44 replies to this topic

[hytga]

sip how do u hack a linux kernell? teach me plz

[Sip]

The term "hack" and "hacking" when dealing with computers has two big definitions. Most non computer people seem to know the bad connotation associated with a "hacker". The dictionary says:

hack:
intransitive verb COMPUTING get into a computer system: to explore and manipulate the workings of a computer or other technological device or system, either for the purpose of understanding how it works or to gain unauthorized access.

and for "hacker" they say:

hacker  noun
1. COMPUTING somebody accessing another’s computer: somebody who uses computer expertise to gain unauthorized access to a computer system belonging to another, either to learn about the system or to examine its data

2. COMPUTING computer enthusiast: somebody who is very interested or skilled in computer technology and programming

When I say to "hack" the linux kernel, I mean it more as in get to know how everything works, change source code, recompile it to work differently, or add/remove functionality from it. Basically, any serious computer science or computer engineering operating systems class you take at a university should make you do all that.

If you want to learn how the linux kernel works, there is a LOT of books, texts, sites, and info on it and it's not something I can teach you in a post or two. Best way is to get your hands on some real linux distribution, install EVERYTHING (including the source code), and start experimenting with it.

If you mean to "hack into the kernel" to gain unauthorized access, that's a totally different story. And it really depends on what it is you are trying to do and what kind of access you have to the system. Most often, it's not easy doing anything bad to a system if you don't have root access. So most attemptes at "hacking" are to get root access (administrator) somehow. Once you are root, you can do pretty much anything.

[Siamanto]

That's not true. Even C compilers are written in C!  At some point someone had to write a really basic C compiler in assembler ... sure.  But then you write you fancy C compiler, compile it using the basic compiler, and then recompile yourself with yourself and Voila!

Thank you for agreeing that C compilers are NOT ENTIRELY written in C and are NOT necessarily written in C: assembler is required!
But that is irrelevant! The point was that, the fact that an OS - i.e. a low level software - is written in C, does not mean that a user product - i.e. higher level software - should be written in C. In fact, very low level software is written in assembler or microcode in firmware, so should we develop Web products in microcode? You can if you have the time!
It's about Software Development Management and Cost and is not necessarily a technological issue: the human resource factor and time to market/cost and...are the main arguments!
There's no one solution fits all!
For the record, you can also write a basic Java compiler in C - or assembler or microcode - and later write a fancy Java compiler in Java! What's the point and where is the challenge?

But as far as Java, it's a lot like Visual Basic vs C ... if you really want to be a true nerd, you'll get to use what really matters. If you just want to be a developer and release "fine" products that follow sound OO design practices, object reuse, platform independence, etc etc, you'll use Java.

Did anybody suggested or stated the contrary? Am I missing the point? Thanks!

When the time comes to REAL performance and real power, Java ain't got nothin on C! Especially when you are dealing with real hardware.  How the heck does one read or write to a specific memory address, do memory mapped IO, or write a low level driver in Java?  I guess the answer is using libraries written in C or assembly

Of course! Nobody will write firmware in Java! The use cases are totally different! Isn't that obvious?
It was about high level "end user" products! Drivers are not classified as such and obviously can hardly be OS independent!

I definitely think Java is headed towards being a fancy version of what Visual Basic was.  What I can't stand is how they are teaching it now as the standard language to COMPUTER ENGINEERING studnets!!!! These poor kids show up and have no idea how to hack a linux kernel, write low level driver code, mess with memory managers, or even do basic sockets in C!

Java is headed towards any other Programming Language that exists: retirement. All technologies are temporary!
It's not about technolgies. I don't have time for a thorough comparison of Java vs C/C++ vs VB. I will simply remind:
The advantage of Java vs. C/C++ is not necessarily in the syntax or the semantics of the respective languages.
Java is preferred because:
1- There exists an extensive amount of Java libraries: when developers "talk" the same language, it's more likely that they will build inter-compatible systems/tools!
2- An extensive number of JSRs and standard APIs
3- It promotes object re-use without the overly complicated paradigms of C++
4- Better garbage collection and dynamic memory management. Memory leaks are the most common problem!
5- Thanks to the standard APIs and libraries, it's faster to release
6- Better portability...
7- ....

"Automatic garbage collection?"  Who the heck needs automatic garbage collection when you're writing code for an mp3 player, cel phone, or in the main controller of a car engine?    Though I understand wanting to teach Java to more "software engineering" type students ... but for God's sake, NOT to hard core hardware and embedded systems guys!

