Re: Moodle in hebrew
Micha Silver wrote: Just installed moodle with the hebrew language pack (from the moodle site). I see that the RTL is not perfect and some strings are not translated yet. Is anyone still working on this? (I seem to remember improving the translation was a candidate for the Hamakor prize some years ago). A Moddle translation project asked for and got a grant from ISOC and then canceled partcipation for undisclosed reasons. An earlier non complete translation is what you've been using. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: implement timers in a multi thread application
Rafi Cohen wrote: Hi, I need some advice implementing timers in a multithread aplication on linux (suse 9.3, kernel 2.6.11). The application is written in C. At first, I chose to use alarm() with a handler for SIGALRM, but after some tests I concluded tht it's not wise to use this in a multithread application, since it seems that each call to alarm() affects or even cancels the previous one, even though the previous call was in another thread. Do I miss here something? may this still be used in a multithread application? Yes, I know about getitimer and setitimer, but I do not understand how to adjust the timer, for example if I want a timer of 20 seconds. If this is the solution you suggest, please provide an example that shows how to use those functions. Generally, any advise concerning this issue will be most appreciated. Thanks, Rafi. POSIX timers are per proccess not per thread. setitimer will still give you one timer per process (OK, actually it will give you 3 timers but each runs in a different time domain and you are only interested in wall clock timer it seems so you still have only one). To do what you want you need to write you on timers heap implmentation based on the single timer you have. You are advised to use a special timers thread for this that blocks with sigwait() for SIGALRM and mask out all other threads from receving SIGALRM if you want anything near sane application. Generally threads and signals mix badly and in POSIX a timer is implmented as a signal :-( Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: IO stats per process
Danny Lieberman wrote: I have a simply question, namely How many disk block reads and writes is a process doing over a given period of time, and what is the minimum, maximum and average stats This seemingly trivial question (which can easily be answered in a Windows operating system using the system performance monitor) appears to be a gray area at best. I know there is work on io-accounting in the new versions of the kernel -- but a) I dont have the luxury of updating the kernel and b) these are stats I would expect to see in user space like vmstat or top in any version of the kernel. Still waiting for an answer. I am sure one of the people on the list will know the answer. The reason there is no good answer to your question is because it seems on it's face to be very well defined, but it really isn't. I'll try to explain why: In Linux processes do not talk to block devices directly (unless using direct IO calls, which is rare). Processes make system calls that call on the file system drivers that call on the buffer cache which in turn uses the block device drivers to write/read from actual disk hardware. The problem is, that all this does not happen in a synchronious manner - the process can make a write system call that will put data in the buffer cache and the kernel will decide to flush those buffer to disk 3.5 seconds later (perhaps even after the proccess has ended!), along with half a dozen other buffers from other processes. Do you charge the proccess for this IO activity and how? In a similar manner: it is possible that a write will go to cache, a second write (from same or different proccess) will update the buffer in the cace and only the end buffer will be written to disk. When the buffer will written to disk at the end - who do you charge for the IO - the first proccess, the second proccess, both? Since the correct answer to these and similar question is completly situation dependendant, a general tool that does what you want does not exist. If you will describe in detail your need we can advise on a good way to acheive it. Hope this helps, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Ethical question..
Shachar Shemesh wrote: Gilad Ben-Yossef wrote: So much as Linus can speak for the entire gang of kernel copyright holders (and probably even if not because of estopel) Isn't estopel only relevant once you tried to trial such claim and fail? I doubt any such thing has happened (i.e. - a kernel developer trying to sue for GPL violation over an IOCTL and losing). Even if it did happen, that same developer couldn't sue again over that claim, but for another developer that would merely be precedence, not estopel. I'm not lawyer, but I believe estoppel is not limited to claims raised in a court case. That is indeed one form of estoppel, but not the only one. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Ethical question..
Shachar Shemesh wrote: I offered instead that they write a GPL module that allows access to what they need in the kernel through IOCTLs and move their binary-only blob to userspace. This was acceptable to them and I believe it's the correct legal solution myself (correct me if I'm wrong) Well, if you want to be sure, I would suggest that they try to do a couple of things: 1. Set in stone the kernel-userspace API, and document it inside the kernel module. This also means that the API must not change when the kernel interfaces they export do. This is both a QA thing and a legal thing. 2. Try to make it as much a complete self-contained API as possible, preferably one that is useful to other people. There is no need for this. Kernel module communicate with user space applications (open or otherwise) via the system call interface (IOCTL, mmap, open...). Linus has made is specifically clear, in comment placed in the Linux kernel source COPYING file, that program making use of the kernel function via standart system call interface are NOT derived work. So much as Linus can speak for the entire gang of kernel copyright holders (and probably even if not because of estopel),your employers seems to be in the clear. Shameless plug This might be a good time to mention that Ravia offices and Codefidence have launched a service centered about providing commerical product makers that utilize Open Source software with the legal and technical tools they need to uphold the Open Source licenses while maintaining control over their own generated software and other IP /shamless plug = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: FireFox and Bank haPoalim
Michael Sternberg wrote: Hello. For everybody who's tired of looking on reversed hebrew of Bank haPoalim personal account web interface I have found a wonderful extension for FireFox that solves this problem: http://www.effie.co.il/mozvuvu/home.html enjoy :) P.S. I do not understand why Bank haPoalim still did not fixed their problem with reverse Hebrew in FireFox, its there for at least three years.. Probably because they don't an effie of Firefox users or site user at large :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Fwd: source routing in extensions.conf
Tzafrir Cohen wrote: Gilad? The mailing list manager config is broken. I'm going to fix it this weekend. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: How to bind privileged ports in a non-root process?
Nadav Har'El wrote: Hi, I've run into a problem which it seems like thousands of people must have come across before, but strangely, I can't figure out a good solution. I have a program which I want to bind port 80. Normally, in Linux, you must be root to bind privileged ports (below port 1024), but I DO NOT want to run this program as root - I don't want to allow it to touch files which doesn't belong to it, or give it any other of the superpowers of the superuser. How do I do that? The obvious solution to this problem - which is what most Unix servers do - is to start the server as root, and after opening the port do a setuid(). But I can't do this: the program (written in Java, by the way) opens the socket itself, and can't call setuid() (there's no such thing in Java) or inherit a socket (no such thing in Java too...). So basically, I want to tell linux to allow this process (or any process, for that manner) to bind any port. I tried using capabilities, but didn't get it to work (does anyone know if this feature still exists in modern kernels??), and I tried looking for a relevant sysctl and couldn't find one (which really suprised me). So can anybody suggest an idea on how to let a non-root process bind port 80 on Linux? In theory at least - start your program from a small SUID root excutable that uses the prctl() PR_SET_KEEPCAPS operation to enable your proccess to not clear capabilities when switching UID's. Give out all capapbilities but CAP_NET_BIND_SERVICE, change real, effective and saved UID to a user one and exec your program. In thoery at least, this should work for Linux 2.6.14 and above. Never tried it myself. Assuming this does not work (capabilites handling in Linux is more of a work in progress, really early in progress actually), I can offer the following kludge: Prepeare a small library that intercepts call to the libray socket() and bind() calls and if, according to the parameters, the socket/bind is for the low port socket, instead of calling the actuall bind/socket of the library, will simply return the apropriate saved file descriptor. For an example of how to do this see: http://www.codefidence.com/src/bindtodevice.c Now, start the program from a SUID root exeutable that will bind to the socket as some known file descriptor (say 3). Then give up root and exec your program, making sure to alter the enviornment passed to it such as LD_PRELOAD enviornment variable is set to force your library to load. It's a horrible hack, I admit. But it will most probably work. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Shared Library entry point
Ami Chayun wrote: Hi all, I have a shared library, and I want a specific function to be called once the library is loaded. Dlls has the notorious DllMain function. Is there a method of achieving the same in an .so file? 5.2. Library constructor and destructor functions Libraries should export initialization and cleanup routines using the gcc __attribute__((constructor)) and __attribute__((destructor)) function attributes. See the gcc info pages for information on these. Constructor routines are executed before dlopen returns (or before main() is started if the library is loaded at load time). Destructor routines are executed before dlclose returns (or after exit() or completion of main() if the library is loaded at load time). The C prototypes for these functions are: void __attribute__ ((constructor)) my_init(void); void __attribute__ ((destructor)) my_fini(void); Shared libraries must not be compiled with the gcc arguments ``-nostartfiles'' or ``-nostdlib''. If those arguments are used, the constructor/destructor routines will not be executed (unless special measures are taken). http://www.faqs.org/docs/Linux-HOWTO/Program-Library-HOWTO.html#INIT-AND-CLEANUP -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: tarball repository
Shachar Shemesh wrote: PTxdist: http://www.pengutronix.de/software/ptxdist/index_en.html Personally, I found ptxdist to be too hard coded for the things already included in it, and not customizable enough for other things. I may not have spent enough time on it, considering that I had already built something along those lines myself by the time I became aware of it, so take this criticism with a grain of salt. Shrug sounds like a case of NIH to me. I've used Ptxdist in numerous projects that went into products of Fortune 500 companies and while it was not without it's quirks it was Good Enough (TM). Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: tarball repository
Shachar Shemesh wrote: PTxdist: http://www.pengutronix.de/software/ptxdist/index_en.html Personally, I found ptxdist to be too hard coded for the things already included in it, and not customizable enough for other things. I may not have spent enough time on it, considering that I had already built something along those lines myself by the time I became aware of it, so take this criticism with a grain of salt. Shrug sounds like a case of NIH to me. I've used Ptxdist in numerous projects that went into product of Fortune 500 companies and while it was not without it's quirks it was Good Enough (TM). Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: tarball repository
Ira Abramov wrote: right now it's a bit of a spaghetti mess and there's no central repository of all the packages at all the versions (older tags of the CVS tree stil use older packages). I was wondering if there has already been a solution for something like this, like a simple portage system that supports downloading from random sources (rather than a centralized one) and be able to hold several versions of the same packages as needed, or should I invent a wheel from scratch afterall? PTxdist: http://www.pengutronix.de/software/ptxdist/index_en.html -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
[Shameless plug] Re: Linux kernel training
Michael Sternberg wrote: Hello Can you, please, recommend me courses on Linux kernel development ? I found something at Interbit http://www.interbit.co.il/index.php?action=showid=98 how good are them ? Can't say anything about Interbit, but Codefidence (of which I'm a co-founder) is provding courses on the Linux kernel through High tech colleage. Since I'm the editor/co-auther of the study material and also pass the course personally from time to time I am not objective, but judging from the feedback we get after each course, they are quite good, if I do say so myself: http://www.hi-tech.co.il/college/hitechin.asp?cc=coursepagecourse_code=913 We also teach courses in Embedded Linux and similar subjects. In addition, Oron Peled (Hi Oron!) also teaches Linux kernel courses, I believe through UniExpert and maybe also IBM: http://users.actcom.co.il/~oron/oron/docs/kernel/index.html I think a similar course also takesw place in Sela colleage, but I'm rather vague as to details. Hope this helps, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: modem - air card sierra wirless 850
eli elizur wrote: Hi guy's, any one ave experience with configuration of modem - air card sierra wireless 850 ? I do, although I don't remember much except that it's a pcmcia card which has a USB controller, to which internal hub connected an internal USB serial modem. You're supposed to get all that configured, do an AT dial and talk PPP with whatever at the other end. There's some magical AT command that you need to send it to turn on and really work but alas I do not remember what it is. Hope this helps, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Are there non-free versions of Linux?
