| |
| |
| |
|
Page: 1 2
<jimm> how can I find out if there are/were mem leaks in 2.4.27?
<szk> I will try ssh'ing in a little bit
<szk> It does not work on other distributions either.. Works fine in windows.
<szk> czr: the system just totally hard locks. Cant ctrl alt backspace or anything
<szk> maybe when I compile my own kernel i enable smp, for my hyperthreading.. that might have something to do with it
<MrParanoia> strange things happen with building kernels, selecting * for Wireless LAN makes the ipw2200 option go away - why?
<jimm> read the config file and see
<MrParanoia> what is causing this in 2.4.14? - WARNING: /lib/modules/2.6.14/kernel/net/ieee80211/ieee80211_crypt_tkip.ko needsunknown symbol wireless_send_event
<MrParanoia> err 2.6.14
<jimm> is 2.6.14 released?
<MrParanoia> yep
<jimm> what program is generating that error message?
<jimm> s/error//
<MrParanoia> not sure since i && many together
<jimm> meetoo
<MrParanoia> my guess is make modules_install
<jimm> make would not at all produce that message.
<jimm> find out what program does.
<MrParanoia> make modules gave me this so the other one came from make modules_install : *** Warning: "wireless_send_event" [net/ieee80211/ieee80211_crypt_tkip.ko] undefined
<jimm> hint: what does make actually do?
<MrParanoia> compiles
<jimm> it does??
<jimm> are you sure?
<MrParanoia> ok, so it builds the modules
<jimm> how?
<jimm> what make does is this:
<MrParanoia> it compiles c code into object code?
<MrParanoia> C/C++
<jimm> - determines from the make file a relation between files, such that one given file is "source" to another file's "result"
<jimm> - compares a result file to a source file to determine from both file's time stamp whether the result is up to date, or out of date
<MrParanoia> yeah, and the modules end with the .o extension which is object code that the compiler spits out
<jimm> - if it is up to date, make does NOTHING
<MrParanoia> yep
<MrParanoia> i noticed that
<jimm> - if it is out of date, make calls ANOTHER PROGRAM which builds the result from the source.
<deadcat> my 2.6.14 kernel waits at the ext3 step. i dont know why. i am using same fs settings from .12 and .14
<jimm> make has NOTHING that actually does ANY building.
<jimm> it p***es the job to OTHER programs.
<jimm> therefore:
<MrParanoia> ok. i didn't know that - i ***umed that it called the compiler
<jimm> your error/warning messages are being produced from programs that make calls, NOT from make itself
<MrParanoia> at any rate - the build failed because i got an error - maybe from mkinitrd which follwed make modules_install?
<jimm> YES, that's exactly what it does: it CALLS the compiler. but it is NOT the compiler.
<jimm> so... you have to determine which program (that make invokes) produces the error message you get
<jimm> (note: exploration in this area will also familiarize you with the set of compiler tools, inasmuch as you're essentially compiling a C program)
<MrParanoia> jimm> there's a lot of things i need to do - i wish i had the time to go more indepth
<jimm> well, until you do, you might not find a substantial number of your own problems
<jimm> anyways, once you find the exact program that produces the message you're getting, you can check the DIAGNOSTICS section of the man page to find the meaning of the message
<MrParanoia> unfortuneatly i have other things that need attention - so the fewer the problems, the more i'll enjoy linux
<jimm> note that you are free to run any OS you want :)
<jimm> unfortunately, time is short... a phenominon we all understand... fortunately, there are paid consultants :)
<MrParanoia> i know that this is freeware but when just about everything has a problem i don't have time to do anything useful with the stuff - as for paid tech support, i have paid tech support for my new laptop and i STILL can't get answers
<jimm> did your laptop come with linux installed on it?
<jimm> in your case, you can get more information if you are moreorless familiar with the parts of the compiler and can deternine which program generated a message...
<jimm> are you sure you typed the entire message? sometimes the program is identified at the start of the warning/error message line
<jimm> at this point we can't even tell whether the message is related to compiling or linking
<MrParanoia> this is what i got after running make modules_install: http://pastebin.com/411579
<jimm> hmm, it looks like depmod is producing the message
<jimm> but that's a warning message, which should not cause make to terminate early... can you find an -error- message? or is that the only output?
<MrParanoia> that's all
<jimm> how did you get the kernel source itself?
