Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Hi all. Here is the latest update. We got the following two patches tested by Ceibal :: a) 0001-TEST-See-if-this-works-for-the-mesh-issue-on-signed-.patch This patch, adds /etc/modprobe.d/libertas.conf to the initramfs image (unarchiving, adding and archiving during OOB stage). Unfortunately, this does not seem to work. We could debug it more, but I fear that that would inevitably mean modifying the initramfs a fair bit. Since, we are talking at the kernel level, modifying initramfs even a bit too much, would require exponential more testing. Lucking, we have a simple, clean, understood and guaranteed-to-work solution. b) 0001-olpc-configure-Reloading-libertas-and-dependent-usb8.patch (As already stated in earlier mails) This patch reloads the libertas (and the dependent usb8xxx driver) during intial boot, which mounts the secondary-storage-based-filesystem. This time, the rules from /etc/modprobe.d/* are read, and the mesh is appropriately disabled via the libertas_disablemesh KLM parameter. So, it would be good if this is included upstream. For brevity, I am re-attaching the two patches. Thanks and Regards, Ajay 0001-TEST-See-if-this-works-for-the-mesh-issue-on-signed-.patch Description: Binary data 0001-olpc-configure-Reloading-libertas-and-dependent-usb8.patch Description: Binary data ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Ajay Garg ajaygargn...@gmail.com writes: I tried once again the Case 2; and upon resume-from-suspend, the icons appeared fine at my end. OK, that's good and bad. Good in the sense that my patch may not be to blame; bad in that you encountered a problem that couldn't be reproduced. It may come back later and bite us twice as hard. If I face the unusual result for Case 2 again, I will post the /var/log/messages, which may be scrutinized. Please also check for and save core dumps. Not sure they'll be generated by default for system daemons (I'm thinking NetworkManager in particular). If they aren't, we should investigate how to enable them globally in our development builds. Manuel Kaufmann pointed out (in another thread) that there's olpc-log. We (= AC) should take a closer look at it [1] and see if we can extend it to provide a complete snapshot for debugging. Having a single command that always saves all of the information that may be needed would reduce the chance of forgetting anything. Sascha [1] https://dev.laptop.org/git/projects/olpc-netutils/tree/usr/bin/olpc-log -- http://sascha.silbe.org/ http://www.infra-silbe.de/ pgpv5Ic3PGm6C.pgp Description: PGP signature ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Paul Fox p...@foxharp.boston.ma.us writes: yes. i think we hope there won't actually be any more of those, but if there are, that patch will be there. Great, thanks! Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ pgpeYEFdWOZk1.pgp Description: PGP signature ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
[CC -= sugar-devel] Martin Langhoff martin.langh...@gmail.com writes: This makes things rather tricky -- currently to get a kernel driver param such as this one you need to install it _on the machine where you build your kernel rpm_. This is at best awkward -- the kernel build environment better be a chroot :-/ Ouch. That means we'll have to build our own kernel, and even with a manual procedure (inside a specially prepared chroot). The reason we were still using the /sys/class/net/eth0/lbs_mesh hack was that we didn't want to fork the upstream (OLPC) kernel. Kernel updates usually have a good reason, so they should spread automatically to all systems in the field. Maybe we should consider something like modifying the initrd (initramfs?) from within some script in olpc-os-builder. Append the module config file to the existing, already installed initrd. We'd also need to add in some hook that runs when the kernel RPM gets updated (which, I assume, means that the initrd gets replaced as well) so that live updates retain our modification. Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ pgphiCVesTrzN.pgp Description: PGP signature ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Are there any deployments that still want the mesh to work? It might be simpler to pull it out now, given that only XO-1 supports it. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Mon, 2012-05-14 at 18:32 +1000, James Cameron wrote: Are there any deployments that still want the mesh to work? It might be simpler to pull it out now, given that only XO-1 supports it. +1 Given that the mesh layout was intended to work with the XS's Active Antenna which isn't being developed anymore. Guess the 2 AAs that I have are collector items now. Jerry ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Mon, May 14, 2012 at 2:02 PM, James Cameron qu...@laptop.org wrote: Are there any deployments that still want the mesh to work? It might be simpler to pull it out now, given that only XO-1 supports it. FWIW, Py, Uy, don't want or use it. AU uses only xo-1.5's so no mesh. +1 from me to disable this by default on the XO-1. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Mon, May 14, 2012 at 5:11 AM, Anish Mangal an...@activitycentral.com wrote: On Mon, May 14, 2012 at 2:02 PM, James Cameron qu...@laptop.org wrote: Are there any deployments that still want the mesh to work? It might be simpler to pull it out now, given that only XO-1 supports it. FWIW, Py, Uy, don't want or use it. AU uses only xo-1.5's so no mesh. +1 from me to disable this by default on the XO-1. I am partially for this as well, but we need to think it through. cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Mon, May 14, 2012 at 2:32 AM, James Cameron qu...@laptop.org wrote: Are there any deployments that still want the mesh to work? It might be simpler to pull it out now, given that only XO-1 supports it. The XO-1 mesh is significantly more reliable than marvell wifi's implementation of ad-hoc. It should stay enabled by default until we have a less broken ad-hoc implementation, in the interests of having serverless collaboration. Daniel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Thanks Martin and Jerry. I followed the following steps :: a) Listed lsinitrd /boot/initrd.img. It did not show any /etc/modprobe.d/libertas.conf. b) Extracted /boot/initrd.img into a temporary folder. Obviously, no /etc/modprobe.d/libertas.conf was seen. c) Generated the required /etc/modprobe.d/libertas.conf with the content options libertas libertas_disablemesh=1 in the tmp folder. d) Re-compressed the temp-folder to /boot/initrd.img. e) Listed lsinitrd /boot/initrd.img. This time, /etc/modprobe.d/libertas.conf was listed. f) Rebooted XO-1. g) Mesh-icons were still observed :( h) Note that I did this, without OOB (i.e. I was just wanting to test that whether this would work at all). I guess that there is something else we might be missing. So, I think that the only solution left now, is to put the rmmod/modprobe hack in olpc-configure. Would it be an acceptable upstream solution (the change in olpc-utils) ? Martin (Langhoff) / Daniel (Drake) ? Looking forward to comments and replies. Thanks and Regards, Ajay On Sun, May 13, 2012 at 3:10 AM, Jerry Vonau jvo...@shaw.ca wrote: On Sat, 2012-05-12 at 15:20 -0400, Martin Langhoff wrote: On Sat, May 12, 2012 at 4:38 AM, Ajay Garg a...@activitycentral.com wrote: I tried the rmmod/modprobe hack in olpc-configure, and it worked (obviously because, this time the /etc.modprobe.d/libertas.conf could be fetched/read from persistent storage). Mesh-icons were no more visible in the neighborhood-view (both during reboot, and resume-from-suspend). But I too felt that this is more of a hack, and not a clean solution. Well, as you pointed out in later emails, the initramfs generation will pick it up from there. I'd forgotten about that -- sorry. So what you want to do is get libertas.conf into some package (olpc-utils) or install it from an OOB module -- the xo1 module is a good candidate. In fact I'm liking the OOB module option. I think we would accept a patch to OOB's xo1 module, where based on an option it creates this file. Just have to make sure it's created before dracut is used to generate the initramfs. Hi Martin: Don't think OOB is used to generate the initramfs, from the kernel spec file: # set to 1 to build the initramfs during kernel-build time # set to 0 to build the initramfs during %post on the host %define buildinitramfs 1 Doesn't that mean the kernel rpm provides pre-build initrdversion.img and actrdversion.img? Doesn't OOB just sign what is in the kernel rpm? Think you would need to re-create initrd.img as needed then place the revised initrd.img into the image via OOB, to be signed by OOB. Jerry ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Sat, May 12, 2012 at 5:40 PM, Jerry Vonau jvo...