Geoffrey S. Mendelson wrote: On Sun, Jul 30, 2006 at 09:20:53AM +0300, Oleg Goldshmidt wrote: I would be curious to know how this is possible technically *and* legally, i.e., without violating the kernel license (GPL). Both Caldera (now called SCO) and SuSe sold distributions that were licensed per seat. I have a copy of post SuSe purchased by Novell, Novell branded Linux that requires an activation key. They allow you to download it and provide a trial key (which I never used), but sell you one later. They are relying on the fact that companies which buy such things will pay license fees. It's perfectly legal, the GPL only covers the Linux Kernel and other software, but nowhere is one required to include only GPL or FREE (as in beer) licensed software in a distribution. That BTW is the difference between Open Office and Star Office. Star Office includes a support contract and some software that is not free (in any form). Up till now you were making perfect sense. The GPL has never been tested in court. No one knows if it would really stand up to a barage of high priced lawyers. This isn't even true (the GPL has been tested in court a few times in Germeny). More to the point, the strengh of a legal document is not judged in how it will hold up in court, but rather in how good is it in making sure the question will never reach a court because the offender will fold. Having a license that protects intelectual property of such high value used by so many rich corporations as the entire collection of software under the GPL never reach a court (in the US) means only one thing - that a barage of high priced lawyers have looked at it and time and time again recommended their clients to fold. Trust me, I work as a consultant to/with lawyers that provide legal advise for corporation of questions of licensing. The GPL does just what it was planned to do and does it very well, *including* having some parts so murky and undefiend on purpose as to scare off good lawyers reccomending their clients to do stuff that technically might be legal. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistance was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: schedule_timeout vs interruptible_sleep_on_timeout
Chava Leviatan wrote: I have a kernel thread which needs to be woke-up every 50 mili seconds (un-conditionaly) to be used as a periodic timer Are you sure a simple kernel timer is not a better option? why do you need a new kernel thread for this? The kernel version I am using is 2.4.18 . I have made a small research and found 2 possibilities: schedule_timeout interruptible_sleep_on_timeout. What wrong with? msleep(50*HZ/1000); Not sure if it's available in the ancient 2.4.18 version though. I made some experiements with both of them , and under a medium load (24Mbit) , it seems that the interruptible is more accurate. The schedule has deviation of several miliseconds every 50/100 samples (I totally took a batch of 150 samples - each sample is a 50 mili interval ). The interruptible_sleep, has also that deviation , but less frequent (once per batch), and the overall picture shows that it is more accurate. Make sure to give your kernel thread real time scheduling priority (SCHED_RR or SCHED_FIFO). Making sur eno driver blocks interrupts for long time is also a good idea :-) On the other hand, I have read that the interruptible_sleep is less stable and gets into race conditions. The question is : Did everyone tried to work with both of those APIs? If the interruptible_sleep is called unconditionaly , does it still un-recommended for use ? The race condition people mention with interruptible_sleep_on_timeout is related to using it with a condition. msleep is still better, though. Hope this helps, Guilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistence was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: diff regexps
Adam Morrison wrote: Apparently, diff uses basic regular expressions. Indeed, that was also why my regexp did not work - I was trying to use extended regular expressions, whereas diff only support basic ones. Luckily, for GNU versions of grep/diff, the only difference between basic and extended regexps is that special characters of extended mode (namely {, } |, $, ^ etc) need to be prefixed with a back slash for their special meaning to be used. Yes, backwards from you normally would expect, a $ is just a dollarm but a \$ is a marker for end of line. In the end, this is the diff line used: # diff -pBbNaur -X dontdiff this_kernel/ that_kernel/ -I '$Id\|$Header\|$Date\|$Source\|$Author\|$Revision' Note that this did *not* weed out all uses of the CVS keywords, only those that happend in a block of chanes where all the changed lines matched the regexp, so $Log ... $ lines, for example, could not be catched using this technique. Luckily, after this diff line the number of files with those lines was small enough to allow manual trimming by using an exclude by file technique. Thanks very much to Ehud, Shachar, Oleg, Adam and everyone else! Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 Resistence was futile. -- Danny Getz, 2004. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
diff regexps
Howdie, I am trying to diff a vanilla vs. a vendor supplied linux kernel source tree. Some dweeb at the vendor has put the Linux kernel into CVS, causing every line with $Id, $Revision, $Date, $Source etc. in the Linux kernel to mutate. I'm trying to generate a diff that doesn't include these random changes but does include the real changes that occurs between the two trees. Now, diff has an option, -I, via one can provide a regexp and blocks of changed lines that all contain the regexp will not get to the output. The problem is that I tried various combiniations and none worked: diff -X ti_dontdiff -pBbNaur -X dontdiff this-kernel/ that-kernel/ -I '\$Id' -I '\$Header' -I '\$Date' -I '\$Source' -I '\$Auther' and also: diff -X ti_dontdiff -pBbNaur -X dontdiff this-kernel/ that-kernel/ -I '\$Id|\$Header|\$Date|\$Source|\$Auther' but to no avail. Anyone care to help a poor regexp challanged bugger? Thanks, Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel module question
Michael Sternberg wrote: Hi, I'm writing kernel module for 2.6 linux. The purpose is to intercept all writes to disk devices and to notify user mode application about sector/length of write. I want to change function pointers of request_fn and make_request_fn to my own routines. Sounds like a horrible and racy plan. I'm sure there's a far better way. The question is what is a difference between this two functions? I did it for make_request_fn and checked with application that writes to some sector on disk. But in kernel I got write to different location. Make_request_fn gets bio as parameter so I got writes info from it, but request_fn gets only queue as parameter. AFAIK, make_request_fn is the function that produces the request queue = it queues requests for a device. request_fn is the consumes the queue - it's the function where the driver reads the queue and handle the requests. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Getting the registers of the program in a signal handler
Howdie list, I'm trying to build the be-old, end-all exception signal signal handler for SIGSEGV, SIGILL, SIGFPE and their ilk. One thing that will be useful to do in such a handler is dump the state of the CPU registers when and where the program segfauled etc. Sadly, despite zealous code reviews and much googling I have not being able to find a way to do this. Before I embark on writing a hairy kernel module that exports the register information caught by the kernel when the exception was caught via /proc (Yikes!) or some other ugly kludge, I'd rather be interested to hear if there is a Better Way(tm). Any pointers, readme, howto's, links or just plain suggestions would be very welcome. Thanks! Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Getting the registers of the program in a signal handler
Muli Ben-Yehuda wrote: On Sun, Jul 02, 2006 at 02:12:39PM +0300, Gilad Ben-Yossef wrote: I'm trying to build the be-old, end-all exception signal signal handler for SIGSEGV, SIGILL, SIGFPE and their ilk. One thing that will be useful to do in such a handler is dump the state of the CPU registers when and where the program segfauled etc. Sadly, despite zealous code reviews and much googling I have not being able to find a way to do this. IIRC, you want to set your signal handler with sigaction(), with sa_sigaction sent to your handler. Then the third parameter to sa_sigaction is the context, including the register context. See getcontext() for the definition of the context structure. Damn! I was sure that there is some simple way to get it from the signal handler using sigaction, but I looked at the various fields of siginfo_t and overlooked the context argument because it was a void * type. Silly me... Yet again, thank Muli and all :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Recommendations for XP under Linux
Aharon Schkolnik wrote: I have started a new job and find that I will have to use some applications which will run only under Windows (XP). I would like to run Linux on my desktop with a virtual XP machine running under it. In the (relatively distant) past I have use VMware for this. At the time, I convinced my employer to shell out the money for the VMware license. I have been informed that getting the money out of my current employer may be difficult and time consuming. In any event, I need to get some solution up and running right away, so they don't start pressuring me to put Windows on my desktop. I understand that VMware has a free player and that it is possible to hack it into installing XP. Is this as big a pain as it looks ? Has anyone done it ? Are there other possible solutions ? What do you recommend ? Use qemu (http://www.qemu.org/) or change jobs ;-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Recommendations for XP under Linux
Shachar Shemesh wrote: Gilad Ben-Yossef wrote: Use qemu (http://www.qemu.org/) or change jobs ;-) Unless qemu changed the fundamental way it is built since last I checked it, I don't think it is the right solution for Aharon. Last time I looked at it, It was SLOW. VERY slow. Not changed, expanded. Google Qemu accelerator module. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Recommendations for XP under Linux
Hetz Ben Hamo wrote: 3. Use QEMU to install the guest OS and then use VMWare player to use your guest OS. You'll still need the guest tools ISO though.. Interesting question, because I know you (Hetz) experimented with qemu a lot - what are the reasons to not just continue to use qemu instead of vmware? I'm not asking about ideology here, just wondering if you believe qemu to have stability perfomance or other problems compared to vmware client? Thanks, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: The select(3p) API
Marc A. Volovic wrote: Quoth Daniel Feiglin: My strategy is to have each fd read into a bit of shared memory, post a got something semaphore and return. You do it with 5 little if() paragraphs (error handling aside), reset the read bit mask (always the same) and back to select. How many consumers? If you post a sem, someone is waiting on it... I assume your consumer then employs the retrieved data and posts it back? Sounds like you're using a semaphore for what ought to be done via a POSIX condition variable. Just my 2cs, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: VMware guest net problems?
Marc A. Volovic wrote: Quoth Ira Abramov: Today I installed a similar setup on a faster machine, host is FC4, Not that I am casting the first stone, but anyone using FC as a production machine should have his orchideae removed. I wrote an email saying the same thing (although the word orchideae was missing from it for some reason ;-) and then decided Ira knew better already. The Fedora Core line is a non production distribution *by design*. Cheers, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Migration threads, SMP and HyperThreading: a question
Rafi Gordon wrote: I have ran ps aux |grep migration on 2 machines, one with FC4 on x86 (NON smp kernel) and one on FC4 on x86_64 (SMP kernel). Each of these machines has a single CPU. On the first (x86) machine, there was no result. On the second (x86_64) there were 2 instances of the migration thread. What does this say? Is it had to do with hyperthreading ? I doubt it. It says that on the SMP kernel your OS two virtual cores (a name I just invented to describe how the OS see's two CPU's although there is only one with hyperthreading) and therefore it fired up a migration kernel thread for each CPU it sees. If I remember well , you can know if an Intel x86/x86_64 processor supports hyperthreading by looking at the flags in /proc/cpu ; if ht exists, it supports hyperthreading (Though I am not fully shure in this point). And both these machines had the ht flags in /proc/cpuinfo. Without an SMP kernel hyperthreading is practicly turned off. Is it has to do with the fact that one machine ran NON-smp kernel and had zero migration threads while the other machine ran SMP kernel and had 2 migration threads? Yes, the machine with the SMP kernel can see and operate that two virtual cores of the hyperthreaded CPU. Without SMP support, they cannot. Will I have 2 migration threads if I will install an SMP kernel on the x86 machine ? Yes. What does this migration thread do, in short? Load balance tasks between CPU's. The scheduler prefers to give each the same CPU for each of it's runs to make the best use of various caches. It is the job of the migratiopn threads to consider the possability that a task will be better served on a different CPU and migrate it there. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 * *Please note the updated phone numbers! * * The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Asterisk - can a station register as either AIX or SIP?
Ira Abramov wrote: I started going through the different SoftPhone options and I noticed an odd thing. most hard phone support only SIP, but there are also some excellent softphones out there that support only AIX2. If I have an employee that uses a hardphone at the office but wants to connect with an AIX softphone from home, he can't do that with the same extension number? the way I read it, an extension can ONLY be one or the other. so in essence I'd have to define two extensions for that guy and put them in a ring group, plus set the IAX extension to give the CID of the main one. did I get it right? also, if I want to have two phones with the same extension number, do they register with the same number or do I have to give them different extension numbers and create a hunt/ring group? In astrerisk extensions and channels (clients) are two different thigns. There is no problem having an extension that ring two, three of 576 different channels of different types. Your problem is that you try to use the FreePBX front end which, as I told you last we talked about it, makes some things very easy at the cost of making the rest impossible. To define an extension that rings two different channles, one SIP and one IAX do: exten = 201,1,Dial(IAX2/[EMAIL PROTECTED]SIP/[EMAIL PROTECTED],30,m) For example. ehhm... I think it's time I dive into an Asterisk-specific ML, eh? :) I tried once to have an asterisk-il mailing list (actually it was called foss-telco but it was mainly asterisk). That didn't catch on quite well. Time to try agsin? Cheers, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 * *Please note the updated phone numbers! * * The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Asterisk - can a station register as either AIX or SIP?
