Messages in this thread | | | Date | Sat, 26 Feb 2000 15:09:49 -0500 | From | Mike Panetta <> | Subject | Re: /dev/console should it always be available? |
| |
Unfortunately I dont have a serial port on the hardware either... It was very poorly designed... (IMO) I will try linking /dev/console to dev null or the patch I was sent to do "console=null". Hopefully this will work.
Oh and BTW... I did not have this problem with a 2.0.x kernel. Was any of the conesole code changed? Currently I have the problem with 2.3.40+ I have not tested anything else. (No 2.2.x kernels)
Thanks, Mike
On Sat, Feb 26, 2000 at 02:25:03PM -0500, Ricky Beam wrote: > On Sat, 26 Feb 2000, Mike Panetta wrote: > >Should /dev/console always be available even if there is no > >video card or keyboard on a machine? I keep getting these > > As a rule, yes. However, intel based systems tend to have serious problem > without a video card in there. Sparcs and most alphas (at least the server > hardware) uses the serial port as the console in the absence of a video > console. > > I look at this as a serious flaw in the x86 hardware design, but seeing > as x86 hardware is generally designed for "desktop" applications, I don't > hold it against anyone. (BTW, have you ever tried running NT on a machine > without a video device? It's not pretty unless you have a specialized > version.) > > >messages in the logs about init not being able to open(/dev/console) > >when I remove the video card from an embedded box I am working on. > > You can compile in serial console support and use "console=ttyS0" > to set the console to the serial port. There were patches as far > back as 1.2 for serial consoles on x86 hardware. > > >when I do this strange things start to happen like it cannot properly > >umount /dev/hda5 (not root partition) on shutdown, and mingetty starts > >to respawn to fast. This is related to a previous message I sent to the > >list about a week ago reguarding the ext2 corruption messages without > >a video card in the machine. Since then I have upgraded all the libraries > >to be compilant with the rquirements in Documentation/Changes and the > >corruption has gone away. Now I am stuck with unclean mounts and other > >random occurences... > > Bad programs making bad assumptions... from a purist standpoint, one > should check the return values from printf() and things [no one ever > does, tho'. I've seen some programs peg the CPU because of it.] > > >What am I to do? > > Make sure /dev/console points to a useable device at all times. One could > write a quicky program to check /dev/tty1, /dev/ttyS0, /dev/null, etc. and > replace /dev/console with a symlink. (Of course, one could argue the kernel > console driver should've done that at boot :-)) > > I quote devices.txt: > > 5 char Alternate TTY devices > 0 = /dev/tty Current TTY device > 1 = /dev/console System console > ... > > That says to me the kernel should always have a valid /dev/console (even if > it has to send it to /dev/null.) > > --Ricky > > PS: programs not having a stdin, stdout, and stderr is a nasty problem. >
--
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |