=== maco_ is now known as maco
Leon_NardellaAnyone working on the ex4 freezes in Jaunty? I'm using the latest mainline build from the kernel-ppa and it seems to be free of this bug.02:59
* Leon_Nardella forgot to mention this is that bug that manifests under heavy IO03:00
=== _Purple_ is now known as _Purple_away
=== _Purple_away is now known as _Purple_
Kamping_KaiserHi, could someone explain the difference between https://launchpad.net/ubuntu/+source/linux and http://packages.ubuntu.com/search?keywords=linux-image-generic (or http://packages.ubuntu.com/search?keywords=linux )10:00
Kamping_Kaiserthe LP link indicates that hardy-security and hardy-updates share the same kernel versions, but p.u.c has them marked as different versions still10:01
=== asac_ is now known as asac
=== Omegamoon is now known as Omegamoon|away
maxbKamping_Kaiser: In the LP link, you are looking at the "linux" source package. In the p.u.c link, you are looking at the "linux-meta" source package11:50
maxbYes, it's a little confusing that the "linux" binary package is built by the "linux-meta" source package, not the "linux" source package11:51
Kamping_Kaisermaxb, ah. so on LP i need to be looking at 'linux-meta' instead?11:51
Kamping_KaiserI should probably move a bug report then, since it relates to the linux-meta packages not linux.11:53
maxbWell, that depends which package you're actually interested in :-)11:53
Kamping_Kaiser350 bugs against linux-meta. criky. not sure moving the bug in there is going to help it13:06
Kamping_KaiserHow can I mark https://bugs.launchpad.net/bugs/342638 as affecting linux-meta not linux ?13:10
ubot3Malone bug 342638 in linux "linux-image-{generic,386} in -security uninstallable without -updates repository." [Undecided,New] 13:10
erle-did anyone try to build kernel with make-kpkg?14:01
=== k-s[AWAY] is now known as k-s
asacis there a hackish way to force a different vendor/product id for a usb device (e.g. so i can test a driver with a new device without rebuilding the kernel)?14:25
Kamping_Kaiserhex editor + kernel module? hackish enough? ;)14:27
asacgood idea14:28
asactoo bad ... cant find the id ;)14:33
asacis there anything i can look for in the module to find the right spot?14:33
asacits cdc_ether ;)14:34
erle-are the "mainline" kernels patched?14:34
erle-i thougt that are vanilla kernels14:34
asacheh. i think i found it :)14:36
asacnow i am scared ;)14:36
anubhavasac: why not just recompile the module with your vend id and prod id14:38
asacanubhav: hmm. whats the best way to just build a module?14:38
anubhavoffcourse you will need sources14:39
asaci dont have a built tree here ;)14:39
anubhavyou are building against the standard ubuntu kerenel?14:39
asacanubhav: i want to do that because i this thing also has some drivers that are not in mainline14:40
asacso yes14:40
asaci want to build a patched cdc_ether.ko14:40
asacagainst current jaunty14:40
asacdarn ... now i cannot get the raw file from kernel.ubuntu.com ... gives me a 0 bytes download14:42
asacgood now it worked14:43
erle-shouldn't a vanilla kernel build with a config of the ubuntu mainline kernel? (same kernel version!!)14:44
asacanubhav: so i have the patched module source ... what i can do now?14:45
amitkerle-: mainline kernels are stock linus' trees with the ubuntu config14:46
erle-amitk, i know14:46
erle-but it doesn't build here ...14:46
erle-Makefile:529: /usr/src/linux-source-2.6.29/arch/xen/Makefile: No such file or directory14:46
erle-make[1]: *** No rule to make target `/usr/src/linux-source-2.6.29/arch/xen/Makefile'. Stop.14:46
erle-make: *** [conf.vars] Error 214:46
erle-i have no idea, why14:46
erle-i don't care about xen14:46
asacamitk: do you have any clue? i need to just build a single kernel module file i have here 14:47
amitkasac: cdc-ether is compiled in?14:47
* amitk checking14:47
asacamitk: its a module ... i need to build it with more device ids14:48
asacamitk: http://paste.ubuntu.com/144184/14:48
amitkasac: yes, it can be done AFAIK. Just looking for the right /sys file to put that info in14:49
asacthe other idea was to use hexedit ;)14:49
asacmy first hope was that i can force a different vendor/product id on that device somehow14:49
asacjust to test14:49
amitkasac: /sys/module/cdc_ether/drivers/usb:cdc_ether/new_id14:50
erle-now i use the sourcecode from the mainline thing and the config from the mainline thing and it still does not build ...14:50
asacamitk: what can i do with that?14:51
amitkasac: still looking for the exact incantation14:51
erle-does nobody have an idea?14:51
amitkerle-: it doesn't always build and it is too early on a weekend for me to know why :)14:51
erle-it says he cannot build xen or something14:52