Shachar Shemesh wrote: Gilad Ben-Yossef wrote: I tried once to have an asterisk-il mailing list (actually it was called foss-telco but it was mainly asterisk). That didn't catch on quite well. Time to try agsin? I suggest that yes. I also suggest that you actually announce it this time around :-) It was annouced to people I knew where intrested in Asterisk and VoIP and the time. Sendind what seemed like an off topic annoucement to Linux-IL didn't seem like a good idea at the time, but this was probably a mistake :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 * *Please note the updated phone numbers! * * The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
[Announcement] Asterisk-IL mailing list
Howdie, Due to popular demand, a mailing list for the discussion of all things Asterisk and FOSS VoIP and telephony in Israel has been set up. Feel free to forward this to people that you might think will be interested and interesting to join. This list supersedes the old and now dead foss-telco mailing list. Yes, you will need to re-register if you were on the old list. To subscribe to the mailing list, simply send a message with the word 'subscribe' in the message body to the -request address of the list To: [EMAIL PROTECTED] To send an e-mail to the mailing list, write to the following address: To: [EMAIL PROTECTED] To cancel the subscription for the mailing list, simply send a message with the word 'unsubscribe' in the message body to the -request address of the mailing list To: [EMAIL PROTECTED] Have a nice day, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 * *Please note the updated phone numbers! * * The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Xen installation, Go #1
Muli Ben-Yehuda wrote: .. but that requires a proprietary kernel module, which is we know, is The Root of All Evil. So does VMware. And Qemu runs just fine with kqemu. Slower, yes, but good enough for many applications. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 * *Please note the updated phone numbers! * * The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Xen installation, Go #1
Muli Ben-Yehuda wrote: Since you need to support RH 7.2 and 7.3, both of which are based on 2.4, and since Xen 3.0 dropped 2.4 support, you need to either go the VT route (which necessitates new hardware, as well as being on the bleeding edge) or drop the requirement to support 7.2/3, or use something else for the time being. That something else is usually VMWare. You meant to say Qemu, of course ;-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 * *Please note the updated phone numbers! * * The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Manageable core files
Michael Sternberg wrote: How can I manage core dump files ? I'm looking for some way to name them differently, so next one will not be written over the previous and maybe to give some information in core dump file name (version of application, date of core dump creation, etc..). It can be from crashing application (some callback that will be run just before crash) or from some external watchdog application.. Is there any way to perform above in Linux ? /proc/sys/kernel/core_pattern rom the proc(5) man file: This file (new in Linux 2.5) provides finer control over the form of a core filename than the obsolete /proc/sys/ker- nel/core_uses_pid file described below. The name for a core file is controlled by defining a template in /proc/sys/ker- nel/core_pattern. The template can contain % specifiers which are substituted by the following values when a core file is created: %% A single % character %p PID of dumped process %u real UID of dumped process %g real GID of dumped process %s number of signal causing dump %t time of dump (secs since 0:00h, 1 Jan 1970) %h hostname (same as the nodename returned by uname(2)) %e executable filename A single % at the end of the template is dropped from the core filename, as is the combination of a % followed by any charac- ter other than those listed above. All other characters in the template become a literal part of the core filename. The maxi- mum size of the resulting core filename is 64 bytes. The default value in this file is core. For backward compatibil- ity, if /proc/sys/kernel/core_pattern does not include %p and /proc/sys/kernel/core_uses_pid is non-zero, then .PID will be appended to the core filename. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 * *Please note the updated phone numbers! * * The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: One More Detail
Alon Altman wrote: On Thu, 11 May 2006, Marc A. Volovic wrote: We start amusing presentation at 3pm Dinner at 3pm on a Monday? I must have missed something here. Seriously, are we talking 3pm Monday? I sort of expected dinner to be held at 19:00++ ? Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: msgrcv without removing the message
Ori Idan wrote: I need to sleep until a message is aviliable in a message queue. I can do it using msgrcv, however it will remove the message from the queue. Is there a way to do it without removing the message? I juast want to wake up and not read the message yet. Move to POSIX message queues instead of Sys V message queues and use mq_notify(): http://www.die.net/doc/linux/man/man3/mq_notify.3.html AFAIK POSIX message queues are supported in Linux circa 2.6.6 or some such. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: BlueTooth dongle
Ilya Konstantinov wrote: Hi, Can anyone recommend a device to add BlueTooth support to my PC (via a USB dongle, I guess -- are there any other popular options?) that's sold in Israel? Naturally, decent Linux support is what's most important to me. Even if the device you use is nothing special, as long as it works on Linux 2.6, do tell. I simply don't want to buy one, bring it home and get disappointed. I have a Circle Data USB BT dongle. It works great. -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.3.7515563 ext. 201 | Fax:+972.3.7515503 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 * *Please note the updated phone numbers! * * The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Who do I have to sleep with to get an FXO card?
Ira Abramov wrote: Is that it? Nobody importing Digium? Sangoma? All I need are two FXSs The problem with Digium in Israel is that Dimitel has gotten an exclusive distribution rights in Israel and they have the typical service of a Israeli company run by engineers - none. For Sangoma just call Gideon Hack, 04-6391385. He's Sangoma CTO and he's (a new) Israeli. He will love to help you. while we're on the subject of VOIP. any good source for VOIP phones? Tikal networks: http://www.tikalnetworks.com/ You just need to ask the right people, see? :-) Oh, and welcome to the VoIP circle. From now on you can call me via SIP:[EMAIL PROTECTED] No need to bleed to bezeq or the cells anymroe :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: linux upnp client
Erez D wrote: hi does anybody know of a UPNP client for linux ? http://upnp.sourceforge.net/ I have good experience (for some definition of good) with building uPNP like clients based on it. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 The journey of a thousand miles begins with a single oy. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: * question
Erez D wrote: I tried configuring asterisk to let me enter a number, and if i do not, to ring the phone. this way i can call my home and turn on my boiler for 15 minutes (1900) or my air-condition (1901), or burn the house down (1902). but if someone else calls me,he hears 'wait a moment' and after 3 seconds it rings my home phone. and then voicemail here is part of my extensions.conf: [incoming] exten = 500,1,ResponseTimeout(3) exten = 500,2,BackGround(wait-moment) Try replacing BackGround with Playback here, see if it helps. exten = 1902,1,System('/scripts/burn_house_down.exe') A Windows executable. How fitting. g :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: asterisk questions
Shachar Shemesh wrote: Tzafrir Cohen wrote: a. a mail message can trigger just about anything on the remote side Does asterisk support the voicemail notification light that exists on many phones? Shachar On many phones - yes, but not on all of them. Almost all SIP phones work, whereas analog phone depend on the exact method of notifying the phone. -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Initrd woes after upgrade from vanilla Debian Testing
Shachar Shemesh wrote: IIRC, they are using a new mechanism to generate an initrd replacement (this is NOT an initrd in the usual sense of the word, but something understood by the kernel directly). Sadly, I don't know the precise details of the change. Perchance one of the kernel devs on this list can elaborate further? This is just a guess, but I believe they are switching from initrd to initramfs. It's a CPIO archive (read: tar file) that is included as part of the kernel at compile time that Linux 2.6 kernels will extract into a internal tmpfs file system and will attempt to run init from first. If this succeeds, this will be the boot proccess (that will usually do initrd style things and switch_root() to the real root file system after done). Otherwise a backwoards compatible boot occurs. This approach will allow in the future to throw out all the code in the kernel that has anything to do with finding the root file system, mounting NFS as root fs, kernel DHCP and IP setting (aka autoconfiguration) and replacing them will early user space - user space code compiled and linked against a mini-libc like library called klibc. This has been 60 seconds about the new initramfs :-) Cheers Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Initrd woes after upgrade from vanilla Debian Testing
Shachar Shemesh wrote: Gilad Ben-Yossef wrote: This approach will allow in the future to throw out all the code in the kernel that has anything to do with finding the root file system, mounting NFS as root fs, kernel DHCP and IP setting (aka autoconfiguration) and replacing them will early user space - user space code compiled and linked against a mini-libc like library called klibc. That much I have heard before. What I totally fail to understand is why this is any different than initrd. Reasons I don't see any difference: - The kernel did not have to find initrd either - it was loaded by the boot loader Which means you need boot loader to support initrd loading. Hell, it even means you need a boot loader - both of which are not obvious in embedded systems, for example. Another similar scenario is when using the newly (for some definition of newly) feature of kexec - with initrd you need to worry about allocating, copying and passing the right address to the initrd in case of a kexec() call, which is possible but a little messy. With initramfs you don't need to think about it even. - Initrd contained the code to do all the above too. If there ever was any such code inside the kernel (and I know that at least DHCP was there), initrd could have made it redundant anyways. Tht is true - but because initrd was sometime limiting, it was more difficult to demand all the users to use it instead of the kernel, so kernel support for these things stayed in there. - Iniramfs will save us the need to compile cramfs into the kernel statically, but instead will require us to compile cpio and tmpfs statically. tmpfs is always compiled into the kernel if you want shared memory to work on your kernel and most of us do. In addition, the initrd data would actually cause two copies of the data in it to be present in memory for items in it that are in use - one in the initrd blocks themselves, one in the page cache for actual use. With initramfs the entire data sits in the page cache anyway, so it's more conservative of memory. - The initramfs would still need to be generated after the kernel is compiled, or else we go back to requiring an initrd IN ADDITION to the initramfs, to store all the modules relevant for mounting root. Yeah, I know. That one bugs me two. I think we need a update initramfs make target for Kbuild. Should be fairly simple to do. In short, I don't understand why a new technology was necessary. Hope I've shed light on some of the reasons. I wonder if enough people are interested in a talk about early user space. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Message queue. can it get blocked?
[EMAIL PROTECTED] wrote: Hi there, Having a message queue designed so that each queue (message type) is assigned to one process for reading. How do you handle a situation where the reading process of a given queue doesn't function and messages still arrive to that queue? 1. Will the queue fill up untill blocking the entire resource? 2. Can the blocking issue effect other message queues supposebly used by other applications? 3. Can you identify from a different process that there is a queue filling up? 4. Can you clear the overflown queue? Which message queue implementation - POSIX mq, Sys V mailboxes, something you wrote? But the simplest answer is to not use the same queue for multiple readers - use different message queues. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Message queue. can it get blocked?
guy keren wrote: if anyone is using sys V message queues, they should scrap them ;) With this notion i tend to agree. The problem is that while Linux (2.6+) does implement POSIX message queues which are oh so better, their implmentation is so un documented that it ain't even funny. I had to go look in the source to figure out how to delete a POSIX message queue left after a program has run (ipcrm style). It's still better then Sys V mailboces, though :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Connecting Asterisk to an analog smart phone?
