View previous topic :: View next topic |
Author |
Message |
jsnorman Tux's lil' helper
Joined: 24 Feb 2005 Posts: 131
|
Posted: Sat Sep 16, 2006 12:44 am Post subject: Rant on bug reports |
|
|
I have tried in the past to be a good Gentoo citizen, submitting bug reports when I run into .. bugs.
Yet bug reports never seem to get taken seriously and there seems to be a desire to just close them, or ignore the problem as someone else's issue (the java 1.5 bug is a good example of that).
The latest annoyance - after repeatedly having emerge --sync crash with kernel oops, I reported a bug.
To my surprise, the bug was closed because I use Nvidia's driver! (thereby "tainting" the kernel).
I expect this from Microsoft and commercial vendors, but seriously not from Gentoo.
I am sure the bug is due to my use of the cutting edge glibc and I am not really all that annoyed by the bug - I reported it to help out Gentoo so the developers could focus on what is likely some kind of issue with the latest glibc. I highly doubt on the other hand that it has anything to do with Nvidia. This seems more of a political statement than a rational response to the bug report. And frankly I am sick of the rabid GPL supporters who see no other way.
Sure, if I have time I will try to boot and not load the nvidia module, and run emerge --synch from a VT. But it is this kind of annoyance that has convinced me not to submit any more bugs. I hope the developers who are listening will try to take user bug reports a little more seriously since from everything I can see we are supposed to be submitting the reports whenever we have an issue that seems like a bug. |
|
Back to top |
|
|
mark_alec Bodhisattva
Joined: 11 Sep 2004 Posts: 6066 Location: Melbourne, Australia
|
Posted: Sat Sep 16, 2006 2:55 am Post subject: |
|
|
Moved from Other Things Gentoo to Gentoo Chat. Not a support request.
Would you mind posting a link to the bug reports so that others can look over them to see if they were indeed closed unjustifiably. _________________ www.gentoo.org.au || #gentoo-au |
|
Back to top |
|
|
jsnorman Tux's lil' helper
Joined: 24 Feb 2005 Posts: 131
|
Posted: Sat Sep 16, 2006 6:49 am Post subject: |
|
|
Sure thing, though I am not really trying to harsh on the particular developer who closed this bug. It really seems more of a global comment since I search bug reports kind of regularly running ~amd64 and some svn packages. I see it in other people's bugs much more; I will have to do some backtracking to find some examples and I will add them here tomorrow. Also, I would add that many I have submitted several other bugs that were taken very seriously and fixed quickly, so obvious it depends on who gets it.
Here is the "tained kernel" response: https://bugs.gentoo.org/show_bug.cgi?id=147641
Here is the Java bug I referred to: https://bugs.gentoo.org/show_bug.cgi?id=65937. Started in 2004. Problem was essentially solved (conceptually) in comment 20 in 2005. But the bug seemed to stagnate for months with no responses or work being done on it, or at least no updates. Very few people cared to respond even though for many users this was a critical bug (problem solution implemented 6/30/2006). 2 years for a critical bug, with user reports being almost ignored (or at least there was no response) is surprising.
Jeff |
|
Back to top |
|
|
Earthwings Bodhisattva
Joined: 14 Apr 2003 Posts: 7753 Location: Germany
|
Posted: Sat Sep 16, 2006 8:07 am Post subject: |
|
|
You are asked to remove the nvidia module, try to reproduce the kernel oops and reopen the bug. What's so bad about that? _________________ KDE |
|
Back to top |
|
|
anello Guru
Joined: 17 Jul 2005 Posts: 557 Location: EU -> DE -> Stuttgart
|
Posted: Sat Sep 16, 2006 9:09 am Post subject: |
|
|
Earthwings wrote: | You are asked to remove the nvidia module, try to reproduce the kernel oops and reopen the bug. What's so bad about that? |
ditto. He's just trying to find a solution and wants to get the options out of the way that he can't 'control'. No signs at all of disproportionate GPL enthusiasm.
PS: I'm refering to the 'tainted kernel' bug. _________________ Antonino Catinello | http://catinello.eu |
|
Back to top |
|
|
toralf Developer
Joined: 01 Feb 2004 Posts: 3925 Location: Hamburg
|
Posted: Sat Sep 16, 2006 9:18 am Post subject: |
|
|
OTOH Jakub Moc is IMO very rigid in closing bugs. |
|
Back to top |
|
|
Fabiolla Apprentice
Joined: 04 Mar 2004 Posts: 277 Location: somewhere
|
Posted: Sat Sep 16, 2006 11:24 am Post subject: |
|
|
Well, another thing is the long time for a simple version bump
Until now (2006-09-16) no response.
Comment two is from me, and also a second user can confirm that the unofficial ebuild works.
What more can a user do than update the ebuild self, install application in local overlay and write a comment in bug report?
greetings |
|
Back to top |
|
|
jsnorman Tux's lil' helper
Joined: 24 Feb 2005 Posts: 131
|
Posted: Sat Sep 16, 2006 2:14 pm Post subject: |
|
|
Earthwings wrote: | You are asked to remove the nvidia module, try to reproduce the kernel oops and reopen the bug. What's so bad about that? |
My reaction was to the "tainted" kernel bit. The kernel is "tainted" because of license issues. It is not like Nvidia puts out crappy drivers or anything. Is there any real reason to suspect that the nvidia-driver module has anything to do with a kernel oops generated during rsync? Really? I have been programming for almost 20 years now, and that is not the first thing that would come to my mind.
