Re: [leaf-devel] What about removing moddb?
Am 31.03.2016 um 19:26 schrieb kp kirchdoerfer: > Hi Erich; > > Am Donnerstag, 31. März 2016, 19:00:45 schrieb Erich Titl: >> Hi KP >> >> Am 31.03.2016 um 17:31 schrieb kp kirchdoerfer: >> ...> >> >>> The name "hwdetect" is misleading. The more interesting part of hwdetect >>> now is loading modules from /etc/modules rather than loading hardware. >>> This way I can add pppoe to /etc/modules, run hwdtect (of "f" from lrcfg >>> menu) and get pppoe and modules it depends on (ppp*) loaded without >>> reboot. This is allows quick testing etc.. >> >> I believe there is a simpler script in /usr/sbin to accomplish this. >> Look for install_modules. >> You need not modify /etc/modules. > > Good suggestion - I believe it' too new. > > A command line option for help could be useful though. (see below) > > Other than that it works. > > But the way it is, I doubt this does work from lrcfg menu? Well I never edited /etc/modules from the lrcfg menu, to be honest I mainly use one single menu point save :-( > > Now what about merging hwdetect (with the ability to read from /etc/modules) > with install_modules (using just the CLI) ? What for? > This way we can test from CLI without /etc/modules and with changing > /etc/modules and testing from lrcfg menu? I believe it is easier to type install_modules sd-mod than - edit /etc/modules enter sd-mod - save - run lrcfg and choose (what is this again) ... ahhh f) Find & load modules for hardware Maybe we could build a webconf page which allows to just enter the module name to be loaded. More difficult is the unloading process, detecting what else to unload, e.g. what modules are not needed when one is removed. cheers ET -- Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] What about removing moddb?
Hi Erich; Am Donnerstag, 31. März 2016, 19:00:45 schrieb Erich Titl: > Hi KP > > Am 31.03.2016 um 17:31 schrieb kp kirchdoerfer: > ...> > > > The name "hwdetect" is misleading. The more interesting part of hwdetect > > now is loading modules from /etc/modules rather than loading hardware. > > This way I can add pppoe to /etc/modules, run hwdtect (of "f" from lrcfg > > menu) and get pppoe and modules it depends on (ppp*) loaded without > > reboot. This is allows quick testing etc.. > > I believe there is a simpler script in /usr/sbin to accomplish this. > Look for install_modules. > You need not modify /etc/modules. Good suggestion - I believe it' too new. A command line option for help could be useful though. (see below) Other than that it works. But the way it is, I doubt this does work from lrcfg menu? Now what about merging hwdetect (with the ability to read from /etc/modules) with install_modules (using just the CLI) ? This way we can test from CLI without /etc/modules and with changing /etc/modules and testing from lrcfg menu? kp /usr/sbin/install_modules -h logger: invalid option -- h BusyBox v1.24.0 (2016-03-25 10:35:04 CET) multi-call binary. Usage: logger [OPTIONS] [MESSAGE] Write MESSAGE (or stdin) to syslog -s Log to stderr as well as the system log -t TAG [ 1577.055461] EXT4-fs (sda2): couldn't mount as ext3 due to feature incompatibilities Log using the sp[ 1577.064539] EXT4-fs (sda2): couldn't mount as ext2 due to feature incompatibilities ecified tag (defaults to user name) -p PRIO Priority (numeric or facility.level pair) install_modules: installing -h modprobe: invalid option -- h BusyBox v1.24.0 (2016-03-25 10:35:04 CET) multi-call binary. Usage: modprobe [-alrqvsDb] MODULE [SYMBOL=VALUE]... -a Load multiple MODULEs -l List (MODULE is a pattern) -r Remove MODULE (stacks) or do autoclean -q Quiet -v Verbose -s Log to syslog -D Show dependencies -b Apply blacklist to module names too -- Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] What about removing moddb?
Hi KP Am 31.03.2016 um 17:31 schrieb kp kirchdoerfer: ...> > The name "hwdetect" is misleading. The more interesting part of hwdetect now > is loading modules from /etc/modules rather than loading hardware. > This way I can add pppoe to /etc/modules, run hwdtect (of "f" from lrcfg > menu) > and get pppoe and modules it depends on (ppp*) loaded without reboot. > This is allows quick testing etc.. I believe there is a simpler script in /usr/sbin to accomplish this. Look for install_modules. You need not modify /etc/modules. cheers ET -- Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] What about removing moddb?
