I really really appreciate your help. That is a very stupid bug, I should review my code more carefully before I sent out for help. Sorry about that, I guess somehow my brain just messed up. Anyway, I did turn on the CONFIG_PROC_FS. I also mount the usb by "mount -t usbdevfs none /proc/bus/usb"; and if I "cat /proc/filesystems" it does show "usbdevfs". However, is there a "inputdevfs" and is there a similar command line such as "mount -t inputdevfs none /proc/bus/input"? I don't see anything like "CONFIG_INPUT_DEVICEFS"? I am using 2.4.18 kernel. Thanks John -----Original Message----- From: Jan-Benedict Glaw [mailto:jbglaw@xxxxxxxxxx] Sent: Thursday, April 15, 2004 2:22 PM To: Linux/MIPS Development Subject: Re: questions about keyboard On Thu, 2004-04-15 14:08:54 -0600, Xu, Jiang <Jiang.Xu@xxxxxxxxxxxx> wrote in message <F71A246055866844B66AFEB10654E7860F1B10@xxxxxxxxxxxxxxxxxxxxxxx>: > Thanks for the reply, I did some testing and found some interesting > things: > > 1. Everytime I push the key on the keyboard, I can see something out > from /dev/input/event0 by "cat /dev/input/event0". However, I don't > see a directory named "proc/input", did I miss configure something in > the kernel? Maybe your kernel isn't configured with CONFIG_PROC_FS=y ? Maybe it's not mounted? > 2. read() does work and is a blocking read. However, if I use > select, then it does not work. > Select() never detects the state change. > Here is the sample code I am using: > { > int test_fd = -1; No need to initialize - you're assigning a value before accessing it. > fd_set rfds; > struct timeval tv; > > tv.tv_sec = 1; > tv.tv_usec = 0; tv_* need to be set before *every* select () invocation, not only once. > test_fd = open("/dev/input/event0", O_RDONLY); > if( test_fd < 0 ) > exit(0); > > while( 1 ) > { > FD_ZERO(&rfds); > FD_SET(test_fd, &rfds); > retval = select( 1, &rfds, NULL, NULL, &tv ); That's wrong. The "1" should be "fd + 1". > if( retval ) > printf("\nDetects something...."); A negative retval would also be != 0 ... > } > } > > What could be wrong? Most probably it's the hardcoded "1" with the select. That is, select only looks at all fd's which are smaller than one, so only fd=0 would be testes, but this one isn't in the set, so... MfG, JBG -- Jan-Benedict Glaw jbglaw@xxxxxxxxxx . +49-172-7608481 "Eine Freie Meinung in einem Freien Kopf | Gegen Zensur | Gegen Krieg fuer einen Freien Staat voll Freier Bürger" | im Internet! | im Irak! ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));
