bad sector

more sidux trickery and unbelievable stupidity

2009-09-20T21:39:00.004-05:00

You would think that with entire riot regarding sidux where many of the users finally took it upon themselves to yell at the developers in the forums for acting deceptively, their actions would change.

That's not the case.

I bumped up an old thread of mine asking the kernel packager for sidux named Steffan-Lippers Hollman to enable a kernel option named TIMER_STATISTICS. This would allow you to view which threads and processes cause the most wakeups per second via powertop.

slam, another ~~puppet~~ developer of sidux promptly answered my response with something reminiscent of blind faith, putting full trust in a kernel packager that thinks TIMER_STATISTICS causes a measurable performance penalty, enough to increase power consumption.

An explanation was given as to why their reasoning was unbelievably stupid, but apparently they don't answer in relation to your posts if they are wrong - so the thread moved, locked, then deleted. (link to deleted post)

Speaking of deleted! It seems that sidux has also removed an essential part of their distro support, smxi. The entire sub-forum was deleted in, what seems like, another step in their plan to quell the voice in their "community".

Users also seem to be splitting with sidux and switching to other distributions like Arch. How can you tell, damentz? Well, take a look at the sidux front page now. They no longer receive donations frequently enough to post a widget expressing their income with pride. In order to subvert this quick observation, they show only the top 3 donations (which is, in essence, very useless for data in any statistics).

In an also very childish manner, they ban all users that mention a few keywords (such as smxi, liquorix, or help) in their forums or irc chat. However, they are not very clever so it's easy to work around these artificial rules that bear benefits to nobody.

Continuing an observation to their juvenile behavior, the following thread should have never happened and exhibits their inability to handle their users.

http://sidux.com/PNphpBB2-viewtopic-t-16940.html

That thread could have been reduced to 5 posts or less if the developers actually read the posts. There are many instances through out this thread which show that many of them actually do not comprehend what is written, and out of desperation, a separate unrelated response is posted that attacks the original author indirectly with lies and other undomesticated thoughts.

To put it simply, the following quote describes sidux in a very concise manner:

"We're a ban first, don't ask questions anyway kind of an organization." - sidux

new branches and honky sources

2009-09-20T21:21:00.004-05:00

Out of thin air, through the kinetic energy of pure boredom, I have cast a spell accelerating the tumor of another branch from the Liquorix tree called past.

Past is not a branch to remember your relatives birthdays or to dwell on things that just depress you, but to support the previously stable version of the Linux kernel so you can stay supported with ugly hacks like fglrx and provide 3 months lapse before any other unusual module or bug prevents you from using the currently stable kernel.

You can use this branch by adding past to your deb line, like this:

deb http://liquorix.net/debian sid main past

At the moment, this branch currently contains my old 2.6.30 kernels with a large set of backports to fix some logic with the CFS process scheduler (ingo's sched-core-for-linus branch).

I also wrote a script to help automate the publishing of raw sources to http://liquorix.net/sources. The script itself is located in that directory for anyone interested.

#sidux support is garbage

2009-07-05T00:32:00.003-05:00

<damentz> cheater-conrad, what did you need?
<damentz> cheater-conrad, if you run compiz --replace from a terminal, do you get anything about gtk-window-decorator?
<etorix> damentz: not here pls
<etorix> damentz: hes already there
<damentz> etorix, you're splitting support unnecessarily which makes something that should be supported, unsupported
<etorix> damentz: ...
<damentz> it's really just a package set, why should we split it into another chatroom subsection
* ChanServ gives channel operator status to etorix
<damentz> it's the same thing as helping someone with recordmydesktop
* etorix sets ban on *!*damentz@*.austin.res.rr.com
<etorix> damentz: not here pls

Please discard any attempts at receiving any real help in the #sidux channel or forum, the signal to noise ratio has dropped lower than Ubuntu's. In fact, the sidux team created a forum specifically for "unsupported" items (really things they are too stupid to understand or burst their ego bubble) which has recently become much more helpful than the absurd amount of split topics which make getting correct support impossible.

For anyone interested in this distribution, check out #smxi or #linux-smokers-club on irc.oftc.net for an accurate overview of the benefits this distribution may offer. The official sidux irc channel has acquired a case of bull shit and can no longer be trusted to make decisions beneficial to most users.

Also, some of you may have noticed that the kernel of sidux has been steadily losing features and performance due to bad configuration choices on the kernel maintainer's end...

Try adding the kernel described at http://liquorix.net if you value your desktop usability. It features patches from Con Kolivas, zen-sources, and a configuration designed for low latency performance, a little like the sidux kernel but much more throughput which is essential for gaming.

Included is the ramzswap module which allows you to send your swap blocks to a virtual device stored in ram. All data entering this device is passed through an LZO function, effectively reducing the size of the cache stored in ram.

Besides this the BFQ scheduler is currently the basic disk scheduler until io-controller replaces the current defunct disk scheduling infrastructure. To minimize the effect of greedy disk copying, a patch from CK reduces the effect of bulk data writes and reads to allow fair use of non volatile media.

If there are any questions, ask me on #smxi!

brood before the storm

2009-06-18T18:22:00.002-05:00

Hey everyone, I try to keep my posts as quintessential as possible but I have not blogged anything new lately - here's an update to quell the silence.

The Liquorix project is proceeding well except for the zen maintainers who rotate according external pressure. The current maintainers are not updating patch sets on time or maintaining the correct patch sets, vital ones like madwifi. Due to these circumstances, I'll be learning over the weeks the tricks and mysteries to seamless kernel patching with git from foreign and unusual sources to part this dependency.

