Re: The time has come for a kernel interfacing library layer

[Joerg Sonnenberger <joerg@xxxxxxxxxxxxxxxxx>]

On Sun, May 08, 2005 at 02:52:53PM -0700, Matthew Dillon wrote:
> 
> :On Sun, May 08, 2005 at 11:33:53AM -0700, Matthew Dillon wrote:
> :>     Well, if we go the cerror() route for the compatibility layer, that
> :>     means this layer cannot just be mapped but instead must run through
> :>     a relocation / ELF pass.
> :
> :I think we can even avoid that.  We push a stack frame to return into
> :cerror() onto the stack which the compat layer can return too. We have
> :to keep that in mind when doing the argument calculation, but that
> :should be easy to handle.
> :
> :Joerg
> 
>     The problem is where does the function address of cerror come
>     from ?  This library is implementing a system call, the function returns
>     to the caller, not to cerror.  I definitely do not want to hack up 
>     the stack frame.

I wanted to put an intermediate stack frame there. We need it anyway for
the internal compatibility handling in libc, even it adds another stack
frame for the "normal" case.

Joerg