Re: lrmi vs new kernels vs libx86
Sorry for re-triggering an old thread, On Sun, 8 Mar 2009 21:07:29 +0100, David Paleino wrote: > On Sun, 8 Mar 2009 16:49:33 +0100, Evgeni Golov wrote: > > > [..] > > David, is there any chance that libx86 will be updated someday? Esp > > because upstream of v86d has an updated 0.10 in his git at > > http://repo.or.cz/w/v86d.git and Debian's v86d is not using it in > > favour of not build duplicate code. > > > > All other (incl David), is there any interest in forking libx86 and > > using it globally instead of fixing that ftbfs 7 times? > > I prepared a package with an updated LRMI: > > http://alioth.debian.org/~hanska-guest/apt/unstable/libx86_1.1+ds1-3.dsc Which now FTBFS on anything non-x86. Grin. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=533259 Matthew, could you have a look at that package again? lrmi.h (from LRMI 0.10) encloses everything into #if defined(__i386__) && (defined(__linux__) || defined(__NetBSD__) \ || defined(__FreeBSD__) || defined(__OpenBSD__)) thus the cited bugreport. *But* even removing that #if, it would FTBFS, because we're not building x86-common.c anymore (since it has been merged into lrmi.c), but x86emu used symbols from there! So the current build only works on i386. I don't know what exactly the x86-specific code is, I tried some patching but nothing came out of it. Really, we should split a x86-common.c out of the new lrmi.c, but I don't even know where to start. Any help is greatly apreciated :) Kindly, David -- . ''`. Debian maintainer | http://wiki.debian.org/DavidPaleino : :' : Linuxer #334216 --|-- http://www.hanskalabs.net/ `. `'` GPG: 1392B174 | http://snipr.com/qa_page `- 2BAB C625 4E66 E7B8 450A C3E1 E6AA 9017 1392 B174 signature.asc Description: PGP signature
Re: lrmi vs new kernels vs libx86
Hi David, On Sun, 8 Mar 2009 21:07:29 +0100 David Paleino wrote: > I prepared a package with an updated LRMI: > > http://alioth.debian.org/~hanska-guest/apt/unstable/libx86_1.1+ds1-3.dsc > > All the people involved: would you please testbuild your packages against it? > Thank you. Sorry that I didn't write yet, was somehow busy and forgot about this issue. Now I have tested -3 from your home and it works fine for me with v86d. Still no luck with including it into the initrd, but that is not your fault (it also happens when I build with the internal copy of lrmi/x86emu). Matthew, any plans on releasing a new version with David's patches upstream? Regards Evgeni pgplrvDbUKvSA.pgp Description: PGP signature
Re: lrmi vs new kernels vs libx86
Hi!
On Sun, 2009-03-08 at 16:49:33 +0100, Evgeni Golov wrote:
> dear maintainers of packages that contain lrmi.{c,h},
>
> today Lucas has reported #518725 - atitvout FTBFS because of missing
> *_MASK defines.
> Seeing that bug and remembering fun with lrmi myself, I thought I can
> have a look how many other packages will FTBFS.
> The following packages contain lrmi.{c,h}, and the ones with the *
> FTBFS:
> read-edid*
Some time ago I filed a bug with a patch to switch it to use libx86.
.
> svgalib*
And I ported svgalib to libx86 long time ago, so it's not using the
embedded lrmi. It also builds locally, do you have the failing logs?
