Toni Mueller wrote: > Hello, > > I'm getting a reproducible panic on a remote 3.9-stable system and have > managed to talk someone into taking photographs with his cell phone > camera that show the panic and 'ps' and 'trace' info. The image quality > is generally poor, however. > > The crash occurs when using smartmontools (from ports). > > Would this be acceptable for filing a crash report, or is this out of > your scope?
The trick to answering a question like this is to imagine yourself as an OpenBSD developer. At this point, 4.0 is done. Out the door to the presses. You are now actively working on what will become 4.1. NOW, someone offers to show you a problem with 3.9, and a kernel panic at that. What is your interest? Definite curiosity. After all, you are an OpenBSD developer, when someone finds a problem, it is nice to make sure it is eliminated. However, it sure would be nice if the person asking about a reproducible problem would try reproducing it on -current code to see if it has already been solved, or if it is something new. Now they tell you they have labored for seconds to grab a digicam and produce photos, which will take you much more time to actually get into readable form on your screen than it took the person who is experiencing the problem to create them. And, you know from experience the photos are going to be either huge, so they take forever to download and are a pain in the butt to view, or tiny and fuzzy and unreadable. Once you make sense of the photo, you will very possibly find the problem is either a custom kernel or bad hardware or something that was solved six months ago. What is your interest now? almost Zero. Unfortunately, the difference between "almost zero" and "zero" was the question: is it a 10megapixel image (i.e., "is their camera better than mine?"), or is it a fuzzy, mostly unreadable phone-cam picture (or both). People are curious about which way you are about to annoy them. You gave that away. :) In general, especially with reproducible errors, try to reproduce them on -current. And then, forget the digicam. ONCE you have a developer's interest, ok, fine, maybe that particular developer will be interested in looking at your fuzzy and/or fat digicam pics to save time on later iterations of the resolution process, but you are unlikely to get their interest initially with a photograph. In general, I think it is usually best if the person ASKING for help expects to spend many times the effort and time they expect the person GIVING free help spend on a problem. Transcribe your photos to text, then file a proper bug report. If it is really reproducible, grab a serial cable, reproduce on -current, and skip the transcription. If you aren't willing spend a little time to do that, how important do you think the OpenBSD developers are going to consider your problem? How much time do you think they will expect you will spend verifying the fix? (and in case you are wondering, yes, I've transcribed quite a few panics, often while sitting on an 80386/25 that is now used as an uncomfortable chair). (and no, this is not aimed at Toni personally, but rather, at people who really don't think about the "other side" of the troubleshooting process. Toni was nice enough to ask before slinging photos around and expecting magic to happen. I'm just taking this opportunity to make a long-winded answer.) Nick.