Shachar Shemesh wrote: hi list, Most standard PBXs support something called smart phones. These are phones that connect over an analog line, but receive special information directly from the PBX, and can display the name of the person behind an incoming connection, transfer calls without the traditional flash, etc. Is there a way to connect such a phone to a machine running Asterisk? Does it require a special adapter? The protocol is called ADSI and it is fully supported by Asterisk on Zaptel channels (e.g. not Voip or digial channels). For full details see: http://www.voip-info.org/wiki/view/ADSI And congrats on going with Asterisks - it's a great tool. Hope this helps, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
[JOB OFFER] Codefidence Ltd. is looking for top notch programmers
Codefidence Ltd., a provider of knowledge and support for organizations making use of Open Source technologies, is looking for top notch programmers for it's expanding operations. Preamble Can you explain why lines starting with the From can sometime be seen in articles and newspaper ads? Do you know why Caller ID will only be displayed before the second ring of the phone in Israel? Can you find out what does the following C line does on a PowerQUICC processor? asm(.long 0x7d821008); /* twge r2, r2 */ [ Google use assumed and encouraged to answer the questions above ] If you answered yes to the questions above, if you enjoy a good challenge and if you can prove to us that you have what it takes we might have a job just for you - please read on. Terms ~ Job location is in Codefidence offices in Netanya (some travel may be required from time to time). From 80% to a full time job - a must. *Partial* work from home an option. A working environment that demands learning new stuff all the time - and allocates the resources for you to do so. Examples of the of work you will be expected to do as part of these jobs include: - Help customers with questions about the use of Linux, GNU tool chain etc. programs and APIs (how to build and compile X?, How to use pthread condition variables? How to map IO mapped memory to my process?) - Write new and port existing device drivers for the Linux kernel. - Create and manage automated custom distribution building frameworks for the embedded and appliance markets - Write efficient, secure portable and documented network server programs as reference code for new protocols or driving large scale distributed content distribution systems. Requirements (in order of importance) ~ - Be reliable, honest and dependable - Great people skills - Fast learner - Ability to both work alone and guide a team - Mastery of C and one of Perl, Python and/or a UNIX shell - Experience with *one or more* of the following areas: Linux kernel programming, embedded systems, real time constraints, TCP/IP communications, Linux from Scratch or similar systems, VoIP or telephony. - Fluent English Nice to have - Active participant as a programmer in a system level Open Source project - huge advantage. - Recovering BOfH. - Doesn't mind traveling abroad. - Background with one or more of: Crosstool, uClibc, Ptxdist, Asterisk, GNU tool chains building, ACE, net-snmp, Busybox Please note that background of C#, JAVA, PHP or VB, without *other* additional qualification is not enough to gain our interest. Please spare our time and your own and don't apply. Interested applicants - please send your resumes to [EMAIL PROTECTED] Thanks, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kinnernet 2005 (OT)
[EMAIL PROTECTED] wrote: Anyone here involved in this, or know anyone who was? http://kinnernet2005.editme.com/ I'm trying to find Israelis involved in this intersection of art and technology. What's wrong with? http://kinnernet2006.editme.com/peoplelist Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kinnernet 2005 (OT)
Eli Marmor wrote: Gilad Ben-Yossef wrote: [EMAIL PROTECTED] wrote: Anyone here involved in this, or know anyone who was? http://kinnernet2005.editme.com/ I'm trying to find Israelis involved in this intersection of art and technology. What's wrong with? http://kinnernet2006.editme.com/peoplelist In our last meeting (during ISOC conference several days ago) you claimed that your password is not active anymore; Please re-check it. It seems that your password is correct and active. It didn't 1 week ago - I could not log in to any restricted area, which is why i assumed the peoplelist is public. It seems to be working now, though. As to Michael: Contrary to me and to you, he doesn't have access to the who's list. So, I made a booboo. Just add it to the the list of other booboo's I've made. Since I've written about KinnerNet in my blog it's not a big secret anyway. Hint for others: Google is not blocked from accessing the who's list; So if you want to know if a specific person was there, just type his full name (with '') in Google, and see if Kinnernet is returned among the results (you may also specify editme.com as the URL to search). The personal pages *are* public, so just look for KinnerNet and $NAME in Google to find out if someone is going or not. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Crossover office - Writing hebrew in IE
Noam Meltzer wrote: Did any of you succeeded to write Hebrew in Crossover Office IE? Can you please enlighten me of how to achieve this goal? See here: http://www.lingnu.com/support.html And thank Shachar Shemesh/Lingnu for creating it. PS. I don't use IE and I have no idea if it works, but I know the page exists and it sounded like it will help you. Cheers, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 `Y'know, I was just saying to myself, Self, I said to myself, you really need an enterprise datacenter architecture that leverages middleware based on robust frameworks. Wow, they must have been reading my mind!' -- Anonymous. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: modify_ldt() system call invokation
Rafi Gordon wrote: Hello, Processes can change their Local Descriptor Table using modify_ldt() system called. see man modify_ldt or : (for x386 look at include/asm-i386/unistd.h: #define __NR_modify_ldt 123) However, usually there is no need to manipulate the LDT with this system call. I saw in wine a call to modify_ldt() (libs/wine/ldt.c) I was wondering in which occasions is this system call being invoked: did anybody had encountered (or used himeslf) a call to modify_ldt() in other apps (emulators maybe ? ). If memory serves me right the old LinuxThreads threading library used modify_ldt on x86 to create thread local variables in the absense of TLS (The Thread Local Storage kind, not the encryption scheme) that NPTL relies on. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: dvb card for linux
Erez D wrote: gmail does not arrive at the list. --- original message (sort of) - hi i'm looking for a dvb card for my linux box anyone knows where i can get a GOOD linux compatible dvb card ? (or things i should be aware of when i buy one). can it work direcly with the dish's LNB or do i need somthing in-between. I had a very positive experience with a Technisat SkyStar2 PCI DVB card. Open Source drivers exists and work great (note that this is was 2.4 days, have no idea on the status of the drivers with 2.4) and can be found on the linuxtv.org site. I never tried to connect to a Yes supplied dish but assuming it is a run of the mill dish you just plug the coax to the card. Cheers, Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Maximum number of interruprs a CPU can receive without dropping/losing any
Rafi Gordon wrote: A practical example which had driven me to post this question originally: I want to know when I should use NAPI (polling) instead of interrupts in a network card so that I won't loss interrupts. (and data , as a result). NAPI is not polling. In fact, you missed the whole point of NAPI. Let me explain: There are two extremes in handling interrupts in general and in network devices specficially: interrupts and polling. Interrupts works great if you system does a lot of things, including network, polling works best if you're system is cheifly dedicated to networking. The problem is most systems in the world are somewhere in the middle. What NAPI does is to find, in an efficient manner, the middle ground - under NAPI the network device start in interrupt mode. When a packet arrives (and thus an interrupt is received), NAPI will turn off the interrupts of the device and will poll the device in cycles comprising of pre-defined load of work (which you can configure to set priority of different devices) until all packets have been read. Then it will re-enable interrupts. What you get is interrupt behaviour for rare traffic and polling behaviour for frequent traffic with the added benefit that by controlling the amount of work performed in each cycle you can dictate to the system how much resources to allocate to each device and have the drops (if any) occur as early as possible in the network stack - in fact before reaching the network stack at all. I could probably give a whole lecture about this subject. If anyone is interesting to hear about this subject let me know and we'll make it a Telux lecture or soemthing. Cheers, Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
[Commercial] Kgdb Pro support package now available in Israel
[ This is a commercial offering of a support package for a FOSS Linux kernel debugger. I *think* this is on topic here, but if not I apologize in advance. let me know and promise not to spam you further. ] Kgdb Pro is a maintainted, tested and easy to deploy package of the Linux source level kernel debugger Kgdb by LinSysSoft (Amit Kale's company, if you know the name). Codefidence now distributes and supports it in Israel. Details at: http://www.codefidence.com/kgdbpro.html Thanks you for taking the time to read this, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Go-Linux 2005
Michael Vasiliev wrote: On Wednesday October 19 2005 19:41, Muli Ben-Yehuda wrote: On Wed, Oct 19, 2005 at 07:23:31PM +0200, Lior Kaplan wrote: Hi, I saw at http://www.pc.co.il/linux2005 that there is a logo for Israel Linux User Group (IGLU ?!). No, ILUG, Israeli Linux User Group. No relation to IGLU or Hamakor or anything else that is community oriented. I have to admit that I found that confusing explanation quite...confusing. It's very simple. People and Computers have created together with Sun a group called JUG (or something similar) for the Java Users Group. JUG is basically an israeli Java users club sponsered promoted and driven by Sun and other commercial vendors. PC then tried to pull the same stunt with Linux, with Oracle in the lead. ILUG is basically a group of marketing people from various commercial entities and some CTOs and the like of big organizations using Linux. The leader of this group is a certain marketing executive from Oracle which is a nice eniough person but wouldn't recognize a Linux if one bit it in the behind. As far as I could tell, this group only exists for the single purpose of participating in the People and Computer events sponsered by the vendors. They don't seem to have any other activity, web site or forum/mailing list except some sporadic mailing between the organization once a year to organize the event. In the past attempts of helping this group gain some substance lead by Shachar Shemesh and Eli Marmor by organizing lectures and such has died out for lack of interest, depite their best efforts. Gilad. -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Forcing the use of specific library directories during link and compile
Shachar Shemesh wrote: Does anyone have any idea how to make sure the standard search paths are simply not looked at? Recompile the GCC you are using from scratch and give it a fixed library/include patgh which is different from the system one. The only problem you'll have left is with package that explictly do a -I path or a -L path with something hard coded or deduced the wrong way. There are actually very few that hard code path in this way (I know of a single package) but there are several packages that use a conf-bin idoticity: the installation of the library leaves a binary that can be called with options and emits the path to the include/ libs. Then when you try to false cross compile the configure runs the binary and gets path for your normal system. But really the best solution here is not to invent the wheel - google for ptxdist. It is highly recomended for the task of micro distro making. Cross compiled or otherwise. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.212.2026643 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Dell Latitude 110L or IBM R50e?
For linux of course... Price is the same (30$ difference). The Dell comes with 3 years warrenty, the IBM with 1. Dell has a working free driver for the modem, IBM doesn't but I don't use the modem. My gut feeling says to go with the IBM. Anyone has any Linux (or otherwise) experience to share with either of them? Thanks, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: disk defragmentation
Moshe Akirav wrote: Hi all, I was wondering why don't I need to defragment my disk, Everyone says it is not necessary, Does anyone know the reason ? There is no reason. Fragmentation can and will occur. It's true that due to file system design (I'm specifically relating to ext3/2 here) fragmentation may occur more slowly then with other systems/file systems, but it will occur. And no, there is no reliable tool to defragment an ext2/3 partition. Don't know about other file systems though. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: disk defragmentation
Moshe Akirav wrote: Gilad Ben-Yossef wrote: Moshe Akirav wrote: Hi all, I was wondering why don't I need to defragment my disk, Everyone says it is not necessary, Does anyone know the reason ? There is no reason. Fragmentation can and will occur. It's true that due to file system design (I'm specifically relating to ext3/2 here) fragmentation may occur more slowly then with other systems/file systems, but it will occur. And no, there is no reliable tool to defragment an ext2/3 partition. Don't know about other file systems though. Gilad Thanks for the reply, Can you point on a document describing ext3/2 design which I can understand from, why fragmentation accur more slowly on these type on FS. The short answer: Ext2fs also contains many allocation optimizations. Block groups are used to cluster together related inodes and data: the kernel code always tries to allocate data blocks for a file in the same group as its inode. This is intended to reduce the disk head seeks made when the kernel reads an inode and its data blocks. When writing data to a file, Ext2fs preallocates up to 8 adjacent blocks when allocating a new block. Preallocation hit rates are around 75% even on very full filesystems. This preallocation achieves good write performances under heavy load. It also allows contiguous blocks to be allocated to files, thus it speeds up the future sequential reads. These two allocation optimizations produce a very good locality of: * related files through block groups * related blocks through the 8 bits clustering of block allocations. The long answer: Design and Implementation of the Second Extended Filesystem http://e2fsprogs.sourceforge.net/ext2intro.html -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: OT: how do i get an ulta-small PC?
