[9fans] gsoc linuxemu project help

[9fans] gsoc linuxemu project help

[Zhao Shuai]

[-- Attachment #1: Type: text/plain, Size: 339 bytes --]

Hi all,
I am interested in the GSoc project: Linuxemu improvement.
But the project description seems too general. I don't know from
which aspect can we make improvement. There is a TODO file in
the linuxemu3 source directory and I find TLS, futex, VDSO listed.
Can these TODOs make a gsoc project?
Thanks.



--

Regards,
Zhao

[-- Attachment #2: Type: text/html, Size: 549 bytes --]

Re: [9fans] gsoc linuxemu project help

[Anthony Sorace]

it seems like a reasonable start to me, at least, but i don't know as
much as i could about the internals of linuxemu (i haven't really
looked inside since russ's initial version). the current maintainer is
frequently found in #plan9 on irc.freenode.net; i'd encourage you to
pop in there and see if you can catch him to ask.

Re: [9fans] gsoc linuxemu project help

[cinap_lenrek]

[-- Attachment #1: Type: text/plain, Size: 1671 bytes --]

Does creative masoshism count as GSoC project?  I dont know :)

Hm...  These points all belong to the big topic, getting modern linux
distro binaries (NTPL stuff) to work.  This would be a good thing
because I'm stuck on some old debian sarge that was just moved to the
archives.

Step one would be to get the simple single-threaded stuff to run like
echo and cat.  The part point here is TLS.  I had a somewhat working
plan9 driver for it but never packaged it up because i wanted the
whole thing working before submitting a kernel patch. You may later
progress to multithreaded apps wich use futex syscall.

The way I work on linuxemu is randomly trying stuff out, see where it
crashes...  try to understand why it crashes...  implement/fix
syscalls and try again.  Sometimes it easy and sometimes you scratch
your head for a several months.  Its hard to estimate the time needed
to get X running because you never know what crazy optimization shit
the linux/libc guys come up with next to make you suffer.

Here is always a huge risk of failure in linuxemu because all the
details are in Ulrich Dreppers head only or encrypted with
c-preprocessor-ifdef-encryption in the glibc-code so getting help is
very hard and most of the stuff you have to find out yourself (this
may cause brain damage over time).

I cant (officialy) mentor you as I'm short of time and have no
scientific background or something, but I will try to give you all
infeormation/code and support I have...  Its good to hear that someone
starts taking over some work!  :) Just drop me an email.  I may be
in irc from time to time too but dont count on it.

--
cinap

[-- Attachment #2: Type: message/rfc822, Size: 4042 bytes --]

[-- Attachment #2.1.1: Type: text/plain, Size: 339 bytes --]

Hi all,
I am interested in the GSoc project: Linuxemu improvement.
But the project description seems too general. I don't know from
which aspect can we make improvement. There is a TODO file in
the linuxemu3 source directory and I find TLS, futex, VDSO listed.
Can these TODOs make a gsoc project?
Thanks.



--

Regards,
Zhao

[-- Attachment #2.1.2: Type: text/html, Size: 549 bytes --]

Re: [9fans] gsoc linuxemu project help

[Zhao Shuai]

[-- Attachment #1: Type: text/plain, Size: 1994 bytes --]



>Does creative masoshism count as GSoC project?  I dont know :)
>
>Hm...  These points all belong to the big topic, getting modern linux
>distro binaries (NTPL stuff) to work.  This would be a good thing
>because I'm stuck on some old debian sarge that was just moved to the
>archives.
>
>Step one would be to get the simple single-threaded stuff to run like
>echo and cat.  The part point here is TLS.  I had a somewhat working
>plan9 driver for it but never packaged it up because i wanted the
>whole thing working before submitting a kernel patch. You may later
>progress to multithreaded apps wich use futex syscall.
>
>The way I work on linuxemu is randomly trying stuff out, see where it
>crashes...  try to understand why it crashes...  implement/fix
>syscalls and try again.  Sometimes it easy and sometimes you scratch
>your head for a several months.  Its hard to estimate the time needed
>to get X running because you never know what crazy optimization shit
>the linux/libc guys come up with next to make you suffer.
>
>Here is always a huge risk of failure in linuxemu because all the
>details are in Ulrich Dreppers head only or encrypted with
>c-preprocessor-ifdef-encryption in the glibc-code so getting help is
>very hard and most of the stuff you have to find out yourself (this
>may cause brain damage over time).
>
>I cant (officialy) mentor you as I'm short of time and have no
>scientific background or something, but I will try to give you all
>infeormation/code and support I have...  Its good to hear that someone
>starts taking over some work!  :) Just drop me an email.  I may be
>in irc from time to time too but dont count on it.
>
>--
>cinap

If nothing else in linuxemu can make a GSoc project, I would like to give NTPL a try.

Now I wonder how does Linuxemu work? I guess it parses the linux binary and whenever
encounter a system call, emulate using Plan 9's function. I haven't dig into the source
code, just guess.

