On Fri, Mar 06, 2009 at 08:58:08PM -0500, Daniel Ouellet wrote: > Stefan Sperling wrote: >> On Fri, Mar 06, 2009 at 06:07:00PM -0500, Daniel Ouellet wrote: >>> Insan Praja SW wrote: >>>> Hi, >>>> On Sat, 07 Mar 2009 03:17:57 +0700, FRLinux <frli...@gmail.com> wrote: >>>> >>>>> On Fri, Mar 6, 2009 at 7:12 PM, Insan Praja SW >>>>> <insan.pr...@gmail.com> wrote: >>>>>> Hi Misc@, >>>>>> on a i386 kernel recent build (6th march), I got panic. It says: >>>>> Hello, >>>>> >>>>> As far as I know, home built kernel is not supported, you need to try >>>>> out a snapshot instead and see if it works. >>>>> >>>>> Cheers, >>>>> Steph >>>> You are right, but I always had a backup of last working kernel, >>>> and that is what I use now. But this panic happens and I like to >>>> report it to see if anyone else experiencing the same panic, with >>>> home build kernel or snapshot. It's a generic kernel, anyway, I >>>> hope I can contribute in some other way, you know.. like testing >>>> diff or finding bugs. >>>> I also use sendbug(1) to report the panic. >>>> Thanks, >>> You just don't built home build kernel at all. This is really not >>> linux here. You can configure all you want on it as is. >> >> So what if I want debug symbols to produce meaningful traces >> from kernel core dumps with gdb? Then I have to compile with >> DEBUG="-g" to get a bsd.gdb. Then I have a self-compiled kernel >> already. > > That wasn't the question, but again, if you know that you need "-g" and > are looking at kernel core dumps then you wouldn't asked questions about > it on misc@ would you? Stay on the topic as it was asked. And it sure > wasn't a question about the core dump used with "-g" was it? But related > to icmp. > >> And what if I'm testing diffs posted to t...@? >> When testing diffs you usually don't only run them for 5 minutes. >> You usually run them for as long as you can. > > Then your question would have been on tech@ related to a spefici diff as > well from tech@ too, but it wasn't. > >> I guess these faq entries are there to stop people from tweaking >> the config so hard that their machine cannot boot anymore, and >> then reporting this as a bug. They don't exist to stop people who >> somewhat know what they are doing from reporting things they find >> in kernels they've compiled themselves. > > They are there to make sure valid tests are done on generic kernel as is > and valid meaning full reports are sent in that can be reproduce by > others and get fix. Not to asked a free for all home built kernel from > anyone. > >> And note that there have recently been changes in the way pf >> keeps track of icmp, so this may well be a valid report. > > Could sure be I give you that. However, still true that snapshot is the > way to go and see the results. This is not one of these is it? There > isn't a snapshot for the 6 ready yet anyway. > > However there is a commit already for icmp on pf as well: > > http://marc.info/?l=openbsd-cvs&m=123638870222588&w=2 > > It may well address this issue for sure, or it may not. > > The idea and intend still stand that it's not for everyone. Good one are > important and useful and this may have been one of them. > > And if the same problem still exists then with a snapshot, I am sure > someone will be more then happy to look into it. > > Hope this help to provide a bit more details as to what the intent of > the faq are and what the spirit of my suggestion was. > > Fell free to disagree, that's fair. >
Sorry, I don't get it a non-developer tries to educate a developer about how kernel crashes should be reported? Sorry most of your standpoints are just wrong. Sure people are encuraged to run snapshot kernels but selfbuilt kernels are fine as long as they're built from a unmodified GENERIC config. Let us developers take care of yelling at those people who send in bad bug reports because we're acctually the people who may fix it in the end. -- :wq Claudio
