KernelBugTriage

From FedoraProject

Revision as of 09:03, 11 January 2012 by Sgruszka (Talk | contribs)

Jump to: navigation, search

Contents

Quick links to bug lists

Possibly fixed in CURRENTRELEASE (Stick for prodding sleeping bugs with)

If the bug is against a previous point release kernel, and the reporter hasn't re-tested (this type of bugs must be set in NEEDINFO_REPORTER state):

Thank you for taking the time to report this bug. Updates to the kernel have been released since it was first reported. If you have time to update the package and re-test, please do so and report the results here. You can obtain the updated package by typing 'yum update kernel' or using the graphical updater, Software Update. If the problem no longer exists then please close this bug or it will be closed in a few weeks if there is no additional information lodged.

Installer bugs

If the bug refers to problems when installing Fedora in any of the possible ways CD, DVD, USB, network, etc (this type of bugs must be set in NEEDINFO_REPORTER state):

Thank you for taking the time to report this bug. Bugs involving the installation process cannot usually be resolved after release date. We therefore ask reporters to use one (or both) of the following options:
Test with an Alpha, Beta or Release Candidate for the next version. These can be found at http://fedoraproject.org/get-prerelease and by testing these we can work to resolve any installation issues so that the final release is free of such bugs.
However if the problem no longer exists then please close this bug or it will be closed in a few weeks if there is no additional information lodged.
Important.png
get-prerelease link
The get-prerelease link sometimes is not available or have not updated development releases.

Video Subsystem bugs

The various video subsystems that have kernel modesetting support would just prefer their bugs to be moved over to the xorg-x11-drv-{intel|nouveau|ati} component. There, their triagers can gather needed info and debug the issue. If the problem is in the kernel, those maintainers can fix it, no need to leave the bug assigned to the kernel.

This bug is in a video subsystem that has a kernel part. We track and work on these bugs via the driver package name instead of leaving them assigned to the kernel

NEW state bugs

If the bug is new and (this type of bugs must be key-worded in either WHITEBOARD or BLOCKER): Hello,

Thank you for taking the time to report this bug. I have added the proper keyword to this bug to bring it to the attention of the <subsystem-name> subsystem maintainer.
Can you attach the following information to this bug:

Then pick the info that is useful in this case:

  • smolt profile
  • dmesg output from boot before issue.
  • dmesg output after issue.
  • dmesg output from boot with 'quiet' option replaced with 'debug'
  • lspci output
  • lsusb output

For other necessary information check:

Rebase Regressions

The kernel is often rebased to the latest stable kernel release and pushed as an update. This fixes a large amount of bugs, but it inevitably introduces a few regressions. If a bug is reported and it is determined that it worked with the previous kernel release, then a rebase regression entry should be added to the whiteboard field in the bug. This entry takes the form of:

rebase-regression-<kernel version>

E.g. if there is a regression when moving to the 3.1 kernel, it should be:

rebase-regression-3.1

The latest working Fedora kernel version and the earliest broken version should be noted in the comment section if possible.

This will allow the kernel developers to track regressions introduced by a rebased kernel. That will also help facilitate our communication with the upstream kernel developers, as we can report these bugs to them to be included in the regression lists that are published upstream.

Bugs with TAINTED modules

Certain bug reports refer to a kernel module that is not open source software (mark this bugs with the keyword TAINTED):

Thank you for taking the time to report this bug. Sadly, your report contains information about a not open-sourced module. This means you have loaded a module into your kernel that we can't debug, making it difficult to isolate the issue. Can you reproduce the issue without loading the tainted <module-name> module?.
If you cannot duplicate the bug without the tainted module loaded it will be closed.
However, you can report it to the vendor of your non-free module.