Of course you need! The same memory leak will exist on the embedded system and is far more critical because of the limited resources.
For the record, if you think that garbage collection is not required for an MP3 player, how would it hurt?
Again, it's about Software Cost and Compatibility!

One last question: where is the problem? what's your point?

Time to go out! Have a nice day!

[Sip]

The problem is that if all the applications I use migrate to Java I will probably denounce computers all together and go become a monk in Tibet.

Unless I see processors being developed that can execute native Java byte code, anytime I hear Java I am going to immediately think WASTED performance.

I am just looking at my system right now and I got around 3 Ghz pumping through the processor and a GIG of ram and it's still not performing that much better than my 486 with less than 1/20 of all those resources. Why? Because of all this wasted layers ... virtual machines and hardware abstraction layers and inter operable this and that.

Basically, my problem is, that with all these widely varying architectures for computers, I do not think having a unifying software platform and forcing every architecture to execute the same "byte code" is a good thing. It sounds good in theory, but in reality it SUCKS.

[DominO]

Siamanto, come on, it isn`t the first time I will be using Linux or installing it in my system, sometimes ago I already installed it as my first system, but left it out for some reasons(like the screen refresh rate problems I had).

I know about the basic burner that comes with Linux, but I was more expecting a Nero type full featured software supporting many CD and DVD burning options.

Domino,
Your posts and your comment about "the 'nerdness' of using Open Source" led me to think that you wanted to explore yourself!
If your goal is to simply burn CDs then Linux has a native CD burner called "cdrecord." It is very likely that Mendrake  includes a GUI interface to cdrecord: Redhat9 does and very likely Fedora!
Check:
  http://en.tldp.org/H...ting-HOWTO.html

If looking to experiment with a GUI for cdrecord, check:
  http://nicestep.sourceforge.net/



It is a good practice to opt for solutions that are OS independent, open platform and can grow gracefully. The following simple guideline should help you stay away from "dead-ends" and "closed doors!"
1- Pure Java implementation should be preferred
2- Web based presentation is preferred to desktop GUI i.e. Swing - the Java GUI - and OS native GUI. Avoid Java applets; it's a dead end technology!
3- All peripherals and devices should support TCP/IP i.e. Ethernet and/or WI-FI. Wireless G is the preferred standard!
USB should be a "last resort" choice (i.e. "un pis aller") but may be acceptable!
4- ...

When selecting an Open Source component/tool, take into consideration the following - incomplete, but simple - indicators:
1- The type of release: aplha, beta or production (The information is provided)
2-  The number of developers
3- The frequency of releases and the last release date
4- How widely used they are used
5- ...
The above indicators do not necessarily accurately reflect the realities; but are helpful!

[DominO]

Oh and I already once installed Debian on a 68K Mac(an 68LC040 Mac upgraded to 68040). It took my years, but finaly it worked.

I still have one of these, I wonder if it worth doing that again.

EDIT: I forgot to add, I installed that one up on another machine. http://www.openbsd.org/mac68k.html But for this one, I won`t tell you how much time it took. It is embarassing.

Edited by QueBeceR, 29 January 2005 - 05:57 PM.

[Azat]

Javas days are numbered.

I have to tell you that the number of people who are moving to .NET will severely impact the Java community out there and developers who stay on Unix, Java, just C will have a harder and harder time finding jobs as time goes by.

[DominO]

Javas days are numbered.

I have to tell you that the number of people who are moving to .NET will severely impact the Java community out there and developers who stay on Unix, Java, just C will have a harder and harder time finding jobs as time goes by.

Another thing Bill Gates has succeeded in.

[Azat]

For a good reason Domino. It is very easy to work with. Unlike Java the developers can be of lower caliber yet the applications that they developer will work faster and better than those developed in Java and resources are less expansive. As an executive who has to make decisions on what platform to use to have a solid application at the lowest price possible it is hard to go with much else but .NET. (And I am strictly talking about business applications and not college or small startup company applications)

[Siamanto]

The problem is that if all the applications I use migrate to Java I will probably denounce computers all together and go become a monk in Tibet.

Unless I see processors being developed that can execute native Java byte code, anytime I hear Java I am going to immediately think WASTED performance.

I am just looking at my system right now and I got around 3 Ghz pumping through the processor and a GIG of ram and it's still not performing that much better than my 486 with less than 1/20 of all those resources.  Why?  Because of all this wasted layers ... virtual machines and hardware abstraction layers and inter operable this and that.