@shaw.ca wrote: Don't think OOB is used to generate the initramfs, from the kernel spec file: You are correct as to current state of play. It _used_ to be generated during the OOB run. Now it is build during the kernel build. This makes things rather tricky -- currently to get a kernel driver param such as this one you need to install it _on the machine where you build your kernel rpm_. This is at best awkward -- the kernel build environment better be a chroot :-/ So currently Ajay will need to rebuild the kernel rpm in a chroot where he has installed that libertas.conf he desires. Luckily, we don' t mess with this too much. At some point however, I hope that hacking in the initrd gets a bit more practical. cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Thanks Martin. Very neatly explained :) I tried the rmmod/modprobe hack in olpc-configure, and it worked (obviously because, this time the /etc.modprobe.d/libertas.conf could be fetched/read from persistent storage). Mesh-icons were no more visible in the neighborhood-view (both during reboot, and resume-from-suspend). But I too felt that this is more of a hack, and not a clean solution. So, I ventured out exploring dracut. I created a initramfs image on my home/work laptop, hosting Fedora-14, via the command :: dracut test.img and then listed the contents of it lsinitrd test.img I saw that /etc/modprobe.d/* files were a part of test.img. So, I think that these files _are_ included as part of the initramfs image, as per say. So, Is this observation (that /etc/modprobe.d/* files are included, as per say) in line with what is expected? If yes, that is kind of a relief, since that would mean that only /etc/modprobe.d/libertas.conf is not being included (in initramfs that is). Thanks and Regards, Ajay On Fri, May 11, 2012 at 3:33 AM, Martin Langhoff martin.langh...@gmail.comwrote: On Thu, May 10, 2012 at 1:23 PM, Ajay Garg a...@activitycentral.com wrote: If I boot with /security/develop.sig folder in my pendrive, a) mesh-icons are observed in neighborhood-view, both during reboot and resume-from-suspend. Welcome to the initramfs stage of your journey! When the laptop needs activation, it loads a different initramfs that among other things loads the libertas module. You need to get your /etc/modprobe.d/ files into the initramfs. For your vanilla build, look into dracut-modules-olpc. If you're hoping to get this integrated into a build with an alternative initramfs (hint: Ceibal) that build will have a custom version of dracut-modules-olpc. That is the right way. When the laptop is in secure mode, olpc.fth is _ignored_, so no chance to set a kernel cmdline there. An easier alternative might be to check for those flags under /sys, and if they are there, rmmod/modprobe libertas for example in olpc-configure (so during early boot). m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Sat, May 12, 2012 at 2:08 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Martin. Very neatly explained :) I tried the rmmod/modprobe hack in olpc-configure, and it worked (obviously because, this time the /etc.modprobe.d/libertas.conf could be fetched/read from persistent storage). Mesh-icons were no more visible in the neighborhood-view (both during reboot, and resume-from-suspend). But I too felt that this is more of a hack, and not a clean solution. So, I ventured out exploring dracut. I created a initramfs image on my home/work laptop, hosting Fedora-14, via the command :: dracut test.img and then listed the contents of it lsinitrd test.img I saw that /etc/modprobe.d/* files were a part of test.img. So, I think that these files _are_ included as part of the initramfs image, as per say. So, Is this observation (that /etc/modprobe.d/* files are included, as per say) in line with what is expected? If yes, that is kind of a relief, since that would mean that only /etc/modprobe.d/libertas.conf is not being included (in initramfs that is). Well, just inspected lsinitrd olpcrd.img (assuming olpcrd.img _is_ the initramfs image in the signed build :D ) olpcrd.img, in fact, does not contain any /etc/modprobe/* files. Am looking into exploring customized dracut-modules-olpc.. Thanks and Regards, Ajay Thanks and Regards, Ajay On Fri, May 11, 2012 at 3:33 AM, Martin Langhoff martin.langh...@gmail.com wrote: On Thu, May 10, 2012 at 1:23 PM, Ajay Garg a...@activitycentral.com wrote: If I boot with /security/develop.sig folder in my pendrive, a) mesh-icons are observed in neighborhood-view, both during reboot and resume-from-suspend. Welcome to the initramfs stage of your journey! When the laptop needs activation, it loads a different initramfs that among other things loads the libertas module. You need to get your /etc/modprobe.d/ files into the initramfs. For your vanilla build, look into dracut-modules-olpc. If you're hoping to get this integrated into a build with an alternative initramfs (hint: Ceibal) that build will have a custom version of dracut-modules-olpc. That is the right way. When the laptop is in secure mode, olpc.fth is _ignored_, so no chance to set a kernel cmdline there. An easier alternative might be to check for those flags under /sys, and if they are there, rmmod/modprobe libertas for example in olpc-configure (so during early boot). m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Martin, Some observations :: a) All observations (appearance/non-appearance of mesh-icons) is inert to the presence/absence of the following packages :: * dracut * dracut-modules-olpc * dracut-modules-ceibal b) I could only see one initramfs image, on any of the signed/unsigned builds :: * olpcrd.img (with initrd.img being a symbolic link to it) So, some queries :: a) Is the (singular) RAM FS image (olpcrd.img) generated for every kind of build (but loaded into memory only when ALL of the following are met) :: * laptop is secured * image is signed * there is no developer key, either in pendrive, or on SD at /security/develop.sig. b) I could not find the place where this (singular) RAM FS image is generated, in the osbuilder (presumably via dracut). Doing some greps, gave me the following results :: ## [ajay@localhost bleeding-edge]$ grep -r -i -s initrd . ./modules/xo1_5/kspost.50.xo15-tweaks.inc:${DN}${PN}\initrd.img expand$ to ramdisk ./modules/signing/preimage.40.sign-os.sh:if [ -e $fsmount/boot/initrd.img ]; then ./modules/signing/preimage.40.sign-os.sh:./sign-os.sh $okey $fsmount/boot/initrd.img $fsmount/boot/runrd.zip ./modules/signing/preimage.10.extract.sh:if [ -e $fsmount/boot/initrd.img ]; then ./modules/signing/preimage.10.extract.sh:cp $fsmount/boot/initrd.img $tgt/data.img ./modules/signing/README: - if found, the initramfs at /boot/initrd.img (or /boot/olpcrd.img) will be ./modules/signing/README:found at /boot/initrd.img will be signed into runos.zip and runrd.zip ./modules/xo1/kspost.50.xo1-tweaks.inc:# FIXME: old olpc.fth looks for olpcrd.img, but we now use initrd.img ./modules/xo1/kspost.50.xo1-tweaks.inc:[ -e /boot/olpcrd.img ] || ln -s initrd.img /boot/olpcrd.img [ajay@localhost bleeding-edge]$ grep -r -i -s olpcrd . ./modules/signing/preimage.10.extract.sh:elif [ -e $fsmount/boot/olpcrd.img ]; then ./modules/signing/preimage.10.extract.sh:cp $fsmount/boot/olpcrd.img $tgt/data.img ./modules/signing/README: - if found, the initramfs at /boot/initrd.img (or /boot/olpcrd.img) will be ./modules/xo1/kspost.50.xo1-tweaks.inc:# FIXME: old olpc.fth looks for olpcrd.img, but we now use initrd.img ./modules/xo1/kspost.50.xo1-tweaks.inc:[ -e /boot/olpcrd.img ] || ln -s initrd.img /boot/olpcrd.img ./modules/xo1/kspost.50.xo1-tweaks.inc:${DN}${PN}\olpcrd.img expand$ to ramdisk ## Looking forward to a reply (especially to the first query). Thanks and Regards, Ajay On Sat, May 12, 2012 at 2:26 PM, Ajay Garg a...@activitycentral.com wrote: On Sat, May 12, 2012 at 2:08 PM, Ajay Garg a...@activitycentral.comwrote: Thanks Martin. Very neatly explained :) I tried the rmmod/modprobe hack in olpc-configure, and it worked (obviously because, this time the /etc.modprobe.d/libertas.conf could be fetched/read from persistent storage). Mesh-icons were no more visible in the neighborhood-view (both during reboot, and resume-from-suspend). But I too felt that this is more of a hack, and not a clean solution. So, I ventured out exploring dracut. I created a initramfs image on my home/work laptop, hosting Fedora-14, via the command :: dracut test.img and then listed the contents of it lsinitrd test.img I saw that /etc/modprobe.d/* files were a part of test.img. So, I think that these files _are_ included as part of the initramfs image, as per say. So, Is this observation (that /etc/modprobe.d/* files are included, as per say) in line with what is expected? If yes, that is kind of a relief, since that would mean that only /etc/modprobe.d/libertas.conf is not being included (in initramfs that is). Well, just inspected lsinitrd olpcrd.img (assuming olpcrd.img _is_ the initramfs image in the signed build :D ) olpcrd.img, in fact, does not contain any /etc/modprobe/* files. Am looking into exploring customized dracut-modules-olpc.. Thanks and Regards, Ajay Thanks and Regards, Ajay On Fri, May 11, 2012 at 3:33 AM, Martin Langhoff martin.langh...@gmail.com wrote: On Thu, May 10, 2012 at 1:23 PM, Ajay Garg a...@activitycentral.com wrote: If I boot with /security/develop.sig folder in my pendrive, a) mesh-icons are observed in neighborhood-view, both during reboot and resume-from-suspend. Welcome to the initramfs stage of your journey! When the laptop needs activation, it loads a different initramfs that among other things loads the libertas module. You need to get your /etc/modprobe.d/ files into the initramfs. For your vanilla build, look into dracut-modules-olpc. If you're hoping to get this integrated into a build with an alternative initramfs
