Author here, I posted this in Show HN but someone clearly beat me to it. So I'll repost my blurb from there.
Various patterns for safer C programming have been cargo-culting around the industry for decades. Because the language evolves intentionally slowly, these patterns rarely get folded into the language as first-class constructs and are passed down through the generations in a sort of oral tradition of programming.
lib0xc leverages GNUC extensions and C11 features to codify safer C practices and patterns into real APIs with real documentation and real testing. Reduce your casts to and from `void *` with the `context_t` tagged pointer type. Enable type-checked, deferred function invocation with `call_t`. Interrogate structure descriptors with `struct_field_t`. Stop ignoring `-Wint-conversion` and praying you won't regret it when you assign a signed integer to an unsigned integer and use `__cast_signed_unsigned`. These are just a few of lib0xc's standard-library-adjacent offerings.
lib0xc also provides a basic systems programming toolkit that includes logging, unit tests, a buffer object designed to deal with types, a unified Mach-O and ELF linker set, and more.
Everything in lib0xc works with clang's bounds-safety extensions if they are enabled. Both gcc and clang are supported. Porting to another environment is a relatively trivial effort.
It's not Rust, and it's not type safety, but it's not supposed to be. It's supposed to help you make your existing C codebase significantly safer than it was yesterday.
My employer holds the copyright and has permitted its release under the MIT license.
Two notes: GCC has its "access" attributes which can give you similar bounds safety as clang.
Please see also my experimental library. https://codeberg.org/uecker/noplate/ While I do not had enough time to polish it yet, I think it provides some very nice interfaces with improve type and bounds safety, and are also rather convenient.
Also I wonder what parts are redundant if you have FORTIFY_SOURCE ?
(And thank you for working in this topic. If you continue, please reach out to us)
> This might be a dumb question, but using this + clang bounds-safety, whats the difference between this and something like Zig or Odin.
I really need to learn more about Zig, but from what I know, there are still worlds of possibilities that a modern, well-designed language offers over something like lib0xc. Zig's ability to evaluate any expression at compile-time is one such example.
But generally, lib0xc gives you bounds-safety everywhere it can. Languages like Zig and Rust give you type-safety to their own degrees, which I think is a superset.
> What do you think C would need in order to reach the user experience of those languages?
Not really having direct user experience, it's hard for me to say. But if I what I can give you is a list of features that would make large parts of lib0xc irrelevant:
1. Protocols/traits
2. Allocating from a caller's stack frame (think, returning the result of `alloca` to the caller)
3. printf format specifiers for stdint.h types and for octet strings
4. Ability to express function parameter lists as structures
5. New sprintf family that returns a value which is always less than or equal to the size passed (no negative values)
Basically, I think that the C standard should be working aggressively to cut down on the use cases for heap allocation and `void *`. And I think that the bounds safety annotations should become first-class language features.
> I really need to learn more about Zig, but from what I know, there are still worlds of possibilities that a modern, well-designed language offers over something like lib0xc.
Doesn't Apple have a nice `defer { }` block for cleanup? Did you include that in lib0xc? I didn't see in on your README.
I think defer has been included in the next round of working group proposals for C2y, but I don't think Apple's clang has it. Maybe it's there as a language extension and I just didn't see it.
What lib0xc has is some cleanup attributes that you can apply to variables to e.g. automatically free a heap allocation or close a file descriptor, at end of scope. Personally, I like variable annotations much more than defer for these uses, but they accomplish the same thing. I've also found that using those attributes inherently pushes your code to make ownership more explicit. I personally stopped being terrified of double-pointers and started using them for ownership transfers, which eliminates a large class of bugs.
> I've also found that using those attributes inherently pushes your code to make ownership more explicit. I personally stopped being terrified of double-pointers and started using them for ownership transfers, which eliminates a large class of bugs.
This is very interesting. Do you have a practical example?
