Date: Mon, 5 Jun 2017 00:44:32 +0800 From: blubee blubeeme <gurenchan@gmail.com> To: Hans Petter Selasky <hps@selasky.org> Cc: Julian Elischer <julian@elischer.org>, FreeBSD current <freebsd-current@freebsd.org>, freebsd-ports@freebsd.org Subject: Re: [Help] Linux low level data structures < - > FreeBSD low level data structures Message-ID: <CALM2mEnSpYYWKLJ45cjnMQ7V6vZGZveiXAB2sq5RzReZm2x7bQ@mail.gmail.com> In-Reply-To: <dd404fe5-9e7f-62b7-f111-03d9c57d45e0@selasky.org> References: <CALM2mE=NkWFJZWeCYM3LWcVWeGSyeomqMLqvDm7XowXsRoPS1Q@mail.gmail.com> <16f2c369-2758-9bd8-9dcb-5e1b400d2336@elischer.org> <CALM2mEnQJn5s4CggtbBPZr9MewYsdXtR=J24ZoFvpnW8u=wJdw@mail.gmail.com> <dd404fe5-9e7f-62b7-f111-03d9c57d45e0@selasky.org>
next in thread | previous in thread | raw e-mail | index | archive | help
=EF=BC=A0Hans I've noticed these things about Linux. A lot of stuff relies on knowing the internals of the data structures instead of working with the API but I don't want to start all that stuff up. The goal is to lift up FreeBSD w/o tearing down Linux and as u know needing to be glued to the Linux data structures, while they always seem to be in flux is one reason why I want to steer clear from the linuxkpi stuff. I cam across the webcamd thing a while ago, it was written when Linux was not moving so fast if I remember correctly. This is a big challenge but I think once it's up and running it will bring some additional resources to FreeBSD that can bring the desktop world up to snuff with it's server acumen. Thanks for the tips though! Best, Owen On Mon, Jun 5, 2017 at 12:07 AM, Hans Petter Selasky <hps@selasky.org> wrote: > Hi Owen, > > The most comprehensive open-source wrappers for Linux kernel APIs in the > FreeBSD kernel is found at: > > https://github.com/FreeBSDDesktop/freebsd-base-graphics/ > tree/drm-next/sys/compat/linuxkpi > > The most comprehensive open-source wrappers for Linux kernel APIs in > FreeBSD user-space is found at: > > http://www.freshports.org/multimedia/webcamd > > I recommend you start with webcamd. Download the source code and look at > the sources. See if there is something you think you can improve. It is > much easier to debug using GDB and if something goes wrong it can easily = be > restarted. > > When you have a working solution for webcamd, try to use the same solutio= n > with the kernel. > > There is no simple 1:1 mapping for Linux APIs to FreeBSD APIs. Take for > example linux/list.h . You might think that a list can be re-implemented > and mapped to our sys/queue.h. But oh-no. Linux kernel code depends on al= l > the characteristics of list.h, how the structure members are stored and > named, and how foreach() macros complete with NULL or non-NULL in the > iterator. Further you'll find resistance among the Linux kernel developer= s > to support compilers different from GCC. I recently found a piece of code > in a header file, which works with GCC and causes a panic() when compiled > with clang. I reported it to one of the Linux kernel team members and the= y > didn't care about it. Even if you get everything compiling it doesn't it > will work :-( > > --HPS > > > On 06/04/17 16:35, blubee blubeeme wrote: > >> Hi Julian >> >> My goals are to port the Linux graphics stack over to FreeBSD w/o relyin= g >> too heavily on the linuxkpi stuff. That's cool for a lot of use cases bu= t >> it just seems a bit too brittle. >> >> It is a very large I understand the task will not be easy but I am willi= ng >> to do the work, even from scratch if necessary although some help would = be >> appreciated. >> >> I've been watching the Linux DRM project grow and while the top levels h= as >> changed, it's been a very long time since the actual low level stuff has >> been changed. Most of the diffs have shown changes in the >> [linux/driver/gpu/drm] layer which relies heavily on the >> [linux/include/drm] that does a lot of the heavy lifting here's a link t= o >> the latest version files: >> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin >> ux.git/tree/include/drm?h=3Dv4.12-rc3 >> >> here's a diff of the latest version of the [linux/driver/gpu/drm] : >> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin >> ux.git/diff/?id=3Dv4.12-rc3&id2=3Dv4.12-rc2&dt=3D2 >> >> The diffs in the drivers change constantly but the lower level stuff >> hasn't >> changed as much. >> >> Doing some of the lower level translation to native FreeBSD style data >> structures then the upper part could be easily migrated, even with >> something as using AST to translate the headers to their FreeBSD >> equivalent >> without worrying about inadvertently breaking something or having major >> diffs that needs people to actively look at maintaining. >> >> That's a high level overview of my plan and what I'd like to achieve. Wi= ll >> it be easy, most likely not but once it's done FreeBSD will be just fine= . >> >> Hope that helps clarify things for anyone who is interested. Any >> assistance >> would be greatly appreciated. >> >> Best, >> Owen >> >> On Sun, Jun 4, 2017 at 9:26 PM, Julian Elischer <julian@elischer.org> >> wrote: >> >> On 4/6/17 7:07 pm, blubee blubeeme wrote: >>> >>> Hello >>>> >>>> Is there anyone on either of these lists that have experience with bot= h >>>> linux low level data structures and their equivalents on FreeBSD? >>>> >>>> For instance the linux header file: >>>> <linux/agp_backend.h> >>>> >>>> which includes the header file: >>>> <linux/list.h> >>>> >>>> Then looking at that file: >>>> <linux/types.h> >>>> <linux/stddef.h> >>>> <linux/poison.h> >>>> <linux/const.h> >>>> <linux/kernel.h> >>>> >>>> >>> You are going to have to be a lot more specific about this. >>> I have worked in several places where they use s shim layer to make Lin= ux >>> basic services work on freeBSD. >>> usually a mix of functions, macros and inlines. >>> However you need to narrow down your questions a bit as the POSSIBLE >>> scope >>> of your question is too large for anyone to attempt an answer. >>> >>> Remember that both systems are POSIX inspired so outside the kernel the= re >>> are many more simlarities than one might be led to expect, >>> but you need to be way more specific. >>> It's even possible to write kernel code to run on both, but it is usual= ly >>> domain specific. >>> >>> >>> >>> I'll be doing a lot of work trying to find these FreeBSD equivalent of >>>> these types of files to port some code. >>>> >>>> Does anyone here have experience with something like this? Is there an= y >>>> other projects that maps these low level data structures from >>>> Linux <-> FreeBSD, etc? >>>> >>>> Best, >>>> Owen >>>> >>>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CALM2mEnSpYYWKLJ45cjnMQ7V6vZGZveiXAB2sq5RzReZm2x7bQ>