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Sat, May 12, 2012 at 4:38 AM, Ajay Garg a...@activitycentral.com wrote: I tried the rmmod/modprobe hack in olpc-configure, and it worked (obviously because, this time the /etc.modprobe.d/libertas.conf could be fetched/read from persistent storage). Mesh-icons were no more visible in the neighborhood-view (both during reboot, and resume-from-suspend). But I too felt that this is more of a hack, and not a clean solution. Well, as you pointed out in later emails, the initramfs generation will pick it up from there. I'd forgotten about that -- sorry. So what you want to do is get libertas.conf into some package (olpc-utils) or install it from an OOB module -- the xo1 module is a good candidate. In fact I'm liking the OOB module option. I think we would accept a patch to OOB's xo1 module, where based on an option it creates this file. Just have to make sure it's created before dracut is used to generate the initramfs. A bit late, but we may be able to land it for 12.1.0 (or later for 12.2.0). cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Sat, 2012-05-12 at 15:20 -0400, Martin Langhoff wrote: On Sat, May 12, 2012 at 4:38 AM, Ajay Garg a...@activitycentral.com wrote: I tried the rmmod/modprobe hack in olpc-configure, and it worked (obviously because, this time the /etc.modprobe.d/libertas.conf could be fetched/read from persistent storage). Mesh-icons were no more visible in the neighborhood-view (both during reboot, and resume-from-suspend). But I too felt that this is more of a hack, and not a clean solution. Well, as you pointed out in later emails, the initramfs generation will pick it up from there. I'd forgotten about that -- sorry. So what you want to do is get libertas.conf into some package (olpc-utils) or install it from an OOB module -- the xo1 module is a good candidate. In fact I'm liking the OOB module option. I think we would accept a patch to OOB's xo1 module, where based on an option it creates this file. Just have to make sure it's created before dracut is used to generate the initramfs. Hi Martin: Don't think OOB is used to generate the initramfs, from the kernel spec file: # set to 1 to build the initramfs during kernel-build time # set to 0 to build the initramfs during %post on the host %define buildinitramfs 1 Doesn't that mean the kernel rpm provides pre-build initrdversion.img and actrdversion.img? Doesn't OOB just sign what is in the kernel rpm? Think you would need to re-create initrd.img as needed then place the revised initrd.img into the image via OOB, to be signed by OOB. Jerry ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Hi all. (Unfortunately) A new use-case has come into effect :: If I boot with /security/develop.sig folder in my pendrive, a) mesh-icons are observed in neighborhood-view, both during reboot and resume-from-suspend. b) /var/log/messages show the loading of msh0. However, there are no exceptions/crash-traces therein. c) Also, I see the presence of a folder /sys/class/net/msh0. However, if there is no security/develop.sig file in my pendrive, a) No mesh-icons are seen in neighborhood-view, during reboot or resume-from-suspend (as expected). b) /var/log/messages has no mention of msh0 (as expected). c) There is no folder /sys/class/net/msh0 (i guess that is what is expected). Is my observation in line with anyone's observation on olpc/sugar? (Note that the issue was reported by UY deployment; and it could be reproduced at our end as well). Kindly reply with comments. Thanks and Regards, Ajay On Sat, May 5, 2012 at 11:27 PM, Ajay Garg ajaygargn...@gmail.com wrote: Hi Kevin. Following are the steps and observations :: a) Booted with mesh-disabled (via options libertas libertas_disablemesh=1). b) In Neighborhood-View, wifi and three adhoc network icons were present. c) Then, I turned wifi-radio off, discarded network history. Now, only three adhoc network icons were present in Neighborhood-View. d) After some time, turned wifi-radio on. Now, wifi and three adhoc network icons could be seen. e) Again, I turned wifi-radio off, discarded network history. Again, only three adhoc network icons could be seen in Neighborhood-View. f) Rebooted. g) Only three adhoc network icons were seen in Neighborhood-View. h) After some time, turned wifi-radio on. Now, wifi and three adhoc network icons could be seen in Neighborhood-View. So, it works fine I guess (note that, mesh-icons were NOT seen anytime in the above steps) :) Regards, Ajay On Sat, May 5, 2012 at 6:43 PM, Kevin Gordon kgordon...@gmail.com wrote: On Sat, May 5, 2012 at 6:26 AM, Ajay Garg ajaygargn...@gmail.com wrote: Well, I tried once again the Case 2; and upon resume-from-suspend, the icons appeared fine at my end. So, the problem was at my end. So, it seems that nothing needs to be done right now. If I face the unusual result for Case 2 again, I will post the /var/log/messages, which may be scrutinized. If someone else faces the same, she/he may do the same. Till that time, I guess nothing needs to be done. Ajay: Since you've already installed the patch, could I be lazy and ask you a favour? In the scenario where you have booted with mesh disabled, could you go over to the Sugar control/panel settings Network applet, turn the radio off, discard network history, turn the radio back on, wait a bit, and see what icons appear in the neighbourhood view? Curious to see what happens here. I'll do some more testing using this kernel/patch on both gnome and sugar next week, but thought if you had a few seconds I could get a 'head start' at looking at this situation. If not, no big deal - what's done is already really neat! Thanks. Cheers, KG Sorry Sascha, and everyone. Regards, Ajay On Sat, May 5, 2012 at 2:15 PM, Martin Abente martin.abente.lah...@gmail.com wrote: just in case, did you try this combination: libertas(without the patch) + disable_mesh.sh(removed) + resume-from-suspend (?) IF NM still crashes then we have been looking in the wrong direction, if not then we need to look deeper into the patch probably (@silbe ;)). On Fri, May 4, 2012 at 10:20 PM, Ajay Garg ajaygargn...@gmail.com wrote: On Sat, May 5, 2012 at 3:42 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Did you remove the disable mesh.script for the testing? Yes. Both from :: a) my custom added in '/etc/init.d/NetworkManager'. b) '/etc/powed/postresume.d/disable_mesh.sh' Regards, Ajay El may 4, 2012 10:39 p.m., Ajay Garg ajaygargn...@gmail.com escribió: On Sat, May 5, 2012 at 2:03 AM, Sascha Silbe si...@activitycentral.com wrote: Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend /
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Oops.. please ignore my previous mail. Instead, this is what the issue is (seems this issue has put me in mental sleep; I am making stupid, idiotic mistakes) :| Anyways, here it is .. (again, please ignore my previous mail completely) If we do the following :: a) In OpenFirmware CLI, do enable-security my XO-1 serial number b) And there is NOT any /security/develop.sig file in my pendrive while booting. c) There is NOT any /security/develop.sig file on the XO-1, it is observed that :: i) mesh-icons are observed in neighborhood-view, both during reboot and resume-from-suspend. ii) /var/log/messages show the loading of msh0. However, there are no exceptions/crash-traces therein. iii) Also, I see the presence of a folder /sys/class/net/msh0. PLEASE NOTE THAT IF ANY OF a), b) OR c) CONDITIONS ARE NOT MET, THE MESH - ICONS ARE NOT OBSERVED (AS IS VERY MUCH EXPECTED). Is my observation in line with anyone's observation on olpc/sugar? (Note that the issue was reported by UY deployment; and it could be reproduced at our end as well). Kindly reply with comments. Thanks and Regards, Ajay On Thu, May 10, 2012 at 10:53 PM, Ajay Garg a...@activitycentral.comwrote: Hi all. (Unfortunately) A new use-case has come into effect :: If I boot with /security/develop.sig folder in my pendrive, a) mesh-icons are observed in neighborhood-view, both during reboot and resume-from-suspend. b) /var/log/messages show the loading of msh0. However, there are no exceptions/crash-traces therein. c) Also, I see the presence of a folder /sys/class/net/msh0. However, if there is no security/develop.sig file in my pendrive, a) No mesh-icons are seen in neighborhood-view, during reboot or resume-from-suspend (as expected). b) /var/log/messages has no mention of msh0 (as expected). c) There is no folder /sys/class/net/msh0 (i guess that is what is expected). Is my observation in line with anyone's observation on olpc/sugar? (Note that the issue was reported by UY deployment; and it could be reproduced at our end as well). Kindly reply with comments. Thanks and Regards, Ajay On Sat, May 5, 2012 at 11:27 PM, Ajay Garg ajaygargn...@gmail.com wrote: Hi Kevin. Following are the steps and observations :: a) Booted with mesh-disabled (via options libertas libertas_disablemesh=1). b) In Neighborhood-View, wifi and three adhoc network icons were present. c) Then, I turned wifi-radio off, discarded network history. Now, only three adhoc network icons were present in Neighborhood-View. d) After some time, turned wifi-radio on. Now, wifi and three adhoc network icons could be seen. e) Again, I turned wifi-radio off, discarded network history. Again, only three adhoc network icons could be seen in Neighborhood-View. f) Rebooted. g) Only three adhoc network icons were seen in Neighborhood-View. h) After some time, turned wifi-radio on. Now, wifi and three adhoc network icons could be seen in Neighborhood-View. So, it works fine I guess (note that, mesh-icons were NOT seen anytime in the above steps) :) Regards, Ajay On Sat, May 5, 2012 at 6:43 PM, Kevin Gordon kgordon...