guy keren wrote: does anyone know if there are such things for sale, either in israel or via the internet? i imagine such a thing might cost several hundreads of dollars, and i'd rather buy something pre-built, then build my own - since it should be able to sustain movement shocks (after all, i'll be taking it in a car and drive with it all around, so i don't want to have loose wires due to a bump on the road, etc). You can buy an evalutation kit from Compulab in Haifa of their 686CORE. It fit your needs, except that you'd have to find some small case for it. AFAIK it costs around 1500 US$. There is also the issue of their not quite kosher use of binary only drivers for the flash cards compiled into the kernel supplied with the machine. Other then that it does what you want. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: devfs - gone
Shachar Shemesh wrote: Hi all, The new kernel (2.6.13) has removed support for devfs. While I certainly understand why devfs is not on by default, it seems that removing it altogether is a bit harsh. Devfs has been a great way to quickly (say - inside an initrd) get a clear picture of what hardware is available, as well as get the relevant devices required to run said hardware. With devfs gone, installers, some embedded configurations, and possibly others, may find their time much harder. If you used devfs to find out what hardware you're running on, then you should consider using sysfs instead. It's an official part of 2.6 anyway. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: devfs - gone
Shachar Shemesh wrote: Gilad Ben-Yossef wrote: If you used devfs to find out what hardware you're running on, then you should consider using sysfs instead. It's an official part of 2.6 anyway. Yes, I know. It's just that sysfs's way of giving you the major/minor numbers, and leaving you with the question of ok, so which DEVICE is it? was annoying. I think I'll just mount a ramdisk on /dev and create the nodes in real time (i.e. - oh, so sysfs says that 3,0 is the CD? mknod /dev/cdrom b 3 0). Congratulations! You just re-invented udev (sort of). Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Kernel mode disk I/O. Take 2. (Was: File I/O within kernel threads?)
Gilboa Davara wrote: Hello all, Couple of weeks ago I started a thread called File I/O within kernel threads? which ended with a call for numbers. (Read: Is I/O within kernel performance increase justifies the development overhead?) Last week I took my user-land simulator (that takes a file, reads it, and creates and sends two raw multiplexed streams out of it); and converted it into a kernel module. The performance increase was nothing short of staggering: From ~1100Mbps to ~1700Mbps. (Which is fairly close to the theoretical limit of 2 GbE NICs.) While both versions could be further optimized, it's clear that the kernel version has an very impressive performance lead. Considering these numbers, I opt to choose a Kernel mode solution. Not so fast, cowboy :-) How did your user space simulator sent the files? If you've just did a read/write pair you should be using Linux's sendfile() system call (assuming no proccessing is required on the content of the file) and I expect you'll find you don't need a kernel moduel after all. My question is simple: While it's fairly easy to do normal buffered I/O from kernel mode (using flip_open/close and -read/-write methods) for the life of me I cannot seem to find how to do O_DIRECT I/O from kernel mode. More-ever, as I need a single huge file on a dedicated computer, I don't really need the FS/VFA/pagecache over-head, AFAIC I rather do raw I/O directly on-to the target block device. Can anyone point me into the right direction? I'm currently trying to drill down the kernel source but it's a daunting task, to say the least. Again, I fully understand the implications of my request; I understand that I'm bending Linux sideways, etc. I don't think you bend Linux sideways. I think you're trying to bend the situation to be such that you'll get a good excuse to show managment to let you write kernel code ;-) Nothing wrong about that of course, at least not in my book, but while you can fool management, don't try to fool yourself. Anyway, if you don't know how to use O_DIRECT from inside the kernel, why don't you take a look at the part in teh kernel that implments O_DIRECT from user space and do the same thing? Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: PC-to-phone VoIP
Shlomo Solomon wrote: What makes Skype evil? The fact that they strongly oppose interoperatibility, by encrypting their protocol and making it extremely hard to reverse-engineer it. It's the same story as it is with any Maybe that's about to change. Here's some fresh news posted only a couple of hours ago: Introducing SkypeNet API Wouldn't it be cool if your friends using other IM applications could contact you on Skype? We want Skype to be the worlds largest open IM system on the Someone is scared shitless from Google Talk. You just have to love Google :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: PC-to-phone VoIP
Geoffrey S. Mendelson wrote: On Thu, Aug 25, 2005 at 11:22:47AM +0300, Gilad Ben-Yossef wrote: You just have to love Google :-) No, you don't. They are becoming the next Microsoft. While they use linux, how many of their wonderful products are open source? None, they are a propritery software company, but they have a stated policy of not tying their users to propritery protocols and they seem to abide by it very well and I find it refreshing compared to the practices of other comapnies, Microsoft included. So they are not saints, but they are not devils either. They invested $100,000,000 in BPL, which will destroy all long range radio communictions including shortwave radio, ham radio, VHF lowband television and all of the IDF communitcaions where you see the long antennas. I have no idea what BPL is so I can't form an opinion. A friend of mine from the U.S. pointed out that Google Earth uses the Arabic names for places in Israel that have both Hebrew and Arabic names. This has caused quite a stir in the Jewish community in Silicon Valley. Either choice they would have made they would have made someone angry. They are planing on blanketing the world with free wifi, putting out of business the small business that are doing the same thing, and preventing places that offer wifi such as coffee houses from turning it off when they want customers to come and buy drinks, not use free wifi and buy nothing or stay for hours keeping other customers out. I don't see any problem with blanketing the world with free wifi. Actually, I'm all in favour of them doing that. The New York times has an article about how their hiring practices have destoyed the job market in that area for startups. Personaly I think that's good, maybe more people will invest in startups here instead. Once again, I appluade their dedication to excellence and hope they will raise the bar for other companies hiring practices even further. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: It's about time: a delay of 1 microsec in a user space
Rafi Gordon wrote: Hello, Thanks. You said If you really meant 1 microsecond (as opposed to 1 millisecond) Yes ,I mean it. In short: it's for a video-audio app; this app can be termed soft (or semi) real time; The terminology is not important at all. The fact is that there are scenarios when the timing is critical; I mean a packet that arrived at a certain time should not be sent immediately but after some sleep interval; this if for keeping a specified rate , which is important for a good quality. So sometimes a microsec sleep is needed, sometimes 5 or more. (Maybe it would have been better to phrase my question as sleeping in a resolution of microsecond instead of 1 microsec). Since you mention packets and I assume you mean network I seriously doubt that your time limits are anyone near the microsecond ranges. I might be wrong though (if you're talking about some RF packet and not Ethernet) , and if so then be aware writing this as a user space application in Linux is a no go: Average premption latency on a x86 platform is somewhere near 10 microseconds. The worst case interrupt latency of same is 2 microseconds but the worst case is ~40 microseconds. This is *after* all the relevant preemtive kernel patches, lock break and so on has beebn appllied. [1] If you cannot miss a bit and you're working in sub 5 micro area and you wish to use Linux you need to use some sort of an interrupt pipline - RTAI/Adeos is my personal recomendation. [1] A little old but gives a good view, espeically since 2.6 is worse, niot better then 2.4 + patches in this regard: http://www.linuxforum.dk/2002/program/udstiller/montavista/slides/MontaVista_Real_Time_Linux_CPH_March2.pdf may I ask ehat is the recommnded way to write a a busy wait loop? If I am not wrong there is no noop command in C ; is there a noop is asm to achieve this ? or simply a loop which just do dummy operation like incrementing some integer ? arch/i386/lib/delay.c in your nearest kernel 2.6 source. Cheers, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Oracle GPL's OCFS
Amos Shapira wrote: Oracle announced in LinuxWorld that it has GPL'ed their Oracle Cluster File System 2. http://lxer.com/module/newswire/view/41239/index.html Can someone in the know how it compare to other technologies available for Linux block device clustering/sharing today? What functionality does it add that isn't provided by the others? No clue. And whether it can be expected to be integrated in the Liux kernel trunk? In the link you provided Andrew Morton, a.k.a the right hand of god, is quoted saying this will be integrated into a 2.6.14 release. -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Transfering PPP over a serial link, oh my...
Hello list, I hope there is someone here twisted enough to know the answer to my question :-) Suppose I have a bunch of PPP packets and suppose I want to push those unto a serial line (tty device) which is connected to a modem. The tty and modem have already been configured to the correct baud rate, dialed, got a connect string and all. Is there any magic IOCTL which I need to user unto the tty device to convince it to transport these PPP packets over the tty such that the host in the other side will actually care to answer. The magic words seems to be PPP line decpline and such but I'm having a hard time googling and understandng what dark magic pppd does. Your help is truely appricated. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Transfering PPP over a serial link, oh my...
Nadav Har'El wrote: On Wed, Aug 03, 2005, Gilad Ben-Yossef wrote about Transfering PPP over a serial link, oh my...: I hope there is someone here twisted enough to know the answer to my question :-) I'm sorry that I don't have any pppd-hacking experience, so my answer will be mostly guessing and quite likely, worthless. Well, it seem you do. Your hint solved the problem. Details in a coming $(UNIVERSITY city)ux lecture... ;-) I am guessing that you need to switch the device driver into the raw line discipline, meaning that no interpretation is done on the bytes you send. In particular, you don't want backspaces and such to do anything special, newlines converted to carriage-return-linefeeds, or anything of this sort. So consider doing the equivalent of stty raw -echo. stty raw line 3 was what needed in the end. I am forever in you debt. Gilad. -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Transfering PPP over a serial link, oh my...
Oron Peled wrote: It seems you tried Nadav's (semi-correct) answer about raw line discipline before I sent my answer... stty raw line 3 was what needed in the end. I know. I looked at the tty.h file to find out that N_PPP is 3 to begin with Since I hate to be wrong, I dug into the kernel sources after all, and asm/termios.h: ... /* line disciplines */ #define N_TTY 0 #define N_SLIP 1 #define N_MOUSE 2 #define N_PPP 3 #define N_STRIP 4 #define N_AX25 5 #define N_X25 6 /* X.25 async */ ... Now look who's 3... you just used stty(1) to send the relevant ioctl and selected the ppp (not raw) line discipline. The real puzzle is: - How all the PPP state behaved? I transported ALL the PPP packets, including the initial LCP and ICP headers intact. I never said I started shuffeling packets to the other side from the middle of the connection. - Where did cooked/raw disappeared (they must be controled via some other ioctl while in the N_TTY discipline, but I'm tired for searching the sources for now...) It didn't dispear at all - it's an orthogonal setting and in fact I needed both stty line 3 and stty raw to get it working, so you were BOTH right - I just didn't see your answer in time so I figuted it out myself :-) Thanks to all who helped. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: NAPI performance ; tg3 and broadcom driver
My question is: 1) does anybody have an experience with this technique ? Yes. what is the threshold (of k packets for a second) from which it became efficinet to use NAPI over usual non-NAPI solutions ? (I am talking about Xeon processor ~2.4 Ghz , but also data on other prcoessors can help) That's the wrong question. The right question is - what is the threshhold of packet per seconds (or interrupt per seconds) that I should set NAPI to go to polling mode? NAPI is adaptive, that's the entire idea - you can set the threshhold yourself. 2) Specically regarding NAPI and Broadcom cards: On many distrubutins, the Broadcom BCM5700 family of Network Drivers uses the tg3 driver. ( Tigon3 ethernet driver, in /linux/drivers/net/tg3.c) On broadcom website there is a driver for linux that they wrote for Linux. http://www.broadcom.com/drivers/downloaddrivers.php (the BCM57xx Drivers). Is there any advantage of using the Broadcom drivers to using the tg3 driver ? (and does the kernel version - 2.4 or 2.6 - has any importance in this respect?) Basically, Broadcam wrote the BCM57xx driver first and it... ehm... did not live up to kernel maintainers taste, so they hacked it up until it looked like something that they are willing to accept into the tree, whereupon it got renamed as tg3. The only problem is that the Broadcom people update their BCM version with newer versions of the chipset form time to time. Basically, my rule to use either one is this - if your hardware works with the tg3 drivers use that. Otherwise, use the BCM one. Cheers, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: NAPI performance ; tg3 and broadcom driver
Marc A. Volovic wrote: Quoth Rami Rosen: Best thing will probably test also with a new kernel, with both drivers ,tg3.c and the bcm5700, in all scenarios. Admittedly, we have not played with 2.4.x with the tg3 driver, but all tests we did on 2.6.x (x 10) indicate that the tg3 fails to hold its own when hammered by many interrupts. Did you enable NAPI? Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: File I/O within kernel threads?
