SST Driver for Windows NT 4.0

Release Notes: SST Driver for Windows NT 4.0

Release Notes for Version 4.22

• Fixed RasDial() to only delay IOCT_GET_MODEMSTATUS for the Analog Modem Pool.
  
  
• Fixed in eqn_t1mp_ioctl() when canceling a TARB strcat() was used and I changed to strcpy() because a memory dump from a customer showed the problem.
  
  
• Fixed hang with SMP machine, IRP_MJ_DEVICE_CONTROL needed to set bd_lock_flag to true.

Release Notes for Version 4.20

• Added support for the SST-4P Universal Low Profile Board.

Release Notes for Version 4.19

• Fixed a memory leak with connecting/disconnecting PRI calls sme_findacb() had MEM_COMP that needed to be compared with the len.
  
  
• Added in ATMEL fix to add a RNG_WAIT and RNG_CHK states for going from "good" to "bad" with quick DC drops on port modules.
  
  
• Added code to the ioctl_serial_set_dtr to set overload signals.
  
  
• Added code to the ioctl_serial_clr_dtr to set overload signals.
  
  
• Added code to the ioctl_serial_set_rts to set overload signals.
  
  
• Added code to the ioctl_serial_clr_rts to set overload signals.
  
  
• Removed code in the primative open area. T1OpenPort had calculation errors relating to low baud rates and wrapping data within the buffer.This code continued to put modems in red X states when connecting at low baud rates.
  
  
• Fixed eqnSSTView.exe to display the correct time for the present SSTlogr file.

Release Notes for Version 4.18

• Modified expansion bus configuration algorithm to test if the SSP 64 state machine is ok. If it is not, then toggle the ring clock and repeat expansion bus configuration. This is known as the ATMEL fix (for 25' expansion bus cables). Once the ring is correctly established, restore input registers to their correct values in case they have been corrupted while the SSP 64 state machine was in a bad state.
  
  
• Added support for the CP16-DB9 panel with the SST-16P.

Release Notes for Version 4.15

• Windows 2000 introduces the concept of a named space for certain kernel resources, including terminal services. Name space qualifiers were added to mib agent filemapping (EqmMALNT.dll) to allow the terminal services to function properly.
  
  
• DMP modems hung in a "red light/green light" state. Fixed bugs in the DTR timeout logic and extended a timeout (to avoid false positives) allowing the DMP code to accurately detect hung ports and schedule them for a modem code download. Added a delay in the modem download ioctl to wait for a pending TARB to finish before erroneously exiting with an error so that these scheduled downloads would complete successfully (specifically, the ioctl was not waiting for a pending change in the DTR signal -- from the open -- to complete which caused the modem download ioctl to fail). If the TARB does not finish in the allocated time it is canceled (see item 4 below).
  
  
• Fix for a problem when canceling Read IRPs. The previous driver was mistakenly processing the transmit IRP queue instead of the receive IRP queue.
  
  
• Fix for processing of WaitMask. WaitMask event bits were being cleared prematurely. This prevents the central polling logic from getting a chance to process the event bits properly. Successful reads were incorrectly clearing information that resulted in WaitCommMask ioctls to not be completed.
  
  
• When an attempt to download the modem code fails, and we detect a pending TARB, we now cancel the pending TARB and retry to download the modem code. Unless the TARB is cancelled, a modem hung in this fashion cannot be reset as all attempts to do so will fail because of the pending TARB.
  
  
• Bits indicating that there is a pending control signal change request are also cleared when the TARB is canceled. Without cleaning up these state bits, the port could fail on subsequent calls even though the reset and download were successful. There may be similar problems elsewhere in the DMP code (i.e., canceling a TARB to allow the download to complete does not necessarily clean up other state information that could cause undesirable behavior in the future).
  
  
• Removed an unnecessary 5 second delay before properly detecting an inbound digital call. Previously, when a DMP took an inbound call, there would be a delay of up to 5 seconds before the "Ring" indication would be generated. This pause is only needed to allow a modem that was reset to finish initialization. If the modem was not reset, this pause is unnecessary.