In C++ you can implement such a thing using destructors, which are guaranteed to run in reverse order on scope exit even in the presence of exceptions. Alexei Alexandrescu's Scopeguard did this (in the 90s I think, long before C++11). But in standard C, there's no mechanism that this could be attached to (especially if you want to use "C exceptions", a.k.a. setjmp()/longjmp()).
Maybe the compilers they support all have non-standard extensions that allow something like this though?
Not really. You'd still be able to address memory as bytes. The problem with void * is that it strips all bounds information by its nature. Most of the time when you're passing a void * without an associated length (e.g. context pointers, objects that you pinky-swear are of a certain type), it indicates a failure in the language. That's the stuff I think needs to be eliminated.
Have a look on libre C SDK's for the GBA and read about how some data it's set. Ditto with another set of archs where some simple C89 it's being ported.
Glad to see you’re still doing great stuff, and also very glad to see your new employer supports such things, especially compared to our old employer! Part of why I retired around the same time you left was because I wanted to make and share things.
Every time I look at how easy for people to use this kind of thing but people tends not to, remind me if so-called "memory safety" is a real concern anyway.
there are no good reasons we don't do this in the standards themselves, C, C++, and POSIX should all be working on editions that add safer APIs and mark unsafe APIs as deprecated, to start a long term migration. we know how to do this, we've had a lot of success with this. there are real engineering concerns, sure, but they're not reasons to not do it. compilers and library chains can retain support for less safe variants for plenty of time.
The reason this wasn't done by the standards committees is that they spent decades refusing to admit there was even a problem they could help fix. And if there was a problem, it was easily avoided by just writing better code. And if writing better code wasn't enough, well it was certainly too expensive to provide as a debug option. And if it wasn't too expensive to provide as a debug option, the implementors should really lead the way first. And on and on.
The C committee at least seems to get it now. The C++ committee still doesn't, led in large part by Bjarne.
This is a misrepresentation based on a misunderstanding on how standardization works. The C standard committee has long recognized the need for better safety and carefully made it possible so that C could be implemented safely. But the process is that vendors implement something and then come together during standardization so that it is compatible, not that the standardization is the government that prescribes top-down what everybody has to do. Vendors did not bother to provide safer C implementations and safety features (such as bounds checking) did not get much attention from users in the past. So I am happy to see that there is now more interest in safety, because as soon as there solutions we can start putting them into the standard.
(We can do some stuff before this, but this is always a bit of a fight with the vendors, because they do not like it at all if we tell them what to do, especially clang folks)
Anything needs to be demonstrated and used in practice before being included in the standard. The standard is only meant to codify existing practices, not introduce new ideas.
It's up to compiler developers to ship first, standardize later.
That produces a bit of a chicken and egg probablem for a stdlib overhaul. Compilers and libc implementations don't have a strong reason to implement safer APIs, because if it is non-standard then projects that want to be portable won't use it , but it won't get standardized unless they do add safer APIs.
So the best hope is probably for a third party library that has safet APIs to get popular enough that it becomes a de facto standard.
I think the real failing is that new language features then must be prototyped by people who have a background in compilers. That's a very small subset of the overall C community.
I don't have any clue how to patch clang's front end. I'm not a language or compiler person. I just want to make stuff better. There needs to be a playground for people like me, and hopefully lib0xc can be that playground.
This is great. Other things needed for a great C development environment are a standardized build process plus build tools and a standardized packaging system.
The title looks very promising. I’ve added this library to my to-do list to take a deeper look at it. Using this standart library within restricted safe subset of C++ can be a strong opponent for Zig (at least for myself).
Haven't really verified that it works with C++, but I tried my best to guard the stuff I knew would be problematic with #if __cplusplus. Happy to have a PR that makes C++ happier with it.
I truly hope something like this catches on. There is so much low hanging fruit in both the C and C++ standard libraries. Spatial memory could be 90% solved in both languages by mandating the use of safe interfaces.
> I'm not sure MSVC could compile the GNU extensions used.
The supported platforms only list Linux and Mac. Notably missing from that list is any of the BSDs (not counting what may or may not remain of BSD under the Mac hood).
