Re: P1666 surprising result with immediate notification

John, Bishnupriya,

personally, I would also prefer straight Verilog semantics.

Wrt. the backwards compatibility: There is already the
SC_SIGNAL_WRITE_CHECK mechanism in place in the PoC simulator. Any
implementation is free to add an something similar to this.

  The check itself could be implemented fairly efficient in the PoC's
trigger() function of the event, where all processes are considered
anyhow. In the immediate case, the implementation would then check if
the triggered process equals the currently running process.

  The cost for this would be roughly one pointer comparison per
sensitive process. Since the notification already involves quite some
work (looping over all processes, check their sensitivity, modify
lists,…), it should be acceptable. Since the check would be needed for
the new semantics anyhow, a warning could be required as well.

  I've not prototyped this to obtain some numbers, though.

  If there's a strong performance concern, we could still add an
explicit notify_immediate() function to sc_event, that does not performs
the check and results in undefined behaviour if the notifying process is
sensitive to the event.

Thoughts?

Greetings from Oldenburg,
  Philipp

On 19/01/11 20:01, john.aynsley@doulos.com wrote:
> Bishnupriya,
>
> I agree with you technically, but I was making a slightly different
> compromise. My thoughts were that the PoC simulator should keep its
> current bad behavior for the time being for backward compatibility, but
> that immediate self-notification should be deprecated in order to
> discourage applications from doing it. The compromise I am making is
> that in some future version of SC a process might actually be forbidden
> from making an immediate notification to an event to which it is
> statically sensitive or following next_trigger, rather than the
> notification simply doing nothing. If that is not acceptable, then my
> proposal collapses.
>
> Personally I would also be okay with straight "Verilog semantics",
> meaning that an immediate self-notification would do nothing (not
> deprecated, no warning), and live with the possibility of backward
> incompatibilities.
>
> John A
>
>
> -----Bishnupriya Bhattacharya <bpriya@cadence.com> wrote: -----
>
> To: "john.aynsley@doulos.com" <john.aynsley@doulos.com>, David Black
> <dcblack@xtreme-eda.com>, Philipp A Hartmann
> <philipp.hartmann@offis.de>, P1666 Technical WG
> <systemc-p1666-technical@eda.org>
> From: Bishnupriya Bhattacharya <bpriya@cadence.com>
> Date: 01/19/2011 05:29PM
> Subject: RE: P1666 surprising result with immediate notification
>
> John,
>
>
>
> I agree with the Verilog semantics, but am not sure of the means to
> achieve that. You say
>
>
>
> “ Deprecating any immediate notification of an event to which the
> calling process is statically sensitive or already sensitive using
> next_trigger”
>
>
>
> This does not seem right to me. Rather the immediate notifications
> as above should be valid, but will not have any effect in terms of
> rescheduling the process in the current eval. An implementation can
> issue a warning in these cases if it wants – of the nature that
> behavior has changed, these notifications will no longer have any
> effect on scheduling the currently executing process.
>
>
>
> Otherwise we will have the sitn. that a process that is statically
> sensitive to an event, and also immediately notifies that event to
> trigger ANOTHER process will start getting this warning stating it
> is doing something wrong, which it is not. The problem is not with
> the process immediately notifying the event, but with the
> implementation wrongly interpreting that to schedule the currently
> executing process.
>
>
>
> Thoughts?
>
>
>
> Thanks,
>
> -Bishnupriya
>
>
>
>
>
> *From:*john.aynsley@doulos.com [mailto:john.aynsley@doulos.com]
> *Sent:* Wednesday, January 19, 2011 3:26 PM
> *To:* David Black; Philipp A Hartmann; Bishnupriya Bhattacharya;
> P1666 Technical WG
> *Subject:* RE: P1666 surprising result with immediate notification
>
>
>
> All,
>
> One plank of SystemC is the co-routine semantics. The more I think
> about it, the more I think that the PoC behavior is an anomaly. A
> running process should not notice an immediate notification to itself.
>
> I would vote for
>
> * Keeping the PoC behavior as it is (for backward compatibility)
> * Deprecating any immediate notification of an event to which the
> calling process is statically sensitive or already sensitive using
> next_trigger.
>
> * An immediate notification of an event to which the calling process
> is not statically sensitive but is subsequently made sensitive using
> wait(ev) or next_trigger(ev) is permitted
>