Basically, my problem is, that with all these widely varying architectures for computers, I do not think having a unifying software platform and forcing every architecture to execute the same "byte code" is a good thing. It sounds good in theory, but in reality it SUCKS.

It is true that the multiple layers of software have a negative impact on the usage of hardware resources and the overall performance, but that is a natural trend that will continue. Sostware has become more complex and it would be hard to develop and maintain future Software Systems without further architectural concepts and abstractions/layers!
In any case, it has nothing to do with "bytecode" or Java - or XML in the near future! In fact, the same is true about C++and compiled Java. Also, some/most of the performance deterioration is due to poor engineering: hardware costs less than good engineering! Unfortunate but true!

Have a nice day!


PS
I have two laptops with similar hardware characteristics; one has Redhat9, the other XP. I get a much better and stable performance on Redhat; despite the fact that I'm far more "abusive" with Redhat.

[Siamanto]

Javas days are numbered.
I have to tell you that the number of people who are moving to .NET will severely impact the Java community out there and developers who stay on Unix, Java, just C will have a harder and harder time finding jobs as time goes by.

Azat,
So what's the magic number? Shall we start the countdown? All technologies have a "numbered days!"
We've been hearing the above "prediction" for a while now! The Industry was - and remains - in such a "desperate" and "confused" situation that anything would have been considered as the universal panacea...There will be some "corrections" - in the sense of the stock markets - and the enthusiasm will fade away!
From a technical point of view, .NET is somewhere between J2SE and J2EE and can only replace J2SE and in some cases J2EE - for small to medium size projects/businesses. The .NET architecture/APIs are not robust enough to handle large projects and enterprise grade solutions!
For the record, in my previous messages, by "Java" I meant J2SE only!
In any case, for anyone who is considering to migrate to Linux or keep a mixed environment, .NET is not the preferred solution because it is NOT OS INDEPENDENT!


PS.
Considering the fact large scale projects last longer than small size projects, it may the case that Java will live longer than .NET!

For a good reason Domino.  It is very easy to work with.  Unlike Java the developers can be of lower caliber yet the applications that they developer will work faster and better than those developed in Java and resources are less expansive.

Yes, it is easier to use than J2EE, but not J2SE! The industry have misused J2EE by using it in small/medium size projects! Later, they complained about it's complexity! How unusual!
J2EE 3.0 is supposed to correct the issue and allow for lighter solutions. In any case, J2EE is not rivaled by .NET for large and enterprise grade projects! It's more complex for good reasons!

As an executive who has to make decisions on what platform to use to have a solid application at the lowest price possible it is hard to go with much else but .NET. (And I am strictly talking about business applications and not college or small startup company applications)

Mostly in small to medium size businesses! Microsoft have so far failed - i.e. for the last 10+ years - to provide enterprise grade solutions. I admit that Microsoft platforms became more reliable and stable, but they are still unreliable for highly mission critical and enterprise grade solutions.
I have an XP that I use occasionally, each time I try something a little complex it either fails or crashes! Microsoft is high maintenance, compared to UNIX! I don't want to say that Linux is as robust and stable than UNIX, but much more than Microsoft!



Have a nice day!

[DominO]

I have an XP that I use occasionally, each time I try something a little complex it either fails or crashes!

My Windows XP hasn<t crashed once for months... I don`t use any NASA softwares on it, but uses Maple, Mathematica. And hasn`t neither crashed with many GNU softwares like some for heavy video compressing.

To say the truth, comparing Mandrake 8 and 9(hasn`t tried 10 yet), for me they were more unstable than Windows XP, and I am even not talking about video drivers issue that was making the system crash and lose X Windows.

[Sip]

I'll reply in detail soon but I just wanted to say that I hope we are nerdy enough that we won't go down this line of "XP crashes" versus "Linux crashes" because in my experience, both kernels are quite stable.

Whenever you say "crash", I hope you would be kind enough to specify what you mean by a crash ... a driver, an application, a random system stall/reset etc. And to add on top of this, if it is a random "crash", it is almost never the fault of the OS either in case of linux or XP.

Edit #1: Also, making statements like "Mandrake performs better than XP" definitely needs to be elaborated on!!!! As in DOING WHAT? What other services are running in the background and how is this performance measured (gut feeling is also valid here ).

Edit #2: By the way, I don't need to be convinced that Linux/Unix are far more superior to windows in multi-user enviornments. But Windows is very hard to beat in the single user "PC" context IMO.

Edited by Sip, 30 January 2005 - 02:08 PM.

[Azat]

hmmmmm
Much of what you say is true, but I do like to add some comments.

.NET is OS independent. There is a .NET compiler for the palm, Linux and I think Sun UX and HP UX are just around the corner.

large scale project. I have to say coming from Candle and IBM I would never use java or .NET for enterprise applications(very large apps). c/c++ is the only way to go. However for the past year+ and continuing for another 18 months I am in the middle of a fairly large mainframe/vax/linux replacement project. We are moving everything over to .NET and SQ 2000 on Win 2003 and so far things are going very smooth.

Today we do have a portion of our application on .NET and win 2000 Enterprise servers and really have no issues with them and do about 90 million dollars worth of business off those servers.

[Siamanto]

I'll reply in detail soon but I just wanted to say that I hope we are nerdy enough that we won't go down this line of "XP crashes" versus "Linux crashes" because in my experience, both kernels are quite stable.

As I have already mentioned, it is true that Microsoft platforms became far more stable than they used to be, at least for desktop applications - and some distributed - applications. My focus was the distributed ones! Let's come back to that later!

Whenever you say "crash", I hope you would be kind enough to specify what you mean by a crash ... a driver, an application, a random system stall/reset etc.  And to add on top of this, if it is a random "crash", it is almost never the fault of the OS either in case of linux or XP.
Edit #1: Also, making statements like "Mandrake performs better than XP" definitely needs to be elaborated on!!!! As in DOING WHAT? What other services are running in the background and how is this performance measured (gut feeling is also valid here ).

For the record, my statement was:
"I have an XP that I use occasionally, each time I try something a little complex it either fails or crashes! Microsoft is high maintenance, compared to UNIX! I don't want to say that Linux is as robust and stable than[as!] UNIX, but much more than Microsoft!"
1- It says "fails or crashes"
2- It had no mention of Mandrake. Maybe, you were replying to Domino!

Yes, it was a hasty statement and now that you asked for an explanation, I'll do my best to explain myself. Let's call it "lazy explanation" - as in lazy evaluation!

Let me first - briefly - describe the major components of my System/Physical Infrastructure.
I have a wireless network that includes:
1- A File/Print Server that I use for backups, storage and printing: it's a Linux box (Redhat9.) Connected to a router that is Wireless Bridged to the main wireless router.
2- A "Cpu Server" that I use for CPU intensive tasks/Applications. A dual processor running Redhat Enterprise Server. Ethernet 100M wired connection to the main wireless router.
3- A laptop that I use as my main "desktop" running Redhat9. Wireless Client.
4- A laptop that I use occasionally running XP. We'll come back to it later. Wireless Client.
5- Two network hard drives that are managed by the File/Print Server - they don't have an embedded OS! Ethernet 100M connection to the same router as the File/Print Server.
6- A Multimedia Server - a Roku HD1000 - powered by an embedded Linux connected to my Stereo/Entertainment System. Wireless Bridged.
6- I'm planning to add a Universal Controller with embedded Linux. Wireless Client.
http://www.hometoys....nto/ipronto.htm
7- I'm planning to add a UPnP Media Broadcaster/Streamer. Did not make my choice yet because I only know of two products: one in development and the other is architecturally "closed!"
I'm considering: HMS300 - in development - by Arcadyan:
http://www.arcadyan....igital_Home.htm
Also Linux based. Wireless Connection.

I use the Xp laptop for the following:
1- Validate Microsoft Office Documents before I share them with others. Often I create the documents on Linux, but I prefer validating them. For the record, there has been FEW instances when the Linux proved to be more compatible!!!! But it's irrelevant!
2- RealPlayer/Rhapsody. Soon to be handled by a hardware Media Streamer/Broadcaster i.e. #7 above.
3- Occasionally scan/Photoshop.

From time to time, I try to find a better usage of the under used Xp laptop. I tried to use it as a Backup Manager and it miserably failed because it could not properly handle the networked resources. It proved to be so unreliable and failed so frequently that I had to kill the project.
Bacula - an open source - handles it smoothly. Bacula is not easy to configure and requires a good technical background, but all distributed solutions involving heterogeneous environments do!
I tried to use it as a Media Server and again it failed for the same reasons.
etc. etc.

