All JP1.3 extenders have a bug in them....

unclemiltie · Post by **unclemiltie** » Tue Apr 14, 2009 11:00 pm

This may explain some random crashes. Thanks to AllanW who not only reported this but disassembled the code and chased down the bug in the 3039 extender. I used the same code in all of the remotes so this impacts all of the JP1.3 extenders.

I'm in the process of releasing new extenders for all of the JP1.3's anyway to support IR 8 and the setup code validation and a couple of new things, so this was good timing.

The details:

1: When adding a 16-bit number, you have to ADD the low byte and then ADD with Carry the high byte (ADC) so that any carry from the low add makes it into the high byte. unfortunately, I was ADDing the high byte first and then ADCing the low byte. oops. So this bug was data dependent. The code was in the part of the extender that loaded macros, so it could make some odd macros appear and things would go haywire.

2: In that same block, I was decrementing a 16-bit pointer (DECW) and then jumping based on that being zero. Allan found an errata in the S3F80 users manual that says that combination can result in strange behavior. I'm not sure we ever saw anything bad here, but it made sense to change the code to ensure that nothing bad happens.

Quite the sharp eye, credit goes out to Allan.

I'm going to China on Friday, some of my usual testers have the builds of the extenders now and have been testing. I've tried the Atlas OCAP and the Comcast DVR and all seems well. Hopefully by the time I return I'll have the results of the testing and I'll be able to release the new extender versions.

-bill

xnappo · Post by **xnappo** » Fri Oct 02, 2009 3:27 pm

So I think I am seeing this bug and therefore am looking to update to 2.10 - however I really want to use RM-IR. (I am seeing weirdness where macros don't work right, but if I move the order of the macro definitions around in memory then they do - irregardless of using IR or RMIR btw).

I am a bit confused as to the status - it seems like it should work with an RDF4 file, but doesn't support the additional checking?

xnappo

unclemiltie · Post by **unclemiltie** » Fri Oct 02, 2009 11:28 pm

don't understand the question. The bug has been fixed in 2.10 (and all of the other JP1.3 extenders) and there is support for IR 8 in the RDF's now. What do you see not working? Can you give me an example?

(I don't use RMIR so I'm at a bit of a disadvantage)

gfb107 · Post by **gfb107** » Sat Oct 03, 2009 5:56 am

Maybe xnappo is asking if RMIR supports the RDF 4 enhancements. The answer to that question is that some are and some aren't, which is also true of "normal" RDF features.

If you want to use RMIR and find that a feature you use is missing or broken, please post in the RM to-do thread or start a bug report thread. Features are far more likely to get implemented or fixed when there are users actively trying things and providing feedback.

xnappo · Post by **xnappo** » Sat Oct 03, 2009 9:12 am

gfb107 wrote:Maybe xnappo is asking if RMIR supports the RDF 4 enhancements. The answer to that question is that some are and some aren't, which is also true of "normal" RDF features.

I don't really care all that much about the enhancements - the readme for the extender says that you must use IR 8.0. So I didn't know if that meant RMIR won't work or just means IR 7.x won't work.

I will give it a shot - once you get used to maintaining in RMIR it is hard to think about going back.

xnappo

gfb107 · Post by **gfb107** » Sat Oct 03, 2009 9:24 am

RMIR is tolerant of all RDF 4 enhancements, even those that pertain to features not yet implemented.