[ Ingo, Thomas - see the whole series on linux-kernel ]

On Thu, 15 Nov 2007, Ulrich Drepper wrote:
quoted text
>
> This part adds support for sys_indirect on x86 and x86-64.

The thing is, not all system calls can do this.

Some system calls are magic, and don't just take the arguments in 
registers: they also care about the actual stack pointer and the whole 
pt_regs struct when returning to user mode. 

So this does need more infrastructure: some way of marking which system 
calls cannot be executed indirectly.

The magic system calls are things like:

 - sys_iopl() - this one changes the eflags value restored on iret
 - execve/clone/vfork() - need direct access to pt_regs
 - vm86() - does magic with the stack, cares about pt_regs
 - sigreturn - magic pt_regs accesses again

and there may be others I have forgotten about.

Calling these system calls from C code will just corrupt the kernel stack, 
and is a big big no-no.

		Linus
-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/