
extern "C" char _binary_hello_manifest_end;
int main(int argc, char** argv)
{
const ptrdiff_t len = &_binary_hello_manifest_end - &_binary_hello_manifest_start;
printf("Hello world: %*s\n", (int)len, &_binary_hello_manifest_start);
}
The symbols are the exact characters/bytes. You could also declare them as char[], but it would result in problems down the road. E.g. for the printf call.
The reason I am calculating the size myself is because a.) I don't know whether the buffer is guaranteed to be zero-terminated and b.) I didn't find any documentation on interfacing with the *_size variable.
Side-note: the * in the format string tells printf that it should read the length of the string from the argument and then pick the next argument as the string to print out.
You can insert any data you like into a .comment section in your output binary. You can do this with the linker after the fact, but it's probably easier to place it in your C++ code like this:
asm (".section .comment.manifest\n\t"
".string \"hello, this is a comment\"\n\t"
".section .text");
int main() {
....
The asm statement should go outside any function, in this instance. This should work as long as your compiler puts normal functions in the .text section. If it doesn't then you should make the obvious substitution.
The linker should gather all the .comment.manifest sections into one blob in the final binary. You can extract them from any .o or executable with this:
objdump -j .comment.manfest -s example.o
Have you thought about using standard packaging system of your distro? In our company we have thousands of packages and hundreds of them are automatically deployed every day.
We are using debian packages that contain all the neccessary information:
Full changelog that includes:
authors;
versions;
short descriptions and timestamps of changes.
Dependency information:
a list of all packages that must be installed for the current one to work correctly.
Installation scripts that set up environment for a package.
I think you may not need to create manifests in your own way as soon as ready solution already exists. You can have a look at debian package HowTo here.

Change OpenSSL library internal name with 'chatr' command on HPIA

I need to change the internal name of libcrypto.so.0.9.8 shared library to libcrypto.so for a specific purpose. I am unable to do so with the chatr command which only displays the internal name.
There is a restriction that I am unable to re compile the shared library with +h option giving the internal name, which was my initial idea.
Thanks in advance.
Use the LD_PRELOAD environment variable. It allows you to interpose libraries.
Basically you setup a symlink, in a directory you control, named libcrypto.so.0.9.8, which points to the library you want to use, (I guess): /lib/libcrypto.so. Aim LD_PRELOAD at the symlink.
LD_PRELOAD will NOT work with setuid programs in HPUX.
You did carefully verify that all of the entry points you use in your code are in libcrypto.so?
Since I have no real idea what is going on this may not be an ideal solution. It is a best guess.