Gilboa Davara wrote: Muli, I well aware of the controversy surrounding FS access from kernel modules and I accept, that in general, kernel modules should be using the FS for storage. However, in essence, I'm using the *wrong* tool for the right job: I shouldn't be using Linux on a i386/x86-64 in the first place; I should be using a network OS with a network chip. However, Linux/x86 uses (relatively) cheap hardware and has massive driver support and a kernel modules is (again, relatively) easy to write and modify. Oh... and Linux is easy to bend :) After doing some contemplating I decided that I don't really need access to an FS; or actually, all I need a is huge cyclic buffer with fast sequential R/W and force-able sync. If anything the VFS layer will only slow me down. I wonder if the raw character code is code enough to be yanked out and used for this project? Now you're re-inventing RelayFs :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: restricting memory mapping
Boris Zingerman wrote: We have some driver that performs zero-copy DMA to userspace allocated buffers. The problem is that the device cannot perform DMA to RAM pages with physical addresses above 4G ( this is heavily memory equipped computer) My question is it somehow possible to restrict memory mapping for shared memory segments ( which are used as DMA destination) to be physically under 4G boundary ? The short answer, use either bounce buffer (slow) or pci_set_dma_map. The long answer: http://www.linux.com/howtos/IO-Perf-HOWTO/overview.shtml Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: File I/O within kernel threads?
Gilboa Davara wrote: Umm... As far as I can see, realyfs uses memory buffer for storage, which is major no-no in my case: At 50-200MB/sec I'll deplete the system RAM within minutes (even on AMD64) and as far as I can see, there's no obvious way to commit the buffers into static storage. Current me if I'm wrong, but relayfs was designed for fast Kernel/User transactions and not Kernel/Kernel transactions, right? I'm looking for 50% match open solution that can be bent to suite my rather weird requirements. As the saying goes: Writing you own FS really cuts down on your Doom3 time! :-) Gilboa, I'm sorry if I seound harsh, but I don't think you udnerstand your own needs. Let's try to analyze them together - you need to store large amount of data from the network for proccessing by a further entity. What that data is exactly doesn't matter, but we will note that you might need to do non trivial handling of the data (encryption). There obviously is no question of delay involved - you're talking about writing to disk, using a FIFO to buffer information etc so obviously there is no serious real time contraints on the data. What you really care about therefore is - *througput*, NOT delay. For this, crossing the kernel/userspace and related context switch is NOT a problem! The solution boils down to this: 1. You need some way to store information temporary in the kernel (because it's getting first to the kernel). 2. You then need to handle it (assembly/encryption) and write it to disk. 3. After the data is on disk, you have a different proccess that sends it to some thrid party, but we really don't care about that. So long as you didn't perform 2, you want the information to wait in a temporary buffer (I first wrote bugger here, which is rather funny ;-) until you're done with it so store the data in a temporary in kernel buffer (or bugger, if you prefer ;-), have a user space proccess read it from there, proccess it, encrypt it, write it to file. Have yet another proccess send it to the thrid party (possibly using the Linux sendfile syscall for efficincy). Now you can write that Temporary buffer layer in kernel that can easily be read by user space: your self but you'll be just wasting DOOM3 cyclesw. This is exactly RelayFS. The rest are simple user space programs (and if you want to see an example on how to use it see Karim's most excellent LTT tool that can use RelayFS: http://www.opersys.com/LTT). Hope this helps, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: File I/O within kernel threads?
Gilboa Davara wrote: Let's try to analyze them together - you need to store large amount of data from the network for proccessing by a further entity. What that data is exactly doesn't matter, but we will note that you might need to do non trivial handling of the data (encryption). I should add the encryption optional, depending on load and source. (And more important, how fanatical is the client) (There's no way in hell, I'll be able to process and encrypt two OC48 links in real time...) I believe you can, but this WILL require much more explanation... So long as you didn't perform 2, you want the information to wait in a temporary buffer (I first wrote bugger here, which is rather funny ;-) until you're done with it so store the data in a temporary in kernel buffer (or bugger, if you prefer ;-), have a user space proccess read it from there, proccess it, encrypt it, write it to file. Have yet another proccess send it to the thrid party (possibly using the Linux sendfile syscall for efficincy). Now you can write that Temporary buffer layer in kernel that can easily be read by user space: your self but you'll be just wasting DOOM3 cyclesw. This is exactly RelayFS. The rest are simple user space programs (and if you want to see an example on how to use it see Karim's most excellent LTT tool that can use RelayFS: http://www.opersys.com/LTT). Interesting... that might work. Let me first point out that once the cells/frames have been processed, I don't care much for timing. (Which bodes well on your solution). However, I'm very tight on CPU and memory bandwidth. (Even a dual Opteron machine with two memory banks tends to suffocate at a certain point.) No matter what I do I just can't afford to add memcpy's to my system. Understandable attitude, but it may be wrong. Take a look at this paper from last year OLS for example: http://www.linuxsymposium.org/proceedings/reprints/Reprint-Ronciak-OLS2004.pdf These guys from Intel thought having a zero copy receive path for network packets where the card will DMA stright into the user space program buffer will be a big win due to saving a memcpy. They implemented and tested. Results? perfomace was *worse* from losing the extra copy, not better. It turned out the extra copy actually pre-loaded the cache and gained more then it costs. Does this fit your scenario? I have no idea. But there is a lesson here: don't assume anything. Build a quick pilot and measure. You may very well find out that your bottle necks are completly different areas (for example - are your network drivers interrupt driven? you might very well find that your system gets into live lock on interrupts before any issues stemming from memcpy of data). As the man said, pre optimization ios the root of all evil. Here's how I see it: Kernel 1: Device - SKB - Reassembly - Disk. (I can even save the third memcpy [Reassembly - Disk] I go rewrite the world under me) User: Device - SKB - Reassembly ( - ?) Relayfs - User: write(2) - Kernel: sys_write (copy_from_user) - Disk. As I already said, you can use Linux sendfile() to avoid the last copy if you're not messing with the data after it reached the disk. Wont help the decryption case, unless you also happen to have a a hardware encryption engine, which is a good idea anyways. Oh... Thanks for the help. I appreciate it. Thanks for the interesting subject :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: File I/O within kernel threads?
Gilboa Davara wrote: I need to file I/O operations from within kthread context (Multiple fast network streams, no time to push it upwards into user-space) As far as I remember, flip_open/close and -read/-write fs methods must be called from a process context. While I can move the I/O task itself using schedule_task to be executed under keventd, I fear that the long I/O waits will starve the other pending tasks. Did I miss anything? Any other ideas? A kernel thread is indeed proccess context, but since you brought it up already and assuming we're talking 2.6, the correct way to do is use work queues (and hence schedule_work) instead of schedule_task. This will give you the option to start your own work queue (and hence your own kernel thread) thus not having to worry about starving other tasks. And while we're on the subject, you might want to take a look into using a work queue instead of a kernel thread all together - it's a very nice interface to do these sort of things. To learn more about Work queues see here (page 23): http://lwn.net/images/pdf/LDD3/ch07.pdf On another different and related note, if you're consider doing IO from kernel sapce (which is a terribly bad as you've pointed out already but I wonder if you know just *how* difficult to debug and racy it is) for fear of perfomance, a saner approach that may or may not fit your bill is using RelayFS: http://sourcefrog.net/weblog/software/linux-kernel/relayfs.html Hope this helps, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: registration August Penguin 2005
Rafi Gordon wrote: Hello, I had heard about the August Penguin 2005 in 4/8/05 and want to attend it (for the first time). What is the procedure for it? Is it enough to come in that day and buy a ticket ; or is there some earlier registration on the web/email/etc No need to pre-register. You jsut buy a ticket on the spot for 30 NIS (Hamakor members and friends enter free of charge). I did not saw details about it in August Penguin 2005 web site (http://august.penguin.org.il) The site will be updated shortly. Right, Ido? :-) Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: cloop vs. compressed filesystems
Shachar Shemesh wrote: Hi all, I have a wonder I would like to raise, in the hope that someone has an insight. Knoppix invented a special loopback device called cloop. This is a compressed loopback device, which means that you can connect an standard file system (say, ISO-9660 cdrom), and it will be compressed and then decompressed on access. When you load a Knoppix filesystem, the entire filesystem is placed on the CD in a cloop/iso format, and decompressed on the fly upon demand. Linux, however, also supports compressed file systems. The most known of those is cramfs, but there are also others (better), such as squashfs. The questions is - why is cloop better? What's the difference between compressing at the block level and compressing at the filesystem level. Is there any CD related performance difference that causes one to be better than the other for certain uses? For the sake of this discussion I'm assuming that both cases offer the same functionality. I.e. - it's obvious that with cloop I can use any filesystem I want, but I'm assuming that if I didn't use cramfs, I would use ISO-9660, which offers very similar functionality. I'm also assuming that I am not bothered by cramfs's limitations of 8 bit UIDs (or am using squashfs, which doesn't suffer from those). There are two main differences: First, cloop was invented out of tree by some bloke who's name escapes me. it went none maintained for a long time untill it was picked up by Knopper who I believe is the current maintainer. cramfs, on the other hand, was born as a in tree project written by none other then Linus himslef. I don't think he knew, or cared, that a different project does soemthing that amount to the same thing when he did. Because crmafs is an in tree project and well maintained it is more in use IMHO, at least in the embeddded world, whereas cloop is more a live cd scene trend. Note that the above is a social difference, not a technical one. it might have been initally caused by cramfs inability to handle CDROM sized storage, but now that we have squashfs that is no longer an issue. The technological difference is, as you stated, that one works on the block device level and the other on the file system level. If you take into account, as you do, that you work with a read only pre-fabricated (that is you use a tool such as mkisofs or mkcrmafs to create the image) the only difference is that you seem to go through a lot more motions and maybe even copies with ISO9660 over cloop then cramfs to get what is basically the same thing but in the case of a live cd this is rather moot - the bottle neck is in the hardware, anyway. For embedded flash based system the case is different. Hope this helps, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: cloop vs. compressed filesystems
Shachar Shemesh wrote: Gilad Ben-Yossef wrote: Hope this helps, Gilad It does, but truthfully, http://tree.celinuxforum.org/CelfPubWiki/SquashFsComparisons helps even more. It shows the numbers but does not tell the story :-) To summarize: CramFS, and more particularly, SquashFS, have lower overheads in terms of CPU usage, and are thus better for cases where performance is CPU bound. It is also the cleaner solution. cloop is better at random access, and is therefor better where CPU performance is not the bottleneck, or when the access metric is highly random in nature. How did you come about to this conclusion? Both random access tests quoted at that page show squashfs as faster then cloop, unless I'm confused again. Another interesting difference is that there is no mode for running cloop as a non-module. I'm sure one can be added, but it requires the file name to bind to at load time, so I'm not sure that's really feasible in cases, such as a live CD, where you don't have the file at boot time. In the embedded field, where there is not much use for a modular kernel, that is a great minus. As I said, cloop is an out of tree solution. Anyway, it's very easy to solve without a hard coded file name via an ioctl or proc file interface if one was so inclined - exactly like the loop device and losetup utility do it, although most users are not aware of this since mount calls losetup in the background... :-) Cheers, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Off topic - open vs. closed sources Re: compiling expect sources or easy to use c++ engine?
