yep. have exactly that:

void initial_thread_cb_skas(void (*proc)(void *), void *arg)
{
    sigjmp_buf here;

    cb_proc = proc;
    cb_arg = arg;
    cb_back = &here;

    block_signals();
    if(sigsetjmp(here, 1) == 0)
        siglongjmp(initial_jmpbuf, 2);
    unblock_signals();

    cb_proc = NULL;
    cb_arg = NULL;
    cb_back = NULL;
}


Jeff Dike wrote:

>dbahi@enterasys.com said:
>
>
>>actually - 2.4.26-3um looks like this already.
>>
>>
>
>Yeah, if I had spent a bit more time looking at the stack, I would have noticed
>that the symptom was wrong for that patch.  What's happening is that you are
>taking a timer interrupt while on the initial process stack, which can't be
>allowed because interrupts should only happen on a kernel stack.
>
>Make sure your initial_thread_cb_skas (in arch/um/kernel/skas/process.c)
>blocks signals like so:
>
>	block_signals();
>	if(sigsetjmp(here, 1) == 0)
>		siglongjmp(initial_jmpbuf, 2);
>	unblock_signals();
>
>				Jeff
>
>
>

--
db