[Kernel-packages] [Bug 1703339] Re: Unable to install Ubuntu on the NVMe disk under VMD PCI domain

2017-09-06 Thread Keith Busch
The intended use for VMD is as a PCIe storage domain, so we expect only to see root ports, switches and NVMe SSDs in the VMD domain. Linux VMD is implemented just as a PCIe host bridge driver in order to reuse existing device drivers, though, so you could theoretically connect any pci device in

[Kernel-packages] [Bug 1626679] Re: NVMe triggering kernel panic followed by "bad: scheduling from the idle thread!"

2017-02-17 Thread Keith Busch
A couple issues here. The nvme driver was ported from 4.5, but the block layer was based on 4.4, so there was a mismatch in how to handle timeouts. That was this launchpad: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1597908 But also, there is a bug handling legacy IRQ that only

[Kernel-packages] [Bug 1597908] Re: linux-kernel: Freeing IRQ from IRQ context

2016-07-28 Thread Keith Busch
Kamil, For your issue, this is potentially a bug specific to the 4.5 driver. We used to poll once a second, which would hide legacy irq issues during initialization, but 4.5 removed it. A patch[*] to 4.6 removed all use of legacy interrupts, and should have been applied to 4.5 stable (they were

[Kernel-packages] [Bug 1597908] Re: linux-kernel: Freeing IRQ from IRQ context

2016-07-28 Thread Keith Busch
Hi Kamil, It looks like the kernel issue is resolved. Your drive just appears to be in a bad state and the driver reacts accordingly. I think this launchpad's issue is resolved and should request the fix go forward. -- You received this bug notification because you are a member of Kernel

[Kernel-packages] [Bug 1597908] Re: linux-kernel: Freeing IRQ from IRQ context

2016-07-01 Thread Keith Busch
Thanks for the quick build! I forwarded this onto the original reporter and will send the results when I hear them. We'll get it tested though might be next week for the results due to holiday weekend. -- You received this bug notification because you are a member of Kernel Packages, which is

[Kernel-packages] [Bug 1597908] Re: linux-kernel: Freeing IRQ from IRQ context

2016-06-30 Thread Keith Busch
This system crashes making apport-collect not possible after the fact, though I confirm it is a bug. As the upstream nvme driver maintainer, I can recommend either which driver commits need to be reverted, or which kernel commit needs to be cherry-picked (preferring the latter :)). Here is a

[Kernel-packages] [Bug 1382221] Re: please add linux nvme block driver commit

2014-11-10 Thread Keith Busch
** Tags removed: verification-needed-trusty ** Tags added: verification-done-trusty -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux in Ubuntu. https://bugs.launchpad.net/bugs/1382221 Title: please add linux nvme block driver

[Kernel-packages] [Bug 1382221] [NEW] please add linux nvme block driver commit

2014-10-16 Thread Keith Busch
Public bug reported: This is a bug in a linux kernel driver, I'm basing off ubuntu trusty tree: http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-trusty.git;a=summary We need an unpstream commit from the NVMe driver to prevent a kernel crash. It's this one here:

[Kernel-packages] [Bug 1382221] Re: please add linux nvme block driver commit

2014-10-16 Thread Keith Busch
We already confirmed the bug and fix upstream during 3.14 merge. I should have marked the commit for stable. I (embarrassingly) wrote the original code that committed this bug upstream, but I also wrote and committed the fix! :) ** Patch added: Fixes the bug, attaching patch committed upstream