[Linux-streams] Re: LiS Problem

[David Grothe]

Try applying this patch to mod.c. The loadable module support (which I did not write) looks somewhat broken to me. This patch essentially disables most of the feature. There seemed to be some attempt to add non-loaded modules to the autopush lists. I'm not sure what caused them to be loaded. But the manner in which this was being done looks bogus to me and I don't have the time to wade into it in depth.

The effect of the patch will be that a module must be loaded via insmod or the like in order to be entered into an autopush list. I think that it would still be OK to I_PUSH an unloaded module, but I am not sure.

If this patch sounds like it will break anything that anyone is doing then perhaps someone with more time for it could look into the problem and come up with a better fix for it.

Thanks,
Dave

At 11:11 AM 9/12/2003 Friday, John Wenker wrote:

Dave:

Sorry to bother you again, but I forgot to mention three things in my last
e-mail.

(1) In LiS-2.8 (the version where autopush is working), the module ID in
lis_apush_init is indeed 3 instead of 7.  The '3' _does_ correspond to the
lis_fmod_sw index for my module.

(2) My module does not have an entry in config.h in LiS-2.8 (the driver
yes, but not the module), but does in LiS-2.16.

(3) In LiS-2.8, LIS_LOADABLE_SUPPORT is defined, and hence execution never
gets to the assert instruction that is failing in LiS-2.16.  In LiS-2.16
the LIS_LOADABLE_SUPPORT symbol is not defined.  Probably irrelevant, but
it is a difference nonetheless.

John

diffs
Description: Binary data