| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | Debugging Modules after 2.6.3 | 
|  | 2 | ----------------------------- | 
|  | 3 |  | 
|  | 4 | In almost all distributions, the kernel asks for modules which don't | 
|  | 5 | exist, such as "net-pf-10" or whatever.  Changing "modprobe -q" to | 
|  | 6 | "succeed" in this case is hacky and breaks some setups, and also we | 
|  | 7 | want to know if it failed for the fallback code for old aliases in | 
|  | 8 | fs/char_dev.c, for example. | 
|  | 9 |  | 
|  | 10 | In the past a debugging message which would fill people's logs was | 
|  | 11 | emitted.  This debugging message has been removed.  The correct way | 
|  | 12 | of debugging module problems is something like this: | 
|  | 13 |  | 
|  | 14 | echo '#! /bin/sh' > /tmp/modprobe | 
|  | 15 | echo 'echo "$@" >> /tmp/modprobe.log' >> /tmp/modprobe | 
|  | 16 | echo 'exec /sbin/modprobe "$@"' >> /tmp/modprobe | 
|  | 17 | chmod a+x /tmp/modprobe | 
|  | 18 | echo /tmp/modprobe > /proc/sys/kernel/modprobe | 
| Robert P. J. Day | 0cadfc0 | 2008-02-03 15:27:38 +0200 | [diff] [blame] | 19 |  | 
|  | 20 | Note that the above applies only when the *kernel* is requesting | 
|  | 21 | that the module be loaded -- it won't have any effect if that module | 
|  | 22 | is being loaded explicitly using "modprobe" from userspace. |