This and last week has been painful though, I'm retaking the Red Hat exam to get RHCE this time (no flaw). This exam didn't incur any stress penalties I'm aware of, but the laptop I work on most frequently simply decided to break the die of the nvidia GPU. So, apparently this is common for Dell laptops manufactured last year and I'm an unlucky victim.

I also scored an awesome job (which will remain undisclosed) that I enjoy working at - thus most of my daily energy is spent the first 10 hours I'm awake. Falling asleep 2 hours earlier is saving my life though, I'll keep that up so I can kick ass and chew bubble gum when I get home :).

So, I'm satisfied. Payment for my education and a logical future share space under an umbrella I control.

mastering regular expressions

2009-05-27T22:03:00.006-05:00

damentz@damentz64:~/.xchat2/xchatlogs$ grep -A10 foot4 -i oftc-#smxi.log
May 27 00:09:46 <h2-gw> ok, problem: convert <a href="#foot4">[4]</a> to <sup id="link-4"><a href="#footer-4">4</a></sup> where the number varies
May 27 00:09:59 <h2-gw> this will be across files
May 27 00:10:04 <damentz> hahaha
May 27 00:10:06 <h2-gw> this is an easy one
May 27 00:10:12 * razberrie fails
May 27 00:10:15 <damentz> h2-gw, sed?
May 27 00:10:24 <h2-gw> I was doing those all alst weak to update bad html to xhtml clean
May 27 00:10:42 <damentz> h2-gw, but you can't swap data in one line
May 27 00:11:03 <h2-gw> well, you just said you know regex, and since I was doing this all last week, I'd say you can do it
May 27 00:11:10 <damentz> oh i see
May 27 00:11:11 <damentz> hmm

nice challenge

echo '<a href="#foot4">[4]</a>' | sed -r 's:(<a [a-zA-Z=]+)"([a-zA-Z#]+)[^\[]+\[([0-9]*)](<[^>]+>):<sup id="link-\3">\1"\2er-\3">\3\4</sup>:g'

<sup id="link-4"><a href="#footer-4">4</a></sup>

music digression

2009-05-13T01:48:00.003-05:00

I thought I would skip my typical and maybe predictable rantings to introduce one song by my favorite musical artist, Bluetech.

One song in Bluetech's newest album, The Divine Invasion, continues to ring in my mind whenever I study. To me, it articulates the elation after suddenly understanding a concept that perpetually eluded me for minutes on end (ya, most things don't take me long).

The first half indulges into your curiosity until the second half, where you are suddenly shocked by an understanding of the world so tangible that you can hold and form it in your hands like a small cup of sealed play-do.

Follow this link if you're curious: http://www.last.fm/music/Bluetech/_/Holding+Space?autostart

Lazy Debian

2009-05-04T23:41:00.005-05:00

Is it fair to bash package maintainers that do not update their packages on time? I don't think so, currently many people are losing their jobs and can't afford to work on freedom projects. But one maintainer promised to update a package that interests me, preload.

So I grabbed an ubuntu package for the newest version, 0.6.4 when writing this (the one in debian sid is 0.4.5 - old!). Simply debianized it a bit and uploaded it to my repo - awesome.

Granted, the package I created includes a useless man page:

help2man is required to generate this file.

I honestly installed this package to build preload! It was a dependency, how could I forget?

What point am I making? Well, I fixed the problem and uploaded a new package - what's taking the primary maintainer so long?

Here's a link to the official bug report: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=500651

Harold Hope, a character I really respect, noticed this trend with the precision of a crayon (still very precise :) ). Debian is beginning to deteriorate a little. Packages are taking ages to propagate into sid and it doesn't feel nearly as bleeding edge as it used to be. When you flip the quarter and view the other side of the problem, maybe there isn't a problem in the first place.

KDE4 hit sid and maybe three users disagreed that this desktop environment is radically disappointing and not ready for end users production. On top of that, gnome was updated greedily to 2.24 when 2.26 was released so the gnome and gtk libraries are perpetually broken and missing dependencies or solving sudoku instead of suggesting solutions.

To fan the flames of breakage, a new version of Xorg hit sid but there were no corresponding hal fdi definitions to manage the touchpad. Of course, all the instructions are written in wikis, but remembering to echo data into /etc/hal/fdi/policy/11-x11-synaptic.fdi is not pleasurable.

Well, I'm predicting that after a month, sid should be ok. That's a brash assumption, but when the economy becomes a variable that directly effects what you see, feel, and experience... it's easy to think everything is going down the the tubes like a clogged urinal full of trident and paper towels.

blog art and new things

2009-05-01T02:13:00.004-05:00

Hi everyone, today I looked at my nearly abandoned blog and thought that it represented a crowd that is too ordinary. So, I changed the background a little to express the way I think sometimes. It's not ordinary, just typically not used in a plain sight like you currently see.

In a short change of events, I'm no longer using KDE4 on my desktop, neither XFCE. NVIDIA's memory management between 2d pixmaps and 3d textures is (probably not anymore) a mess with KDE4. Performance would grind to a halt with the kwin desktop compositor enabled or texture greedy OpenGL applications running in concurrency. I was impatient for NVIDIA to fix their drivers, so I dove directly into gnome - smart plan.

What does gnome do right? Well, the widgets, panels, and any tool using gtk, grab minimal graphics resources. Because of this, I can run compiz fusion seamlessly across two monitors without much of a single spike or delay in desktop rendering.

KDE4 on the other hand seems to be accelerating its already 2d desktop using compositing features of the xorg video driver to even render just the plain desktop without the use of kwin's desktop effects. This methodology is regressive and incurs more penalty on video driver than necessary.

