|
Maybe because neither data loss nor data corruption are that great? Here's another link: http://lmgtfy.com/?q=xfs+arm+bug 
Also, the ARM compiler chain is suspect - we've got evidence that gcc from 4.6 through to 4.9 miscompile various different bits of XFS code (kernel and userspace) due to a combination of bad optimisations and broken (and unfixed) kernel asm code (e.g. the do_div() implementation).
IOWs, it's not a good investment in time to be chasing ARM specific XFS corruptions until the known problems with the toolchain and platform support are fixed up first.
And now guess what? The compiler you're using is a scruffy 4.6:
- root@opennms:~# /var/git/H3/OrangePI-Kernel/brandy/gcc-linaro/arm-linux-gnueabi/bin/gcc --version
- gcc (crosstool-NG linaro-1.13.1-2012.02-20120222 - Linaro GCC 2012.02) 4.6.3 20120201 (prerelease)
- Copyright (C) 2011 Free Software Foundation, Inc.
- This is free software; see the source for copying conditions. There is NO
- warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
Copy code
While choosing XFS on big iron is sometimes a good idea (when using Mainline kernel I would always prefer btrfs/zfs these days) it's a really bad choice in your environment. And since you provide an OMV distributions the data of users relying on your work is at high risk. Please stop including dangerous 'features'!
|
|