Intel Atom CPU can oops because of bug listed in Intel errata AAH41 and AAE44

Bug Description

Binary package hint: linux-source-2.6.24

http://download.intel.com/design/processor/specupdt/319536.pdf page 33 describes errata AAE44 which can cause oops at seemingly random places.

Upstream commit 211b3d03c7400f48a781977a50104c9d12f4e229 reduces the window of opportunity for this bug to occur

Committed:

http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-hardy.git;a=commit;h=7b618b8b711d53ac628d8e789c3343efee2ff466

This has been committed to netbook branch only.

Main has CONFIG_DEBUG_RODATA enabled which will oops i386 flavours with this patch (this is being investigated).
Howerver, CONFIG_DEBUG_RODATA is disabled for lpia (for Atom based systems) so this patch is acceptable.

Tested on N270 and Z530 CPUs.

This is actually commited to the Hardy master branch as a general issue.

We have also decided to pull this change up to Lucid until we can confirm it is not affected. We have a number of isssues showing up which seem to be fixed up by mem=nopentium, suspiciously similar to this issue.

This bug was fixed in the package linux - 2.6.32-15.21

---------------
linux (2.6.32-15.21) lucid; urgency=low

[ Andy Whitcroft ]

* Revert "(pre-stable) drm/i915: Increase fb alignment to 64k"
* Revert "[Config] lenovo-sl-laptop -- enable"
* Revert "ubuntu: lenovo-sl-laptop -- git tip (b19a08f81f)"
* armel -- cramfs module will no longer be built
* d-i -- make all modules optional
* rename the debug packages to match archive standard
- LP: #527837
* lenovo-sl-laptop is no longer built

[ Colin Ian King ]

* Disable 4MB page tables for Atom, work around errata AAE44
- LP: #523112

[ Colin Watson ]

* ubuntu: dm-raid4-5: Depend on XOR_BLOCKS
* ubuntu: fsam7400: Depend on CHECK_SIGNATURE

[ Jesse Barnes ]

* SAUCE: drm/i915: don't change DRM configuration when releasing load
detect pipe
- LP: #488328

[ Loïc Minier ]

* [Config] armel Update versatile initrd configs
- LP: #524893
* SAUCE: [um] Don't use nx_enabled under UML
- LP: #524849

[ Manoj Iyer ]

* [Config] added new config option CONFIG_SR_REPORT_TIME_LIMIT

[ Mario Limonciello ]

* SAUCE: v3 - Add Dell Business Class Netbook LED driver

[ Rafael J. Wysocki ]

* SAUCE: PM report driver and device suspend/resume times.

[ Surbhi Palande ]

* Revert "[Upstream] e1000e: enhance frame fragment detection"
- CVE-2009-4538
* Revert "[Upstream] e1000: enhance frame fragment detection"
- CVE-2009-4536

[ Tim Gardner ]

* [Config] Enabled CONFIG_LEDS_DELL_NETBOOKS=m
* SAUCE: (pre-stable) netfilter: xt_recent: fix buffer overflow
* SAUCE: (pre-stable) netfilter: xt_recent: fix false match

[ Upstream Kernel Changes ]

* Revert "(pre-stable) eCryptfs: Add getattr function"
* Fix potential crash with sys_move_pages
* futex_lock_pi() key refcnt fix
* futex: Handle user space corruption gracefully
* futex: Handle futex value corruption gracefully
* Fix race in tty_fasync() properly
* hwmon: (w83781d) Request I/O ports individually for probing
* hwmon: (lm78) Request I/O ports individually for probing
* hwmon: (adt7462) Wrong ADT7462_VOLT_COUNT
* ALSA: ctxfi - fix PTP address initialization
* drm/i915: disable hotplug detect before Ironlake CRT detect
* drm/i915: enable self-refresh on 965
* drm/i915: Disable SR when more than one pipe is enabled
* drm/i915: Fix DDC on some systems by clearing BIOS GMBUS setup.
* drm/i915: Add HP nx9020/SamsungSX20S to ACPI LID quirk list
* drm/i915: Fix the incorrect DMI string for Samsung SX20S laptop
* drm/i915: Add MALATA PC-81005 to ACPI LID quirk list
* usb: r8a66597-hcd: Flush the D-cache for the pipe-in transfer buffers.
* i2c-tiny-usb: Fix on big-endian systems
* drm/i915: handle FBC and self-refresh better
* drm/i915: Increase fb alignment to 64k
* drm/i915: Update write_domains on active list after flush.
* regulator: Fix display of null constraints for regulators
* ALSA: hda-intel: Avoid divide by zero crash
* CPUFREQ: Fix use after free of struct powernow_k8_data
* freeze_bdev: don't deactivate successfully frozen MS_RDONLY sb
* cciss: Make cciss_seq_show handle holes in the h->drv[] array
* ioat: fix in...

This was upstream in the 2.6.29 timeframe so it already exists in the lucid and karmic kernels, there shouldn't be a need to pull it in for those?

http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-lucid.git;a=commit;h=211b3d03c7400f48a781977a50104c9d12f4e229

@Robert -- sadly not, we are finding in the Lucid kernels that a number of netbooks are triggering what appear to be this issue dispite the work around being in place. We should note the work around is not guarenteed. We found that mem=nopentium sorted out those machines which is what the new fix does.

This bug was fixed in the package linux - 2.6.24-27.69

---------------
linux (2.6.24-27.69) hardy-proposed; urgency=low

[Colin Ian King]

* SAUCE: Disable 4MB page tables for Atom, work around errata AAE44
- LP: #523112, #322867

[Stefan Bader]

* Fix build failure in bnx2
- LP: #435185

[Upstream Kernel Changes]

* bnx2: Add PCI ID for 5716.
- LP: #435185
* bnx2: Add PCI ID for 5716S.
- LP: #435185
-- Stefan Bader <email address hidden> 2010年3月18日 00:13:51 +0000

error :

E: /var/cache/apt/archives/linux-headers-2.6.24-27-generic_2.6.24-27.69_i386.deb: impossible d'analyser (déréférencer) le lien symbolique existant « /usr/src/linux-headers-2.6.24-27-generic/include/linux/compiler-intel.h »