If you think I'm joking, forciby disable the composite extension via /etc/X11/xorg.conf, and drag any window across the desktop. You _should_ be able to notice areas of the desktop that are not redrawn fast enough (if you are mentally slow or outrageously stupid, this effect shouldn't be a problem).

Externally, outside the scope of performance, gnome is just simple. I used to hate gnome for it's apparent take on simplicity, maybe too easy. But hating on a product for working correctly with sane defaults sounds a little backward. In fact, Ubuntu 9.04 showcases what sane defaults can achieve... just go read the reviews or run the livecd through a virtual environment - you'll see.

I personally think KDE 4.2 is bullocks, a waste of time. I'll, of course, take a look at KDE 4.3 when it hits Debian Sid - but it is currently too slow to be taken seriously as a competitor to Gnome at its current stable release.

VirtualBox 2.2 - Compiz in Linux Guests

2009-04-12T01:01:00.011-05:00

Ok, so you read this story title - here's a brief example:

Cool right? Though, I don't really see anything useful coming out of this besides an epic distraction you can impress friends with. The only need for 3D acceleration in a virtual machine would be through the Windows XP operating system where a lot of applications still refuse to run under WINE, such as in the following video.

With feature after feature being actively developed at a threatening pace, what will VMWare end up doing with their products? VMWare Workstation offers exactly the same functionality as VirtualBox 2.2, maybe less... so I don't see them monetizing that in the next year at all.

As closure, I found a new TV series to track called The IT Crowd. I've already watched the first episode from season one and it's pretty funny, but there is a chance it may not grow on me and I'll ditch it after three episodes.

Oh well.

RHCE

2009-04-09T00:46:00.002-05:00

Quick update from the stillness of an unpopular blog: I'm taking the RHCE exam Friday, should be fun >:)

The Stallman Predicament

2009-03-23T02:33:00.007-05:00

A new article was written by Richard Stallman named, The Javascript Trap. He goes on to explain that the code transmitted to users that request data as part of the original document are usually not free software even if the code is in fact, the source.

What he's referring to is obfuscript, or code that has been compacted by renaming variables and removing unnecessary white space. This code is really useless to most code hackers and doesn't include any comments or sensibly named variables to be effectively malleable.

Furthermore, he writes a suggestion on dealing with this code by adding stylized comments that link to the original, uncompacted source. This idea works for legitimate companies that honestly want to ensure that there is no question as to how their web applications work and clobber any myths for shady activity.

The problem is that this doesn't solve the dishonest or malicious web sites that use scripts to exploit unsuspecting victims or unsolicited traffic. It would be easy for someone to simply link to some false source code (or even the original, who cares!) to bypass the non-free check for the new web browsers that implement licensing checks in JavaScript and other code.

To give credit, Richard merely suggested that this would highlight the ethical contrasts of websites, but not solve malicious scripts from running in your browser directly. Addons or extensions such as NoScript already had this objective in mind.

I simply see this system being abused to create a false sense of security in the future. There maybe many loop holes to tricking your browser into interpreting the newly downloaded javascript is seriously free software but there is no proof of concept that has been publicly announced anywhere to my knowledge to prove this concept wrong.

It will be interesting to follow the outcome of this story and watch to see whether this proposal bears fruits. Time will tell and I'll be watching.

source: http://www.gnu.org/philosophy/javascript-trap.html

gcc -Os vs -O2: linux kernel

2009-03-07T02:05:00.001-06:00

So, I've been reading up about the size optimization (CC_OPTIMIZE_FOR_SIZE) after finding that the Fedora distro config enabled this option by default.

According to the LKML, this option is deprecated as an option that will "optimize" the performance of your builds of the Linux kernel. Here's an interesting quote from David Lang:

I started useing Os several years ago, even when it was hidden in the embedded menu becouse in many cases the smaller binary ended up being faster.

in reality this was a flaw in gcc that on modern CPU's with the larger difference between CPU speed and memory speed it still preferred to unroll loops (eating more memory and blowing out the cpu cache) when it shouldn't have.

Possibly the only reason this option remains default on large or important distributions is because they simply forgot it. I mean, if there's no reason to keep this on unless you're stupid enough to build the kernel with a deprecated compiler.

I actually have my own statistics. I compressed the debian package rules for the pidgin package in debian sid on an old pentium 3 laptop which has 256kb of L2 cache. This directory is small but after repeated benches, the results are still as important as a long 2 hour benchmark:

# CONFIG_CC_OPTIMIZE_FOR_SIZE=n
real 0m2.470s
user 0m2.251s
sys 0m0.134s

# CONFIG_CC_OPTIMIZE_FOR_SIZE=y
real 0m2.522s
user 0m2.279s
sys 0m0.159s

Well, the results look small, but this time can add up after a while. The funny part is the actual image of the kernel is only 12% smaller - what a waste. Within the LKML, people actually suggest that this size option is used to reduce the livecd image - hahaha.

Though, it is possible that this option is on for another reason but I'm doubtful that this should stay on.

liquorix kernel repository

2009-02-24T15:24:00.000-06:00

Ok, so I finally set up a repository with some extra hosting space from a very good friend.

To set it up, you can try something like this:

echo "deb http://liquorix.net/debian sid main" > /etc/apt/sources.list.d/lqx.list

Of course, the location and spelling is up to you.. but I'm pretty sure spelling counts.

To install the liquorix kernel and all it's modules (only kernel image and headers at this point) you can run:

apt-get install a*2.6-liquorix-686

Cool right? You'll get some authentication errors which I'll solve in the future as they're not high priority at this point. The kernel has also been included in the smxi script, so you can find it through the advanced kernel options area.

