svn commit: r212076 - head/lib/libthr/thread

[David Xu]

Kostik Belousov wrote:
> On Thu, Sep 02, 2010 at 04:34:58PM +0000, David Xu wrote:
>> Kostik Belousov wrote:
>>> On Wed, Sep 01, 2010 at 02:18:33AM +0000, David Xu wrote:
>>>> Author: davidxu
>>>> Date: Wed Sep  1 02:18:33 2010
>>>> New Revision: 212076
>>>> URL: http://svn.freebsd.org/changeset/base/212076
>>>>
>>>> Log:
>>>>  Add signal handler wrapper, the reason to add it becauses there are
>>>>  some cases we want to improve:
>>>>    1) if a thread signal got a signal while in cancellation point,
>>>>       it is possible the TDP_WAKEUP may be eaten by signal handler
>>>>       if the handler called some interruptibly system calls.
>>>>    2) In signal handler, we want to disable cancellation.
>>>>    3) When thread holding some low level locks, it is better to
>>>>       disable signal, those code need not to worry reentrancy,
>>>>       sigprocmask system call is avoided because it is a bit expensive.
>>>>  The signal handler wrapper works in this way:
>>>>    1) libthr installs its signal handler if user code invokes sigaction
>>>>       to install its handler, the user handler is recorded in internal
>>>>       array.
>>>>    2) when a signal is delivered, libthr's signal handler is invoke,
>>>>       libthr checks if thread holds some low level lock or is in critical
>>>>       region, if it is true, the signal is buffered, and all signals are
>>>>       masked, once the thread leaves critical region, correct signal
>>>>       mask is restored and buffered signal is processed.
>>>>    3) before user signal handler is invoked, cancellation is temporarily
>>>>       disabled, after user signal handler is returned, cancellation state
>>>>       is restored, and pending cancellation is rescheduled.
>>>> +static void
>>>> +thr_sighandler(int sig, siginfo_t *info, void *_ucp)
>>>> +{
>>>> +	if ((actp->sa_flags & SA_SIGINFO) != 0)
>>>> +		(*(sigfunc))(sig, info, ucp);
>>>> +	else {
>>>> +		((ohandler)(*sigfunc))(
>>>> +			sig, info->si_code, (struct sigcontext *)ucp,
>>>> +			info->si_addr, (__sighandler_t *)sigfunc);
>>>> +	}
>>> I do not think this is very important, but freebsd old-style signal
>>> handler fourth argument is usually the faulted %eip value. This is
>>> most likely irrelevant for any source that is linked with libthr.so
>>> new enough to contain this change.
>> Isn't the si_addr in siginfo a fault address ? I remembered I saved
>> the fault address in ksiginfo_t which is converted to userland
>> siginfo, and fault address should be there. what's wrong here ?
>>
> Oops, sorry, I miscalculated the position of the arguments :(.

Sorry, I think I also misunderstood you too. :(
I think kernel still has some compatible problems. I just skimmed
it again, and I found a problem.

In RELENG_4, I found sys/i386/i386/machdep.c has following code in
sendsig():


	sf.sf_signum = sig;
	sf.sf_ucontext = (register_t)&sfp->sf_uc;
	if (SIGISMEMBER(p->p_sigacts->ps_siginfo, sig)) {
		/* Signal handler installed with SA_SIGINFO. */
		sf.sf_siginfo = (register_t)&sfp->sf_si;
		sf.sf_ahu.sf_action = (__siginfohandler_t *)catcher;

		/* fill siginfo structure */
		sf.sf_si.si_signo = sig;
		sf.sf_si.si_code = code;
		sf.sf_si.si_addr = (void*)regs->tf_err;
	}
	else {
		/* Old FreeBSD-style arguments. */
		sf.sf_siginfo = code;
		sf.sf_addr = regs->tf_err;             <-------
		sf.sf_ahu.sf_handler = catcher;
	}

the sf.sf_addr is assigned by tf_err here.

In later branch ( I don't know which ), it seems it uses ksi.ksi_addr,

       /* Old FreeBSD-style arguments. */
                 sf.sf_arg2 = ksi->ksi_code;
                 sf.sf_addr = (register_t)ksi->ksi_addr;
                 sf.sf_ahu.sf_handler = catcher;



the tf_err may not be equal to ksi_addr! This may need to be fixed.