The problem is, I spent 30 minutes just preparing the report for this bug. To shut down my system (which, thanks to the great work the Gentoo developers have done, I do RARELY), modify my startup scripts to not load nvidia and boot to a VT rather than X, etc., and then go through the process to resubmit the bug is kind of a lot of stuff to ask a user to do based on what seems from my perspective to be a political/license issue more than anyting. I am very happy to spent time diagnosing stuff if there is a real reason but this seemed more to me like jumping through hoops.
If the oops I gave him references code that he determined may actually implicate the nvidia driver, and he looked at it and made that determination and therefore was asking me to test removing nvidia-drivers for a legitimate reason - my mistake and I am quite happy to help with that, and I completely withdraw my rant in that case.
As it is, though, I can get 1100fps, run any game, and since the latest nvidia-drivers have not had a graphics related problem at all. And how rsync would trigger a fatal interaction at the kernel level with nvidia??? I suppose anything is possible, it just seems like there are more likely culprits to do testing on - like my 20060805 cvs version of glibc?
I probably should have added this to the original rant - but another reason for frustration was the new limits on what you can actually put in the "box" on a bug report. I could not even FIT my output from emerge --info. Had to save it to a file and attach it! And then in another bug I could not upload the full output from a gcc fault (5MB limit on file uploads) - I had to host it on my file server. Is gentoo really that short on file space? |
|
Back to top |
|
|
dleverton Guru
Joined: 28 Aug 2006 Posts: 517
|
Posted: Sat Sep 16, 2006 2:49 pm Post subject: |
|
|
jsnorman wrote: | My reaction was to the "tainted" kernel bit. The kernel is "tainted" because of license issues. It is not like Nvidia puts out crappy drivers or anything. Is there any real reason to suspect that the nvidia-driver module has anything to do with a kernel oops generated during rsync? Really? I have been programming for almost 20 years now, and that is not the first thing that would come to my mind.
|
The problem is that, since the nvidia module is closed source, the developers have no way of knowing whether it has any problems. It's true that the crash probably wasn't related to video, but since the module is in the kernel, any bugs can corrupt any part of the system. The upstream kernel developers would have told you the same thing - they just don't have time to hunt down problems that they might have no way of finding or fixing. |
|
Back to top |
|
|
mark_alec Bodhisattva
Joined: 11 Sep 2004 Posts: 6066 Location: Melbourne, Australia
|
Posted: Sat Sep 16, 2006 4:02 pm Post subject: |
|
|
jsnorman wrote: | The problem is, I spent 30 minutes just preparing the report for this bug. To shut down my system (which, thanks to the great work the Gentoo developers have done, I do RARELY), modify my startup scripts to not load nvidia and boot to a VT rather than X, etc., and then go through the process to resubmit the bug is kind of a lot of stuff to ask a user to do based on what seems from my perspective to be a political/license issue more than anyting. I am very happy to spent time diagnosing stuff if there is a real reason but this seemed more to me like jumping through hoops. |
It is actually very easy to stop the nvidia kernel from loading and X from starting. Just comment a line in /etc/modules.autoload.d/kernel-2.6 and `rc-update del xdm default`. As has been mentioned, removing the nvidia module is needed, because if it is indeed a kernel bug, then upstream would not touch it if you had a binary module running. _________________ www.gentoo.org.au || #gentoo-au |
|
Back to top |
|
|
drwook Veteran
Joined: 30 Mar 2005 Posts: 1324 Location: London
|
Posted: Sat Sep 16, 2006 4:22 pm Post subject: |
|
|
or the entire minute it'd take to switch to the nv/vesa driver if you wanted to keep X up while trying to reproduce... |
|
Back to top |
|
|
ciaranm Retired Dev
Joined: 19 Jul 2003 Posts: 1719 Location: In Hiding
|
Posted: Sat Sep 16, 2006 4:40 pm Post subject: |
|
|
jsnorman wrote: | My reaction was to the "tainted" kernel bit. The kernel is "tainted" because of license issues. It is not like Nvidia puts out crappy drivers or anything. Is there any real reason to suspect that the nvidia-driver module has anything to do with a kernel oops generated during rsync? Really? |
Uh, yes. Nvidia put out extremely crappy drivers that routinely cause random corruption to other parts of the kernel. A broken driver is by far the most likely cause of a kernel showing strange problems when under heavy interrupt load -- of which rsync is a perfect source. |
|
Back to top |
|
|
Carlo Developer
Joined: 12 Aug 2002 Posts: 3356
|
Posted: Sat Sep 16, 2006 6:26 pm Post subject: |
|
|
toralf wrote: | OTOH Jakub Moc is IMO very rigid in closing bugs. |
Right. OTOH bug wrangling is not an easy or rewarding job. Nevertheless: Who has a problem with developers actions or feels treated incorrectly is asked to contact the user relations team.
Fabiolla wrote: | What more can a user do than update the ebuild self, install application in local overlay and write a comment in bug report? |
Become active.
jsnorman wrote: | My reaction was to the "tainted" kernel bit. The kernel is "tainted" because of license issues. |
It is tainted because of a binary blob in it, so ist can not be properly debugged. That has nothing to do with the license. _________________ Please make sure that you have searched for an answer to a question after reading all the relevant docs. |
|
Back to top |
|
|
|