<jimm> (note, I don't understand why building a kernel should involve depmod in any way... depmod should really be run at each boot to create the module dependency file)
<izo> hi how can i remove hacked kernel module from kernel ?
<izo> it doesnt show via lsmod
<izo> bu ti found it via module_hunter
<jimm> there is rmmod
<izo> i know but the module name is funny
<jimm> but if you've been compromised... you might want to format hd and restore from recent backup
<izo> WVS|$1v_v,h-Q`h4
<jimm> quote it
<izo> did that
<izo> didtn help
<jimm> why?
<izo> i belive rmmod was compromised as well
<jimm> but if you've been compromised... you might want to format hd and restore from recent backup
<izo> well i'd prefer to get rid of that stuff rist
<jimm> so get rid of it... but you're talking about forensics, which is tedious and therefore expensive
<jimm> if rmmod was compromised, how do you know what else was?
<jimm> or was not?
<c14n> i'm stuck on an issue with a driver here, as of unregister_ioctl32_conversion/register were removed as of 2.6.13
<c14n> now 2.6.14
<c14n> I want to register cmd, handler
<c14n> and unreg cmd
<izo> jimm i dont but i'ld like to investigate :-)
<jimm> there's nothing wrong with that
<c14n> (for 64bit)
<jimm> if you have that kind of time, go for it :)
<jimm> but if the machine involved was serving a purpose, you might want to replace it and get it going... then you can explore the machine in question all you want without worrying about outage
<izo> well good thing we had backup server
<jimm> hmm, yes... now I see why you want to investigate.. because if it happened to your server, it could just as easily happen to the backup, and in the same way
<izo> well not eaxtly the hacked one was old redhat the backup is new debian
<izo> anyway thnx
<jimm> is the backup working well now? including necessary data?
<izo> yeah
<jimm> back it up. soon. or better... now.
<jimm> make sure you can restore onto some machine from a backup
<c14n> you guys know what handler is for 64bit?
<jimm> you mean for C kernel api?
<c14n> yes
<jimm> no...
<c14n> damn
<c14n> Time to do
<c14n> dig*
<jimm> but can you get an api listing?
<c14n> i'm going to look through this arch/x86_64/ia32/ia32_ioctl.c
<c14n> i'm trying to fix some drivers
<jimm> ia32? will that help you with 64bit?
<c14n> Because of the late, What: register_ioctl32_conversion() / unregister_ioctl32_conversion()
<c14n> Why: Replaced by ->compat_ioctl in file_operations and other method
<c14n> vecors.
<jimm> it looks like you'll have to look at how to use those
<c14n> Those are removed, i need to find a new way to implement the 'register' and unregister functions.
<jimm> what version of debian do you have? what kernel?
<jimm> err, izo:
<c14n> I should try and find another function in the deprecated kernel that used this before
<c14n> and see what it has changed to to support a proper register and unregister
<jimm> excellent idea...
<c14n> I figure if I want to do it properly I should follow an examplke
<c14n> example*
<jimm> that should lead almost directly to a solution
<c14n> Indeed.
<c14n> Fixing ATI drivers for 64bit as well.
<c14n> ah ha!
<jimm> omg, fixed it already?!
<c14n> found a function
<c14n> which does exactly what i was looking for
<jimm> nice :)
<c14n> hmm still the same in this one, but i wonder
<c14n> wonder if 2.6.12 has more of that function within it
<c14n> nope
<c14n> ah ha
<c14n> 2.6.11 has a ton of those conversions in it
<c14n> :)
<c14n> rock on
<c14n> figured it out
<jimm> nice :)
<magyar> does 2.6.14 containes pptp VPN p***through for netfiler?
<b-_-d> where is the info on why or why not a patch gets into the kernel
<c14n> b-_-d: They get in if they're required most of the time
<c14n> and or signed off by osdl members
<b-_-d> i saw on google reiser4 will not be in 2.6.15
<b-_-d> i want to know why
<c14n> He probably does not update fast enough.
<b-_-d> two week window is too small
<c14n> If it's not ready for a patch release sure
<b-_-d> ok
<b-_-d> i will just google for answers
<Jason-Bourne> reiser4 will be coming in 2.8.10.2.w00t.lol.^_^
<rebel> are there any kernel-land functions for opening a file?
<MrParanoia> got one problem solved in 2.6.14
Return to kernel
or
Go to some related
logs:
chatzone
lisp
6502 "test suite"
|
|