Bug state transitions

  • A bug marked as MODIFIED has patches in testing and should not have their status changed. An exception to this is when a bug has been in MODIFIED state for some time (this usually indicates that the issue was fixed in an update, and no-one ever closed the bug. If in doubt, ping the reporter to retest with the latest build).
  • If a bug has been in NEEDINFO for several months with no follow-up from the reporter, there's not a great deal we can do. In this situation, closing the bug is the only recourse available (which does actually tend to 'wake up' some reporters who then reopen the bug).
  • When marking bugs as duplicates, we want the one with the most information to be the one that remains open, even if this means a higher numbered bug gets closed.
  • Don't close bugs marked beginning with [mmconf] or [msi] (bugs beginning with [$something] are usually specifically marked so developers can quickly see the main issue within the bug)

Basics

  • If there's an upstream (http://bugzilla.kernel.org) bug that matches
  • If the bug is against an already released Fedora (Ie, Fedora 7) and there isn't much information to go on "my machine locked up" for eg, request that the user try to reproduce the problem using the kernel-debug kernel. The additional debugging checks may trigger some clues.
  • If the bug you are triaging contains a patch, please add [PATCH] to the beginning of the summary line of the bugzilla.
  • Sometimes a user will attach a kernel panic as a jpeg, or in worst case, as a tarball of their /var/log/messages.
    • It saves the kernel team time if the kernel oops parts of these are transcribed/cut out and pasted into the bug as text.
  • Additionally, making sure that text attachments of bugs have their MIME type set to text/plain can save some time.
  • You can ask reporters to try and duplicate with the latest nightly live composes: http://alt.fedoraproject.org/pub/alt/nightly-composes/
  • You can ask reporters to try the newest kernel available in koji: http://koji.fedoraproject.org/koji/packageinfo?packageID=8
  • If the bug appears to be video/kernel modesetting related, reassign it to the xorg-x11-drv-{intel,ati,nouveau} component.
  • If the reporter is using a custom compiled kernel, close the bug as a duplicate of https://bugzilla.redhat.com/show_bug.cgi?id=126342

Common Problems

Point reporters to: KernelCommonProblems

Bug assignment

For certain categories of bug, assigning/cc'ing people responsible for the relevant subsystem is a good idea. When reassigning Add kernel-maint@redhat.com to Cc:

Subsystem Reassign to Add to Cc: Other notes.
intel video xorg-x11-drv-intel
radeon video xorg-x11-drv-radeon
nouveau video xorg-x11-drv-ati
Installer bob@fedoraunity.org, kanarip@kanarip.com, jbwillia@math.vt.edu
Dell-specific linux-bugs@dell.com
ACPI acpi-bugzilla@lists.sourceforge.net
ELF interpreter <FIXME>@redhat.com
PATA
SATA jgarzik@redhat.com <FIXME?>
USB zaitcev@redhat.com
SCSI
device-mapper lvm-team@redhat.com
Sound
Firewire fenlason@redhat.com stefan-r-rhbz@s5r6.in-berlin.de
Wireless linville@redhat.com sgruszka@redhat.com
iwlegacy wireless driver (iwl3945 and iwl4965) sgruszka@redhat.com linville@redhat.com
Ralink wireless driver (rt2x00) sgruszka@redhat.com linville@redhat.com
Realtek wireless driver (rtlwifi) linville@redhat.com larry.finger@lwfinger.net
Broadcom wireless driver (b43) linville@redhat.com larry.finger@lwfinger.net
Broadcom wired drivers (bnx2, bnx2x, b44, tg3) mcarlson@broadcom.com
CIFS jlayton@redhat.com steved@redhat.com, staubach@redhat.com
EXT2/EXT3/JBD/XFS esandeen@redhat.com (Or pretty much any filesystem stuff goes to Eric)
BTRFS jbacik@redhat.com
CPU frequency scaling davej@redhat.com anton@redhat.com
3D (DRM/AGP) airlied@redhat.com Change component to xorg-x11-drv-{intel,ati}
udev related kay@redhat.com
USB cameras hdegoede@redhat.com
KVM fedora-virt-maint@redhat.com
Xen fedora-virt-maint@redhat.com ketuzsezr@darnok.org
ptrace (and utrace) oleg@redhat.com
inotify eparis@redhat.com
IMA (integrity measurement) eparis@redhat.com
DMAR (Intel IOMMU) dwmw2@infradead.org
r8169 ethernet driver romieu@fr.zoreil.com
PCI / PnP resource allocation bhelgaas@google.com
Video4Linux mchehab@redhat.com