here I talk
Kernel 2.6.33.3-soultrain released for Debian/Ubuntu (Soultrain)
Well first of all, sorry for this late release, but had some stuff to handle which at that point in time where more important. Kernel 2.6.33.3 was released last week and it comes like always with a bunch of bug fixes, improvements and new stuff added. I’m not gonna go into them, because there are too many, but just form the top of my head I will mention one small fix, which I noticed myself, and that is the thinkpad_acpi modules which didn’t work properly in the older released. By this I mean that enabling/disabling WLAN and Bluetooth didn’t work. You will have them on, but coudn’t switch them off using FN key plus F5 key. This part seems that it was fixed and now in kernel 2.6.33.3 it works like it should. In rest, I can’t tell you much, besides the fact that it works, I didn’t notice any downgrade in performance or anything similar.
Now lets take a small look at the ChangeLog:
r8169: clean up my printk uglyness
x86/gart: Disable GART explicitly before initialization
KVM: x86: Fix TSS size check for 16-bit tasks
KVM: Increase NR_IOBUS_DEVS limit to 200
KVM: fix the handling of dirty bitmaps to avoid overflows
KVM: MMU: fix kvm_mmu_zap_page() and its calling path
KVM: VMX: Save/restore rflags.vm correctly in real mode
KVM: allow bit 10 to be cleared in MSR_IA32_MC4_CTL
KVM: Don’t spam kernel log when injecting exceptions due to bad cr writes
KVM: SVM: Fix memory leaks that happen when svm_create_vcpu() fails
KVM: VMX: Update instruction length on intercepted BP
sched: Use proper type in sched_getaffinity()
ext4: fix async i/o writes beyond 4GB to a sparse file
powerpc: Fix SMP build with disabled CPU hotplugging.
md: deal with merge_bvec_fn in component devices better.
module: fix __module_ref_addr()
lockdep: fix incorrect percpu usage
modules: fix incorrect percpu usage
ACPI: EC: Limit burst to 64 bits
ACPI: EC: Allow multibyte access to EC
xfs: check for more work before sleeping in xfssyncd
xfs: fix locking for inode cache radix tree tag updates
xfs: Non-blocking inode locking in IO completion
perf_events, x86: Implement Intel Westmere/Nehalem-EX support
e1000e: stop cleaning when we reach tx_ring->next_to_use
iwlwifi: fix scan race
iwlwifi: clear all tx queues when firmware ready
USB: cdc-acm: Fix stupid NULL pointer in resume()
USB: cdc-acm: Update to new autopm API
ecryptfs: fix error code for missing xattrs in lower fs
eCryptfs: Decrypt symlink target for stat size
ecryptfs: fix use with tmpfs by removing d_drop from ecryptfs_destroy_inode
ALSA: hda – Add position_fix quirk for Biostar mobo
b43: fall back gracefully to PIO mode after fatal DMA errors
b43: Allow PIO mode to be selected at module load
i2c-i801: Add Intel Cougar Point device IDs
ahci: AHCI and RAID mode SATA patch for Intel Cougar Point DeviceIDs
ata_piix: IDE Mode SATA patch for Intel Cougar Point DeviceIDs
ALSA: hda – enable snoop for Intel Cougar Point
ALSA: hda_intel: ALSA HD Audio patch for Intel Cougar Point DeviceIDs
x86/PCI: irq and pci_ids patch for Intel Cougar Point DeviceIDs
IPoIB: Fix TX queue lockup with mixed UD/CM traffic
PCI: kill off pci_register_set_vga_state() symbol export.
pci: Update pci_set_vga_state() to call arch functions
9p: Skip check for mandatory locks when unlocking
ocfs2: Change bg_chain check for ocfs2_validate_gd_parent.
ocfs2: set i_mode on disk during acl operations
quota: Fix possible dq_flags corruption
fix NFS4 handling of mountpoint stat
x86-64, rwsem: Avoid store forwarding hazard in __downgrade_write
x86: Fix breakage of UML from the changes in the rwsem system
x86-64: support native xadd rwsem implementation
x86-64, rwsem: 64-bit xadd rwsem implementation
x86: clean up rwsem type system
x86-32: clean up rwsem inline asm statements
dm ioctl: introduce flag indicating uevent was generated
x86, cacheinfo: Enable L3 CID only on AMD
x86, cacheinfo: Remove NUMA dependency, fix for AMD Fam10h rev D1
x86, cacheinfo: Calculate L3 indices
x86, cacheinfo: Add cache index disable sysfs attrs only to L3 caches
intel-agp: Switch to wbinvd_on_all_cpus
x86, cacheinfo: Fix disabling of L3 cache indices
x86, lib: Add wbinvd smp helpers
Revert “x86: disable IOMMUs on kernel crash”
x86/amd-iommu: enable iommu before attaching devices
x86/amd-iommu: Use helper function to destroy domain
x86/amd-iommu: Pt mode fix for domain_destroy
drm/radeon/kms: add FireMV 2400 PCI ID.
