Firstly, the distro is Scientific Linux 6.1. on a bare metal supermicro server.
I've just reverted to an older kernel via yum due to some issues with a megaraid driver. The kernel I've downgraded to is 2.6.32-131. The kernel installed fine, but some subsequent package installations (post kernel downgrade) failed. The error relates to an ldconfig call:
ldconfig: /etc/ld.so.conf.d/: hwcap index 0 already defined as nosegneg
It seems that each kernel places a .conf file in /etc/ld.so.conf.d/ and the hwcap nosegneg flags conflict. To resolve, I just removed the .conf from the unused kernel.
This option is not really well documented anywhere on the net, except in relation to Xen kernels.
Can someone tell me in detail, preferably with examples:
-What does nosegneg mean, and how does it work?
-What is the implication of using the wrong flag?
-Why would it change between kernels?
-Can I consider this a reportable bug, given that a kernel downgrade causes all later package installations to fail?