@gmail.com wrote: On Sat, May 5, 2012 at 6:26 AM, Ajay Garg ajaygargn...@gmail.com wrote: Well, I tried once again the Case 2; and upon resume-from-suspend, the icons appeared fine at my end. So, the problem was at my end. So, it seems that nothing needs to be done right now. If I face the unusual result for Case 2 again, I will post the /var/log/messages, which may be scrutinized. If someone else faces the same, she/he may do the same. Till that time, I guess nothing needs to be done. Ajay: Since you've already installed the patch, could I be lazy and ask you a favour? In the scenario where you have booted with mesh disabled, could you go over to the Sugar control/panel settings Network applet, turn the radio off, discard network history, turn the radio back on, wait a bit, and see what icons appear in the neighbourhood view? Curious to see what happens here. I'll do some more testing using this kernel/patch on both gnome and sugar next week, but thought if you had a few seconds I could get a 'head start' at looking at this situation. If not, no big deal - what's done is already really neat! Thanks. Cheers, KG Sorry Sascha, and everyone. Regards, Ajay On Sat, May 5, 2012 at 2:15 PM, Martin Abente martin.abente.lah...@gmail.com wrote: just in case, did you try this combination: libertas(without the patch) + disable_mesh.sh(removed) + resume-from-suspend (?) IF NM still crashes then we have been looking in the wrong direction, if not then we need to look deeper into the patch probably (@silbe ;)). On Fri, May 4, 2012 at 10:20 PM, Ajay Garg ajaygargn...@gmail.com wrote: On Sat, May 5, 2012 at
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Thu, May 10, 2012 at 1:23 PM, Ajay Garg a...@activitycentral.com wrote: If I boot with /security/develop.sig folder in my pendrive, a) mesh-icons are observed in neighborhood-view, both during reboot and resume-from-suspend. Welcome to the initramfs stage of your journey! When the laptop needs activation, it loads a different initramfs that among other things loads the libertas module. You need to get your /etc/modprobe.d/ files into the initramfs. For your vanilla build, look into dracut-modules-olpc. If you're hoping to get this integrated into a build with an alternative initramfs (hint: Ceibal) that build will have a custom version of dracut-modules-olpc. That is the right way. When the laptop is in secure mode, olpc.fth is _ignored_, so no chance to set a kernel cmdline there. An easier alternative might be to check for those flags under /sys, and if they are there, rmmod/modprobe libertas for example in olpc-configure (so during early boot). m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Thu, 2012-05-10 at 23:11 +0530, Ajay Garg wrote: Oops.. please ignore my previous mail. Instead, this is what the issue is (seems this issue has put me in mental sleep; I am making stupid, idiotic mistakes) :| Anyways, here it is .. (again, please ignore my previous mail completely) OK, no problem. If we do the following :: a) In OpenFirmware CLI, do enable-security my XO-1 serial number Now you booing the signed zip files. b) And there is NOT any /security/develop.sig file in my pendrive while booting. c) There is NOT any /security/develop.sig file on the XO-1, Your still booting signed zip files. If you were to boot with your develop.sig present does the disabling of the mesh work? it is observed that :: i) mesh-icons are observed in neighborhood-view, both during reboot and resume-from-suspend. ii) /var/log/messages show the loading of msh0. However, there are no exceptions/crash-traces therein. iii) Also, I see the presence of a folder /sys/class/net/msh0. PLEASE NOTE THAT IF ANY OF a), b) OR c) CONDITIONS ARE NOT MET, THE MESH - ICONS ARE NOT OBSERVED (AS IS VERY MUCH EXPECTED). Is my observation in line with anyone's observation on olpc/sugar? (Note that the issue was reported by UY deployment; and it could be reproduced at our end as well). How are you introducing the libertas_disablemesh=1 option in the unlocked boot? Are you using modprobe.d or passing a boot arguement in /boot/olpc.fth? Jerry ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
just in case, did you try this combination: libertas(without the patch) + disable_mesh.sh(removed) + resume-from-suspend (?) IF NM still crashes then we have been looking in the wrong direction, if not then we need to look deeper into the patch probably (@silbe ;)). On Fri, May 4, 2012 at 10:20 PM, Ajay Garg ajaygargn...@gmail.com wrote: On Sat, May 5, 2012 at 3:42 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Did you remove the disable mesh.script for the testing? Yes. Both from :: a) my custom added in '/etc/init.d/NetworkManager'. b) '/etc/powed/postresume.d/disable_mesh.sh' Regards, Ajay El may 4, 2012 10:39 p.m., Ajay Garg ajaygargn...@gmail.com escribió: On Sat, May 5, 2012 at 2:03 AM, Sascha Silbe si...@activitycentral.com wrote: Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend / resume was triggered (e.g. idle suspend vs. lid or power switch)? I tried 3 times, and it happened every time. (I tried with the lid approach every time, under identical conditions and set of procedures.). I don't see anything in the patch or the module params code that would explain this behaviour. If it's reproducible, I'll have to dive into it and debug a bit... It is reproducible every time (at least at my end). :) Thanks for testing, BTW! My pleasure :) Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ Regards, Ajay ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Well, I tried once again the Case 2; and upon resume-from-suspend, the icons appeared fine at my end. So, the problem was at my end. So, it seems that nothing needs to be done right now. If I face the unusual result for Case 2 again, I will post the /var/log/messages, which may be scrutinized. If someone else faces the same, she/he may do the same. Till that time, I guess nothing needs to be done. Sorry Sascha, and everyone. Regards, Ajay On Sat, May 5, 2012 at 2:15 PM, Martin Abente martin.abente.lah...@gmail.com wrote: just in case, did you try this combination: libertas(without the patch) + disable_mesh.sh(removed) + resume-from-suspend (?) IF NM still crashes then we have been looking in the wrong direction, if not then we need to look deeper into the patch probably (@silbe ;)). On Fri, May 4, 2012 at 10:20 PM, Ajay Garg ajaygargn...@gmail.com wrote: On Sat, May 5, 2012 at 3:42 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Did you remove the disable mesh.script for the testing? Yes. Both from :: a) my custom added in '/etc/init.d/NetworkManager'. b) '/etc/powed/postresume.d/disable_mesh.sh' Regards, Ajay El may 4, 2012 10:39 p.m., Ajay Garg ajaygargn...@gmail.com escribió: On Sat, May 5, 2012 at 2:03 AM, Sascha Silbe si...@activitycentral.com wrote: Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend / resume was triggered (e.g. idle suspend vs. lid or power switch)? I tried 3 times, and it happened every time. (I tried with the lid approach every time, under identical conditions and set of procedures.). I don't see anything in the patch or the module params code that would explain this behaviour. If it's reproducible, I'll have to dive into it and debug a bit... It is reproducible every time (at least at my end). :) Thanks for testing, BTW! My pleasure :) Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ Regards, Ajay ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Sat, May 5, 2012 at 6:26 AM, Ajay Garg ajaygargn...@gmail.com wrote: Well, I tried once again the Case 2; and upon resume-from-suspend, the icons appeared fine at my end. So, the problem was at my end. So, it seems that nothing needs to be done right now. If I face the unusual result for Case 2 again, I will post the /var/log/messages, which may be scrutinized. If someone else faces the same, she/he may do the same. Till that time, I guess nothing needs to be done. Ajay: Since you've already installed the patch, could I be lazy and ask you a favour? In the scenario where you have booted with mesh disabled, could you go over to the Sugar control/panel settings Network applet, turn the radio off, discard network history, turn the radio back on, wait a bit, and see what icons appear in the neighbourhood view? Curious to see what happens here. I'll do some more testing using this kernel/patch on both gnome and sugar next week, but thought if you had a few seconds I could get a 'head start' at looking at this situation. If not, no big deal - what's done is already really neat! Thanks. Cheers, KG Sorry Sascha, and everyone. Regards, Ajay On Sat, May 5, 2012 at 2:15 PM, Martin Abente martin.abente.lah...@gmail.com wrote: just in case, did you try this combination: libertas(without the patch) + disable_mesh.sh(removed) + resume-from-suspend (?) IF NM still crashes then we have been looking in the wrong direction, if not then we need to look deeper into the patch probably (@silbe ;)). On Fri, May 4, 2012 at 10:20 PM, Ajay Garg ajaygargn...@gmail.com wrote: On Sat, May 5, 2012 at 3:42 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Did you remove the disable mesh.script for the testing? Yes. Both from :: a) my custom added in '/etc/init.d/NetworkManager'. b) '/etc/powed/postresume.d/disable_mesh.sh' Regards, Ajay El may 4, 2012 10:39 p.m., Ajay Garg ajaygargn...@gmail.com escribió: On Sat, May 5, 2012 at 2:03 AM, Sascha Silbe si...