drm/radeon/kms: fix rs600 tlb flush
drm/radeon/kms: fix tv dac conflict resolver
drm/radeon/kms: disable the tv encoder when tv/cv is not in use
drm/radeon/kms: more atom parser fixes (v2)
drm/i915: Add no_lvds entry for the Clientron U800
hwmon: (sht15) Properly handle the case CONFIG_REGULATOR=n
hwmon: (sht15) Fix sht15_calc_temp interpolation function
ALSA: usb – Fix Oops after usb-midi disconnection
ALSA: hda – add a quirk for Clevo M570U laptop
Input: sparse-keymap – free the right keymap on error
Input: wacom – switch mode upon system resume
NFSv4: fix delegated locking
NFSv4: Fall back to ordinary lookup if nfs4_atomic_open() returns EISDIR
sched: Fix sched_getaffinity()
sched: sched_getaffinity(): Allow less than NR_CPUS length
x86: Fix double enable_IR_x2apic() call on SMP kernel on !SMP boards
x86, hpet: Erratum workaround for read after write of HPET comparator
x86: Handle overlapping mptables
x86-32, resume: do a global tlb flush in S4 resume
drm/radeon/kms/combios: verify dac_adj values are valid
drm/radeon/kms: fix washed out image on legacy tv dac
drm/radeon: R300 AD only has one quad pipe.
drm/edid/quirks: Envision EN2028
ARM: 6031/1: fix Thumb-2 decompressor
mac80211: Handle mesh action frames in ieee80211_rx_h_action
iwlwifi: need check for valid qos packet before free
ath9k: fix double calls to ath_radio_enable
CIFS: initialize nbytes at the beginning of CIFSSMBWrite()
cifs: Fix a kernel BUG with remote OS/2 server (try #3)
s390: disable change bit override
libata: disable NCQ on Crucial C300 SSD
pata_via: Add VIA VX900 support
backlight: mbp_nvidia_bl – add five more MacBook variants
ALSA: hda – Add MSI blacklist for Aopen MZ915-M
ALSA: hda – Add ASRock mobo to MSI blacklist
ALSA: hda: Fix 0 dB offset for Lenovo Thinkpad models using AD1981
ALSA: mixart: range checking proc file
include/linux/kfifo.h: fix INIT_KFIFO()
readahead: fix NULL filp dereference
raw: fsync method is now required
HID: fix oops in gyration_event()
tty: release_one_tty() forgets to put pids
genirq: Force MSI irq handlers to run with interrupts disabled
WATCHDOG: iTCO_wdt: TCO Watchdog patch for additional Intel Cougar Point DeviceIDs
WATCHDOG: hpwdt – fix lower timeout limit
sunxvr500: Ignore secondary output PCI devices.
sparc: Fix regset register window handling.
mac80211: tear down all agg queues when restart/reconfig hw
mac80211: move netdev queue enabling to correct spot
mac80211: fix PREQ processing and one small bug
setup correct int pipe type in ar9170_usb_exec_cmd
iwlwifi: range checking issue
iwlwifi: counting number of tfds can be free for 4965
iwlwifi: fix regulatory
Freezer: Fix buggy resume test for tasks frozen with cgroup freezer
x86,kgdb: Always initialize the hw breakpoint attribute
libiscsi: Fix recovery slowdown regression
reiserfs: Fix locking BUG during mount failure
sh: Enable the mmu in start_secondary()
sh: Fix FDPIC binary loader
drm: Return ENODEV if the inode mapping changes
drm/radeon/kms: fix pal tv-out support on legacy IGP chips
drm/radeon/kms: don’t print error on -ERESTARTSYS.
drm/radeon/kms: Fix NULL pointer dereference if memory allocation failed in a simple way
drm/radeon/kms: never treat rs4xx as AGP
oom: fix the unsafe usage of badness() in proc_oom_score()
fat: fix buffer overflow in vfat_create_shortname()
drm: remove the EDID blob stored in the EDID property when it is disconnected
drm/radeon: add new RS880 pci id
To download this kernel please use one of our mirrors:
Mirror US/CA: http://dl.robertalks.com/kernel/2.6.33.3-soultrain/
Mirror EU/CZ: http://mirror.visualserver.org/kernel/2.6.33.3-soultrain/
The basic installation is simple and you will need root access or sudo access, plus dpkg. Here is how (use –force-all to make sure it will install it, it can happen that on older system will run into an error, for example on Ubuntu 8.04, but with –force-all it will be installed and it will work):
sudo dpkg -i –force-all linux-image-2.6.33.3-soultrain_2.6.33.3-1_i386.deb
For those of you who want or need to rebuild the kernel, you can download the source code with the config file from the links above, as I wrote before there wont be anymore included kernel source archive, but you will have a package call linux-source-2.6.33.3-soultrain_2.6.33.3-1_i386.deb which contains the kernel source code.