Enjoy!

lack of updates

2009-02-22T17:00:00.000-06:00

Hey everyone, I will no longer be releasing my kernels through free sites such as mediafire.com.

By next week I should have a repository that anyone can sync to :)

And keep an eye on the domain, liquorix.net - it currently doesn't point to anything so it'll just show a bunch of useless ads though.

Lastly, with testing soon to receiving updates after Lenny was released, the canabix project might pick up: canabix.org. This project will serve as an alternative to sidux tailored more towards debian testing than sid.

Is it worth the creation of an entirely new distribution just to fill the niche for testing? Maybe. Sidux' packages sometimes depend on exclusively unstable packages that won't propogate into testing until maybe a week later. This means that a rebuild of the packages might be necessary if they choose the route of adopting sidux' tools.

From what I've seen though, the entire distribution is sort of stalled until they can understand how the sidux installer operates XD.

kernel-2.6.28-zen11-dmz1

2009-02-19T19:24:00.001-06:00

A quick update to the zen kernel, now is synced to 2.6.28.6 and includes some reiser4 updates.

Enjoy :)

mediafire.com/?vtgmodmnjnm

Yes, I buy music too

2009-02-19T19:06:00.000-06:00

How do you support your favorite artist? You buy their CD, even when it looks a little over priced and the music studio suggests the download instead. Check out what I recieved in the mail today :)

Very neat looking CD slip with simple sewing on the sides and a spray painted phoenix - most likely to represent the first album attempt since Sines and Singularities released 5 years ago.

And surprise, it's signed! Didn't expect that or to know that it is the 563rd rare copy :D

I'm not an expert at album reviews so I'll just link you to the main site which includes fairly lengthy music samples: bluetech - phoenix rising

grub profiling, readahead, preload, and prelink

2009-02-13T14:59:00.000-06:00

Ok, so this post should demystify a few tweaks and what exactly they do.

I've often come across a few tweaking guides that mention this little trick named profile for grub.

Just googling, I see it misused like a throw around option when really it is not a permanent tweak.

profile, the grub option, requires that the readahead package/software be installed. It watches what files are required for daemons to run in the init sequence and creates a new list, sorted by disk block, with an updated list of files over the original.

Thus, you just use this once unless you make serious changes to your boot sequence (new or less services). Profiling also loads some kernel modules into page cache so if a performance loss is noticed after installing a new kernel, rerunning this might be an option.

preload

preload monitors applications that users run, and by analyzing this
data, predicts what applications users might run, and fetches those
binaries and their dependencies into memory for faster startup times.

So there you have it - a post boot preloading of files into page cache. This monitors for frequently used applications and attempts to load their libraries into the kernel's page cache. There shouldn't be any question as to why you would need this - totally dependent on personal preference.

prelink

When building a program, you can build libraries into the binary with a linker, thus making a static binary. These are portable but also come at a price of increased code size. Why build an app with integrated libraries when the shared libraries are accessible on the systems you will be installing it on?

Great, but now you have a performance penalty looking up these libraries for every instance that the executable is executed. What now?

prelink is a dynamic linker that 'prelinks' libraries to executables so that on run, libraries locations are preresolved so that any searches are eliminated on instantiation of the executable.

There, that should wrap up any loose ends if you previously read wrong information from the Ubuntu forums.

next up... how XFCE stole KDE's place as my main desktop environment :O

kernel-2.6.28-zen11-dmz0

2009-02-13T00:55:00.000-06:00

Quick update to the zen kernel, now using 2.6.28.5.

Also, the following is changed:

Kernel is now preemptible, not voluntary

tests of mine showed that with some renicing, you can prevent important applications from being preempted and thus, allow for better QoS depending on what applications you use.

Voluntary kernel preemption didn't require this as much, but latency to essentially everything was noticeably more at the minutest.
Mousepolling is enabled and set to 2ms. By default, the module usbhid is set to use events. For some reasons, games such as Warsow add this terrible lag to any abrupt mouse movement with kernel preemption enabled. Sure, polling is less advanced than events but provides lesser latency on average due to its persistent interrupts (when there is movement).

have fun with this one :)

mediafire.com/download.php?ekgtyzemkgw

kernel-2.6.28-zen9-dmz0

2009-02-07T22:02:00.000-06:00

Another kernel, disabled OPTIMIZE_INLINING for stability issues...

zen9 pulls the latest updates from the stable git of 2.6.28 - updating it to 2.6.28.4 along with some more upstream improvements for btrfs.

link: mediafire.com/?sefnectzqlm

I've switched to using mediafire for all future file storage used on this site due to obnoxious ads and poor performance delivered on divshare.

If anyone else has file storage recommendations, send me an email or leave a comment.

kernel-2.6.28-zen8-dmz2

2009-02-03T16:04:00.000-06:00

Hey guys, I built a new kernel based off of the current master for the 2.6.28 stable series:

http://www.divshare.com/download/6472308-995

Anyway, the usual improvements as dubbed by my previous kernel... except since it's zen, essentially all the features that you would expect from a zen kernel are included.

compcache 0.5.1

If you're curious as to how exactly you use this module, add the line to /etc/modules:

compcache compcache_size_kbytes=(size in kbytes you want to limit your raw size of cache to)

Then you'll want to mount /dev/ramzswap0 as a swap device with priority of 100... you can just put this into fstab.

/dev/ramzswap0 none swap sw,pri=100 0 0

The command swapon -a will automount it and you can enjoy some good 'ol LZO compressed cache :D

Just remember that the size of /dev/ramzswap0 is the size of uncompressed data, not compressed so it's safe to put in an unusually large number there :)

If not, well damn.