Author here. It is not currently in production, but it is part of a project in Azure which will go to production at some point. I'm actually leaving Microsoft next week and fully intend to keep working on it if I can reach an agreement to do so with my new employer.
Look, C is so simple, you can implement a compiler in under 10K lines of code on a new platform. And then spend the your life dodging bugs and come up with a zoo of bizarre macros to keep you safe.
People say that kind of thing on HN every now and then. I have no idea why this idea is around, it's a complete fantasy in my opinion. I say this as someone who mostly uses Linux.
Switching to C++ is relatively easy in an existing codebase. It's in many cases as simple as renaming a file from .c to .cpp. But for writing something from scratch it's better to use Rust.
Renaming c. to .cpp may work with ancient c89 code, but not with anything remotely modern. But while the code then is technically C++, it is not better. I still prefer C for new projects to any other language, because I value short compilation time and reduced complexity. For me, this translates in higher productivity and more fun. With modern tooling, also most C issues are detected early.
How things change. Imagine Microsoft circa 2000 publishing an MIT-licensed source code library targeting "the competition"'s compilers, and including some light humour ("Embiggen C's Pit of Success") in the docs.
Quick question for those who've tried it — does this play with existing C
codebases incrementally, or is it more of a "new project only" situation?
The README didn't make that obvious to me.
It's designed to be incremental. For example, you can do a search for `sprintf` and replace it with `ssprintf`. The function signature is the same. Any instance of printing to a character array just works. Think of the APIs as "the stuff you usually do by hand, but safer".
If you get compiler errors, it means you were printing to a heap-allocated buffer (or a buffer whose bounds you did not know), and you should be propagating bounds and using `snprintf`.
Integer conversion is the same way. If you have something like
int v1;
uint64_t v2;
<stuff happens>
v2 = (uint64_t)v1;
Then you can replace it with
v2 = __cast_signed_unsigned(uint64_t, v1);
and you'll get a runtime trap when v1 is a negative value, meaning you can both enable -Wint-conversion and have defined behavior for when the value in a certain integer type is not representable in another.
Microsoft supports memory safety. Rust is 100% the direction for new projects. But there are existing C codebases that are unlikely to be entirely rewritten in a memory-safe language for various reasons. Such projects can significantly benefit from incremental improvements in memory safety.
It's just an excuse not to use safe languages. And adopting it isn't that easy - one need to learn how the new library works or even rewrite old working and tested code with it.
60 comments:
Author here, I posted this in Show HN but someone clearly beat me to it. So I'll repost my blurb from there.
Various patterns for safer C programming have been cargo-culting around the industry for decades. Because the language evolves intentionally slowly, these patterns rarely get folded into the language as first-class constructs and are passed down through the generations in a sort of oral tradition of programming.
lib0xc leverages GNUC extensions and C11 features to codify safer C practices and patterns into real APIs with real documentation and real testing. Reduce your casts to and from `void *` with the `context_t` tagged pointer type. Enable type-checked, deferred function invocation with `call_t`. Interrogate structure descriptors with `struct_field_t`. Stop ignoring `-Wint-conversion` and praying you won't regret it when you assign a signed integer to an unsigned integer and use `__cast_signed_unsigned`. These are just a few of lib0xc's standard-library-adjacent offerings.
lib0xc also provides a basic systems programming toolkit that includes logging, unit tests, a buffer object designed to deal with types, a unified Mach-O and ELF linker set, and more.
Everything in lib0xc works with clang's bounds-safety extensions if they are enabled. Both gcc and clang are supported. Porting to another environment is a relatively trivial effort.
It's not Rust, and it's not type safety, but it's not supposed to be. It's supposed to help you make your existing C codebase significantly safer than it was yesterday.
My employer holds the copyright and has permitted its release under the MIT license.
Thanks!
Two notes: GCC has its "access" attributes which can give you similar bounds safety as clang.
