DragonFlyBSD

So a little history:

-On the initial import of amd64/x86_64, the functions existed.
-In July 2006 they were removed in 635872e06f82693646a8d979c423433fa839a656
because i386 didn't have them and they are marked obsolete.
-Then in January 2007, they were added for i386 in
e274265043a09fe420ae2182ba74ee43fcfbcbd5.

So that's why they're there for i386 and not x86_64. The question is, do we
want them or not? They are completely removed from IEEE Std 1003.1-2008, but
there's still 3rd-party software that's using those functions. My vote is to
keep them and make them available to x86_64.

The original amd64 getcontext() we had in libc assumed it was a syscall.

The i386 version is libc only. So just bringing back the amd64 one we had won't
work.

Someone will have to write a x86_64 libc only getcontext.S.

Maybe a starting point?

http://leaf.dragonflybsd.org/~alexh/0001-getcontext-x86_64-implementation.patch

We need to bump this up in priority. On x86_64, every ruby package using ruby19-base is failing due to missing getcontext and makecontext on pkgsrc-current.

I dug up my seemingly working implementation of getcontext for x86_64 last weekend. The email to submit@ seems to have been lost. Anyway, the patch is available at

http://leaf.dragonflybsd.org/~profmakx/getcontext.patch

or in my git repository a3cb33319377734976008e240fded1e69ea07272,

I tested it with w3m, which uses boehm-gc, but had some issues with vkernels that stopped me from testing the other programs mentioned here. Heres a patch, feedback is appreciated.

new commit in my git repository ref 1e29c3ebaaf978d191d581087167cd268268c106

I have a patch in my git repository that adds getcontext, setcontext, makecontext and swapcontext. However, testing this with puffs still segfaults the machine so there are issues. I welcome any testing/feedback though.

commit 5f643fb7f2f5b3020964cfd507b9e8120c19535c in git://leaf.dragonflybsd.org/~profmakx/dragonfly.git now contains a patch that adds the functions in the context-family. I tested puffs, w3m and gale (compiling and running) and up to now i did not have any problems.

Would people please pull this patch and test a bit more intensively than I have time for at the moment?