Hi, I've got two odd issues I've encountered that I cannot solve. Both are seen on two different machines (U30, E250), 2.6 kernel (tried: 2.6.12.1, 2.6.12.3, both vanilla sources). Both issues relate to serial console (neither machine has any video).
The first issue is that the kernel seems to feel it necessary to print 70+ lines of whitespace (in groups of two 30 or 40ish packs, with an actual "reset-cursor-to-0,0" after each) when booting. It's making troubleshooting some problems I'm having maddeningly difficult. The activity seems to take place just prior to "PROMLIB: ..." output, and nicely destroys my scrollback buffer. I first suspected the use of initrd, but a non-initrd kernel has the same behavior, so it seems to be in the kernel itself. This behavior occurs regardless of the terminal I use (minicom, cisco 2509). The second issue arrises if the kernel panics during boot (like a stupid system operator forgot to set the root= line correctly in silo.conf ><); the serial console becomes totally unusable. On an ix86 box, I wouldn't care; but on a sparc machine, I want to be able to send a break and use the OBP to recover. This issue, however, means I have to go physically power cycle the box. This happens whether I use a direct serial link with minicom, or my "reverse telnet" via a 2509. The 2509 reports "noCTS noDSR* DTR RTS" after this event (normal operation is "CTS DSR* DTR RTS"). It seems to cause any input I provide after the panic to be queued on my end, since as soon as the machine is back up it drops to OBP if I sent a break signal after the panic. Both issues are highly irritating, the latter especially so when I'm not at the machine's physical location. Any ideas? Thanks, -- Adam [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