Nzer Zaidenberg wrote: but its worth noting that if you will act as you act (espcially at your age) you will be treated as a less then progessional person. more so, you give other open source advocates like me bad reputation Nzer, I stayed out of the argument up till now, because I really don't like trolls, but I have to tell you that reading what and how you both wrote sure makes one of you come out as a less then progessional person (I do assume you meant proffesional, right?) but it sure ain't Oron. You can screw your customer as much as you (and they) want, and it really isn't our business untill you made it one by coming to this public mailing list and asking for help to screw your customers. Don't like what we have to say? though. It's not about open source and software, it's about being a human being. PS. Have you considered the fact that this is a public and archived mailing list? anyone who is about to do business with your company and has two bits for a brain will read this little exchange. I doubt that it bode well for you and your reputation. Have a nice life, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: good news: evolution + exchange server
Ira Abramov wrote: Now that Daniel Robins (ex- Gentoo founding father and Ex- Israeli) started working for MS on their FOSS initiatives (oh yes, they exist), we might even see more of that going on. I wonder how many people will think I've gone mad, but here goes anyway: I believe that in a few years Microsoft will be a major participant in the field of Open Source and Free Software. Just like Sun/IBM/SGI/Oracle it will not mean it will dump all it's propritery technoogy to SourceForge and sign Stallman as the new CEO (unless of course Apple will build a Mac that runs on Intel Chips... oh wait!) but they will be a corporate member of the comuunity just like all the other corporate behemoth. You can laugh, you can cry, but remember - you heared from me first :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: slashdot: m$ praises unix and linux shell clis and prepares to emulate them
Geoffrey S. Mendelson wrote: On Fri, Jun 10, 2005 at 10:44:50AM +0300, Gilad Ben-Yossef wrote: There is only one response to their statement, which is to remind everyone what happend the LAST time M$ tried to do a Un*x like shell: Obviously you have never used Micrsoft's Services for UNIX (free download). I am very very happy to admit that I never did as for several years now I avoid working in that envorinment as much as possible . When I have to usually in order to prepare a Linux corss tool chain for an embedded project for some client, it's done via an SSH to a Cygwin Bash session. It includes a very nice CLI with KSH and C shell, bash is availble from a third party. The thrid party stuff used to be free, now there is $20 download fee. Obviously you you didn't read the link I sent. It describes the origins of exactly that shell that you describe, what M$ said about said shell when it was first announced and... well, without giving out too much of the punch line - demonstrates what happens when M$ marketing Reality Distortion Field hits a massive object :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Help with Asterisk for $$
Aharon Schkolnik wrote: Hi. We have asterisk up and running, but have a strange problem - it has 30 seconds of post-dial-delay (time between dialing and first ring). We are pretty sure it is an asterisk issue, and need someone to help us debug it. We are prepared to pay. Anyone capable and interested ? I can help you for money (check my .sig for the gory details), but since it sounds like a very simple issues to me (perhas I'm wrong) let me try and guess that in your dial plan (extension.conf context) you have defined a wildcard extension with no bound of digits, something similar to this: exten = _.,1,Dial(Zap/${EXTEN}) AND you never set DigitTimeout(seconds) to a sane value (or Set(TIMEOUT(digit)=timeout) in CVS HEAD). So, if this is the case, either bound the number of digits the wild card will accept like so: exten = _NX,1,Dial(Zap/${EXTEN}) or exten = _.,1,DigitTimeout(3) If it's not the case give me a call. Cheers, Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: adding software RAID on the fly to SecurePlatform
Shachar Shemesh wrote: Ira Abramov wrote: is there an official procedure for this or should one improvise? What raid level do you want to add? RAID1. there are two standalone disks in there, they want to add another to mirror one of the existing ones. SCSI? Do NOT use raid-1 on two IDE disks that reside on the same bus, as performance is going to suck big time. I don't know how well it will work with two disks on separate buses, but another disk on one of them. In theory, SecurePlatform is no different than any other Linux. It's just a standard system with a couple of extra software pieces, a couple of which in the kernel. plus there is no compiler, no kernel source readily available, old libraries since it's based on RHL 7.x, and quite a few other changes. That's not a problem. I believe you will find a kernel source on the CD, but if not, just email Check Point and ask for one. They were responsive to GPL compliance requests in the past, and I doubt that has changed. As for the rest, just install another SPLAT on a spare machine, and compile an RPM there. oh, and I didn't read any fine print, but I bet adding GCC on my own to the machine voids the warrenty or something. THEY PROVIDE WARRANTY? Wow, that's excellent news. it's supported as an appliance I am guessing, so I would not want to just go around CP's protections (easy at it may be with a root password) but instead use what they recommend to do. I don't think so. It's an RPM based distro. I'm also not aware that it's supported at all. One more thing - recompiling the (non-existant) kernel source may also violate the warrenty, so that's why I was wondering if the official supported kernel supported MD, or the older method of /etc/raidtab (whatever it was called) or none at all. Like I said, I'm not sure there is a warranty. Even if there is, as you and I both know that it won't break anything, just resort to non-raided system with the original kernel whenever you need support. The general procedure for setting up md on the fly is to set up the raid with one failed disk (being the disk that currently holds the data), copy the data to the raid, and then attaching the last disk to the raid. I've only tried it on raid-1 before, but there is no reason for it not to work on any other raid level. no, except the client wants to do it on the fly with a living system, at minimum downtime. sounds a bit of a problem to me. the howtos indeed tell you to set up MD in failed mode, move over the stuff to it, reboot the system from the new MD, format the old disk (or if you are sane, get a virgin disk) and reattach the failed pats of the raid. they don't discuss doing it in-place by just adding a disk and synching to it. I believe that if you tried to phrase to yourself the phases of what you describe, you will find out that the request makes no sense. The root FS is mounted over /dev/hda1, and you want to move it to /dev/md0. How can you do that without unmounting /dev/hda1? Anyways, that's just one reboot worth of down time. I'm not aware of any way of replacing the root FS of a running system without a reboot. Actually it's rather easy if you have a new enough kernel version (and I guess you don't because you use the kernel version that is supplied with SPLAT) - cp -a /mnt/hda1 /mnt/md0 cd /mnt/md0 pivot_root . mnt/old-root exec chroot . sh dev/console dev/console 21 umount /mnt/old-root man pivot_root for the details. can pivot root, I guess, but that won't help with unmounting the old files, as file in use will prevent you from unmounting the old file system anyways. init 1 init 3 ? But I agree that it is an excersize in futility. You SHOULD reboot to make sure the system comes up right anyhow. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Lightweight graphical environment for Linux
Shachar Shemesh wrote: Hi all, I'm looking for a light weight graphical environment to write a Linux program in. I'm currently aware of X (free, org), and that they have a light weight version (though I don't remember what or where, links would be welcome), and of svgalib. Is there anything besides those two? Also, is there anything that will give me actual widgets? I don't need anything fancy. Simple buttons and a progress bar would more than do. I'll just define what I mean by light. I need it to require close to zero configuration, and to load quickly. The dotted screen X puts up when it loads is something I would rather avoid. The environment is a semi embedded situation. It's not embedded platform, but I have practical reasons I would like to keep memory and space footprints to a minimum. I recommend you look into the combination of TinyX (a.k.a KDrive) and fltk. Compiled with uCLibc they give you a full graphical envrionment in ~10MB of space with a full widget set and zero configuration. There is also embedded QT, but I don't like it as much. And send dash to Shuky :-) Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Accessing eci b-focus 312+ ADSL Router/Modem
Peter wrote: On Sat, 4 Jun 2005, Matan Ziv-Av wrote: No it is not. There are many examples of kernel modules that provide support for modems, network drivers and other devices, which are NOT open source and NOT 'aggregations'. If you don't want to make your 802.11g wireless card work If the modules are distributed apart from the kernel itself (as is usually the case for the modules you speak of), then Linus allows this. Some other I.o.w., if the modules come on disk and are loaded at run time by a script it's ok. You know what ? That's exactly what they are doing. Unbelievable. What is 'distributed' in an embedded context ? An appliance is a device. So, if you build an embedded device that displays text (an eBook of a sort) that has built in the text of Gone with the wind (a copyrighted work) you are exempt from receiving a license to distribute Gone with the Wind from the holder of the copyright because it's an embedded system? Funny, must've missed that in my last read through copyright law... It contains everything that needs to be in it to make it work. Even the batteries sometimes. And Linus and the authors of the libraries and of the utilities (and of busybox) know this and they are ok with this, as far as an open source solution does not exist. Can you show me a single place where Linus says so? Because I can show you several places where Linus publicly states exaqctly the opposite... (http://kerneltrap.org/node/1735) More to the point, both the coyright holders of kernel code and of busybox have sued and won a very similar case of voilations of their rights in regard to same code. The LGPL was designed specifically to allow user applications to be linked against it for this purpose, and most libraries used in embedded are under LGPL. Busybox and the Linux kernel are not covered by the LGPL. http://www.mvista.com/previewkit/ This is a commercial firm that provides advanced linux development packages. Monta Vista provides source with every binary of a GPL program that they provide. What has that got to do with ECI? Hello ?! We are talking about an ECI router box that runs a Monta Vista Linux kernel and busybox, yes ? Don't lose focus on the original topic, I will not follow you in philosophical pursuits. The only problem is that regardless to Monte Vista or ECI, the person that did distribute the code in violation of the license terms is Bezeq. I'm pretty sure MonteVista did not violate the terms. I can only assume that ECI did not violate the terms (but I don't know anything about it because they did not distribute anything to me) but I *know* that Bezeq violated the terms of the license agreement. What I can or can not find on the internet is irrelevant to the fact that ECI have to do specific things if they want to distributes Linux or Busybox, and the other fact, that they do not do those things. This is ridiculous. What do you expect them to do, email you a cdrom ? What we accept them to do is follow the terms of the license to the letter. All it requires is that in the glossy paper manual that they ship with the unit they will add a paragraph that reads: This product contains software that is licensed under the terms of the GNU pulibc license. We hereby grant a non revocable right valid for 3 years since the shipment of this product to any third party to recieve the original source code for this work blah blah.. That's all. What do you mean what has this got to do with the ECI router ?!! Where do you think I got the information about Monta Vista on that box ?!! I ssh'd into it and typed cat /proc/version !!! You have no idea what you are talking about, you are interpreting things in ways that defy objective reality, and I think that threads like this can cause people who later find them in the archives to draw wrong conclusions as to the suitability of Linux for embedded appliances. Your kind of postings on this thread so far has been a disservice to Linux and to the spirit of the GPL in my opinion. Why do you want to shoot the messanger? all he is telling you is that these are the license terms. He happens to be right. Like it or not they are binding. Would you not prefer the people who consider to use Linux in an embedded environment to make an informed judgement? No one is claiming that Bezeq/ECI or MV did some horrible crime. We claim that due to what is probably an oversight somewhere an important paragraph has been ommited from the box manual. Because of this, Bezeq at least is now technically in violation of the copyright of the copyright holders of Busybox and the Linux kernel but I am sure that this have been done by mistake. Why are you so angry? Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Accessing eci b-focus 312+ ADSL Router/Modem
Matan Ziv-Av wrote: Try following this, it is not very hard. 1. The ECI router runs Linux. 2. ECI got a copy of Linux from Montavista. 3. Montavista gave ECI both binary and source, so they comply with the license. 4. ECI gave Matan a binary copy of Linux. Actually, it was Bezeq who gave you the binary, we don't know anything about ECI in this regard. Otherwise you are right. 5. ECI did not give matan the corresponding source, and did not offer to give such source. 6. Therefore, they did not comply with the license. 7. Therefore, they copied Linux without having a legal right to do so. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Accessing eci b-focus 312+ ADSL Router/Modem
Aharon Schkolnik wrote: Hi. Just received an eci b-focus 312+ ADSL Router/Modem. It is installed, configured, up and running. However, I would like to be able to play around with it, and don't know how to access it. The default IP for the modem is 10.0.0.138, the username Admin (which has uid 0, just like root) default password is Admin. Note that the root file system is squashfs which is a compressed read only file system used in embedded devices to save flash space, so you can change anything in the device unless you re-flash the unit. You can however control anything that can be controled via the http GUI - just go to http://10.0.0.138 (username and password are the same). YOu can even save an XML configuration file from the device, edit it and upload it back to the device using the HTTP GUI interface. It's sort of funny in a way that it runs Linux - think for example how many Israeli high ranking Microsoft execs connect to the internet using this Bezeq provided Linux box and don't even realise it... :-) Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Accessing eci b-focus 312+ ADSL Router/Modem
Matan Ziv-Av wrote: On Sat, 4 Jun 2005, Gilad Ben-Yossef wrote: It's sort of funny in a way that it runs Linux - think for example how many Israeli high ranking Microsoft execs connect to the internet using this Bezeq provided Linux box and don't even realise it... :-) I guess they would actually be happy, if they knew. The modem is distributed in violation of the GPL. There is no mention of the GPL, Linux or Busybox, even though binaries of Linux and Busybox are included. There is also no source included, and no written offer to provide the source. I agree that Bezeq is violating the GPL with their distribution of the router. Note that it is Bezeq, not ECI - you get the modem from them. I have already given the matter some thought as part of Hamamor activities. I was planning to make a background check to see if ECI supplies the GPLed parts source code somehow (if so it will still be a violation, but not a very interesting one). If they don't then this is something that we should approach very very carefully. Accusing a major corp with GPL violation (even though it's true) should not be done lightly. For example, to have a standaing in court, we actually need to find someone with code in the GPLed parts of that product - anyone here has code in the kernel/busybox/glibc? Is anyone able to locate on ECI web site any refernce to the GPL components source code? Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: chroot(2) by a user.