Now, this _is_ for debian sid, but you can still steal the configuration out of the deb and build your own with the same sources.

KDE 4.2 in Debian Experimental

2009-01-29T13:01:00.001-06:00

Upgrading to KDE 4.2 in the experimental branch of Debian is not _that_ straight forward, so for the few that requested some help... I'll just make a post that people can look back at and reference later.

So first, let's make a list of packages we already have installed in case you really mess up:

dpkg --get-selections > ~/prekde42

Neat, you just mastered stdout redirection.

Ok, so next, make sure you have the experimental repo in sources.list or sources.list.d/something.list with this line:

deb http://ftp.us.debian.org/debian/ experimental main contrib non-free

Now you'll want to remove all your kde packages with: apt-get purge kde*

This is the tricky part - the kde4 metapackage itself is broken most of the time, so just check what it depends on.

$ apt-cache depends kde4
kde4
Depends: kde4-minimal
Depends: kdeadmin
Depends: kdeartwork
Depends: kdegraphics
Depends: kdeedu
Depends: kdegames
Depends: kdemultimedia
Depends: kdenetwork
Depends: kdeutils
Depends: kdepim
Depends: kdeplasma-addons

You'll definitely want everything but kdeedu, kdegames, so you'll do something like...

apt-get install package0 package1 package2 -t experimental

Don't forget kdm!

apt-get install kdm -t experimental \
&& update-rc.d kdm remove && update-rc.d kdm defaults

That should be enough, everything else is up to you. Some important notes though, you can still install at this time Amarok 1.4.x by not specifying -t experimental when installing it through apt-get.

EDIT: amarok 2 will replace amarok 1.4 as soon as possible when lenny is released.

the better kernel

2009-01-27T23:03:00.001-06:00

Hey everyone,

Sidux is a great distro and the kernel's configuration makes for a very responsive system. The only problem is that proprietary drivers such as fglrx have issues building against it due to the GPL symbols RCU preemption exports.

So, I made a kernel that uses voluntary kernel preemption instead of both preemptible kernel coupled with RCU preemption. Disabling RCU preemption would actually solve the problem immediately but the kernel preemption will cause symptoms of inconsistent FPS in 3d applications when background tasks are a little more aggressive.

Here is the full parity against the slh kernel:

Voluntary Kernel Preemption
btrfs 0.18
4kb stacks
mtrr sanitizer
disabled PentiumPro memory ordering errata workaround (noone should be using that cpu anymore... really. Besides, it lengthens the time for spinlocks - disabling this should reduce the need for a hard preemptible kernel)
Sparse memory model
OPTIMIZE_INLINING
port 0xED IO delay
Tux On Ice hibernation support
disabled hda-intel digital beep
ondemand default cpufreq governer (this is changed immediately to userspace when cpufrequtils is run)

link -> http://www.divshare.com/download/6421159-efd

I will post a 2.6.29 zen kernel when I think it's stable enough for those that really don't want to tinker beyond getting fglrx working... Come to think of it, fglrx doesn't work with new kernels _anyway_ hahahaha.

a cold stab at broadcom's power thirst

2009-01-11T04:44:00.000-06:00

Broadcom's insistence on total control over the use of their chipsets is sort of... forking in the wrong direction.

On the 9th of January, Francesco Gringoli released the firmware he and some others had hacked up in order to progress with internal research. Benefits of this firmware are not apparent since it was simply used for testing and only supports 3 different chipsets.

The sad part is the coupling of intensive reverse engineering for both the driver _and_ the firmware... which you can't even redistribute.

From a company like Broadcom, you would think that their importance in the market would imply support one of these fields... but as usual, they sit back and let politics and laws fuck with their customers.

If further work on this firmware is a success, maybe their cards will finally do something else that create noise and break under ndiswrapper.

migrate ext3 > ext4

2008-12-27T00:13:00.000-06:00

So, you want to migrate from ext3 to ext4... ok. This is fairly simple as long as you deal with non critical partitions such as /boot or / (root).

If you recall back when journaling was a little new, in order to migrate from ext2 to ext3, all you needed to do was enable a journal and mount with the filesystem type of ext3 over ext2.

All examples of partitions will be using /dev/sda1.
You would unmount the filesystem and run tune2fs -j .

So for example:

tune2fs -j /dev/sda1

Then you would modify /etc/fstab and change the filesystem to ext3.

ext4 is a little bit different. Instead of adding a journal, you add features.

Yes, you can actually just mount as ext4 without modifying any features but that ruins the point and is counter intuitive.

arbitrary filesystems

Partitions that are unused for booting like /home are no problem. First unmount then with umount and run tune2fs -l | grep features on it.

If the filesystem is indeed ext3, you'll get something like this:

Filesystem features: has_journal ext_attr resize_inode dir_index filetype sparse_super large_file

Perfect, now you want it to look like this:

Filesystem features: has_journal ext_attr resize_inode dir_index filetype extent flex_bg sparse_super large_file huge_file uninit_bg dir_nlink extra_isize

Shit, what now? Actually you use the same tool to add these options.

for i in extent flex_bg huge_file uninit_bg dir_nlink extra_isize; do tune2fs -O $i /dev/sda1;

You'll want to run a fsck pass on the partition after this...

fsck.ext4 -fDy /dev/sda1

Now mount it once with ext4 with something like this:

mount -t ext4 /dev/sda1

And of course, make the change in /etc/fstab too... I shouldn't have to write this one out.

root partition

Word of caution, don't try this on your root partition if it includes the /boot directory. In other words, if you did not prepare your system with a separate boot partition, don't bother until you are very sure your version of grub can handle it.

