**** BEGIN LOGGING AT Sat Jul 15 02:59:56 2006 Jul 15 11:19:41 anyone ever figure out how EJTAG's FASTDATA works? I've read the spec, but I can't make heads or tails of it. Jul 15 11:44:01 heh, i find the whole concept of the MIPS EJTAG rather confusing Jul 15 11:50:02 not as confusing as doing user loadable microcode instruction macros theres actual a big preformance problem when you are running a opperating system and the cpu is emulating a ppc for one app and a x86 for another , not to mention that the os is writen in its native assembly Jul 15 11:51:33 thats it takes longer for the cpu to swap the microcode sets when switching task that you may as well use 3 cpus Jul 15 11:56:45 is there a dat sheet for it ,, a link ? Jul 15 12:00:42 http://www.mips.com/content/Documentation/MIPSDocumentation/EJTAG/doclibrary Jul 15 12:02:24 :( got to register to view anything there Jul 15 12:03:02 http://www.amslink.com/file/mipsart1.html Jul 15 12:03:11 this is a nice (and short) explanation Jul 15 12:03:24 kool ty Jul 15 12:06:27 k so whats confusing you about it Jul 15 12:07:49 heh, well, i only had a brief look, and the idea of having the core explicitly request debug-handler operations from the EJTAG probe seems a bit confusing Jul 15 12:09:44 no ,, like a mode bit ,, if you initalise it with the jtag and put it in debug mode , they can easaly add extra debug checking bits to the instruction sequence Jul 15 12:11:19 they have a predefined set of them that will control things like break point and conditional debug instructions also Jul 15 12:11:54 can be done with little extra logic on the silicone Jul 15 12:13:25 but mips desing there own cores so they have the choice of how they impliment it , but the concept is fine Jul 15 12:15:41 its quite a good way for high tolarance systems that would use say 3 or 4 of thease doing the same job and using the jtag to compare the results Jul 15 12:17:08 and not only that but if one gives the wrong result it beeing able to pinpoint the error and fix it or work arround it Jul 15 12:20:13 it looks like it takes quite a lot of i/o pads to allow for it 25 by the looks of it Jul 15 12:21:17 but bet its in a bga package so wont realy matter that mutch Jul 15 12:25:24 you could probablay do remote code exicution via the jtag on it , but it would take a long time , could be usefull for getting around a boot flash fail , or loading a boot loader to program the flash Jul 15 12:27:37 that gives it production advantages , like it just gets plugged into one machine and it runs the boundry scan , initalises the flash and runs a self test and returns the result ,, Jul 15 12:32:45 think that is the sort of thing that should be posible even on microcontrolers Jul 15 12:34:08 there a sod to debug in the same way , you end up trying to simulate it than actualy beeing able to see any real data **** ENDING LOGGING AT Sun Jul 16 02:59:57 2006