May the force be with you, Luke 
Hi! Can you compile the next version with rtl8187se module within?
I have a netbook with Ubuntu 10.04 LTS and your kernel performs much better than the stock, but I don’t have wireless without this module.
Thanks!
Well to be honest, module rtl8187 is already build in:
CONFIG_RTL8187=m
CONFIG_RTL8187_LEDS=y
Try to run sudo modprobe rtl8187 with of course “se”
Dear Roberts.
Please correct me if I am wrong, your kernel does not have x64 support.
If I am correct, then I need to ask the question(s); in your own opinion how much of an improvement is the performance of the BFS over the normal scheduler on an x86 machine and how does that compare with an x64 running the default scheduler?
I currently have a quad core processor, but I am planning to build a machine with 2 i7(Quad cores) and as many GPU boards as the motherboard (and my bank account) can afford (I plan to get Nvidia GPUs). This will give me 8 intel cores and (hopefully) between 2000 – 5000 GPU cores. Considering your comments about performance degradation once you pass 16 cores, is BFS a good choice for such a future configuration?
Hi, yes, there is no support for 64bit kernel yet… but regarding the BFS, I can’t tell you for sure. If you wanna know you should try to do a search on google like BFS vs CFS or something like that.
Thats why I said “in your own opinion”,
Thanks.
Well I know, but I’m just saying
Sorry, I don’t know where to post this comment.
I am a Windows user looking for FASTER desktops for older single core machines (usually P3 or P4 and >1 GHz). I have setup a few PCs with a multi-boot environment including WinXP, and a large selection of LINUX distros.
One of the problems with multi-booting is that it is impractical to store data on the system drive. First, the data would overwhelm the system drive. Second the data would NOT be consolidated if it were distributed across the system drives of multiple operating systems. Third, OS revisions, upgrades, and crash recoveries would be complicated if the system drive contains substantial data.
So my data is stored in a drive partition of its own, It may be the last partition on a drive with many partitions, or it may be a drive entirely separate from the drive used for operating systems.
Most of the Linux distributions do not default to automount other drive partitions in an internal drive. Sometimes they don’t even automount an external drive. This usually means that my data is not accessable, and that I will be working in what is essentially a sandbox from the rest of the system.
Sometimes a sandbox is good, but I can’t do everything I want in a sandbox, particulary if some of the things I want to do is organize and cleanup my data.
As a windows user, I am unfamiliar with Linux commands. I can follow instructions posted on the internet to add or change features in a distribution, but don’t know if what I am reading is applicable or correct for a particular distribution.
What I really need is a distro which automounts all other drive partitions that it finds that are NOT recognized as operating systems. Or, at least provides me with a simple check box to automount them.
Having the automount will get me out of sandbox mode, but to make the desktop work efficiently, I also need desktop links to my files and folders on the automounted DATA drive. I realize that these could be widgets, but they are certainly not as intuitive in LINUX as they are in WinXP.
Invariably, whenever I boot an opering system, I wait a minute or so get to the desktop and click on a link or widget to acess a folder, a recent document, or start an application. Eventually, I can lean new menu structures to access applications, but it is certainly quick and convenient to drop a link or widget on the desktop to access my frequently used applications and folders.
With automount and desktop links I’m off and running. Win32 apps that don’t have opensource equivalents can run in VirtualBox on LINUX and in Windows.
This is what I am looking for in a LINUX disto, and I’m trying dozens of them. Some don’t want to install in partition-12 and beyond, but there seems to be no shortage of new distributions each month. So I have many more to try.
Rgds,
fgk002
fgk002,
Your best bet, for the criteria you have stated, is PCLOS, aka PC Linux OS… go to: http://www.pclinuxos.com
I suggest either the Zen-Mini (a minimalist Gnome desktop) or the LXDE desktop ISO distributions.
PCLOS will automount all your partitions and their update system is perfect for new Linux converts.
Take note of my posts regarding the BFS kernel which comes as the default kernel with PCLOS 2010… you’ll probably want to remove it and install the standard kernel. Easily accomplished with Synaptic package manager. After initial setup, use synaptic to install the standard kernel, uninstall BFS and then reboot and let it fix any dependencies by itself.