kernel/TODO

**** Queued for 3.2 ***********************************************************************************

* acpi-ensure-thermal-limits-match-cpu-freq.patch
* revert-efi-rtclock.patch
* block-stray-block-put-after-teardown.patch
* efi-dont-map-boot-services-on-32bit.patch
* epoll-limit-paths.patch (in linux-next via -mm tree)

**** Queued for 3.3 ***********************************************************************************
drm-edid-try-harder-to-fix-up-broken-headers.patch
Bluetooth-Add-support-for-BCM20702A0-0a5c-21e3.patch
ext4-Support-check-none-nocheck-mount-options.patch
ext4-Fix-error-handling-on-inode-bitmap-corruption.patch

**** Other stuff that should go upstream (in decreasing likelyhood) ************************************

* linux-2.6-acpi-video-dos.patch
* linux-2.6-defaults-acpi-video.patch
* linux-2.6-defaults-aspm.patch
* disable-i8042-check-on-apple-mac.patch
* linux-2.6.30-no-pcspkr-modalias.patch
* die-floppy-die.patch
	Fedora policy decisions
	Turn into CONFIG_ options and upstream ?

* linux-2.6-input-kill-stupid-messages.patch
* linux-2.6-silence-acpi-blacklist.patch
* linux-2.6-silence-fbcon-logo.patch
* linux-2.6-silence-noise.patch
	Fedora local 'hush' patches.  (TODO: push more upstream)

* linux-2.6-makefile-after_link.patch
  Rolandware that is used by the debuginfo generation.
  Possibly upstreamable ?

* linux-2.6-serial-460800.patch
	Probably not upstreamable.
	http://marc.theaimsgroup.com/?l=linux-kernel&m=112687270832687&w=2
	https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=126403
	http://lkml.org/lkml/2006/8/2/208

* linux-2.6-acpi-debug-infinite-loop.patch
  Responsible: mjg59

* fix_xen_guest_on_old_EC2.patch
  Ugly for upstream. It's a hack to make old HV's work optimally.
  Eventually we can drop it, but probably not until fixed HV's
  are commonplace. (When?)
  Responsible: Justin.

* add-appleir-usb-driver.patch
  Added September 2009, but had been around even before that.
  " NACKed upstream with "i'll implement this in the ir framework" or something
  and never got around to it."
  Responsible: j-rod

* dmar-disable-when-ricoh-multifunction.patch
  Added October 2010

* linux-2.6-intel-iommu-igfx.patch
  Invert igfx_off/igfx_on option. Around since forever. Upstreamable ?
  Mustard ? Config option ?

* arm-omap-dt-compat.patch
* arm-smsc-support-reading-mac-address-from-device-tree.patch
  reponsible: Dennis

***********************************************************************************

'MUSTARD' patches.  Fedora local patches that are very unlikely to go upstream.

* linux-2.6-32bit-mmap-exec-randomization.patch
  One half of the remaining exec-shield diff.
  - davej bugged Ingo again on Jun 17 2011 about upstreaming, no response.
  - mailed Linus asking for opinions in August 2011. He still hates it.
   "Ugh. Certainly not in this form. That patch is just disgusting." ..
   "I think that with NX and now SMEP, if you really care about security,
    you'd better be using a CPU that supports 64-bit operations anyway."
   As well as the increased randomisation, there's a key part of this patch
   that means that DSOs get loaded in the ASCII-Armor area on 32bit.
   (Addresses have topmost byte == 0)
   See also https://bugzilla.redhat.com/show_bug.cgi?id=734239

* linux-2.6-i386-nx-emulation.patch
  The ugly segment hack part of exec-shield that Linus hates.
  Unlikely to ever go upstream.

* linux-2.6-crash-driver.patch
	Unlikely to go upstream.
	https://bugzilla.redhat.com/show_bug.cgi?id=492803

* linux-2.6-e1000-ich9-montevina.patch
  Intel hates it. Davej has an SDV that needs it.
  Low maintenence, and worth hanging onto until it gets replaced
  with something that fails in a different way.

* utrace.patch
  Hopefully it'll die when uprobes gets upstream.


***********************************************************************************

Spec file/config todos/cleanups

* modules-extra: Do a few more things to make it a bit more robust.
  - Allow for comments in the mod-extra.list file.
  - Don't fail the build if a module is listed but not built (maybe).
  - See if it can be tied into Kconfig instead of module names.

* investigate gzip modules. Everything should support this now?
  Looks like about 70M savings per kernel installed.