@activitycentral.com wrote: Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend / resume was triggered (e.g. idle suspend vs. lid or power switch)? I tried 3 times, and it happened every time. (I tried with the lid approach every time, under identical conditions and set of procedures.). I don't see anything in the patch or the module params code that would explain this behaviour. If it's reproducible, I'll have to dive into it and debug a bit... It is reproducible every time (at least at my end). :) Thanks for testing, BTW! My pleasure :) Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ Regards, Ajay ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Fri, May 4, 2012 at 4:41 PM, Paul Fox p...@laptop.org wrote: sascha wrote: i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586.rpm Thanks! Just to be sure: as it's been merged into olpc-2.6.35, all future official 2.6.35 based OLPC kernel builds will include this patch? hi sascha -- yes. i think we hope there won't actually be any more of those, but if there are, that patch will be there. current and future releases get the patch for free, since it's upstream. (thank you) Just because I like to inquire on what to many is the obvious :-) ... this change is *not* upstream for the 12.1.0 XO1.0 current and future kernels though, correct? KG paul p.s. somehow the git hash i pasted above is incorrect. the correct cherry-pick was this one: - commit 6bdbdbf4a151a3a1333818cd17a7d7795e936041 Author: Sascha Silbe si...@activitycentral.com Date: Wed May 11 14:52:34 2011 +0200 libertas: Add libertas_disablemesh module parameter to disable mesh interface This allows individual users and deployments to disable mesh support at runtime, i.e. without having to build and maintain a custom kernel. Based on a patch by Paul Fox p...@laptop.org. Signed-off-by: Sascha Silbe si...@activitycentral.com Signed-off-by: John W. Linville linvi...@tuxdriver.com - The reason we've not gone the module parameter route so far (in Dextrose 3) is that we didn't want to divert from upstream (OLPC in this case) on the kernel level. If it's included now, that concern is addressed and we can go this route, which IMO is technically the best option. It avoids all possible race conditions and only needs a single configuration file to be set up. Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
kevin wrote: On Fri, May 4, 2012 at 4:41 PM, Paul Fox p...@laptop.org wrote: sascha wrote: i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde 819f.i586.rpm Thanks! Just to be sure: as it's been merged into olpc-2.6.35, all future official 2.6.35 based OLPC kernel builds will include this patch? hi sascha -- yes. i think we hope there won't actually be any more of those, but if there are, that patch will be there. current and future releases get the patch for free, since it's upstream. (thank you) Just because I like to inquire on what to many is the obvious :-) ... this change is *not* upstream for the 12.1.0 XO1.0 current and future kernels though, correct? yes, it is. the change is already upstream -- it just never made it into the olpc-2.6.35 branch, so it wasn't getting into dextrose. paul KG paul p.s. somehow the git hash i pasted above is incorrect. the correct cherry-pick was this one: - commit 6bdbdbf4a151a3a1333818cd17a7d7795e936041 Author: Sascha Silbe si...@activitycentral.com Date: Wed May 11 14:52:34 2011 +0200 libertas: Add libertas_disablemesh module parameter to disable mesh interface This allows individual users and deployments to disable mesh support at runtime, i.e. without having to build and maintain a custom kernel. Based on a patch by Paul Fox p...@laptop.org. Signed-off-by: Sascha Silbe si...@activitycentral.com Signed-off-by: John W. Linville linvi...@tuxdriver.com - The reason we've not gone the module parameter route so far (in Dextrose 3) is that we didn't want to divert from upstream (OLPC in this case) on the kernel level. If it's included now, that concern is addressed and we can go this route, which IMO is technically the best option. It avoids all possible race conditions and only needs a single configuration file to be set up. Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Hi Kevin. Following are the steps and observations :: a) Booted with mesh-disabled (via options libertas libertas_disablemesh=1). b) In Neighborhood-View, wifi and three adhoc network icons were present. c) Then, I turned wifi-radio off, discarded network history. Now, only three adhoc network icons were present in Neighborhood-View. d) After some time, turned wifi-radio on. Now, wifi and three adhoc network icons could be seen. e) Again, I turned wifi-radio off, discarded network history. Again, only three adhoc network icons could be seen in Neighborhood-View. f) Rebooted. g) Only three adhoc network icons were seen in Neighborhood-View. h) After some time, turned wifi-radio on. Now, wifi and three adhoc network icons could be seen in Neighborhood-View. So, it works fine I guess (note that, mesh-icons were NOT seen anytime in the above steps) :) Regards, Ajay On Sat, May 5, 2012 at 6:43 PM, Kevin Gordon kgordon...@gmail.com wrote: On Sat, May 5, 2012 at 6:26 AM, Ajay Garg ajaygargn...@gmail.com wrote: Well, I tried once again the Case 2; and upon resume-from-suspend, the icons appeared fine at my end. So, the problem was at my end. So, it seems that nothing needs to be done right now. If I face the unusual result for Case 2 again, I will post the /var/log/messages, which may be scrutinized. If someone else faces the same, she/he may do the same. Till that time, I guess nothing needs to be done. Ajay: Since you've already installed the patch, could I be lazy and ask you a favour? In the scenario where you have booted with mesh disabled, could you go over to the Sugar control/panel settings Network applet, turn the radio off, discard network history, turn the radio back on, wait a bit, and see what icons appear in the neighbourhood view? Curious to see what happens here. I'll do some more testing using this kernel/patch on both gnome and sugar next week, but thought if you had a few seconds I could get a 'head start' at looking at this situation. If not, no big deal - what's done is already really neat! Thanks. Cheers, KG Sorry Sascha, and everyone. Regards, Ajay On Sat, May 5, 2012 at 2:15 PM, Martin Abente martin.abente.lah...@gmail.com wrote: just in case, did you try this combination: libertas(without the patch) + disable_mesh.sh(removed) + resume-from-suspend (?) IF NM still crashes then we have been looking in the wrong direction, if not then we need to look deeper into the patch probably (@silbe ;)). On Fri, May 4, 2012 at 10:20 PM, Ajay Garg ajaygargn...@gmail.com wrote: On Sat, May 5, 2012 at 3:42 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Did you remove the disable mesh.script for the testing? Yes. Both from :: a) my custom added in '/etc/init.d/NetworkManager'. b) '/etc/powed/postresume.d/disable_mesh.sh' Regards, Ajay El may 4, 2012 10:39 p.m., Ajay Garg ajaygargn...@gmail.com escribió: On Sat, May 5, 2012 at 2:03 AM, Sascha Silbe si...@activitycentral.com wrote: Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend / resume was triggered (e.g. idle suspend vs. lid or power switch)? I tried 3 times, and it happened every time. (I tried with the lid approach every time, under identical conditions and set of procedures.). I don't see anything in the patch or the module params code that would explain this behaviour. If it's reproducible, I'll have to dive into it and debug a bit... It is reproducible every time (at least at my end). :) Thanks for testing, BTW! My pleasure :) Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ Regards, Ajay ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Paul Fox p...@laptop.org writes: i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586.rpm Thanks! Just to be sure: as it's been merged into olpc-2.6.35, all future official 2.6.35 based OLPC kernel builds will include this patch? The reason we've not gone the module parameter route so far (in Dextrose 3) is that we didn't want to divert from upstream (OLPC in this case) on the kernel level. If it's included now, that concern is addressed and we can go this route, which IMO is technically the best option. It avoids all possible race conditions and only needs a single configuration file to be set up. Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ pgpktCWgQvo8W.pgp Description: PGP signature ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend / resume was triggered (e.g. idle suspend vs. lid or power switch)? I don't see anything in the patch or the module params code that would explain this behaviour. If it's reproducible, I'll have to dive into it and debug a bit... Thanks for testing, BTW! Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ pgpji401xnu3W.pgp Description: PGP signature ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Sat, May 5, 2012 at 2:03 AM, Sascha Silbe si...@activitycentral.com wrote: Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend / resume was triggered (e.g. idle suspend vs. lid or power switch)? I tried 3 times, and it happened every time. (I tried with the lid approach every time, under identical conditions and set of procedures.). I don't see anything in the patch or the module params code that would explain this behaviour. If it's reproducible, I'll have to dive into it and debug a bit... It is reproducible every time (at least at my end). :) Thanks for testing, BTW! My pleasure :) Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ Regards, Ajay ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
