ðòïåëôù 


  áòèé÷ 


Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 

  óôáôøé 


  ðåòóïîáìøîïå 


  ðòïçòáííù 



ðéûéôå
ðéóøíá












     áòèé÷ :: Inet-Admins
Inet-Admins mailing list archive (inet-admins@info.east.ru)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[inet-admins] Rel notes for mica 2.7.2.2


  • To: <inet-admins@info.east.ru>
  • Subject: [inet-admins] Rel notes for mica 2.7.2.2
  • From: "Dmitri Kalintsev" <dek@hades.uz>
  • Date: Thu, 10 May 2001 08:15:46 +1000
  • Delivered-To: inet-2317-ak@frog.east.ru
  • Delivered-To: inet-admins@info.east.ru

îÅ Õ×ÅÒÅÎ, ÐÒÁ×ÄÁ, ÞÔÏ ÜÔÏ ÕÖÅ ÒÁÚÄÁÀÔ, ÎÏ ÔÅÍ ÎÅ ÍÅÎÅÅ.

SY,
--
 CCNP, CCDP (R&S)                          Dmitri E. Kalintsev
 CDPlayer@irc               Network Architect @ connect.com.au
 dek @ connect.com.au     phone: +61 39 674 3913 fax: 251 3666
 http://-UNAVAIL-         UIN:7150410    cell: +61 41 335 1634

Date: 05/07/01

Checksum Information
====================
ftp-eng{jtremoli}8% cksum pw2722.ios
1755798784   474939     pw2722.ios
ftp-eng{jtremoli}9%

Overview
========
Portware patch 2.7.2.2 ("patch" = interim release not available on CCO) is
a renumber of engineering release r07f4.  2.7.2.2 is targetted as the next
CCO maintenance release (tentatively named 2.7.3.0) due out at the beginning
of June 2001.

2.7.2.2 contains all of the fixes/DDTS's in r07f4 and in past versions (e.g.
2.7.2.1, 2.7.2.0, etc.)  See the respective Release Notes for additional
details, caveats, etc. regarding past DDTS's.

Caveats/Details
===============
Below are some caveats to keep in mind as well as detailed explanations
on some of the enhancements made in this version and/or previous versions
of portware.

Note #1
=======
The S54 regisiter (General Capability Register) now defaults to S54=200
in 2.7.2.2 and above.  This enables Power Control (CSCdk11008), 0.5 increment
for SQ value (CSCdp40871), and the UART functionality/EC parity bit (CSCds35369).

* = Enabled by default.

S54 [  200]  General capability (bit mapped)

	     0  = All general options disabled

	     1  = Reserved for engineering use

	     2  = Enable Aggressive Client Capping

	     4  = Force 4-point training (See end of file for 
                  a summary on 4-point training.)

	    *8  = Enable Power Control

	     16 = Enable EC Quality Checking

             32 = CAP Rx Rate to 26400 when detecting PC-Tel Modem
                  (See CSCdm45403.)

            *64 = Enable UART Functionality (See CSCdj92333.)

           *128 = Enable 0.5 increment for SQ value via S32 (See
                  CSCdp40871.)

            256 = DSP Reset Enable (See CSCdp94025.)

            512 = Enable V.110 S&X option to be the same as AS5200.
                  (See CSCdr39887.)

           1024 = Enable enhanced S detector.  (See CSCdr65889.)


DDTS's included in 2.7.2.2
==========================
CSCdr94776 DSP death during Phase4 of modem start-up (MICA Answer mode)
CSCdm51072 Hex Death causes Mica to miss heartbeats
CSCdp67697 DSP flu in 2711 - DSP yields low CSR but doesn't utterly fail
CSCdt82242 Modem #UEL fails to connect Bell212 w/ MICA (s64=10)
CSCdt41661 Need memory dump capability from SP
CSCdt54979 Separate ASSERT statements in timer code
CSCds82376 MICA S46 inband busy detect fails if dialtone detect bit is on
CSCdt47453 Redwood messaging has timing issue, needs interrupts disabled
CSCds22750 PW mail processing loop needs init 2nd time
CSCdp78427 Change sh mo op param#38 to display time averaged # of 
           bad frames
CSCds38258 Link Info Parameter #6 not initialized in the right place
CSCdp94723 Link information index 85 not incrementing
CSCds30003 S64=2 on Mica S-register doesn't work
CSCds16196 V.42bis Reset/Disconnect Problem
CSCdr56969 S29=2 or S29=3 can hang the SP
CSCdt21201 V.90 INFO0d bits 28 through 32 are transmitted incorrectly
CSCds21207 DSP death case when Originate MICA tries to connect to vendor 
           modem
CSCds48117 Some link info not updated
CSCds48158 TxQ underrun counter not incremented
CSCds38289 MICA Originate and Answer fail to connect to Bell212 (S64=10)
CSCds35369 EC parity bit in S54 should be on by default
CSCds50046 Mips starvation problem seen in tone_pll routine
CSCds36614 MICA parity over EC feature (CSCdj92333) garbles AT interface
CSCds71259 Additional changes for false busy detect in R2
CSCdp67281 Cannot make a 12K V.110 connection with vendor TA.
CSCdt31649 Code clean up and adding comments into isdn.s
CSCdt31667 K56Flex echo cancellor does not update correctly in data mode
CSCdt31775 Mica Flex does not sending correct L2 level
CSCdt35638 Code in Terminate memory accidentally writes to reserved memory
CSCdt16071 Incorrect use of instructions (POPD/PSHD) may overflow HW stack
CSCdt35668 Pointer is not set properly in SSprime phase reversal detector
CSCdt35717 DTMF input buffer could be initially loaded with incorrect sample
CSCds55455 R2 busy issue needs to be recommited into r272x, main, etc.
CSCds86723 documentation on MICA S13 parity settings is incomplete or 
	     incorrect
CSCdt26092 SET_BIT and CLR_BIT macros missing braces
CSCdt41661 Need memory dump capability from SP
CSCdp72881 S7 value not honored on dialout
+CSCdr40079 decouple max symbol rates for V.90 and V.34
+CSCds39284 Add debug selectablity to link info param # 32