Am Montag, 28. März 2016, 22:13:37 schrieb Erich Titl: > Am 27.03.2016 um 18:26 schrieb kp kirchdoerfer: > > Am Sonntag, 27. März 2016, 16:03:06 schrieb Erich Titl: > >> Hi KP > >> > >> Am 26.03.2016 um 16:55 schrieb kp kirchdoerfer: > >>> H Gents; > >> > >> ... > >> > >>> At last I changed hwdetect to neglect moddb, and to do not use > >>> /var/lib/lrpkg/*.depmod any longer - there is no need to copy modules > >>> to > >>> /lib/modules any more. > >> > >> IMHO we could also ditch hwdetect. There is hardly any difference in the > >> code than in linuxrc. WE might need to add mount_modules to the hotplug > >> script though. > > > > hotplug seems not to be a replacement for hwdetect. > > > > Yes the hwdetect code is in linuxrc, but as I said it could be convenient > > to run hwdetect and load modules *without reboot*. > > What for? It is highly unlikely the hardware configration changes on a > running machine. The only place I could see this is with unknown usb > hardware. But then hwdetect needs to mount modules. The name "hwdetect" is misleading. The more interesting part of hwdetect now is loading modules from /etc/modules rather than loading hardware. This way I can add pppoe to /etc/modules, run hwdtect (of "f" from lrcfg menu) and get pppoe and modules it depends on (ppp*) loaded without reboot. This is allows quick testing etc.. > > Another issue iI found is we are creation, deletion, link and unlink > > /lib/modules/$KVER add various places, which ends up in a link to itself > > > > # ls -al /lib/modules/ > > lrwxrwxrwx1 root root12 Mar 27 18:18 4.4.6-x86_64 -> > > /lib/modules > > This link should not exist anymore. It is a leftover from the old > linuxrc which may have survived. I've made my work on removing moddb traces available as remote repository (remove_moddb) - pls review before it will be merged! Maybe the leftover is solved as well. thx kp -- Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] Delayed umount
Am 31.03.2016 um 10:47 schrieb Andrew: > It just blocks killing on console logout (IMHO it'll be bad if all > remains mounted after logout) sure . It oesn't hurt TERM/KILL signals. Tests have shown problems for some reason. normal background operation works fine. cheers ET -- Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] Delayed umount
It just blocks killing on console logout (IMHO it'll be bad if all remains mounted after logout). It oesn't hurt TERM/KILL signals. 31.03.2016 11:35, Erich Titl пишет: > Am 31.03.2016 um 10:30 schrieb Andrew: >> 31.03.2016 01:32, Erich Titl пишет: >>> Am 30.03.2016 um 17:20 schrieb Andrew: 30.03.2016 17:17, Erich Titl пишет: > ... >> Command can push itself in background. Something like this: >> >> if [ "$1" == "--nofork" ]; then >> umount all fs here >> else >> nohup $0 --nofork & >> endif > nohup is probably not a good idea it might block the killing > > > > > -- > Transform Data into Opportunity. > Accelerate data analysis in your applications with > Intel Data Analytics Acceleration Library. > Click to learn more. > http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 > > ___ > leaf-devel mailing list > leaf-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/leaf-devel -- Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] Delayed umount
Am 31.03.2016 um 10:30 schrieb Andrew: > 31.03.2016 01:32, Erich Titl пишет: >> Am 30.03.2016 um 17:20 schrieb Andrew: >>> 30.03.2016 17:17, Erich Titl пишет: ... > Command can push itself in background. Something like this: > > if [ "$1" == "--nofork" ]; then > umount all fs here > else > nohup $0 --nofork & > endif nohup is probably not a good idea it might block the killing -- Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] Delayed umount
31.03.2016 01:32, Erich Titl пишет: > Am 30.03.2016 um 17:20 schrieb Andrew: >> 30.03.2016 17:17, Erich Titl пишет: > ... >>> There are a few drawbacks with the invocation >>> >>> - umount_delayed must be pushed to the background to terminate >> it may call itself with some parameter (that indicates that it's fork) >> and if parameter is present - do delayed umunt, if not - run itself in >> background with parameter via nohup (to be sure that all is umounted >> even if console is terminated). > Maybe I was not clear enough. The calling process needs to push the > umount command to the background to make sure it can continue its own > process. Else it will wait for the termination of the called process. Command can push itself in background. Something like this: if [ "$1" == "--nofork" ]; then umount all fs here else nohup $0 --nofork & endif > cheers > > ET > > -- > Transform Data into Opportunity. > Accelerate data analysis in your applications with > Intel Data Analytics Acceleration Library. > Click to learn more. > http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 > > ___ > leaf-devel mailing list > leaf-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/leaf-devel -- Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471=/4140 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