[-- Attachment #2: Type: text/html, Size: 2256 bytes --]

Re: [9fans] gsoc linuxemu project help

[David Leimbach]

[-- Attachment #1: Type: text/plain, Size: 2338 bytes --]

2009/3/26 Zhao Shuai <upczhsh@163.com>

>
> >Does creative masoshism count as GSoC project?  I dont know :)
> >
> >Hm...  These points all belong to the big topic, getting modern linux
> >distro binaries (NTPL stuff) to work.  This would be a good thing
> >because I'm stuck on some old debian sarge that was just moved to the
> >archives.
> >
> >Step one would be to get the simple single-threaded stuff to run like
> >echo and cat.  The part point here is TLS.  I had a somewhat working
> >plan9 driver for it but never packaged it up because i wanted the
> >whole thing working before submitting a kernel patch. You may later
> >progress to multithreaded apps wich use futex syscall.
> >
> >The way I work on linuxemu is randomly trying stuff out, see where it
> >crashes...  try to understand why it crashes...  implement/fix
> >syscalls and try again.  Sometimes it easy and sometimes you scratch
> >your head for a several months.  Its hard to estimate the time needed
> >to get X running because you never know what crazy optimization shit
> >the linux/libc guys come up with next to make you suffer.
> >
> >Here is always a huge risk of failure in linuxemu because all the
> >details are in Ulrich Dreppers head only or encrypted with
> >c-preprocessor-ifdef-encryption in the glibc-code so getting help is
> >very hard and most of the stuff you have to find out yourself (this
> >may cause brain damage over time).
> >
> >I cant (officialy) mentor you as I'm short of time and have no
> >scientific background or something, but I will try to give you all
> >infeormation/code and support I have...  Its good to hear that someone
> >starts taking over some work!  :) Just drop me an email.  I may be
> >in irc from time to time too but dont count on it.
> >
> >--
> >cinap
>
>

>
> If nothing else in linuxemu can make a GSoc project, I would like to give NTPL a try.
>
> Is NTPL the same thing as NPTL, and everyone is just repeatedly spelling
their acronyms incorrectly?



>
> Now I wonder how does Linuxemu work? I guess it parses the linux binary and whenever
> encounter a system call, emulate using Plan 9's function. I haven't dig into the source
> code, just guess.
>
>
>
> ------------------------------
> 网易邮箱，中国第一大电子邮件服务商 <http://www.yeah.net>

[-- Attachment #2: Type: text/html, Size: 2962 bytes --]

Re: [9fans] gsoc linuxemu project help

[Sebastian A. Liem]

On 3/26/09, David Leimbach <leimy2k@gmail.com> wrote:
> 2009/3/26 Zhao Shuai <upczhsh@163.com>
>
>>
>> >Does creative masoshism count as GSoC project?  I dont know :)
>> >
>> >Hm...  These points all belong to the big topic, getting modern linux
>> >distro binaries (NTPL stuff) to work.  This would be a good thing
>> >because I'm stuck on some old debian sarge that was just moved to the
>> >archives.
>> >
>> >Step one would be to get the simple single-threaded stuff to run like
>> >echo and cat.  The part point here is TLS.  I had a somewhat working
>> >plan9 driver for it but never packaged it up because i wanted the
>> >whole thing working before submitting a kernel patch. You may later
>> >progress to multithreaded apps wich use futex syscall.
>> >
>> >The way I work on linuxemu is randomly trying stuff out, see where it
>> >crashes...  try to understand why it crashes...  implement/fix
>> >syscalls and try again.  Sometimes it easy and sometimes you scratch
>> >your head for a several months.  Its hard to estimate the time needed
>> >to get X running because you never know what crazy optimization shit
>> >the linux/libc guys come up with next to make you suffer.
>> >
>> >Here is always a huge risk of failure in linuxemu because all the
>> >details are in Ulrich Dreppers head only or encrypted with
>> >c-preprocessor-ifdef-encryption in the glibc-code so getting help is
>> >very hard and most of the stuff you have to find out yourself (this
>> >may cause brain damage over time).
>> >
>> >I cant (officialy) mentor you as I'm short of time and have no
>> >scientific background or something, but I will try to give you all
>> >infeormation/code and support I have...  Its good to hear that someone
>> >starts taking over some work!  :) Just drop me an email.  I may be
>> >in irc from time to time too but dont count on it.
>> >
>> >--
>> >cinap
>>
>>
>
>>
>> If nothing else in linuxemu can make a GSoc project, I would like to give
>> NTPL a try.
>>
>> Is NTPL the same thing as NPTL, and everyone is just repeatedly spelling
> their acronyms incorrectly?
>
>
>
>>
>> Now I wonder how does Linuxemu work? I guess it parses the linux binary
>> and whenever
>> encounter a system call, emulate using Plan 9's function. I haven't dig
>> into the source
>> code, just guess.
>>
>>
>>
>> ------------------------------
>> 网易邮箱，中国第一大电子邮件服务商 <http://www.yeah.net>
>