Please see also my experimental library. https://codeberg.org/uecker/noplate/ While I do not had enough time to polish it yet, I think it provides some very nice interfaces with improve type and bounds safety, and are also rather convenient.
Also I wonder what parts are redundant if you have FORTIFY_SOURCE ?
(And thank you for working in this topic. If you continue, please reach out to us)
This might be a dumb question, but using this + clang bounds-safety, whats the difference between this and something like Zig or Odin.
What do you think C would need in order to reach the user experience of those languages?
> This might be a dumb question, but using this + clang bounds-safety, whats the difference between this and something like Zig or Odin.
I really need to learn more about Zig, but from what I know, there are still worlds of possibilities that a modern, well-designed language offers over something like lib0xc. Zig's ability to evaluate any expression at compile-time is one such example.
But generally, lib0xc gives you bounds-safety everywhere it can. Languages like Zig and Rust give you type-safety to their own degrees, which I think is a superset.
> What do you think C would need in order to reach the user experience of those languages?
Not really having direct user experience, it's hard for me to say. But if I what I can give you is a list of features that would make large parts of lib0xc irrelevant:
1. Protocols/traits
2. Allocating from a caller's stack frame (think, returning the result of `alloca` to the caller)
3. printf format specifiers for stdint.h types and for octet strings
4. Ability to express function parameter lists as structures
5. New sprintf family that returns a value which is always less than or equal to the size passed (no negative values)
Basically, I think that the C standard should be working aggressively to cut down on the use cases for heap allocation and `void *`. And I think that the bounds safety annotations should become first-class language features.
Why not pick a different language if you want different features? Why does C specifically need to change, if there are already Zig, Rust etc.?
Why Must C be safe, rather than people writing safer code in it or transfering to other languages if they cannot be bothered?
> I really need to learn more about Zig, but from what I know, there are still worlds of possibilities that a modern, well-designed language offers over something like lib0xc.
Doesn't Apple have a nice `defer { }` block for cleanup? Did you include that in lib0xc? I didn't see in on your README.
I think defer has been included in the next round of working group proposals for C2y, but I don't think Apple's clang has it. Maybe it's there as a language extension and I just didn't see it.
What lib0xc has is some cleanup attributes that you can apply to variables to e.g. automatically free a heap allocation or close a file descriptor, at end of scope. Personally, I like variable annotations much more than defer for these uses, but they accomplish the same thing. I've also found that using those attributes inherently pushes your code to make ownership more explicit. I personally stopped being terrified of double-pointers and started using them for ownership transfers, which eliminates a large class of bugs.
> I've also found that using those attributes inherently pushes your code to make ownership more explicit. I personally stopped being terrified of double-pointers and started using them for ownership transfers, which eliminates a large class of bugs.
This is very interesting. Do you have a practical example?
In C++ you can implement such a thing using destructors, which are guaranteed to run in reverse order on scope exit even in the presence of exceptions. Alexei Alexandrescu's Scopeguard did this (in the 90s I think, long before C++11). But in standard C, there's no mechanism that this could be attached to (especially if you want to use "C exceptions", a.k.a. setjmp()/longjmp()).
Maybe the compilers they support all have non-standard extensions that allow something like this though?
Yes, because all compilers support a non-standard defer mechanism, its now being considered for inclusion into standard C. [0]
And that suggested defer standard, is already available from GCC 9 and clang 22.
[0] https://www.open-std.org/JTC1/SC22/WG14/www/docs/n3734.pdf
Wouldn't the last case (void *) hurt embedded C development, or retrogaming with direct memory access and pointers?
They said "cut down", not "eliminate."
Not really. You'd still be able to address memory as bytes. The problem with void * is that it strips all bounds information by its nature. Most of the time when you're passing a void * without an associated length (e.g. context pointers, objects that you pinky-swear are of a certain type), it indicates a failure in the language. That's the stuff I think needs to be eliminated.
Have a look on libre C SDK's for the GBA and read about how some data it's set. Ditto with another set of archs where some simple C89 it's being ported.