> * (With an OSCI hat on) Adding a warning to the PoC sim to cover the
> deprecated cases (is that too expensive?)
>
> (IMHO adding a method by which a process can immediately notify
> itself is not such a good idea, unless some compelling use case
> emerges)
>
> Does that work?
>
> John A
>
>
> From:
>
>
>
> Bishnupriya Bhattacharya <bpriya@cadence.com>
>
> To:
>
>
>
> "john.aynsley@doulos.com" <john.aynsley@doulos.com>
>
> Cc:
>
>
>
> David Black <dcblack@xtreme-eda.com>, Philipp A Hartmann
> <philipp.hartmann@offis.de>, P1666 Technical WG
> <systemc-p1666-technical@eda.org>
>
> Date:
>
>
>
> 19/01/2011 03:51
>
> Subject:
>
>
>
> RE: P1666 surprising result with immediate notification
>
>
>
>
>
>
> John,
>
> I had exactly the same assumption as you until I discovered certain
> anomalies in the process control work. Personally, I would be
> happiest with the Verilog rules, which only differs from my rules in
> that a next_trigger for a method process takes effect immediately
> rather than waiting for the method to return. But I worry about
> backwards compatibility. If we can work something out (perhaps along
> the lines of what Philipp suggests in a subsequent email) whereby we
> can both clean up the semantics as well as maintain a backward
> compatibility hook, that might be the right way to go both
> philosophically and pragmatically.
>
> I also agree with Tor about self-triggering processes being a
> nuisance in parallel implementations.
>
> Thanks,
> -Bishnupriya
>
> *From:*john.aynsley@doulos.com [mailto:john.aynsley@doulos.com] *
> Sent:* Wednesday, January 19, 2011 12:56 AM*
> To:* Bishnupriya Bhattacharya*
> Cc:* David Black; Philipp A Hartmann; P1666 Technical WG*
> Subject:* RE: P1666 surprising result with immediate notification
>
> Bishnupriya, All,
>
> Back in 2005 I had assumed that neither a method process nor a
> thread process would have the capability to be sensitive to anything
> while the associated function was actually running, because that is
> how Verilog behaves (which does have immediate notifications). In
> other words, for a thread, I assumed the rule you described below.
> For a method, I assumed next_trigger() created the next trigger
> condition, but that the method process would not become a candidate
> to be triggered until it had returned control to the kernel.
>
> So I think we have 3 sets of rules on the table:
>
> * The "Verilog rules" (which was the intent of the 1666-2005 LRM,
> although not spelled out very explicitly)
> * The de facto PoC sim behavior
> * Bishupriya's rules
>
> My view on this is not principled, but purely practical. Do we fix
> the PoC simulator, or do we fix the rules? Is there code out there
> that relies on the current PoC scheduling rules? My default position
> would be that we have to document the PoC behavior, though I think
> that adopting the Verilog rules would lead to the fewest surprises.
> I think we might argue the case on the grounds of harmonisation
> between languages.
>
> Cheers,
>
> John A
>
>
> -----Bishnupriya Bhattacharya <bpriya@cadence.com> wrote: -----
> To: "Philipp A. Hartmann" <philipp.hartmann@offis.de>, Jerome CORNET
> <jerome.cornet@st.com>, David C Black <dcblack@xtreme-eda.com>
> From: Bishnupriya Bhattacharya <bpriya@cadence.com>
> Date: 01/18/2011 06:56PM
> Cc: John Aynsley <john.aynsley@doulos.com>, P1666 Technical WG
> <systemc-p1666-technical@eda.org>
> Subject: RE: P1666 surprising result with immediate notification
>
> Philipp, John, David, All,
>
> I agree with Philipp's excellent analysis below, except for the
> static sensitivity case.
>
> For a SC_METHOD, it is legitimate that next_trigger(e1) followed by
> notify(e1) will schedule the process in current eval.
>
> // will trigger:
> next_trigger(e1);
> e1.notify();
>
> For SC_THREAD, it is legitimate that notify(e1) followed by wait(e1)
> will NOT schedule the process in current eval
>
> // won't:
> e1.notify();
> wait(e1);
>
> What I don't find legitimate is that static sensitivity is triggered
> by immediate notification. This is because of the hole in SystemC
> that the effective sensitivity of a process is not defined WHILE it
> is running - it is an implementation artifact that the effective
> sensitivity during a process's execution is assumed to be its static
> sensitivity, which results in the method or thread below triggering
> in current eval, if e1 is the static sensitivity of the process.
> (I've raised this before in the context of the process control
> semantics).
>
> // will
> e1.notify();
> wait();
>
> In fact, even if the thread is changed as below to wait on another
> event e2 (instead of waiting for its static sensitivity), it will
> still be scheduled in current eval although e2 was never
> triggered!!!!!!! This is clearly BAD.