sascha wrote: i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586.rpm Thanks! Just to be sure: as it's been merged into olpc-2.6.35, all future official 2.6.35 based OLPC kernel builds will include this patch? hi sascha -- yes. i think we hope there won't actually be any more of those, but if there are, that patch will be there. current and future releases get the patch for free, since it's upstream. (thank you) paul p.s. somehow the git hash i pasted above is incorrect. the correct cherry-pick was this one: - commit 6bdbdbf4a151a3a1333818cd17a7d7795e936041 Author: Sascha Silbe si...@activitycentral.com Date: Wed May 11 14:52:34 2011 +0200 libertas: Add libertas_disablemesh module parameter to disable mesh interface This allows individual users and deployments to disable mesh support at runtime, i.e. without having to build and maintain a custom kernel. Based on a patch by Paul Fox p...@laptop.org. Signed-off-by: Sascha Silbe si...@activitycentral.com Signed-off-by: John W. Linville linvi...@tuxdriver.com - The reason we've not gone the module parameter route so far (in Dextrose 3) is that we didn't want to divert from upstream (OLPC in this case) on the kernel level. If it's included now, that concern is addressed and we can go this route, which IMO is technically the best option. It avoids all possible race conditions and only needs a single configuration file to be set up. Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Did you remove the disable mesh.script for the testing? El may 4, 2012 10:39 p.m., Ajay Garg ajaygargn...@gmail.com escribió: On Sat, May 5, 2012 at 2:03 AM, Sascha Silbe si...@activitycentral.com wrote: Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend / resume was triggered (e.g. idle suspend vs. lid or power switch)? I tried 3 times, and it happened every time. (I tried with the lid approach every time, under identical conditions and set of procedures.). I don't see anything in the patch or the module params code that would explain this behaviour. If it's reproducible, I'll have to dive into it and debug a bit... It is reproducible every time (at least at my end). :) Thanks for testing, BTW! My pleasure :) Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ Regards, Ajay ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Sat, May 5, 2012 at 3:42 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Did you remove the disable mesh.script for the testing? Yes. Both from :: a) my custom added in '/etc/init.d/NetworkManager'. b) '/etc/powed/postresume.d/disable_mesh.sh' Regards, Ajay El may 4, 2012 10:39 p.m., Ajay Garg ajaygargn...@gmail.com escribió: On Sat, May 5, 2012 at 2:03 AM, Sascha Silbe si...@activitycentral.com wrote: Ajay Garg ajaygargn...@gmail.com writes: [...] b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 [...] f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. Interesting. g) Observations e) and f) were observed, _every single time_. OK, I suppose you repeated this often enough and using exactly the same environment and procedures as the other test cases? I.e. you are sure that it's really specific to libertas_disablemesh=0 rather than just occurring at random even without the libertas_disablemesh setting or based on how the suspend / resume was triggered (e.g. idle suspend vs. lid or power switch)? I tried 3 times, and it happened every time. (I tried with the lid approach every time, under identical conditions and set of procedures.). I don't see anything in the patch or the module params code that would explain this behaviour. If it's reproducible, I'll have to dive into it and debug a bit... It is reproducible every time (at least at my end). :) Thanks for testing, BTW! My pleasure :) Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ Regards, Ajay ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Thu, May 03, 2012 at 10:58:42AM +0530, Ajay Garg wrote: == JUST ONE LAST QUERY == Is the disable-mesh-patch the only difference between the following :: kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586 (kernel generated by you) kernel-2.6.35.13_xo1-20111005.1403.olpc.c2bd7b9.i586 (original kernel present on XO-1) The seven characters between olpc. and .i586 are the git hash for the olpc-2.6.35 branch in the git repository git://dev.laptop.org/olpc-kernel These tell you which git hash was used to build the kernel. The hash c2bd7b9 is two patches away from bde819f in the branch log. [1] One patch [3] is what Paul did for you. The other patch [2] is XO-1.5 specific. So for your purposes, yes, it is the only difference. You can see a list of previous official kernels. [4] References: 1. http://dev.laptop.org/git/olpc-kernel/?h=olpc-2.6.35 2. http://dev.laptop.org/git/olpc-kernel/commit/?h=olpc-2.6.35id=5d76efb5df8c6b1d181c47b17b1d1e9a39ef66b6 ov7670: Disable non-YUV modes on XO-1.5 (#11297) Non YUV modes cannot be scaled by the via-camera driver without messing up the colours in the image. 3. http://dev.laptop.org/git/olpc-kernel/commit/?h=olpc-2.6.35id=bde819fe60fdc8ca7c7d6e0552105d0438cc7f89 libertas: Add libertas_disablemesh module parameter to disable mesh interfaceolpc-2.6.35 This allows individual users and deployments to disable mesh support at runtime, i.e. without having to build and maintain a custom kernel. 4. http://dev.laptop.org/~kernels/f14-xo1/?C=M;O=D -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 12:13 AM, Ajay Garg ajaygargn...@gmail.com wrote: Just that, we do not wish to set up a mesh-network, as per say. I think you are missing the background reason here. AFAIK, reasons to disable mesh network on XO-1: - Mesh can easily saturate RF, so dense usage scenarios (schools!) benefit from switching it off. - Easier out-of-the-box interop with later XO-1.x models, where the under a tree network uses ad-hoc 802.11g. By removing all references to the device, we could provide a soft solution Nah, messing with Sugar code is the wrong solution. m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 11:49 AM, Martin Langhoff martin.langh...@gmail.comwrote: On Wed, May 2, 2012 at 12:13 AM, Ajay Garg ajaygargn...@gmail.com wrote: Just that, we do not wish to set up a mesh-network, as per say. I think you are missing the background reason here. AFAIK, reasons to disable mesh network on XO-1: - Mesh can easily saturate RF, so dense usage scenarios (schools!) benefit from switching it off. - Easier out-of-the-box interop with later XO-1.x models, where the under a tree network uses ad-hoc 802.11g. By removing all references to the device, we could provide a soft solution Nah, messing with Sugar code is the wrong solution. I agree. But there is no working lower-level solution :\ Regards, Ajay m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 11:49 AM, Martin Langhoff martin.langh...@gmail.com wrote: On Wed, May 2, 2012 at 12:13 AM, Ajay Garg ajaygargn...@gmail.com wrote: Just that, we do not wish to set up a mesh-network, as per say. I think you are missing the background reason here. AFAIK, reasons to disable mesh network on XO-1: - Mesh can easily saturate RF, so dense usage scenarios (schools!) benefit from switching it off. - Easier out-of-the-box interop with later XO-1.x models, where the under a tree network uses ad-hoc 802.11g. +1, this was the reason By removing all references to the device, we could provide a soft solution Nah, messing with Sugar code is the wrong solution. m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 2:25 AM, Ajay Garg ajaygargn...@gmail.com wrote: I agree. But there is no working lower-level solution :\ Let's fix that. Messing with Sugar won't help you. Earlier in the thread someone pointed to you the scripts to trigger on resume (by powerd). Do those work? Not work? What's the problem there? AIUI, if you - disable mesh on boot - disable mesh on resume, from a powerd-triggered script - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Thanks Martin (a ton !!) On Wed, May 2, 2012 at 12:32 PM, Martin Langhoff martin.langh...@gmail.comwrote: On Wed, May 2, 2012 at 2:25 AM, Ajay Garg ajaygargn...@gmail.com wrote: I agree. But there is no working lower-level solution :\ Let's fix that. Great !!! Messing with Sugar won't help you. Earlier in the thread someone pointed to you the scripts to trigger on resume (by powerd). Do those work? Not work? What's the problem there? The /etc/powerd/postresume.d/disable_mesh.sh doesn't work. My belief is that there is the same problem - race condition between the 'echo 0' script, and NetworkManager. This causes the NetworkManager to crash randomly. AIUI, if you - disable mesh on boot Done. Added the 'echo 0' script in 'start()' method of NetworkManager, so that the effect takes place before NetworkManager starts up. Works like a charm. - disable mesh on resume, from a powerd-triggered script Does not work, as explained above. - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. Would be awesome. I believe this is the one and only complete solution possible :) Could you point me to the suitable (examples) link? I will be heartfully grateful. cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff Regards, Ajay ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 3:22 AM, Ajay Garg ajaygargn...@gmail.com wrote: The /etc/powerd/postresume.d/disable_mesh.sh doesn't work. So we need to understand why it does not work. Is it a race condition? Perhaps it is better fixed as a udev script -- triggering when the device appears. The step that disable_mesh performs is very important. If you don't disable it at that level, you haven't disabled mesh at all and all the problems persist. - disable mesh on boot Done. Added the 'echo 0' script in 'start()' method of NetworkManager, so that the effect takes place before NetworkManager starts up. Works like a charm. Ok. Maybe a udev script is a better place. - disable mesh on resume, from a powerd-triggered script Does not work, as explained above. Right but you MUST find a way to make it work. - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. Would be awesome. I believe this is the one and only complete solution possible :) Careful here! This only hides the device from NM so you don't race with NM. Could you point me to the suitable (examples) link? I will be heartfully grateful. look at the latest olpc-configure in git://dev.laptop.org/projects/olpc-utils m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
