RE: [PATCH net-next] bridge: multicast to unicast

[Jean-Pierre Tosoni]

> -Message d'origine-
> De : linux-wireless-ow...@vger.kernel.org [mailto:linux-wireless-
> ow...@vger.kernel.org] De la part de Stephen Hemminger
> Envoyé : samedi 7 janvier 2017 04:14
> À : Linus Lüssing
> Cc : net...@vger.kernel.org; David S . Miller; bridge@lists.linux-
> foundation.org; linux-kernel@vger.kernel.org; linux-
> wirel...@vger.kernel.org; Felix Fietkau
> Objet : Re: [PATCH net-next] bridge: multicast to unicast
> 
> On Mon,  2 Jan 2017 20:32:14 +0100
> Linus Lüssing  wrote:
> 
> > This feature is intended for interface types which have a more
> > reliable and/or efficient way to deliver unicast packets than
> > broadcast ones (e.g. wifi).
> 
> 
> Why is this not done in MAC80211 rather than  bridge?

OTOH mac80211 has more information to decide whether it is more economic to 
send one multicast or several unicast.
It depends on the bitrate of each station, number of stations and the (not 
necessarily slower) bitrate of multicasts.

RE: [PATCH net-next] bridge: multicast to unicast

[Jean-Pierre Tosoni]

> -Message d'origine-
> De : linux-wireless-ow...@vger.kernel.org [mailto:linux-wireless-
> ow...@vger.kernel.org] De la part de Stephen Hemminger
> Envoyé : samedi 7 janvier 2017 04:14
> À : Linus Lüssing
> Cc : net...@vger.kernel.org; David S . Miller; bridge@lists.linux-
> foundation.org; linux-kernel@vger.kernel.org; linux-
> wirel...@vger.kernel.org; Felix Fietkau
> Objet : Re: [PATCH net-next] bridge: multicast to unicast
> 
> On Mon,  2 Jan 2017 20:32:14 +0100
> Linus Lüssing  wrote:
> 
> > This feature is intended for interface types which have a more
> > reliable and/or efficient way to deliver unicast packets than
> > broadcast ones (e.g. wifi).
> 
> 
> Why is this not done in MAC80211 rather than  bridge?

OTOH mac80211 has more information to decide whether it is more economic to 
send one multicast or several unicast.
It depends on the bitrate of each station, number of stations and the (not 
necessarily slower) bitrate of multicasts.

RE: should RTS init in serial core be tied to CRTSCTS

[Tosoni]

Krzysztof Halasa wrote:

> "Tosoni" <[EMAIL PROTECTED]> writes:
> > As far as I know in the old times this was the *standard*
> way to use a modem
> > (per CCITT V24), and even nowadays many modems can handle
> this method for
> > transmit, to stay compatible with the standard.
>
> I think it wasn't standard for real modems as they were full-duplex
> (even these 1200/75 or what was that) but it was for other devices
> such as current loops (which were frequently half-duplex).
>

It has always been the standard for all modems. The mistake comes from the
fact that the serial ports has been used extensively to drive things which
are *not* modems (say, printers and VT100 consoles on Unix systems). Such
devices did not need the standard-specified RTS function.

CCITT V24 says about RTS: "...this signal drives the DCE and sets it to
transmit data..." (translated from french)
CCITT V24 does not constraint the DCE to being half or full duplex.
CCITT V24 says nothing about using RTS to handle flow control.

> I've seen such devices quite recently, perhaps ~ 10 years ago.
> OTOH I think even "current" PC BIOSes use such signaling.

Even Windows implements the CCITT view of RTS, via a flag named "RTS_TOGGLE"

>
> > Think of radio modems. Some are inherently half duplex.
>
> Sure. But /dev/ttyS* ports are full-duplex, with CRTSCTS or without,
> so they don't use such handshaking.

A full duplex port can be used for both full- and half- duplex. A half
duplex modem cannot be used on a full-duplex-only port. So Linux should be
adapted to the standard, since we cannot adapt the standard to Linux.

> /proc is probably no-no.

Then sysfs sounds good to me.

>
> For such signaling, it would perhaps be better to invent another flag,
> similar to CRTSCTS. The driver would, of course, need some real code
> for that.

