I'm a newbie to kernel programming, and I'm stuck on something, so I'd appreciate some help. I appologize in advance if something similar was asked before, I did not find any relevant post, and could find explanations on the web which were simple enough for someone unexperienced as myself in this field to understand.
I want to experiment with my own version of ext2.
I've got the source files from kernel.org, and made the proper changes. Nothing fancy, just to check something I had in mind.
Now I want to insert it to my linux kernel (ubuntu 2.6.31-14-generic-pae if it matters).
How can I do this?
My (obviously naive) initial thought was to simply use the makefile that comes along with it (after manually setting various flags there so it has obj-m/obj-y where needed) and compile it as a kernel module.
However I keep getting errors during compile time about redifining macros, implicit declarations of functions etc. For example
ext2.h:181:1: warning: "ext2_find_first_zero_bit" redefined
balloc.c:574: error: implicit declaration of function dquot_free_block_nodirty
Obviously this is not the way to go. I guess worst case scenario is compiling the entire kernel again (with my modified ext2 code instead of the original) so it creates the relevant library with my own ext2, and rebooting from the new image. I find it hard to believe this is the best approach.
Is it even possible for a new file system to be inserted as a kernel module?
Myabe I should put my modified ext2 code in /usr/src and somehow compile only the relevant library which contains the current ext2 code?
Anyway, I'd appreciate any help on what should I be doing.
Thank you
Do a search and replace of ext2 with my_awesome_filesystem or some such.
Related
Specifically, my issue is that I have CUDA code that needs <curand_kernel.h> to run. This isn't included by default in NVRTC. Presumably then when creating the program context (i.e. the call to nvrtcCreateProgram), I have to send in the name of the file (curand_kernel.h) and also the source code of curand_kernel.h? I feel like I shouldn't have to do that.
It's hard to tell; I haven't managed to find an example from NVIDIA of someone needing standard CUDA files like this as a source, so I really don't understand what the syntax is. Some issues: curand_kernel.h also has includes... Do I have to do the same for each of these? I am not even sure the NVRTC compiler will even run correctly on curand_kernel.h, because there are some language features it doesn't support, aren't there?
Next: if you've sent in the source code of a header file to nvrtcCreateProgram, do I still have to #include it in the code to be executed / will it cause an error if I do so?
A link to example code that does this or something like it would be appreciated much more than a straightforward answer; I really haven't managed to find any.
You have to send the "filename" and the source of each header separately.
When the preprocessor does its thing, it'll use any #include filenames as a key to find the source for the header, based on the collection that you provide.
I suspect that, in this case, the compiler (driver) doesn't have file system access, so you have to give it the source in much the same way that you would for shader includes in OpenGL.
So:
Include your header's name when calling nvrtcCreateProgram. The compiler will, internally, generate the equivalent of a std::map<string,string> containing the source of each header indexed by the given name.
In your kernel source, use #include "foo.cuh" as usual.
The compiler will use foo.cuh as an index or key into its internal map (created when you called nvrtcCreateProgram), and will retrieve the header source from that collection
Compilation proceeds as normal.
One of the reasons that nvrtc provides only a "subset" of features is that the compiler plays in a somewhat sandboxed environment, without necessarily having all of the supporting tools and utilities lying around that you have with offline compilation. So, you have to manually handle a lot of the stuff that the normal nvcc + (gcc | MSVC| clang) combination provides.
A possible, but non-ideal, solution would be to preprocess the file that you need in your IDE, save the result and then #include that. However, I bet there is a better way to do that. if you just want curand, consider diving into the library and extracting the part you need (blech) or using another GPU-friendly rand implementation. On older CUDA versions, I just generated a big array of random floats on the host, uploaded it to the GPU, and sampled it in the kernels.
This related link may be helpful.
You do not need to load curand_kernel.h yourself and add it to the include "aliases" mechanism.
Instead, you can simply add the CUDA include directory to your (set of) include paths, e.g. by adding --include-path=/usr/local/cuda/include to your NVRTC compiler options.
(I do this in my GPU-kernel-runner test harness, by default, to be on the safe side.)
I was writing much java code lately and got quite used to IDE features like jumping to definitions.
Now I'm working on C (reading some kernel code) and wonder if there is any similar tooling for that language, too.
I am aware of ctags, cscope, and lxr, but they seem to rely much on pattern matching and do not really understand the code.
For example I have some code using struct file, how do I navigate to the definition of that struct in order to determine the available fields?
Even if I know that it is defined in fs.h and it contains a field of type fmode_t defined in types.h as unsigned __bitwise__, is there a way to get the semantics of the single bits? (Probably by finding the FMODE_* constants in fs.h again.)
