porting uml to Hurd

uml-devel Constantine Kousoulos 3 participants 3 messages Dec 14, 2006 – Dec 21, 2006 User-mode Linux (UML)

Share & tools

Constantine Kousoulos

author

2006-12-14 12:25 UTC #0 lore raw

Hello!

i'm interested in porting uml to GNU/Hurd. Since uml is just "another process" (or many processes), in theory, it could be ported to Hurd's user space. Uml's architecture would remain i386.

Can you supply me with more info (docs, urls, directions to source code, etc) on what needs to be done in order to accomplish the port? Technical details would really help me discover if uml can actually be ported to the Hurd.

Thanks,
Constantine

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Jeff Dike

2006-12-20 18:13 UTC #1 lore raw

Quoted 4 lines · Constantine Kousoulos

Can you supply me with more info (docs, urls, directions to source code, etc) on what needs to be done in order to accomplish the port? Technical details would really help me discover if uml can actually be ported to the Hurd.

There's no documentation to speak of wrt to an OS port. There is some outdated architecture porting information at
http://user-mode-linux.sourceforge.net/arch-port.html, some of which is relevant.

Here's a little list off the top of my head, for a skas-like mode -
interception and nullification of another process system calls ability to change memory mappings of another process interception and nullification of signals received by another
process

If you can handle those, you are probably in good shape.

If you can't do the second two, then a tt-like mode might be possible, where you map the UML kernel into every UML process address space. Then those become trivial since "the other process" is a UML process. However, in this case, there are other nasty tricks needed, like making sure that the host processes have different address spaces, but the UML portions are shared. The kernel data obviously needs to be shared across all processes.

Look at the interface declared in arch/um/include/os.h and defined in arch/um/os-Linux/. It's probably unnecessarily wide at this point, and contains Linux-isms which shouldn't be there, but that's my current stab at an OS abstraction interface.

Jeff

Signature · 11 lines

Work email - jdike at linux dot intel dot com

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
User-mode-linux-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Erik Paulson

2006-12-21 19:54 UTC #2 lore raw

Quoted 1 line · Jeff Dike

<...>

Quoted 11 lines

There's no documentation to speak of wrt to an OS port. There is some outdated architecture porting information at
http://user-mode-linux.sourceforge.net/arch-port.html, some of which is relevant.
Here's a little list off the top of my head, for a skas-like mode -
interception and nullification of another process system calls ability to change memory mappings of another process interception and nullification of signals received by another
process

Quoted 3 lines

If you can handle those, you are probably in good shape.
If you can't do the second two, then a tt-like mode might be possible,

What are the options for an OS that can't intercept system calls? On Darwin (the BSD part of MacOS X) there is no PTRACE_SYSCALL - Apple's stance on intercepting system calls is 'use dynamic loader hacks'. They also appear to use sysenter for system calls (and I think int80 in some places too)

Modifying memory is easy on MacOS, in fact it's a lot easier than on Linux.

For system calls, past apporaches have been to patch the MacOS host kernel:
http://www.citi.umich.edu/u/provos/systrace/macosx.html

I suppose a MacOS UML could do some binary translation looking for Linux system calls too.

For a good overview of some of the MacOS functionality, see:
http://uninformed.org/?v=4&a=3&t=sumry

-Erik