My definition of a "complex" application is not a desktop application that executes complex algorithms/tasks such as the ones mentioned by Domino:
1- Often they exercise a limited number of OS modules.
2- Their INTERACTION with the OS is relatively straightforward
3- Their execution model is relatively straightforward
I Characterize a "complex" applications as follows:
1- Distributed
2- Multitasking/Asynchronous execution of different sub-systems
3- IPC
4- Heterogeneous environments
5- Exercise many aspects of the OS
6-...anything I forgot!

I hope that the above did clarify my hasty statement!

Edit #2: By the way, I don't need to be convinced that Linux/Unix are far more superior to windows in multi-user environments. But Windows is very hard to beat in the single user "PC" context IMO.

Yes, Microsoft platforms are good enough for the home and desktop applications!
Honestly, initially, I did not know why Domino was trying to migrate to Linux. After his claim that he's happy with Microsoft Xp, I'm starting to think that he has too much free time to waste if he's considering to replace a system that works for him!

Edited by Siamanto, 30 January 2005 - 09:14 PM.

[Siamanto]

hmmmmm
Much of what you say is true, but I do like to add some comments.

.NET is OS independent.  There is a .NET compiler for the palm, Linux and I think Sun UX and HP UX are just around the corner.

That is what Microsoft advertises! I would like to see a couple of enterprise grade applications before I pay any attention to that claim!
There exists also compilers for C/C++ - and even FORTRAN - for all existing Operating Systems, but - as you know - that does not guarantee portability of non-trivial applications!
I will have to remind that we are not comparing languages but Application Development Frameworks i.e. .NET vs. J2SE/J2EE.

large scale project.  I have to say coming from Candle and IBM I would never use java or .NET for enterprise applications(very large apps). c/c++ is the only way to go.

Let's leave Candle aside: it is not an Application Development Framework and let's focus on .NET vs J2SE/J2EE. Besides, I don't want to upset Sasun!
As for Websphere - IBM's Application Server and flagship product - it is a J2EE Framework!
Again C/C++ has its uses but do you know of any CURRENTLY existing Application Development Framework based on C/C++? C/C++ has still a market in the back-end, system level Software development, but not Application Development!

However for the past year+ and continuing for another 18 months I am in the middle of a fairly large mainframe/vax/linux replacement project.  We are moving everything over to .NET and SQ 2000 on Win 2003 and so far things are going very smooth.

1- Isn't too early to call?
2- I need to better understand understand the Architecture before I make an assessment
3- I wish you success
4- That should not mean that you couldn't have equally succeeded with plain C/C++, J2SE only or with a J2EE !

Today we do have a portion of our application on .NET and win 2000 Enterprise servers and really have no issues with them and do about 90 million dollars worth of business off those servers.

Again, I need to understand the Architecture. $90M does not tell me anything about the Functional and/or Technical Architecture. Nor the use cases.
There exists FORTRAN/COBOL applications that generate comparable revenues. There exists very sophisticated Applications that do not generate $90K

Edited by Siamanto, 30 January 2005 - 09:11 PM.

[Sasun]

Let's leave Candle aside: it is not an Application Development Framework and let's focus on .NET vs J2SE/J2EE. Besides, I don't want to upset Sasun!

That's a good idea. You nerds leave my candle alone

[Siamanto]

Sip,
For the record: Most storage is accessible on all machines: NFS/Samba

Also, as for the performance:
1- I have noticed that Linux handles the network in a more efficient way i.e. download a huge file and compare the performance.
2- Disk IO, page faults etc are better handled on UNIX/Linux
3- Task scheduling is more robust.
etc. etc.

I have to go!

Edited by Siamanto, 30 January 2005 - 09:58 PM.

[DominO]

I did not know why Domino was trying to migrate to Linux. After his claim that he's happy with Microsoft Xp, I'm starting to think that he has too much free time to waste if he's considering to replace a system that works for him!

The reason being that I am slowly changing my existing programs to GNU equivalents. I have enough of searching serial numbers for pirated softwares, or hacks, and even my windows is pirated.

I want an all clean environment and nothing to worry about, so I will give Mandrake another chance. I just hope the graphic drivers are better supported, and since I have a new videocard... I said to myself why not.

I will still have a Windows XP system, for the few programs and games that there is no alternatives, a system that won't be connected to the internet.

[Azat]

Let's leave Candle aside: it is not an Application Development Framework and let's focus on .NET vs J2SE/J2EE. Besides, I don't want to upset Sasun!

hehehehe

Siamanto jan, I did not mean after using Candle and IBM products but rather after working at both companies on Enterprise applications for both.