Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!

Bug 511444

Summary: sys-fs/zfs-kmod-0.6.2-r5 with sys-kernel/gentoo-sources-3.14.4 - upon reading /proc/spl/kstat/zfs/arcstats - BUG: unable to handle kernel NULL pointer dereference at 0000000000000080 in [<ffffffffa04618ac>] __kstat_create+0x66c/0x690 [spl]
Product: Gentoo Linux Reporter: Matthias Maier <tamiko>
Component: [OLD] Core systemAssignee: Richard Yao (RETIRED) <ryao>
Status: RESOLVED INVALID    
Severity: normal    
Priority: Normal    
Version: unspecified   
Hardware: All   
OS: Linux   
Whiteboard:
Package list:
Runtime testing required: ---
Attachments: Full dmesg output with kernel oops
emerge --info
kernel configuration

Description Matthias Maier gentoo-dev 2014-05-25 12:06:05 UTC
With sys-fs/zfs-kmod-0.6.2-r5 (sys-fs/zfs-0.6.2-r5, sys-kernel/spl-0.6.2-r5) I encounter a NULL pointer dereference and Kernel Oops upon reading /proc/spl/kstat/zfs/arcstats (see attached dmesg).

This happens with sys-kernel/gentoo-sources-3.14.4, the currently stable version sys-kernel/gentoo-sources-3.12.13 is fine.

I also encounter some strange memory leak problems - slowly declining MemFree and MemAvailable by constant memory consumption from userland, and kernel (slab, buffer, cache) that I suspect are related to this. So this might be an indicator for a bigger problem with memory management - or memory accounting (which changed a bit in 3.14, e.g. new MemAvailable).

emerge --info and kernel configuration attached.

Reproducible: Always
Comment 1 Matthias Maier gentoo-dev 2014-05-25 12:07:00 UTC
Created attachment 377610 [details]
Full dmesg output with kernel oops
Comment 2 Matthias Maier gentoo-dev 2014-05-25 12:07:46 UTC
Created attachment 377612 [details]
emerge --info
Comment 3 Matthias Maier gentoo-dev 2014-05-25 12:09:21 UTC
Created attachment 377614 [details]
kernel configuration
Comment 4 Matthias Maier gentoo-dev 2014-05-25 21:34:51 UTC
Well, I tried to investigate a bit and I could not reproduce on a test setup. Later, upon completely rebuilding everything the kernel oops (and the memory consumption problem) do not show up any more. So, this was very likely a layer 8 problem.

Sorry for the noise.