Restructuring druntime

Feb 25, 2012

Johannes Pfau

Feb 26, 2012

Kagamin

Feb 26, 2012

Feb 26, 2012

Feb 26, 2012

Feb 27, 2012

Feb 27, 2012

Feb 26, 2012

Feb 26, 2012

February 25, 2012

Restructuring druntime

Posted by Johannes Pfau

Permalink

Johannes Pfau

Permalink

As recently discussed in a pull request it would be great if we could make it easier to port druntime to different architectures. Special casing every C library, every architecture and every OS using version blocks could lead to difficult to maintain code, so we need a better solution.

As far as I can see, the biggest differences are caused by different C libraries, not by different architectures (except 32bit vs 64 bit differences). For example glibc headers seem to be very similar for arm and x86, but bionic vs. glibc brings more differences. Bionic initially didn't support anything wchar_t related, so core.stdc.wchar_ and core.stdc.wctype couldn't work with android at all. And C libraries vary even more in the subset of posix functionality they support.

We need a directory scheme to support
* Different C libraries
  * on different OS (glibc/bsd glibc/linux)
  * on the same OS (glibc on linux, bionic on linux, uclibc on
    linux, ...)
* Different architectures (ARM, X86, MIPS, PPC, SH4, ...)

I propose the following directory layout, but of course better solutions are appreciated:

druntime
|-core
|---stdc
|---sync
|---sys
|-----posix
|-gc
|-gcstub
|-rt
|-gcc
|---arch
|-----glibc
|-------core
|---------stdc
|---------sync
|---------sys
|-----------posix
|-----bionic
|-------stdc
|-------sync
|-------core
|---------sys
|-----------posix
|-----newlib
|-------stdc
|-------sync
|-------core
|---------sys
|-----------posix

Code in the gc, gcstub and rt directories shouldn't depend on special C libraries, small differences can be solved using version() blocks.

Every C library get's it's  own bindings in the gcc.arch package. I'm not sure if those bindings should only provide functionality included in core, or whether these bindings should include additional, C library specific bindings.

The files in core.* would reference the corresponding files in gcc.arch.core.*

The downside of that approach is that we'll have some code duplication, especially for similar C libraries. And merging changes is therefore not that easy as well. However, as the C and Posix headers don't change often, that's probably neglectable.

Here are some examples for that directory scheme:

core/stdc/config.d (Should be shared with dmd druntime?)
-----------------------------------------------
enum LibC
{
    GlibC,
    uClibc,
    dietlibc,
    newlib,
    dmc, //digitalmars, windows
    msvcrt,
    ...,
    unknown
}

//Need some way to define this correctly
enum libC = LibC.GlibC;
enum string libCVersion = ""; //Free form version number
-----------------------------------------------

core/stdc/math.d
-----------------------------------------------
import core.stdc.config;

static if(libC == LibC.GlibC)
{
    public import gcc.arch.glibc.core.stdc.math;
}
else
{
    static assert(false, "C library '" ~ libC ~ "' not supported in " ~
    __FILE__);
}

//Maybe put generic parts here?
-----------------------------------------------

core/sys/posix/dirent.d
-----------------------------------------------
import core.stdc.config;

static if(libC == LibC.GlibC)
{
    /*
     * I admit this looks horrible. We can probably remove the core
     * part, maybe even gcc, so arch.glibc.sys.posix.dirent?
     * or glibc[.core].sys.posix.dirent?
     */
    public import gcc.arch.glibc.core.sys.posix.dirent;
}
else static if(libC == LibC.GlibC)
{
    public import gcc.arch.glibc.core.sys.posix.dirent;
}
else
{
    static assert(false, "C library '" ~ libC ~ "' not supported in " ~
    __FILE__);
}
-----------------------------------------------

Maybe we could use pragma(msg) warnings instead of static asserts, so
it'd be possible to import the modules in any case and use
traits(__compiles) to check for specific symbols/functions.

On Sunday, 26 February 2012 at 13:06:58 UTC, Johannes Pfau wrote: > still have to make sure core.* works. That's an issue. Given the c libraries incompatibilities, what functionality the core.stdc should provide? What to do with functions like backtrace?