Basically, do all of the above but you'll need to make changes to /boot/grub/menu.lst

Find one of the kernel's you have installed, it should look something like this:

title Debian GNU/Linux, kernel 2.6.28-rc9-zen1-dmz1
root (hd0,0)
kernel /vmlinuz-2.6.28-rc9-zen1-dmz1 root=UUID=35c8820a-13c2-44b8-abe1-e2cbcce0ab73 ro quiet vga=791
initrd /initrd.img-2.6.28-rc9-zen1-dmz1

Add the rootfstype=ext4 option to the kernel line to look similar to this.

kernel /vmlinuz-2.6.28-rc9-zen1-dmz1 root=UUID=35c8820a-13c2-44b8-abe1-e2cbcce0ab73 ro quiet rootfstype=ext4 vga=791

Placement of the option obviously doesn't matter. But if for some reason the placement happens to relieve a bit of stress, go ahead living life knowing your character is a little interesting and others would like to know why you need to be so difficult.

If you're running a Debian or Ubuntu base, you can also use the #kopt line along w/ update-grub to manage this all in two steps instead of man handling each kernel.

Add the rootfstype line as so:

# kopt=root=UUID=35c8820a-13c2-44b8-abe1-e2cbcce0ab73 ro quiet rootfstype=ext4

Then run update-grub to automagically update the kernel lines on each bootable entry in menu.lst

Suggestions are welcome to fix any misunderstandings of this article :)

Merry belated christmas!

EDIT: changed one of the sections to a bash loop, I looked over my guide a day ago and noticed how tedious it was.

ext4 migration post

2008-12-26T23:30:00.000-06:00

Hey everyone,

Just letting whoever reads this that I'll be writing a quick guide for migrating ext3 to ext4 and some things to look out for.

I lost my flash drive though, so I'll need something else to test with in the meantime.

EDIT: after nearly going insane thinking to myself 'how the fuck could I lose this', I found the drive on the floor right next to my new CPI coat. They're nearly the same color so I simply proved that camouflage works when it's unexpected.

realism

2008-11-28T02:12:00.001-06:00

Games are supposed to be fun, but why make realism such an important aspect of this?

Here is a neat excerpt i had with a British nut (names are replaced with variants of James):

(09:18:38 PM) James: have you ever played EU III?
(09:20:27 PM) damentz: nope
(09:20:33 PM) damentz: i don't play war games or simulations of such
(09:20:33 PM) James: damn
(09:20:40 PM) James: and you're insane >.>
(09:20:51 PM) damentz: i'm not a fan of games that attempt to reenact something realistically
(09:20:59 PM) damentz: in my mind, games should be artificial by nature
(09:21:22 PM) damentz: and to try to inherent realistic items into a game only reduces the effectiveness of the fun factor
(09:21:33 PM) James: I disagree
(09:21:43 PM) damentz: i know, you have a stiff upper lip
(09:21:58 PM) James: People want to play realistic games
(09:22:14 PM) James: that's why there is hype about graphics and destructibility of enviroments
(09:22:15 PM) damentz: when they want to experience it without being a liability to anyone
(09:22:26 PM) James: what?
(09:22:45 PM) damentz: graphics and destructible environments don't have to be sub sections of the realism planning
(09:22:57 PM) damentz: they could just be a feature that makes it more fun
(09:23:01 PM) damentz: so that's out of the question
(09:23:05 PM) James: physics.
(09:23:13 PM) James: Real-world weapons
(09:23:15 PM) damentz: could detract from gameplay
(09:23:21 PM) James: no
(09:23:24 PM) James: they contribute
(09:23:30 PM) damentz: depends on their application
(09:23:38 PM) damentz: i have yet to see a game that uses physics improperly
(09:23:53 PM) James: is that a bad thing?
(09:25:36 PM) damentz: no that's a good thing, physics is nearly a bonus feature
(09:25:42 PM) damentz: it almost never makes a game worse
(09:25:48 PM) damentz: other aspects of realism can
(09:25:54 PM) James: such as?
(09:26:02 PM) damentz: what is required to make us survive
(09:26:06 PM) damentz: if quake 3 was realistic
(09:26:53 PM) damentz: every hit would kill you
(09:26:55 PM) damentz: nearly
(09:27:04 PM) damentz: and you could not bunny hop to gain speed
(09:27:14 PM) damentz: you could not enjoy the rapid turning of the game and blinding reflexes required
(09:27:29 PM) damentz: when a game is planned, the developers should think, what would make this fun
(09:27:30 PM) damentz: not realistic
(09:27:40 PM) damentz: it's whenever they try to combine all aspects does it get annoying
(09:27:45 PM) damentz: personally i hate call of duty 4
(09:27:52 PM) damentz: or most of that series
(09:28:24 PM) damentz: because they make great strides to make it realistic
(09:28:40 PM) damentz: and thus i know my limitations and how quickly the game could get boring
(09:29:03 PM) James: I was not even refering to this section of gaming
(09:29:10 PM) damentz: that's what war games are
(09:29:11 PM) James: as if the Quake series could be realistic in the first place
(09:29:19 PM) damentz: which is why it's fun
(09:29:20 PM) James: Strategy is not FPS
(09:29:35 PM) damentz: i don't care what genre it is
(09:29:37 PM) James: Strategy is inherently realistic.
(09:29:44 PM) damentz: if the focus is realism, i won't play it
(09:29:46 PM) James: while FPS it is much harder to do so
(09:29:48 PM) damentz: if i do, i probably won't like it
(09:39:21 PM) damentz: fps is easier than strategy
(09:39:43 PM) James: strategy is always far more interesting
(09:40:28 PM) damentz: depends on what you like
...and then we get off subject

For anyone that reads this, what do you think? I might be totally biased since nearly every game that tried to simulate a real life moment of time always bored me or perpetually annoyed me with incomplete gameplay.

thanksgiving's usefulness

2008-11-28T01:55:00.000-06:00

It's just another excuse to get out, enjoy yourself, and consume a selection of food that meets a narrow criteria. Anyone that wants to "deliver", "give", or "make"(not the command that compiles code according to instructions in the Makefile) thanks is just bull shitting you.

