GitLab

The mtu regression seems to have been fixed upstream. We backport.

udev-176 and later needs it

Fixes rx_dropped overflow problem
http://forums.grsecurity.net/viewtopic.php?f=3&t=2933

- set utf8 as default charset for filesystem NLS and FAT
- enable xattr for squashfs
- enable latencytop
- disable IOMega ZIP drives
- disable PCMCIA SCSI drivers
- disable CAN subsystem
- disable CAIF sysbsystem
- disable Power Supply drivers
- disable Voltage and regulators
- disable Dallas 1-wire support

fixes "sleeping while atomic" kernel oops under memory pressure

for Xen

This reverts commit f8d836e1.

This reverts commit 7fb62dd1.

based on grsecurity-2.2.2-3.0.9-201111161802

[x86]
- enable CONFIG_X86_BIGSMP
- set CONFIG_NR_CPUS to 32

[x86_64]
- set CONFIG_NR_CPUS to 32

With CONFIG_NR_CPUS > 8 on x86, the kernel may hang upon encountering
more than 8 logical CPUs without CONFIG_X86_BIGSMP enabled.  x86_64 does
not have this problem, but as a side effect of the CONFIG_X86_BIGSMP
change, needed to have CONFIG_NR_CPUS adjusted there as well.

Discussions on various distribution lists:

Alpine: http://lists.alpinelinux.org/alpine-devel/0496.html
Fedora: https://bugzilla.redhat.com/show_bug.cgi?id=480844
rPath: https://issues.rpath.com/browse/RPL-3022

fixes performance regression

We no need to manually edit every 3rd party module's kernel ver to match
the kernel packages version.

We need to do this because we will at some point allow building in a
chroot whithout the entire aports tree beeing present.

fixes #747