Because it is C not Zig/Odin. The mental/ecosystem cost to use a new language is way way under-estimated in most cases.
Glad to see you’re still doing great stuff, and also very glad to see your new employer supports such things, especially compared to our old employer! Part of why I retired around the same time you left was because I wanted to make and share things.
Every time I look at how easy for people to use this kind of thing but people tends not to, remind me if so-called "memory safety" is a real concern anyway.
there are no good reasons we don't do this in the standards themselves, C, C++, and POSIX should all be working on editions that add safer APIs and mark unsafe APIs as deprecated, to start a long term migration. we know how to do this, we've had a lot of success with this. there are real engineering concerns, sure, but they're not reasons to not do it. compilers and library chains can retain support for less safe variants for plenty of time.
The reason this wasn't done by the standards committees is that they spent decades refusing to admit there was even a problem they could help fix. And if there was a problem, it was easily avoided by just writing better code. And if writing better code wasn't enough, well it was certainly too expensive to provide as a debug option. And if it wasn't too expensive to provide as a debug option, the implementors should really lead the way first. And on and on.
The C committee at least seems to get it now. The C++ committee still doesn't, led in large part by Bjarne.
This is a misrepresentation based on a misunderstanding on how standardization works. The C standard committee has long recognized the need for better safety and carefully made it possible so that C could be implemented safely. But the process is that vendors implement something and then come together during standardization so that it is compatible, not that the standardization is the government that prescribes top-down what everybody has to do. Vendors did not bother to provide safer C implementations and safety features (such as bounds checking) did not get much attention from users in the past. So I am happy to see that there is now more interest in safety, because as soon as there solutions we can start putting them into the standard.
(We can do some stuff before this, but this is always a bit of a fight with the vendors, because they do not like it at all if we tell them what to do, especially clang folks)
Well, there is Annex K which is based on a previous Microsoft effort. Almost universally it is considered terrible and few people implemented it.
Immediately what I thought of when I saw /microsoft.
Not all of the APIs were brain-dead. They just ignored all previous developments and in the proposal they didn't even remove the C++-related language.
The C charter has a rule of "no invention".
Anything needs to be demonstrated and used in practice before being included in the standard. The standard is only meant to codify existing practices, not introduce new ideas.
It's up to compiler developers to ship first, standardize later.
That produces a bit of a chicken and egg probablem for a stdlib overhaul. Compilers and libc implementations don't have a strong reason to implement safer APIs, because if it is non-standard then projects that want to be portable won't use it , but it won't get standardized unless they do add safer APIs.
So the best hope is probably for a third party library that has safet APIs to get popular enough that it becomes a de facto standard.
I think the real failing is that new language features then must be prototyped by people who have a background in compilers. That's a very small subset of the overall C community.
I don't have any clue how to patch clang's front end. I'm not a language or compiler person. I just want to make stuff better. There needs to be a playground for people like me, and hopefully lib0xc can be that playground.
There are only two kinds of standards: ones that prioritize stability and backwards compatibility over usefulness and security, and ones nobody uses.
C and POSIX aren't related to C++ at all.
This is great. Other things needed for a great C development environment are a standardized build process plus build tools and a standardized packaging system.
I think most people who are into c that I’ve met quite like header only libraries. Copy paste as a package manager does have its benefits.
I mostly just install a -dev package on Linux and I am done.
Because that's precisely what is needed: an easy way to ship dependency malware like npn, pip, cargo, etc.
Like it or not, having a little bit of friction prevents pulling in packages with thousands of transitive dependencies.
Thanks! I agree, a better build story for C projects is desperately needed.
Unfortunate naming. I thought this was about https://libxc.gitlab.io/ but there's an extra '0' in the name here.
The title looks very promising. I’ve added this library to my to-do list to take a deeper look at it. Using this standart library within restricted safe subset of C++ can be a strong opponent for Zig (at least for myself).
Haven't really verified that it works with C++, but I tried my best to guard the stuff I knew would be problematic with #if __cplusplus. Happy to have a PR that makes C++ happier with it.