Am Sun, 26 Feb 2012 19:30:03 +0100 schrieb "Kagamin" <spam@here.lot>: > On Sunday, 26 February 2012 at 13:06:58 UTC, Johannes Pfau wrote: > > still have to make sure core.* works. > > That's an issue. Given the c libraries incompatibilities, what functionality the core.stdc should provide? What to do with functions like backtrace? Fortunately backtrace is not part of core.stdc. core.stdc only includes functions which are described by the C(99?) standard and C libraries usually provide most of that functionality. It's more difficult for core.sys.posix as 'lightweight' libcs often decide to not implement some posix modules. If some of the basic functionality isn't supported, there's probably nothing we can do. For the special backtrace & backtrace_symbols case we should get rid of backtrace and use gcc functionality (We already have access to that in gcc.unwind). backtrace_symbols can be implemented using dladdr. dladdr isn't standard C/Posix either, but it's supported by more C libraries. In case dl_addr isn't available, we can always just print the addresses instead of the function names. addr2line works better than backtrace_symbols / dladdr anyway. I had a look at glibc and bionic though and those C libraries provide much more functionality than what's currently included in druntime. Creating complete bindings is probably a waste of time as those 'extra' functions are not portable anyway. Maybe the simplest solution is to add a 'sublibrary' to druntime, libdc or something, then store the normal core.stdc/core.sys modules in a subdir for every C library and compile in only the correct modules using the configure script: druntime |-core |---atomic.d (generic files) |---bitop.d |---[...] |-gc |-gcstub |-rt |-libdc |---glibc |-----core |-------sys |-------stdc |---bionic |-----core |-------sys |-------stdc |---newlib |-----core |-------sys |-------stdc The files in libc/glibc/core/stdc would be in the core.stdc package (not glib.core.stdc), so we just have to compile the correct subdirectory into druntime.

Kagamin, el 26 de febrero a las 19:30 me escribiste: > On Sunday, 26 February 2012 at 13:06:58 UTC, Johannes Pfau wrote: > >still have to make sure core.* works. > > That's an issue. Given the c libraries incompatibilities, what functionality the core.stdc should provide? What to do with functions like backtrace? Backtrace is not stdc :) I'm sorry to interject when I'm not very involved, but libraries implementing the C standard library should not be too different from each other. Maybe some miss some features, but the things they have, should be pretty much the same. I think core.stdc should publicly import the local compiler's stdc. Any incompatibilities among different libraries should be handled by the users of core.stdc, not core.stdc itself, which is supposed to be just a thin layer to get what the current C environment have. People should be using std stuff, not core.stdc, most of the time, specially if they intend to write portable code. -- Leandro Lucarella (AKA luca) http://llucax.com.ar/ ---------------------------------------------------------------------- GPG Key: 5F5A8D05 (F8CD F9A7 BF00 5431 4145 104C 949E BFB6 5F5A 8D05) ---------------------------------------------------------------------- Si por el chancho fuera, se autocomería con chimichurri Worshestershire!

Am Sun, 26 Feb 2012 18:00:49 -0300 schrieb Leandro Lucarella <luca@llucax.com.ar>: > Kagamin, el 26 de febrero a las 19:30 me escribiste: > > On Sunday, 26 February 2012 at 13:06:58 UTC, Johannes Pfau wrote: > > >still have to make sure core.* works. > > > > That's an issue. Given the c libraries incompatibilities, what functionality the core.stdc should provide? What to do with functions like backtrace? > > Backtrace is not stdc :) > > I'm sorry to interject when I'm not very involved, but libraries implementing the C standard library should not be too different from each other. Maybe some miss some features, but the things they have, should be pretty much the same. I mostly agree. The API of the C libraries is the same, what's different is ABI: The C/Posix standards don't define layouts for some structs, for example the 'stat' struct is different between most C libraries. That's the reason we need different bindings for different C libraries. > > I think core.stdc should publicly import the local compiler's stdc. Any incompatibilities among different libraries should be handled by the users of core.stdc, not core.stdc itself, which is supposed to be just a thin layer to get what the current C environment have. People should be using std stuff, not core.stdc, most of the time, specially if they intend to write portable code. > I agree. The question is how that 'publicly import' should actually work. Using real 'public import' statements with static if/version blocks adds some additional clutter. Placing bindings for different C libraries in different directories and just compiling in the correct folder seems to be the easiest solution.

On Sunday, 26 February 2012 at 20:54:27 UTC, Johannes Pfau wrote: > The files in libc/glibc/core/stdc would be in the core.stdc package > (not glib.core.stdc), so we just have to compile the correct > subdirectory into druntime. The structure seems good, but I still suggest to go through deimos project. One can start with partial bindings, which get druntime working, but eventually they should become complete, if someone will have time for it. Does deimos accept partial bindings? Should be a problem, I believe. If the deimos bindings will be worked on later, we'll get them in core* for free. libc/glibc/core/stdc will just reimport the deimos code in a structure adopted for druntime.

Forums