On Fri, Sep 05, 2008 at 05:14:07AM -0400, Chuck Hallenbeck wrote: > A week ago Michael reported a problem after upgrading his Debian system > to kernel 2.6.26-1 with the associated speakup modules. His ltlk could > not handle lengthy outputs, and locked up his system completely. My LTLK has never been able to handle lengthly outputs with any of the speakup kernels I've used, actually. Long directory listings wouldcrash something )I couldn't tell you what), and the system could be crashed with a long directory lising. The keys would become inoperative and the synthesizer would continue to speak until it finally stopped. I'm assuming that the buffer in the synth finally ran out of text to speak, while the computer had crashed long ago. For some reason, I had assumed it was the UART's fault, probably because my Soyo motherboard uses a crappy one that the setserial package couldn't consistantly identify or set as far as CTS/RTS flow control went. The setserial man page mentioned something about junky UARTs, so I figured I had one of them, but was told that SpeakUP had it's own flow control handled in software. I'm not sure I can add anything more as to my own experiences and diagnosis, but I think it's a flow control problem that crops up when the LTLK buffer gets filled past capacity. My Wireless connection has been rather slow for the past few days and may have not overflowed the LTLK's buffer I can say that occasionally the LTLK will reset itself when using the Shane kernel at random times, and appear to be disconnected from what the proc filesystem says. The LTLK will be slow and quiet after the reset while the SpeakUP settings say otherwise. For instance, I keep the volume set at five, but after the LTLK mysteriously resets to maybe volume level three, pressing Speakup+2 will switch the LTLK to volume six, and I have to step it back down again. I'm not sure what causes the reset or crash of the LTLK at random intervals, but it happens usually after pressing Keypad Plus when I'm trying to read the entire screen, and has always reset while in the process of reading the screen...maybe a second or two into the reading of the text. I hope the information is of use in nailing down what's going on. Your problem appears to be the same as mine, and results from flow control problems, or a mismatched buffer size setting somewhere. I wish we could file a class-action lawsuit against some of these companies making it near-impossible to work with their equipment. If accessibility is supposed to be out right, then they're infringing on it with proprietary information reserved for themselves alone. Michael