I think (guessing by the responses) the original problem here is that, if you disable the mesh AFTER NM has taken the device, NM crashes. This a regression bug, considering that this didn't happened in fedora-11 based builds. So the solution here is to find another place to place the script, where it guarantee that the script will be executed before NM does its job at resume time. Another solution is to find out why NM crashes now and why didn't before, but I wouldn't go that way. Cheers, On Wed, May 2, 2012 at 3:28 AM, Martin Langhoff martin.langh...@gmail.comwrote: On Wed, May 2, 2012 at 3:22 AM, Ajay Garg ajaygargn...@gmail.com wrote: The /etc/powerd/postresume.d/disable_mesh.sh doesn't work. So we need to understand why it does not work. Is it a race condition? Perhaps it is better fixed as a udev script -- triggering when the device appears. The step that disable_mesh performs is very important. If you don't disable it at that level, you haven't disabled mesh at all and all the problems persist. - disable mesh on boot Done. Added the 'echo 0' script in 'start()' method of NetworkManager, so that the effect takes place before NetworkManager starts up. Works like a charm. Ok. Maybe a udev script is a better place. - disable mesh on resume, from a powerd-triggered script Does not work, as explained above. Right but you MUST find a way to make it work. - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. Would be awesome. I believe this is the one and only complete solution possible :) Careful here! This only hides the device from NM so you don't race with NM. Could you point me to the suitable (examples) link? I will be heartfully grateful. look at the latest olpc-configure in git:// dev.laptop.org/projects/olpc-utils m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 4:07 AM, Martin Abente martin.abente.lah...@gmail.com wrote: I think (guessing by the responses) the original problem here is that, if you disable the mesh AFTER NM has taken the device, NM crashes. This a regression bug, considering that this didn't happened in fedora-11 based builds. The timings in F11 builds were completely different. Maybe you were winning the race that you're now losing. So the solution here is to find another place to place the script, where it guarantee that the script will be executed before NM does its job at resume time. udev script :-) -- I am pretty sure you can number yourself lower (to run earlier) than the udev script that fires off the new device event to NM. Another solution is to find out why NM crashes now and why didn't before, but I wouldn't go that way. Making NM completely resilient to these race conditions is probably a hard task. cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 10:42 AM, Martin Langhoff martin.langh...@gmail.com wrote: On Wed, May 2, 2012 at 4:07 AM, Martin Abente martin.abente.lah...@gmail.com wrote: I think (guessing by the responses) the original problem here is that, if you disable the mesh AFTER NM has taken the device, NM crashes. This a regression bug, considering that this didn't happened in fedora-11 based builds. The timings in F11 builds were completely different. Maybe you were winning the race that you're now losing. So the solution here is to find another place to place the script, where it guarantee that the script will be executed before NM does its job at resume time. udev script :-) -- I am pretty sure you can number yourself lower (to run earlier) than the udev script that fires off the new device event to NM. Another solution is to find out why NM crashes now and why didn't before, but I wouldn't go that way. Making NM completely resilient to these race conditions is probably a hard task. This is also a temporary solution. There is a kernel patch in 3.1 and greater kernels that allows you to disable mesh as a kernel module parameter. I just played around with the udev script and there definitely seems to be some timing issues even with that. -Jon ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
How hard and sensible do you think it could be to backport that patch? :D (Assuming that touching the kernel is an option for someone, hehe) On Wed, May 2, 2012 at 5:21 AM, Jon Nettleton jon.nettle...@gmail.comwrote: On Wed, May 2, 2012 at 10:42 AM, Martin Langhoff martin.langh...@gmail.com wrote: On Wed, May 2, 2012 at 4:07 AM, Martin Abente martin.abente.lah...@gmail.com wrote: I think (guessing by the responses) the original problem here is that, if you disable the mesh AFTER NM has taken the device, NM crashes. This a regression bug, considering that this didn't happened in fedora-11 based builds. The timings in F11 builds were completely different. Maybe you were winning the race that you're now losing. So the solution here is to find another place to place the script, where it guarantee that the script will be executed before NM does its job at resume time. udev script :-) -- I am pretty sure you can number yourself lower (to run earlier) than the udev script that fires off the new device event to NM. Another solution is to find out why NM crashes now and why didn't before, but I wouldn't go that way. Making NM completely resilient to these race conditions is probably a hard task. This is also a temporary solution. There is a kernel patch in 3.1 and greater kernels that allows you to disable mesh as a kernel module parameter. I just played around with the udev script and there definitely seems to be some timing issues even with that. -Jon ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Martin, just one small query : In one of the earlier mails, it was said that the mac address for msh0 is the same as eth0. So, would blacklisting the (same) device, be feasible? I mean, would that not _also_ disable general wifi network detections? Regards, Ajay On Wed, May 2, 2012 at 12:58 PM, Martin Langhoff martin.langh...@gmail.comwrote: On Wed, May 2, 2012 at 3:22 AM, Ajay Garg ajaygargn...@gmail.com wrote: The /etc/powerd/postresume.d/disable_mesh.sh doesn't work. So we need to understand why it does not work. Is it a race condition? Perhaps it is better fixed as a udev script -- triggering when the device appears. The step that disable_mesh performs is very important. If you don't disable it at that level, you haven't disabled mesh at all and all the problems persist. - disable mesh on boot Done. Added the 'echo 0' script in 'start()' method of NetworkManager, so that the effect takes place before NetworkManager starts up. Works like a charm. Ok. Maybe a udev script is a better place. - disable mesh on resume, from a powerd-triggered script Does not work, as explained above. Right but you MUST find a way to make it work. - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. Would be awesome. I believe this is the one and only complete solution possible :) Careful here! This only hides the device from NM so you don't race with NM. Could you point me to the suitable (examples) link? I will be heartfully grateful. look at the latest olpc-configure in git:// dev.laptop.org/projects/olpc-utils m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 5:32 AM, Ajay Garg ajaygargn...@gmail.com wrote: In one of the earlier mails, it was said that the mac address for msh0 is the same as eth0. Ah, sorry, you're correct, that won't help. So your options are - a kernel module parameter, as Jon proposes, in modprobe.d/ or in the boot commandline - udev rule / script that triggers before the event is sent to NM m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