Thanks for the reply. Noted :)
I truly hope something like this catches on. There is so much low hanging fruit in both the C and C++ standard libraries. Spatial memory could be 90% solved in both languages by mandating the use of safe interfaces.
I'm curious – is MSFT using this in production, or is this a "20% time" project? I'm not sure MSVC could compile the GNU extensions used.
> I'm not sure MSVC could compile the GNU extensions used.
The supported platforms only list Linux and Mac. Notably missing from that list is any of the BSDs (not counting what may or may not remain of BSD under the Mac hood).
Author here. It is not currently in production, but it is part of a project in Azure which will go to production at some point. I'm actually leaving Microsoft next week and fully intend to keep working on it if I can reach an agreement to do so with my new employer.
Look, C is so simple, you can implement a compiler in under 10K lines of code on a new platform. And then spend the your life dodging bugs and come up with a zoo of bizarre macros to keep you safe.
Interesting. I'll be studying this later tonight so I can apply it to my C projects. Especially clang's -fbounds-safety.
Interesting that a project from Microsoft doesn't support MSVC or Windows.
I suspect in 20 years Windows will be a Linux distribution with a compatibility layer.
People say that kind of thing on HN every now and then. I have no idea why this idea is around, it's a complete fantasy in my opinion. I say this as someone who mostly uses Linux.
> I have no idea why this idea is around,
To the best of my fallible knowledge, the notion was first popularized via <http://esr.ibiblio.org/?p=8764>.
Is there anything in here for something like a "slice" or dynamically sized array that carries its length along with it?
Just use compiler option -std=c++20 and use std::span. Don't try reinventing it in C.
If someone needs more than C provides, why on earth would they choose C++?
No rational person is going to want to have to deal with 10x the number of foot guns.
Literally anything when moving from C is better than C++.
Switching to C++ is relatively easy in an existing codebase. It's in many cases as simple as renaming a file from .c to .cpp. But for writing something from scratch it's better to use Rust.
Renaming c. to .cpp may work with ancient c89 code, but not with anything remotely modern. But while the code then is technically C++, it is not better. I still prefer C for new projects to any other language, because I value short compilation time and reduced complexity. For me, this translates in higher productivity and more fun. With modern tooling, also most C issues are detected early.
How things change. Imagine Microsoft circa 2000 publishing an MIT-licensed source code library targeting "the competition"'s compilers, and including some light humour ("Embiggen C's Pit of Success") in the docs.
Quick question for those who've tried it — does this play with existing C codebases incrementally, or is it more of a "new project only" situation? The README didn't make that obvious to me.
It's designed to be incremental. For example, you can do a search for `sprintf` and replace it with `ssprintf`. The function signature is the same. Any instance of printing to a character array just works. Think of the APIs as "the stuff you usually do by hand, but safer".
If you get compiler errors, it means you were printing to a heap-allocated buffer (or a buffer whose bounds you did not know), and you should be propagating bounds and using `snprintf`.
Integer conversion is the same way. If you have something like
int v1; uint64_t v2;
<stuff happens>
v2 = (uint64_t)v1;
Then you can replace it with
v2 = __cast_signed_unsigned(uint64_t, v1);
and you'll get a runtime trap when v1 is a negative value, meaning you can both enable -Wint-conversion and have defined behavior for when the value in a certain integer type is not representable in another.
I thought Microsoft adopted Rust. Are they back pedaling?
Microsoft supports memory safety. Rust is 100% the direction for new projects. But there are existing C codebases that are unlikely to be entirely rewritten in a memory-safe language for various reasons. Such projects can significantly benefit from incremental improvements in memory safety.
This is very cool!
This stuff is amongst my favourite type of engineering.
Practical. Useful. Not sexy. (I am only one of those.)
Bravo!
Hell, I aim to be all of those.
It's just an excuse not to use safe languages. And adopting it isn't that easy - one need to learn how the new library works or even rewrite old working and tested code with it.