>
> // will
> e1.notify();
> wait(e2);
>
> IMO, the LRM should plug this hole and clearly define the state of a
> process's effective sensitivity when it is running. The rules below
> make most sense to me.
>
> 1) process's effective sensitivity is set to null when process
> starts executing
> 2) process can set up dynamic sensitivity by calling next_trigger in
> a method or wait(...) in a thread, which then becomes its effective
> sensitivity with immediate effect
> 3) if process does not set up dynamic sensitivity, its static
> sensitivity becomes its effective sensitivity AFTER method process
> returns or thread process yields to kernel via wait()
>
> Thoughts?
>
> Thanks,
> -Bishnupriya
>
> -----Original Message-----
> From: owner-systemc-p1666-technical@eda.org
> [mailto:owner-systemc-p1666-technical@eda.org] On Behalf Of Philipp
> A. Hartmann
> Sent: Tuesday, January 18, 2011 6:51 AM
> To: Jerome CORNET; David C Black
> Cc: John Aynsley; P1666 Technical WG
> Subject: Re: P1666 surprising result with immediate notification
>
> David, Jerome, All,
>
> I'll try to give the immediate notification issue a shot. Firstly, a
> summary of my understanding of the immediate notification semantics in
> general.
>
> 1) The kernel maintains a set of runnable processes for the (current)
> evaluation phase.
>
> 2) Processes can be sensitive to events
> (either statically or dynamically, which is not important here).
>
> 3) Upon immediate notification of an event, sensitive processes are
> immediately added to the set of runnable processes.
>
> 4) If a process is already in the set of runnable processes, it's
> not added for a second time when triggered. It's a set …
>
> 5) If a process is activated, it's immediately removed from the
> set of runnable processes.
>
> Apart from the general dangers of immediate notifications, three
> corner cases have been identified by David, which are all direct
> consequences of the rules above:
>
> ạ) Processes can be triggered immediately by themselves, resulting
> in multiple invocations in a single evaluation phase.
>
> To trigger itself immediately in the same evaluation phase, a
> process needs to be sensitive to the notified event _before_
> it calls ev.notify().
>
> The reason that this works at all is the removal of the process
> from the runnable set _before_ activating the process (5).
>
> The trigger can be due to a static sensitivity (as John pointed
> out), or due to a next_trigger(ev) in an SC_METHOD:
>
> // will trigger:
> next_trigger(ev);
> ev.notify();
>
> // won't:
> ev.notify();
> next_trigger(ev);
>
> b) wait(ev) is not triggered from within the current process
>
> // won't trigger in the current evaluation phase
> ev.notify();
> wait(ev);
>
> I think, this is expected. The notification is _immediate_, which
> means, that all processes that are _currently_ sensitive to to event
> are triggered (3). The wait(ev) call has not been reached, yet.
>
> We should NOT change this rule, since this may well break existing,
> carefully coded models.
>
> b) The number of activations of processes sensitive to immediately
> notified events can differ.
>
> This is a general artefact of immediate notifications. Their
> effect can depend on the kernel's internal scheduling order in
> which the set of runnable processes is activated.
>
> In David's example, one process is sometimes triggered twice in a
> row, depending on the instantiation order. The cause for this is the
> state of the set of runnable processes, when the event is notified.
>
> Consider the "repetition" case in the reference simulator
> (R being the set of runnable processes):
>
> start - R = {t2,t1}
> run t2 - R = {t1}
> run t1 - R = {} - notify ev - R = {t1,t2}
> run t1 - R = {t2} - notify ev - R = {t2,t1}
> run t2 ...
>
> Compared to:
> start - R = {t1,t2}
> run t1 - R = {t2} - notify ev - R = {t2,t1}
> run t2 - R = {t1}
> run t1 - R = {} - notify ev - R = {t1,t2}
> run t2 ...
>
> While I agree, that all of this can be quite surprising, I think we
> should keep the rules as they are:
>
> · Immediate notification (3)(4) needs to be defined this way, to be
> actually immediate.
> · Self-triggering (especially with next_trigger) may be useful
> especially for method processes.
> · Process order dependence can not be avoided with immediate
> notifications.
>
> Still, it may be helpful to refine the wording in that area.
>

-- 
Philipp A. Hartmann
Hardware/Software Design Methodology Group
OFFIS Institute for Information Technology
R&D Division Transportation · FuE-Bereich Verkehr
Escherweg 2 · 26121 Oldenburg · Germany · http://www.offis.de/
Phone/Fax: +49-441-9722-420/282 · PGP: 0x9161A5C0 · Skype: phi.har
-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.