regards,
guillem
--
To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: lrmi vs new kernels vs libx86
On Sun, Mar 08, 2009 at 09:07:29PM +0100, David Paleino wrote: > Matthew: since you're libx86 upstream, you might be interested in the contents > of debian/patches, I'll remove those as soon as you release a new version > (also, please drop debian/ from upstream tarballs) Thanks, I'll take a look at those. -- Matthew Garrett | mj...@srcf.ucam.org -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Using libx86/x86emu also on x86 (was: Re: lrmi vs new kernels vs libx86)
On Sun, 8 Mar 2009 20:34:12 +0100, Evgeni Golov wrote: > On Sun, 8 Mar 2009 19:24:22 +0100 David Paleino wrote: > > > [.. regarding libx86 ..] > > May I throw in the fact that one sometimes wants x86emu on i386? Yes, I *do* remember your bugreport about it ;) (apropos: sorry if I didn't even reply to it, it just fell off my TODO...) > e.g. v86d works for me when built with x86emu, but not with lrmi, dunno who > to blame though (libx86 built for x86emu is okay too). Maybe v86d is to blame? IMVHO it's *weird* emulating x86 on an x86 arch... anyhow, I'll have a look next weekend (don't have many chances to do Debian-work during the week). David -- . ''`. Debian maintainer | http://wiki.debian.org/DavidPaleino : :' : Linuxer #334216 --|-- http://www.hanskalabs.net/ `. `'` GPG: 1392B174 | http://snipr.com/qa_page `- 2BAB C625 4E66 E7B8 450A C3E1 E6AA 9017 1392 B174 signature.asc Description: PGP signature
Re: lrmi vs new kernels vs libx86
On Sun, 8 Mar 2009 16:49:33 +0100, Evgeni Golov wrote: > [..] > David, is there any chance that libx86 will be updated someday? Esp > because upstream of v86d has an updated 0.10 in his git at > http://repo.or.cz/w/v86d.git and Debian's v86d is not using it in > favour of not build duplicate code. > > All other (incl David), is there any interest in forking libx86 and > using it globally instead of fixing that ftbfs 7 times? I prepared a package with an updated LRMI: http://alioth.debian.org/~hanska-guest/apt/unstable/libx86_1.1+ds1-3.dsc All the people involved: would you please testbuild your packages against it? Thank you. Matthew: since you're libx86 upstream, you might be interested in the contents of debian/patches, I'll remove those as soon as you release a new version (also, please drop debian/ from upstream tarballs) Kindly, David -- . ''`. Debian maintainer | http://wiki.debian.org/DavidPaleino : :' : Linuxer #334216 --|-- http://www.hanskalabs.net/ `. `'` GPG: 1392B174 | http://snipr.com/qa_page `- 2BAB C625 4E66 E7B8 450A C3E1 E6AA 9017 1392 B174 signature.asc Description: PGP signature
Re: lrmi vs new kernels vs libx86
On Sun, 8 Mar 2009 19:24:22 +0100 David Paleino wrote: > On Sun, 8 Mar 2009 19:19:47 +0100, David Paleino wrote: > > > On Sun, 08 Mar 2009 18:12:01 +, Matthew Garrett wrote: > > > > > David Paleino wrote: > > > > > > >I'll start work on liblrmi (i.e. ITP, making it, buildtesting relevant > > > >packages, [..]) if some interest is shown. > > > > > > I'm not sure what the benefit would be over libx86? > > > > IMVHO code bundling is *never* good. > > Or, since the code would only be bundled in *one* package, we could just skip > this "issue" and do everything in libx86. May I throw in the fact that one sometimes wants x86emu on i386? e.g. v86d works for me when built with x86emu, but not with lrmi, dunno who to blame though (libx86 built for x86emu is okay too). -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: lrmi vs new kernels vs libx86
On Sun, 8 Mar 2009 19:19:47 +0100, David Paleino wrote: > On Sun, 08 Mar 2009 18:12:01 +, Matthew Garrett wrote: > > > David Paleino wrote: > > > > >I'll start work on liblrmi (i.e. ITP, making it, buildtesting relevant > > >packages, [..]) if some interest is shown. > > > > I'm not sure what the benefit would be over libx86? > > IMVHO code bundling is *never* good. Or, since the code would only be bundled in *one* package, we could just skip this "issue" and do everything in libx86. David -- . ''`. Debian maintainer | http://wiki.debian.org/DavidPaleino : :' : Linuxer #334216 --|-- http://www.hanskalabs.net/ `. `'` GPG: 1392B174 | http://snipr.com/qa_page `- 2BAB C625 4E66 E7B8 450A C3E1 E6AA 9017 1392 B174 signature.asc Description: PGP signature
Re: lrmi vs new kernels vs libx86
