Monday, December 30, 2013

Emulator Version 0.18 Released

Nigel and I are pleased to announce that version 0.18 of the retro-B5500 emulator has been released. All changes have been posted to the Subversion repository for our Google Code project GitHub project. The hosting site has also been updated with this release, and for those of you running your own web server, a zip file of the source can be downloaded from the GitHub repository [updated 2022-05-07].

This is a bug-fix release. The only significant new feature is the ability to dump the system processor and memory state to a window in the B5500Console and B5500SyllableDebugger interfaces, as described below.

Processor Bugs Fixed

The following corrections were made in B5500Processor:
  1. The TRN (Transfer Numerics) syllable was not clearing zone bits in the destination character string as it should have. In addition, it was not always setting the True-False Flip Flop (TFFF, also known as MSFF) properly.
  2. The TRW (Transfer Words) syllable was incorrectly throwing an Invalid Address interrupt when the source string ended at an address which did not have a valid next address (e.g., the highest memory address, @77777).
  3. A call on cc.fieldIsolate() in the ISO (Field Isolate) syllable had an extraneous parameter. Thanks to Peter Grootswagers for identifying this.


More Corrections for Tape Drive Oscillation

We thought we had all of the issues with tape drives oscillating between rewind and ready states when tape was at BOT resolved in release 0.17, but alas, that proved not to be the case. Part of the problem was that the I/O Unit did not have tape writes implemented yet, but the rewind and interrogate functions are degenerate forms of tape write. As a result, result descriptors for Mod III I/O Units were not being properly constructed in some cases.

We also discovered, under Sid McHarg's advice, that the MCP is capable of detecting Mod III I/O Units automatically, but the problems mentioned above with the way that result descriptors were being constructed prevented that. With these fixes in place, that automatic detection now works properly, and the "SO USE OPTN 2" command described in the blog post for the 0.17 release is no longer necessary.

Dumping the System State

We have not had a good way to look at the complete system state up to now. The SyllableDebugger script provides some means to do this, but it is awkward to use when having to look at many widely-dispersed memory locations, and it's of no use at all when running the standard B5500Console user interface.

With this release, you can now dump the complete system state. In the SyllableDebugger, there is a Dump button near the top of the page. In the B5500Console interface, you can click the MEMORY CHECK indicator. Both methods will open a separate window and format both the processor state and the contents of core memory in that window. You can examine the dump in the window, save the window's contents, or copy/paste the contents into another application (such as a text editor) for further analysis. When you are finished with the dump, simply close its window.

These dumps can be requested any time the emulator is in a powered-on state. You can generate multiple dumps and have multiple dump windows open at a time. The emulator will pause while the dump is taking place and then resume operation. Thus, you can easily take a snapshot of the system state while the MCP is running, and then continue.

Further Work

We are continuing to see occasional flag-bit and invalid-address errors when running the MCP under heavy load. At present, we are working under the assumption that these are due to emulator bugs, and are pursuing them as a priority. The system state-dump feature mentioned above was implemented to facilitate this.

The next major feature scheduled for implementation is writing to tape, along with providing a means to persist tape volume data outside the emulator environment. After that we are considering implementing a more dynamic and user-friendly mechanism for controlling the system configuration, and implementing (finally) the double-precision numeric syllables.

No comments:

Post a Comment