+Notes regarding CSCds40079 and CSCds39284 
==========================================

CSCdr40079: decouple max symbol rates for V.90 and V.34

2.7.2.2 adds two new S registers for control of the V.90 uplink maximum
and minimum symbol rates, making them independent of the V.34 max and min 
symbol rate controls (S41 and S42).  The details are as follows:

Index  Sreg   Default   Configuration Description          Values
-----  ----   -------   ------------------------------- -------------
 72     71       5      V.90 Uplink Maximum Symbol Rate  3: 3000 baud
							 4: 3200 baud
							 5: 3429 baud

 73     72       3      V.90 Uplink Minimum Symbol Rate  3: 3000 baud
						 	 4: 3200 baud
							 5: 3429 baud

Note: These new S registers are read-only unless an ATDEBUGTHISMODEM 
command has been issued.



CSCds39284: Add debug selectablity to link info param # 32

S73 is a new S register which serves as the Link Info #32 information
selector:

S73=1: LI#32 reports (DIL status << 8) | client modem ID; this is
       (byte 1 of LI#51 << 8) ORed with (byte 2 of LI#51);
       see CSCdm92993 for a list of Client Modem IDs;
       see Link Info 51 description in the API for DIL status

S73=2: LI#32 reports DSP TX queue underruns; this counter (internal to
       the DSP) indicates out-of-MIPs condition; this counter is reset
       a the start of every call

*S73=3: LI#32 reports dsp_dma_blowups; this is a 32-bit value
       containing three 10-bit fields (2 MSBs undefined); each field
       is a counter for one of the DSPs; these counters are
       incremented when there is a CP<->DSP data buffer under/overrun;
       these counters are reset only at boot time; this is what gets
       reported in version 2721

S73=0: LI#32 reports a packed version of the other three;
       in this case LI#32 is formatted as:
MSB                                  LSB
  xxxxxxxxxxxxxxxx xx xx xx xx xxxxxxxx
  ---------------- -- -- -- -- --------
                #6 #5 #4 #3 #2       #1
#1 : V.90 client modem ID (8 bits)
#2 : tx_underruns counter saturated at 3 (2 bits)
#3 : DSP #1 DMA under/overrun counter saturated at 3 (2 bits)
#4 : DSP #2 DMA under/overrun counter saturated at 3 (2 bits)
#5 : DSP #3 DMA under/overrun counter saturated at 3 (2 bits)
#6 : reserved for future use (16 bits)

* The default value is S73=3.


Summary on 4-Point Training (S54=4)
===================================
There's 4-point training and 16-point training.  This information
is provided by signal J in phase 3 of a modem connection.  4 point
versus 16-point tells the modems the constellation size used when
the modems are training up (in phase 4).  16-point is used the most,
but if you don't have a very clean line, and you're having problems
syncing up, you can set it to 4-point to see if you're having
problems during training.  With 4-point, you'll have less detection
errors (the points are spread further away, so the detection is more
lenient) than 16-point which has to be more precise.

Additionally, 4-point training is a more robust way of completing the
full duplex train in V.34/V.90/K56Flex.  The regular train uses an
8-point constellation.  When the line is noisy the standard allows
you to switch down to a 4-point constellation which allows you to
operate reliably in a lower SNR environment and complete training.
We have broken this out into an S-register (S54=4) as a debug tool
for customer use.

[END]


 




Copyright © Lexa Software, 1996-2009.