Please don't break threads ;) On Sun, 08 Mar 2009 18:12:01 +, Matthew Garrett wrote: > David Paleino wrote: > > >I'll start work on liblrmi (i.e. ITP, making it, buildtesting relevant > >packages, [..]) if some interest is shown. > > I'm not sure what the benefit would be over libx86? IMVHO code bundling is *never* good. > liblrmi would leave you stuck with x86, whereas using libx86 means that much > of the code will also work on amd64. Ok, I goofed there :) Honestly, liblrmi0 would only be a dependency of libx86 on x86, while it would use x86emu on other arches. Other packages would then depend on libx86, stop. Am I totally wrong? Kindly, David -- . ''`. Debian maintainer | http://wiki.debian.org/DavidPaleino : :' : Linuxer #334216 --|-- http://www.hanskalabs.net/ `. `'` GPG: 1392B174 | http://snipr.com/qa_page `- 2BAB C625 4E66 E7B8 450A C3E1 E6AA 9017 1392 B174 signature.asc Description: PGP signature
Re: lrmi vs new kernels vs libx86
David Paleino wrote: >I'll start work on liblrmi (i.e. ITP, making it, buildtesting relevant >packages, [..]) if some interest is shown. I'm not sure what the benefit would be over libx86? liblrmi would leave you stuck with x86, whereas using libx86 means that much of the code will also work on amd64. -- Matthew Garrett | mjg59-chiark.mail.debian.de...@srcf.ucam.org -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: lrmi vs new kernels vs libx86
Evgeni Golov wrote: >On Sun, 08 Mar 2009 17:17:44 + Matthew Garrett wrote: > >> >All other (incl David), is there any interest in forking libx86 and >> >using it globally instead of fixing that ftbfs 7 times? >> >> You could just send a patch to libx86 upstream, you know... > >The patch for what? >For the *_MASK defines FTBFS? Thats the task of the maintainer. >For updating to lrmi 0.10? Maybe. >For incorporating the fixes from v86d? Wasnt able to incorporate them >to the actual libx86 as v86d used 0.10 as base. Whichever of the above you were planning to put in any forked version and are applicable to distributions other than Debian. I'm happy to merge stuff. -- Matthew Garrett | mjg59-chiark.mail.debian.de...@srcf.ucam.org -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: lrmi vs new kernels vs libx86
On Sun, 8 Mar 2009 16:49:33 +0100, Evgeni Golov wrote:
> Dear debian-devel,
> dear maintainers of packages that contain lrmi.{c,h},
Hello Evgeni,
thanks for this heads up.
> [..]
> The following packages contain lrmi.{c,h}, [..]
> But actually we should stop duplicating code (esp. OLD code - some
> packages have lrmi.c from lrmi 0.6, 0.10 is latest) and using libx86
> only.
> If only libx86 would have latest lrmi code...
>
> David, is there any chance that libx86 will be updated someday? Esp
> because upstream of v86d has an updated 0.10 in his git at
> http://repo.or.cz/w/v86d.git and Debian's v86d is not using it in
> favour of not build duplicate code.
>
> All other (incl David), is there any interest in forking libx86 and
> using it globally instead of fixing that ftbfs 7 times?
My proposal is: is there any interest in a separate liblrmi package, which
others package would Depend on?
Sure, that would need some patching of those 10 packages (to use the
system-wide instead of the bundled one), but I believe it's a saner solution --
and it avoids bundled code at all.
I'll start work on liblrmi (i.e. ITP, making it, buildtesting relevant
packages, [..]) if some interest is shown.
Kindly,
David
--
. ''`. Debian maintainer | http://wiki.debian.org/DavidPaleino
: :' : Linuxer #334216 --|-- http://www.hanskalabs.net/
`. `'` GPG: 1392B174 | http://snipr.com/qa_page
`- 2BAB C625 4E66 E7B8 450A C3E1 E6AA 9017 1392 B174
signature.asc
Description: PGP signature
Re: lrmi vs new kernels vs libx86
On Sun, 08 Mar 2009 17:17:44 + Matthew Garrett wrote: > >All other (incl David), is there any interest in forking libx86 and > >using it globally instead of fixing that ftbfs 7 times? > > You could just send a patch to libx86 upstream, you know... The patch for what? For the *_MASK defines FTBFS? Thats the task of the maintainer. For updating to lrmi 0.10? Maybe. For incorporating the fixes from v86d? Wasnt able to incorporate them to the actual libx86 as v86d used 0.10 as base. Regards -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: lrmi vs new kernels vs libx86
Evgeni Golov wrote: >All other (incl David), is there any interest in forking libx86 and >using it globally instead of fixing that ftbfs 7 times? You could just send a patch to libx86 upstream, you know... -- Matthew Garrett | mjg59-chiark.mail.debian.de...@srcf.ucam.org -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