They could be honest but in reality maybe a little greedy to predictably use Thanksgiving as a holiday to 'make up' or 'apologize' for previous mistakes, intentional or not.

I give thanks when needed and this holiday is only a reminder. Like antithesis to a book of morality, I do what's reasonable when deserved and on time, when it should happen. There is no magic to Thanksgiving and those that really force this mindset should look at other things they might have been gulled into.

I don't have anything else - good night.

Life

2008-11-01T23:09:00.000-05:00

I've been thinking a lot lately over the monotony of life - something so generic, everyone thinks they have a unique view on. What is it that makes us feel special? What do people do for vanity, to just look important?

One of my friends, Nick, invited me to tango dancing somewhere in downtown Austin as an off the wall activity from the usual Halloween parties or get togethers. Sure, I took the bait since I wasn't planning to do anything.

Turns out all of this was an obfuscated idea depending on an unreliable person who Nick didn't even know well. So he planned (don't plan dude!) to wait for immediate/short dance lessons from this girl who he correctly predicted would show up in the first place despite her vague references to frequent appearances in these events. Well, she left immediately to eat dinner and escape work.

Short changed.

I'm sitting here studying exactly how to do this tango cruft (to me it looked like simple movements to background music while "tricks" were all up to you or part of some standard set of "tango moves") thinking, ok, what the hell am I doing here - really. I'm not enjoying myself, everyone here is probably 35 or older, and everyone just seems to be so... scripted. Eaves dropping on a few conversations, I thought that the only substance of discussion was the weather.

Why am I saying this? What pulled me out here to do tango dancing? It's so stupid. I even looked online and watched some clips from youtube on exactly what this was. The dance is so simple I thought that I would bore myself just knowing it.

I ended up walking down 6th street while in total, two random street ricers interjected quick comments of "Jeremy?!" as I went by. Not hard to guess who I might have been dressed up like.

Sometimes I wonder if holidays like Halloween are just for suckers that think they can tell the world they're important at the same time while they overfill their buckets of vanity just so they can tell their friends that they bleed attention.

Then again, I took the route for technical college over the more social and abstract core curriculum. Why? According to my study habits of High School, there is no doubt I would be miserable since I can't focus on subjects that are so impractical to me for days at a time. At some point I'll push it off for a while and watch my grades bomb like a single drop of water racing across my windshield at night as I drive back home alone.

EDIT: It was fun anyway, but I know what it's like to have more fun. Must be the art of progress working against me.

cloud computing won't take off

2008-09-30T21:59:00.000-05:00

Oh shit, I'm making a big statement; cloud computing won't take off.

I was reading an article... more like skimming with the few minutes I have left for today - but I agree with Richard Stallman on his argument towards Cloud Computing.

source: http://arstechnica.com/news.ars/post/20080930-why-stallman-is-wrong-when-he-calls-cloud-computing-stupid.html

What is it that makes Cloud Computing so appealing to many users? Well, for one, they don't have to worry about the mundane maintenance of local software. I know many people that have trouble merely responding to the request of 'right click', but that's another story.

The worries of local software and data backup are sifted under the carpet with the concept of Cloud Computing. Personal documents and other file formats are stored safely in data centers across redundant storage devices to prevent your data from ever being destroyed 100%.

Online applications would be cross platform, the only software requirement being a set of trusted browsers capable of running these online applications as the developers wanted them to.

What about the system requirements for this processing? Most cpu intensive tasks are offloaded to the mass processing farms handling the load of thousands of sessions all simultaneously connected at once. At most, you just need a local cpu or gpu powerful enough to simply render your screen without frusturating your self esteem.

So, what's the problem? Well, it's just not local, and thus, increasingly complex.

First of all, the software is obviously not local. All of a sudden, a new series of dependencies introduce themselves: your ISP, modem, network adapter, network adapter's driver, and the application that interfaces with the server that handles this cloud computing (web browser for you cheese graters).

Hardening your dependency on these third party services are equivalent to tying a ring pop to your elbow and trying to suck it in public. What if you need to bring your laptop to a presentation and one of the local administrators conveniently, quite out of spite, decided to smash his wireless router with a 14 pound hammer?

You might be stifled at the fact that you have no access to your online services that handle data manipulation incase you need to modify your presentation before show time. On the other hand, you may be more intrigued by the pure aspect of lifting a hammer of this magnitude and what incentive this unruly employee had in mind when a simple cut of the power circuit was reasonable enough.

Second, now you're abstracting another application through another piece of software like a wrapper... or virtualization. The reliability of this online application is multiplied by your browsers tendency to crash at immature moments or the online application to follow through with it's intended task.

Google did come out with a new browser to combat the concept of "fast" by optimizing their java virtual machines to compute code several times faster than the current stable builds of Firefox, Opera, and other browsers that I would rather not mention for previous negative experiences. But honestly, who is browser hopping to view their site with the software the web designer intended for best experience?

No one (no one that can aim on a piss break at least)! People stick with what they're comftorable with but look for alternatives when the defects in their favorite software become well too obvious and consistent.

Cloud computing can't take off fast with this attitude. Google should have reengineered or contributed to Firefox's base code instead of building an entirely new interface to Webkit.