Another flag would help to drive modems, yes. But it would not help keeping
RTS low at open, because we need to open the port (which raises RTS) in
order to set the flag. Hence the need for an extra parameter in sysfs.

Best regards,
Jean-Pierre Tosoni

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

RE: should RTS init in serial core be tied to CRTSCTS

[Tosoni]

Krzysztof Halasa wrote:

 Tosoni [EMAIL PROTECTED] writes:
  As far as I know in the old times this was the *standard*
 way to use a modem
  (per CCITT V24), and even nowadays many modems can handle
 this method for
  transmit, to stay compatible with the standard.

 I think it wasn't standard for real modems as they were full-duplex
 (even these 1200/75 or what was that) but it was for other devices
 such as current loops (which were frequently half-duplex).


It has always been the standard for all modems. The mistake comes from the
fact that the serial ports has been used extensively to drive things which
are *not* modems (say, printers and VT100 consoles on Unix systems). Such
devices did not need the standard-specified RTS function.

CCITT V24 says about RTS: ...this signal drives the DCE and sets it to
transmit data... (translated from french)
CCITT V24 does not constraint the DCE to being half or full duplex.
CCITT V24 says nothing about using RTS to handle flow control.

 I've seen such devices quite recently, perhaps ~ 10 years ago.
 OTOH I think even current PC BIOSes use such signaling.

Even Windows implements the CCITT view of RTS, via a flag named RTS_TOGGLE


  Think of radio modems. Some are inherently half duplex.

 Sure. But /dev/ttyS* ports are full-duplex, with CRTSCTS or without,
 so they don't use such handshaking.

A full duplex port can be used for both full- and half- duplex. A half
duplex modem cannot be used on a full-duplex-only port. So Linux should be
adapted to the standard, since we cannot adapt the standard to Linux.

 /proc is probably no-no.

Then sysfs sounds good to me.


 For such signaling, it would perhaps be better to invent another flag,
 similar to CRTSCTS. The driver would, of course, need some real code
 for that.

Another flag would help to drive modems, yes. But it would not help keeping
RTS low at open, because we need to open the port (which raises RTS) in
order to set the flag. Hence the need for an extra parameter in sysfs.

Best regards,
Jean-Pierre Tosoni

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

RE: should RTS init in serial core be tied to CRTSCTS

[Tosoni]

Krzysztof Halasa wrote:

> OTOH I wonder what does the device in question require WRT the
> serial port and WRT RTS line in particular.
> I know there are some half-duplex converters which drive RTS only
> while sending and which require CTS to send.

As far as I know in the old times this was the *standard* way to use a modem
(per CCITT V24), and even nowadays many modems can handle this method for
transmit, to stay compatible with the standard.

Think of radio modems. Some are inherently half duplex.

> They are perhaps a bit broken 
No, no, they apply an old standard. Probably they are old as well.
>
> I don't know if one can expect RTS to stay low all the time. Even
> a simple /sbin/setserial from /etc/rc.* would drive it high for
> a moment. I'm afraid the only way which makes sense may be using
> a customized plug.

It's a pity that Linux (or Unixes) never handled RTS this way.
I feel that the /proc or sysfs solutions are the best to alter this well
established default in this driver. It would not break existing installed
hardware.

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

RE: should RTS init in serial core be tied to CRTSCTS

[Tosoni]

Krzysztof Halasa wrote:

 OTOH I wonder what does the device in question require WRT the
 serial port and WRT RTS line in particular.
 I know there are some half-duplex converters which drive RTS only
 while sending and which require CTS to send.

As far as I know in the old times this was the *standard* way to use a modem
(per CCITT V24), and even nowadays many modems can handle this method for
transmit, to stay compatible with the standard.

Think of radio modems. Some are inherently half duplex.

 They are perhaps a bit broken snip
No, no, they apply an old standard. Probably they are old as well.

 I don't know if one can expect RTS to stay low all the time. Even
 a simple /sbin/setserial from /etc/rc.* would drive it high for
 a moment. I'm afraid the only way which makes sense may be using
 a customized plug.

It's a pity that Linux (or Unixes) never handled RTS this way.
I feel that the /proc or sysfs solutions are the best to alter this well
established default in this driver. It would not break existing installed
hardware.

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