Re: [patch 00/12] Syslets, Threadlets, generic AIO support, v5

[Anton Blanchard <anton@samba.org>]

Hi Ingo,

> this is the v5 release of the syslet/threadlet subsystem:
> 
>    http://redhat.com/~mingo/syslet-patches/

Nice!

I tried to port this to ppc64 but found a few problems:

The 64bit powerpc ABI has the concept of a TOC (r2) which is used for
per function data. This means this wont work:

        if (ah->restore_stack) {
                set_task_stack_reg(new_task, ah->restore_stack);
                task_ip_reg(new_task) = ah->restore_ip;
                /*
                 * The return code 0 is needed to tell the
                 * head user-context that the threadlet went async:
                 */
                task_ret_reg(new_task) = 0;
        }

I think we would want to change restore_ip to restore_function, and then
create a per arch helper, perhaps:

void set_user_context(struct task_struct *task, unsigned long stack,
		      unsigned long function, unsigned long retval);

ppc64 could then grab the ip and r2 values from the function descriptor.

The other issue involves the syscall table:

asmlinkage struct syslet_uatom __user *
sys_async_exec(struct syslet_uatom __user *uatom,
               struct async_head_user __user *ahu)
{
        return __sys_async_exec(uatom, ahu, sys_call_table, NR_syscalls);
}

This exposes the layout of the syscall table. Unfortunately it wont work
on ppc64. In arch/powerpc/kernel/systbl.S:

#define COMPAT_SYS(func)        .llong  .sys_##func,.compat_sys_##func

Both syscall tables are overlaid.

Anton