On Thu, Nov 23, 2006 at 02:19:51PM -0800, Bakul Shah wrote: > > It "should" work, but people sometimes report that it doesn't > > (i.e. when they get the resulting panic). It at least needs to be > > investigated. > > What I suggested works for qemu since kldstat -m works for > aio and kqemu. May be people tried kldload without checking > if the module existed? kldload should do the equiv. of > kldstat -m.
The conditions of the panic are that a user has a module already compiled in statically to their kernel, and mistakenly tries to kldload a second copy of it. Beyond this I do not remember details other than that it is still being reported from time to time. > > > There's still the "stale module" problem. > > You mean an old module that you want to replace? No, a module compiled for a different kernel version (e.g. the user ran cvsup and rebuilt modules but has not yet built or booted the new kernel). Modules are notoriously fragile things, even in -stable. > > No, it only works for things that are modules: > > > > > grep -i compat /usr/src/sys/i386/conf/XOR > > options COMPAT_43 # Compatible with BSD 4.3 [KEEP > > THIS!] > > options COMPAT_FREEBSD4 # Compatible with FreeBSD4 > > options COMPAT_LINUX > > Ah, I see what you mean... The quick way is to default to > options INCLUDE_CONFIG_FILE but the DEFAULTS file has muddied > the situation as its contents are not shown. May be it is > better for config(8) to generate necessary sysctl glue for > options, makeoptions etc. Even better if sysctl kern.config > can used to recreate the config file of a running system.... That's more or less what I was suggesting. There should be a one stop shop for user programs to figure out the full set of running kernel features. Kris
pgpYznj6VmI88.pgp
Description: PGP signature
