On Fri, 29 Sep 2000, Keith Owens wrote:
> >Maybe it still applies to beta2, otherwise I can redo it on beta2.
> >Basically, I suggest to move kdb_vmlist_check() out of vmalloc.c because
> >it makes kdb patch less intrusive into non-kdb-specific code.
>
> I have to agree with Scott on this one. A kdb hook for a particular
> area needs to be in the source where all the other work for that area
> is done. It picks up unexpected kernel changes much faster and tells
> people who are integrating multiple patches that there might be
> conflicts in that area.
Ok, I suppose it makes sense - I do not strongly disagree and also confess
that one of the reasons is that my vx-linux patch also contains other
changes to vmalloc.c and I never trust patch(1)'s "hunking succeeded"
stuff and always check the conflicts manually, just in case :)
Regards,
Tigran
|