But like I said earlier, Google Chrome redefined what fast really means and the most competitive software projects are working to deface their prestige and reclaim their crown as just the better choice.

What most people don't notice is that you're still abstracting a display through another piece of software. It will never be as fast or stable as local software.

Then because of this extra abstraction layer, the browser that all these services would run under becomes a huge target for exploitation. I can't imagine how Internet Explorer will evolve if some important online services become popular freedom resorts from expensive proprietary software.

With all this extra functionality that would need to be implemented, more features go untested and are then subject to exploitation. The web browser is already a large target for 0day exploits. With cross platform compatablity of these online services, hackers wouldn't really need to worry about OS centric code if the objective of their attack would be to simply command some power in user space - enough to join an irc channel and run denial of service attacks.

Well... on Vista, possibly a denial of access by opening millions of UAC prompts.

And because the web browser would be a portal to all these necessary online services, most users wouldn't think of closing it, would they?

So many things wrong with this concept, perhaps we are tackling online applications the wrong way?

Google Chrome, not exactly multiplatform

2008-09-16T22:51:00.000-05:00

Take a look at this: http://digg.com/software/Chrome_is_Windows_inside_which_may_be_a_strategic_error

For anyone reading this thinking, I didn't know Google manufactured metal, this is their attempt at manipulating the market and forcing different browser companies to double check their priorities.

So, in summary, Google Chrome is supposed to solidify the following points:
*reliability
*speed

Steve Gibson from Security Now noticed that under tight memory conditions, the browser masters a flophouse crash and just disappears immediately. Terrible way to introduce stability.

I will give kudos for the creation of a new process per tab/window. This reduces heap fragmentation which kind of plagues firefox 3.0.x to this day. By creating a new process, a new heap is created in preparation for new dynamic allocations.

New process creation takes more cpu time though. On the other hand, I don't think anyone will notice. The downside actually appears when you notice when the process per tab decision comes with less cache sharing. Couple that with Vista and your machine just suddenly fails to a million new degrees. Sometimes the sun looks cold in comparison.

Pretty exciting!

Development of the alternative operating systems comes from their detailed optimization for the windows platform. This makes their browser inherently more difficult to code for cross platform. Instead of using standard libraries that are portable across all OS's, they decided to try an approach that benefits the OS itself, not the developers time.

This will inevitability cause pseudo code duplication and the quality fairness across all supported platforms will not be so even.

lame -_-

hp's 2133 mini note sucks

2008-09-12T18:16:00.000-05:00

As far as an instantaneous prejudice for my opinion, I think I'll go ahead and explain why this particular notebook sucks.

So when you design a mini laptop, the sort that appeals to consumers that use their laptop as a device like a generic device, you would think that the Linux sector would attract a little more attention. Nope.

Here's the first problem, they chose this vendor named VIA for their GPU setup. What's wrong with that? Maybe their apparent lack of a true open source driver solution.

Well you're thinking, who the fuck cares right? I mean, who cares if you install a new distro and you realize that the proprietary driver for some reason is _incompatible_ with your currently installed kernel revision. No big deal, just find some patch to pretty much later fix the kernel interface and downgrade X right?

And don't you love installing your favorite distribution and you realize that the only model of the broadcom wireless doesn't have a functional wireless driver? I mean, not even fundamental support that might work if you glue the screen to your routers antenna for maximum signal strength.

You actually need to use NDISWrapper to use the Windows drivers. The sad thing is that the kernel developers have been working really hard to make sure that this approach is made as difficult as possible. So in about a year or two, either some reverse engineered b43 module will provide fundamental support in worst case scenarios, or you'll just sit there thinking fuck with your hands essentially squeezing your skull hard enough to leave a nice imprint of your decision failure.

People actually bought this thinking there was Linux support. Sure, HP supports Linux when you have to throw an enterprise distribution at it.

Really?

That's a joke. Enterprise distributions use frozen repositories that retrieve security updates. You'll be lucky if that update happens to be an entire upstream backport as was the case with my little experiment with Cent OS (actually sucks btw, don't shoot me). I don't plan on using that device as a kiosk, so maybe it's easier to just grab Dell's Mini Note 9 (ya, almost same name)

And by listening to Linux Action Show, HP hasn't altered their trend with inhumane heat dissipation. Bryan, the host with the deeper voice despite his small stature mentioned that the thing can heat a cup of coffee to a nice steam if you set it on the flipside without a coaster. Actually that's completely wrong but I wouldn't be too wrong if the coffee does steam.

If someone can actually come up with a reason that they would buy this piece of shit, just leave a comment for everyone to read.

in linear thought

2008-09-10T19:25:00.000-05:00

I've made an acute decision to publish my thoughts in a uniformly generic template with erratic undecidable sentences.

Although subsequent posts might be unnaturally difficult to agree with, it's only normal that you won't understand anything. If the ESRB decided to tag my blog, for some reason an unsuitable for you tag would always block your confident perception of the English language, prohibiting you from abusing t.

So most people seem to have trouble appreciating the simple tweaks of life. I'm only worried that the parameters of my argument don't overflow you're short term memory into an eye piercing inducement of sleep.

And no, I don't enjoy writing with deliberate rich vocabulary. That's just your problem to tackle if dictionary.com becomes an official widget in your desktop environment.

Unique vocabulary was conceived to actually reduce the verbosity of writing and elegantly distributing information in a pseudo compressed fashion. The less we have to throw in our mind to think of at once, the simpler reading ends up being for bandwidth rapists and raw index fingered front desk receptionists.

First impressions do nothing unless you can peruse a catalog of posts and find something of interest, so bookmark this if you're interested in future Linux and tech related opinions.