Is this kind of (eclipse-like) direct navigating possible, or do I really have to do much guessing if I do not know all of this before?
If you like eclipse try nsight. It's for cuda, but I think it gives you what you want.
Try QtCreator for the Linux kernel. I wrote how to use it here.
It is the best IDE for Linux kernel development I have ever seen.
I've been looking at examples of C code that is compiled for some lesser known processors (like ZPU) using the gcc cross compiler.
Most of the working examples I see assume a certain arquitecture (Memory map and set of peripherals) and simply give you a recipe to compile for these and they work.
However I can find very little information on what needs to modified if you use the same cpu with a different memory map and set of peripherals.
From what I've read. There are two main files that I need to make sure that are done "right". The linker script that is used and the crt0.o (Which if I need to modify means recompiling the crt0.S which is assembler). On this last one, especially I find very little information on what is actually supposed to do (other that setting up reset there is no clear info, and I'm talking conceptually not for an specific processor. Although something for this would also be useful).
Can any one tell me what is the relationship between a the c files for the code of program (bare metal development), the crt0.S (specially why it is needed) and it's relationship with a working linker script?
PD: Answers of the form "read this book" are welcome and I would love them.
PD: I realize this kind of question is usually vague and closed quickly but I don't know where else to turn, so I ask for a bit of leniency.
I often spend lots of time trying to find out where the exact implementation is located. It gets very frustrating when dealing with some low-level code that might end up somewhere in kernel.
I usually just google or try to guess the location and/or method names, but it is not always very effective.
Is there some methodical way to trace the flow up to the implementation? How do you guys usually do it?
Load the whole the code with relevant dependencies to a graphical IDE (NetBeans can do it, for instance) which can to call graph, declaration-definition jumps, etc. or use LibClang and its wrapper for the text editor of your choice, it is also very good at indexing. At last, you can consider classics, ctags, which can link definition and declaration points.
There used to be a ctrace program that did just that but I don't think it is actively maintained.
Ultimately, it depends what exactly you are trying to achieve. It actually looks like you want to look up a specific function rather than trace it. If that is the case indeed, consider using some kind of source browser: from etags to cscope to OpenGrok.
I learned to program with Pascal in high school, and more recently I decided to get out of the sandbox and try to figure out how my computer actually works. So I installed ubuntu on my iMac (i686) and started learning C, which seemed like a good way to get "under the hood."
One of the basic things I'm trying to figure out is where the kernel ends and the standard libraries begin. A book told me that the linux system calls (which I understand to be the interface between the kernel and the libraries) could be found in the header file unistd.h, so this seemed like a good place to start. But when I tried to find the header on my system (using locate unistd.h), I got this result:
/usr/include/unistd.h
/usr/include/asm-generic/unistd.h
/usr/include/i386-linux-gnu/asm/unistd.h
/usr/include/i386-linux-gnu/bits/unistd.h
/usr/include/i386-linux-gnu/sys/unistd.h
/usr/include/linux/unistd.h
/usr/lib/syslinux/com32/include/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/alpha/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/arm/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/avr32/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/blackfin/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/c6x/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/cris/include/arch-v10/arch/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/cris/include/arch-v32/arch/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/cris/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/frv/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/h8300/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/hexagon/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/ia64/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/m32r/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/m68k/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/microblaze/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/mips/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/mn10300/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/openrisc/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/parisc/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/powerpc/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/s390/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/score/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/sh/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/sparc/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/tile/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/unicore32/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/x86/include/asm/ia32_unistd.h
/usr/src/linux-headers-3.5.0-27/arch/x86/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/arch/xtensa/include/asm/unistd.h
/usr/src/linux-headers-3.5.0-27/include/asm-generic/unistd.h
/usr/src/linux-headers-3.5.0-27/include/linux/unistd.h
/usr/src/linux-headers-3.5.0-27-generic/include/linux/unistd.h
Why the heck are there so many versions of this file--and other header files--in my system? Some of them seem to be for other CPUs (like sparc), so why did ubuntu bother to install them on my computer? And how does the all of this fit with what Eric Raymond calls the SPOT rule: "every piece of knowledge must have a single, unambiguous, authoritative representation within a system." (The Art of Unix Programming, p. 91.)
Thanks in advance for any help. I'm happy to read big books if necessary.
I think these header files are directly from linux-3.5.0-27 source code. Ubuntu developers didn't know what kind of target they are dealing with. Maybe Intel x86/powerPC/ or even a mobile hand set(ARM), so they just copy all the head files and make a simple link.