erle-but there is no xen in the source code which actually should have been built14:52
erle-do you use make-kpkg to build?14:53
erle-maybe it is the tool failing14:53
asacok so its a colon separated list14:58
asacamitk: http://paste.ubuntu.com/144189/14:58
asacthats what i found on google :)14:59
asacnow i have to find the values for everything  before vendor14:59
erle-it build now15:00
erle-but i had to mod the configuration manually ...15:00
asachmm ... not sure if its the same for usb though15:01
amitkasac: do a 'modinfo cdc-ether'15:02
amitkthe format of the alias strings might work15:02
amitkit is a concatenated string15:02
amitkv=vendor, p=productid, etc.15:02
asacecho 'usb:v413cp8147d*dc*dsc*dp*ic02isc0Aip00*' > /sys/module/cdc_ether/drivers/usb\:cdc_ether/new_id15:04
asacbash: echo: write error: Invalid argument15:05
asacecho "0 0 ffffffff ffffffff 413c 8147" > /sys/module/cdc_ether/drivers/usb\:cdc_ether/new_id15:07
asacamitk: that didnt complain15:07
amitkasac: New PCI IDs may be added to a device driver pci_ids table at runtime15:08
amitkas shown below:15:08
amitkecho "vendor device subvendor subdevice class class_mask driver_data" > \15:08
amitkAll fields are passed in as hexadecimal values (no leading 0x).15:08
amitkThe vendor and device fields are mandatory, the others are optional. Users15:08
amitkneed pass only as many optional fields as necessary: o subvendor and subdevice fields default to PCI_ANY_ID (FFFFFFFF) o class and classmask fields default to 0 o driver_data defaults to 0UL.15:08
amitkNote that driver_data must match the value used by any of the pci_device_id15:08
amitkentries defined in the driver. This makes the driver_data field mandatory15:08
amitkif all the pci_device_id entries have a non-zero driver_data value.15:08
amitkasac: so usb should be the same15:08
asacamitk: yeah ... but it didnt really do anything 15:10
asacbut i dont know what to expect15:10
asaci hoped that it showed up in hal ;15:10
anubhavrocksasac: but the string you echo'd is diffrent from what amitk said15:11
anubhavrocksyou mean to say PCI and USB have diffrent conventions?15:12
asacecho "413c 8147" > /sys/module/cdc_ether/drivers/usb\:cdc_ether/new_id15:12
asacamitk: would you think that that triggers udev events automatically?15:13
anubhavrocksasac: i think this should work15:15
amitkasac: it should, if the driver can actually bind to that particular device15:15
asaci am quite sure it would work as its the same device just rebranded for dell15:16
amitkasac: ^15:16
asachmm ... unplugging is difficult for me ;)15:19
asacok trying udev rule15:21
asacany idea what zaurus driver does?15:25
asacok rebooting15:27
asacno change15:33
asacnot sure whz zaurus is always loaded15:34
anubhavrocksblacklist the driver if you don't want it to load15:35
asaci did that15:36
anubhavrocksyour problem is that when you insert your device the zaurus module loads instead of cdc_ether?15:38
asacnot sure. could be ... i think my problem is that just the device/vendor id is not enough15:39
asacwait a sec15:39
anubhavrockscheckk the output of lsusb and make sure you enter the correct Vid Pid 15:39
asaci think i also need to tell new_id that its USB_CDC_SUBCLASS_MDLM,15:39
asacanubhavrocks: well. those are correct ;)15:40
asaci get complain about invalid mdlm descriptors in dmesg15:40
asacso i think thats really the problem15:40
asac16:08 < amitk> echo "vendor device subvendor subdevice class class_mask driver_data" > \15:42
asachow could i encode a subclass there?15:42
anubhavrocksasac: lsusb -v 15:45
asacso back to square oone ... how can i just build this bloody .c file ;)15:45
asacanubhavrocks: the ids are right :2)15:45
asacwait a sec15:45
anubhavrocksto get the subclass etc15:45
asacBus 001 Device 003: ID 413c:8147 Dell Computer Corp. 15:45
asacanubhavrocks: well. the problem is that i dont kinow how to encude subclass in the echo above ;)15:46
asacanubhavrocks: http://paste.ubuntu.com/144220/15:47
asacisnt there a bloody gcc command to make a .ko out of .c ?15:49
asacok so i guess i will go for hexedit option15:51
asac that worked!!15:53
* asac dances!!15:54
asacamitk: thanks :) ... hexeidt was the right thing i needed15:54
asac[ 1541.258987] usb0: register 'cdc_ether' at usb-0000:00:1d.7-6, CDC Ethernet Device, 02:80:37:ec:02:0015:54
* asac back to modemmanager hacking ;)15:55
=== Omegamoon|away is now known as Omegamoon
=== k-s is now known as k-s[AWAY]
pjwafflehi this is a little off topic but it is for kernel purposes... how do you create a apt repo? I need to know so I can setup a repository for testing git version of kernel21:09
=== k-s[AWAY] is now known as k-s
=== k-s is now known as k-s[AWAY]

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!