Alex, Alex Behar wrote: Also, Gilad, a binary will keep its permissions and attributes if copied while root and by using the -a flag to cp, consult the man pages for more details. Of coruse it does! you've missed the entire point of my post - *non-root* users don't get to keep suid bits of binaries they copy to avoid the same scenario I explained with the libraries and giving mortal users the ability to chroot circumvent that. That root can copy binaries and keep their suid bit is obvious - root can give any binary suid bit as he or she pleases, but mortal users cannot. Cheers, Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: chroot(2) by a user.
Shaul Karl wrote: Can you give more details about what you wrote? Shaul Karl wrote: On my Debian machine only root is permitted to chroot(2). What are the cons for having a regular user chroot? Are there any patches floating around to change that? If there exists any dynamically linked SUID root binary in the system What do you mean by dynamically linked SUID root binary? For me, `dynamically linked' is associated with libraries, while a binary is the same as executable and suid is only for executables. In short, what is my mistake? I mean an executable (like the ping program as I wrote), which is not statically linked - it requires libraries (such glibc, for example) to work and it's permissions have the SUID bit set and it's owner is root. (e.g. ping) which a user can contain in a directory such that the What is the difference between what you refer to and a simple cp of an suid executable to somewhere under the user home directory? The SUID bit does not survive a copy. Look: [EMAIL PROTECTED] gby]$ ls -l /bin/ping -rwsr-xr-x1 root root28628 Jan 25 2003 /bin/ping [EMAIL PROTECTED] gby]$ cp /bin/ping . [EMAIL PROTECTED] gby]$ ls -l ./ping -rwxr-xr-x1 gby gby 28628 May 8 09:48 ./ping regular directory structure puts files controlled by the user in directories that would become, for example, /lib or /usr/lib after chroot and then chroot inside it you have effectively given this user root privileges. How all this would give him root privileges? Simple, a library that an application loads is really part of its code. if you can get a SUID executable to load a library YOU wrote you can make it do everything. Just take glibc and replace sprintf with a version you wrote that does sprintf and also creates a SUID copy of /bin/bash in your home directory. Now if you can get a SUID exdcutable to load your library you are in business. Now, if you allow a user to do a chroot he or she can take a suid root owned executable, and *if* they can create around it a directory tree that has, for example, /lib, and put there their own copy of glibc with the change I mentioned, then the second they do chroot and run ping they will have a copy of a root SUID shell in their home directory (assuming ping uses sprintf somewhere, which is reasnoable). There are systems where a regular user can chroot, aren't there? Can you name them? AFAIK there are none. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: August Penguin 2005 - Not on Friday ?
Amir Binyamini wrote: Hello, Thanks for the info; It will be a bit difficult for me (and I assume others) to arrive (in the sense of missing a working day) but I assume that I wil. You don't have to miss a whole working day - you can instead be going to a technology confrence. AP VI is packed with technology and academic content and going to it should be viewed (or at the very least presented to managers ;-) as a very good way to make sure your knowledge is up to day with the current technology trends and it only cost 30 NIS (for non Hamakor members/friends) - it's a bargin! ;-) All the above is true if you are working in a technology related job, of course... Anyhow,it is good to hear that August Penguin was extended to a full day ! Thank you, we debated quite a lot about the day move, but if we want to have even more content then the previous years we simply have no choice other then to do a whole day - and that means giving up on Friday. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: August Penguin 2005 - Not on Friday ?
Lior Kaplan wrote: AP = August Penguin. VI - 4 in the Roman numbering system. Indeed. Sorry for going all FLA on you. Cheers, Gilad PS. Ok, ok... FLA is Four Letter Acronym. Sorry, I couldn't resist. :-) = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: chroot(2) by a user.
Shaul Karl wrote: On my Debian machine only root is permitted to chroot(2). What are the cons for having a regular user chroot? Are there any patches floating around to change that? If there exists any dynamically linked SUID root binary in the system (e.g. ping) which a user can contain in a directory such that the regular directory structure puts files controlled by the user in directories that would become, for example, /lib or /usr/lib after chroot and then chroot inside it you have effectively given this user root privileges. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: interruptible_sleep_on_timeout
Muli Ben-Yehuda wrote: On Mon, May 02, 2005 at 07:31:23AM -0700, Gil Paz wrote: Hi, Does any one knows of limitations of using interruptible_sleep_on_timeout in 2.4 Not sure what you mean? it's a kernel API, meant to be called in certain ways and contexts. It's also a pretty lousy API which should not be used because it's inherently racy - see e.g. http://lwn.net/Articles/68118/ It's still usefull if you want to block for a certain time period, uncondtionality. For everything else there is wait_event_interruptiple. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: interruptible_sleep_on_timeout
Muli Ben-Yehuda wrote: On Tue, May 03, 2005 at 08:49:54AM +0300, Gilad Ben-Yossef wrote: It's still usefull if you want to block for a certain time period, uncondtionality. For everything else there is wait_event_interruptiple. if you want to block unconditionally, wouldn't msleep()/ssleep()/usleep() and friends be a better idea? Of course they are - if you're using a kernel version that has them, such as 2.6, while the question was specifically for 2.4. Yes, those functions has been back-ported to 2.4, but it only happened in 2.4.29 (2.4.28-rc1 to be exact) so most of the 2.4 using world (read: distributionn kernels) don't have them yet. Cheers, Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: interruptible_sleep_on_timeout
Peter wrote: if you want to block unconditionally, wouldn't msleep()/ssleep()/usleep() and friends be a better idea? Of course they are - if you're using a kernel version that has them, such as 2.6, while the question was specifically for 2.4. Yes, those functions has been back-ported to 2.4, but it only happened in 2.4.29 (2.4.28-rc1 to be exact) so most of the 2.4 using world (read: distributionn kernels) don't have them yet. 2.4.26 has usleep and nanosleep. I am using both. 2.4.26 is one of the Debian unstable kernels (also used by Knoppix 3.4). We are talking about kernel API's, not system calls. -- Gilad Ben-Yossef [EMAIL PROTECTED] Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] IL: +972.9.8650475 ext. 201 | Fax:+972.9.8850643 US: +1.360.2275194 ext. 201 | Cel: +972.52.8260388 I am Jack's Overwritten Stack Pointer -- Hackers Club, the movie = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Asterisk
Aviram Jenik wrote: On Friday 29 April 2005 10:33, Patrick Kariuki wrote: Has anyone fully implemented asterisk with their office phone system without any glitch? Yup, we did. Same here. I know of more Israeli deployments and did some myself. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: MOSIX vs OpenMOSIX [2nd attempt]
Quoting David D [EMAIL PROTECTED]: [this is my second attempt. The first one didn't make it to the list] OK, I know how to google, and I have read several documents, but almost all of them talk about ideological (license) differences of the two systems. What I am interested in is the differences in performance, ease of use and administer. Unfortunately, I haven't seen any comperative benchmark except for this one: http://pubs.cs.uct.ac.za/archive/0193/01/SCAW.ppt (html version http://tinyurl.com/9cqe3) Can anyone on this list share his/her own experience and/or recommend on one either MOSIX or OpenMosix (religion and idiology aside) Thank you Q: If you had to choose for a production system between a project written and maintained by CS students as a thesis research problem and a fork of said project maintained by a Byte column writer at his spare time which would you choose? I don't knwo what the answer is, but sometime asking the right questions is the right thing to do. Gilad This message was sent using IMP, the Internet Messaging Program. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: MOSIX vs OpenMOSIX [2nd attempt]
David D wrote: Well your question is a little bit unfair. You can paraphrase it that way: Do you prefer a product written by a team of dedicated MSc and PhD students, under tight guidance and supervision by a world recognized professor over a product written by a I did not mean to say that dedicated MSc and PhD student, nor Prof. Amnon Barak are something to neglect, all I meant to say is that sometime, when one writes code for an academic research project and when as opposed to code written for use in a day in day out production system, the two will not look the same because of the different objective of the two developers. To put it in other ways - I think MOSIX is written by very very talented people doing great cutting edge research work. They also write unmaintainble code that looks like Perl poetry. couple of enthusiasts in their __spare__ time? If you somehow gather from my response that OpenMOSIX is any better, you got me very very wrong :-) The whole point of my rethorical question was to say that the entire MOSIX/OpenMOSIX technical difference issue is moot, not because there is no difference (there is), it is because MOSIX/OpenMOSIX is a great academic excersize - a working academic excersize, but not something I would use except for very specific and narrow taks in controled conditions. (I'm not taking sides here). These are ideological/phylosophical questions. What I am intrested in is objective benchmarks or personal experience. The only benchmark that I saw (see the I don't have any to offer. My experience as an OpenMOSIX developer for a couple of months leads me to believe the MOSIX will be faster. This a notion based on the target of the two maintainers, not on actual benchmark. Stability is a different issue that should be taken into account too. link in my original posting) is in favor of (closed) Mosix. Although I do admit that that was a very primitive benchmark. Mosix is not really closed source. AFAIK the kernel part are bone fida GPL free software and the user space part, while not FOSS according to the definition, are far from being propritery in the sense that most propritery software is - sort of like Qmail, you don't get FOSS grade freedom, but you're far from being a prisoner. Cheers, Gilad PS. In the intrest of full disclosure, you should know that I worked for Qlusters for several months. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]