martin wrote: On Wed, May 2, 2012 at 3:22 AM, Ajay Garg ajaygargn...@gmail.com wrote: The /etc/powerd/postresume.d/disable_mesh.sh doesn't work. So we need to understand why it does not work. Is it a race condition? Perhaps it is better fixed as a udev script -- triggering when the device appears. i think it's almost a guaranteed race. that script essentially does this: while [ ! -f /sys/class/net/eth0/lbs_mesh ] do sleep 0.5 done echo 0 /sys/class/net/eth0/lbs_mesh in other words -- the disable_mesh script will discover the disable node at just about exactly the time that NM discovers the interface. there's also the lbs_disablemesh module parameter, which could be supplied at initial module load. does that not work, for some reason? (i seem to recall there may be a problem with it.) paul The step that disable_mesh performs is very important. If you don't disable it at that level, you haven't disabled mesh at all and all the problems persist. - disable mesh on boot Done. Added the 'echo 0' script in 'start()' method of NetworkManager, so that the effect takes place before NetworkManager starts up. Works like a charm. Ok. Maybe a udev script is a better place. - disable mesh on resume, from a powerd-triggered script Does not work, as explained above. Right but you MUST find a way to make it work. - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. Would be awesome. I believe this is the one and only complete solution possible :) Careful here! This only hides the device from NM so you don't race with NM. Could you point me to the suitable (examples) link? I will be heartfully grateful. look at the latest olpc-configure in git://dev.laptop.org/projects/olpc-utils m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Good News. I managed to get this working (albeit via changes in sugar). The details are at :: http://git.sugarlabs.org/dextrose/mainline/commit/4ac1a5300f4c43608b0f009a23d966d404a15632 Regards, Ajay On Wed, May 2, 2012 at 6:02 PM, Paul Fox p...@laptop.org wrote: martin wrote: On Wed, May 2, 2012 at 3:22 AM, Ajay Garg ajaygargn...@gmail.com wrote: The /etc/powerd/postresume.d/disable_mesh.sh doesn't work. So we need to understand why it does not work. Is it a race condition? Perhaps it is better fixed as a udev script -- triggering when the device appears. i think it's almost a guaranteed race. that script essentially does this: while [ ! -f /sys/class/net/eth0/lbs_mesh ] do sleep 0.5 done echo 0 /sys/class/net/eth0/lbs_mesh in other words -- the disable_mesh script will discover the disable node at just about exactly the time that NM discovers the interface. there's also the lbs_disablemesh module parameter, which could be supplied at initial module load. does that not work, for some reason? (i seem to recall there may be a problem with it.) paul The step that disable_mesh performs is very important. If you don't disable it at that level, you haven't disabled mesh at all and all the problems persist. - disable mesh on boot Done. Added the 'echo 0' script in 'start()' method of NetworkManager, so that the effect takes place before NetworkManager starts up. Works like a charm. Ok. Maybe a udev script is a better place. - disable mesh on resume, from a powerd-triggered script Does not work, as explained above. Right but you MUST find a way to make it work. - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. Would be awesome. I believe this is the one and only complete solution possible :) Careful here! This only hides the device from NM so you don't race with NM. Could you point me to the suitable (examples) link? I will be heartfully grateful. look at the latest olpc-configure in git:// dev.laptop.org/projects/olpc-utils m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 10:18 AM, Ajay Garg ajaygargn...@gmail.com wrote: Good News. I managed to get this working (albeit via changes in sugar). The details are at :: http://git.sugarlabs.org/dextrose/mainline/commit/4ac1a5300f4c43608b0f009a23d966d404a15632 I have a couple of minor observations from the great unwashed - yes me. One is that this solution perhaps wont have any effect when one boots to the Gnome side. Two, this may be build version dependent, as there would appear from some of the discussions in the group that there is an effort for the 12.1.0 builds to more closely link the NM functions on Sugar and Gnome so that settings are identical when restarting the other environment. The echo script in boot startup and the corresponding entry powerd resume , on the other hand, might handle both. Not sure yet, as I'm still playing with WAP variants, and also, I am the newbie :-) Regards, Ajay On Wed, May 2, 2012 at 6:02 PM, Paul Fox p...@laptop.org wrote: martin wrote: On Wed, May 2, 2012 at 3:22 AM, Ajay Garg ajaygargn...@gmail.com wrote: The /etc/powerd/postresume.d/disable_mesh.sh doesn't work. So we need to understand why it does not work. Is it a race condition? Perhaps it is better fixed as a udev script -- triggering when the device appears. i think it's almost a guaranteed race. that script essentially does this: while [ ! -f /sys/class/net/eth0/lbs_mesh ] do sleep 0.5 done echo 0 /sys/class/net/eth0/lbs_mesh in other words -- the disable_mesh script will discover the disable node at just about exactly the time that NM discovers the interface. there's also the lbs_disablemesh module parameter, which could be supplied at initial module load. does that not work, for some reason? (i seem to recall there may be a problem with it.) paul The step that disable_mesh performs is very important. If you don't disable it at that level, you haven't disabled mesh at all and all the problems persist. - disable mesh on boot Done. Added the 'echo 0' script in 'start()' method of NetworkManager, so that the effect takes place before NetworkManager starts up. Works like a charm. Ok. Maybe a udev script is a better place. - disable mesh on resume, from a powerd-triggered script Does not work, as explained above. Right but you MUST find a way to make it work. - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. Would be awesome. I believe this is the one and only complete solution possible :) Careful here! This only hides the device from NM so you don't race with NM. Could you point me to the suitable (examples) link? I will be heartfully grateful. look at the latest olpc-configure in git:// dev.laptop.org/projects/olpc-utils m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 7:57 PM, Kevin Gordon kgordon...@gmail.com wrote: On Wed, May 2, 2012 at 10:18 AM, Ajay Garg ajaygargn...@gmail.com wrote: Good News. I managed to get this working (albeit via changes in sugar). The details are at :: http://git.sugarlabs.org/dextrose/mainline/commit/4ac1a5300f4c43608b0f009a23d966d404a15632 I have a couple of minor observations from the great unwashed - yes me. One is that this solution perhaps wont have any effect when one boots to the Gnome side. What could be the difference therein? (I could change to one, that could work in both) :) Regards, Ajay Two, this may be build version dependent, as there would appear from some of the discussions in the group that there is an effort for the 12.1.0 builds to more closely link the NM functions on Sugar and Gnome so that settings are identical when restarting the other environment. The echo script in boot startup and the corresponding entry powerd resume , on the other hand, might handle both. Not sure yet, as I'm still playing with WAP variants, and also, I am the newbie :-) Regards, Ajay On Wed, May 2, 2012 at 6:02 PM, Paul Fox p...@laptop.org wrote: martin wrote: On Wed, May 2, 2012 at 3:22 AM, Ajay Garg ajaygargn...@gmail.com wrote: The /etc/powerd/postresume.d/disable_mesh.sh doesn't work. So we need to understand why it does not work. Is it a race condition? Perhaps it is better fixed as a udev script -- triggering when the device appears. i think it's almost a guaranteed race. that script essentially does this: while [ ! -f /sys/class/net/eth0/lbs_mesh ] do sleep 0.5 done echo 0 /sys/class/net/eth0/lbs_mesh in other words -- the disable_mesh script will discover the disable node at just about exactly the time that NM discovers the interface. there's also the lbs_disablemesh module parameter, which could be supplied at initial module load. does that not work, for some reason? (i seem to recall there may be a problem with it.) paul The step that disable_mesh performs is very important. If you don't disable it at that level, you haven't disabled mesh at all and all the problems persist. - disable mesh on boot Done. Added the 'echo 0' script in 'start()' method of NetworkManager, so that the effect takes place before NetworkManager starts up. Works like a charm. Ok. Maybe a udev script is a better place. - disable mesh on resume, from a powerd-triggered script Does not work, as explained above. Right but you MUST find a way to make it work. - blacklist the MAC address so NM ignores it you win. Yes, every XO has a different MAC address, but you can read that on first boot of the OS, and write the NM configuration. See the olpc-configure script for examples. Would be awesome. I believe this is the one and only complete solution possible :) Careful here! This only hides the device from NM so you don't race with NM. Could you point me to the suitable (examples) link? I will be heartfully grateful. look at the latest olpc-configure in git:// dev.laptop.org/projects/olpc-utils m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 10:18 AM, Ajay Garg ajaygargn...@gmail.com wrote: Good News. I managed to get this working (albeit via changes in sugar). The details are at :: http://git.sugarlabs.org/dextrose/mainline/commit/4ac1a5300f4c43608b0f009a23d966d404a15632 The patch seems fairly wrong to me. You are hiding the mesh icons in sugar, but the mesh is active. Packet forwarding is still happening. One of the top reasons we stopped using mesh is because it saturates the RF spectrum, which is a bad thing to do when you have many users in a small space (ie: in a school). You had the mesh disable trick working on F11, and (I assume) happy users of that feature. With this, the feature is broken, but you're making the UI look right... cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
It's worth noting that half the battle can be won by overriding the following XO-1 specific line in OLPC OS Builder's kspost.50.xo1-tweaks.inc: gconftool-2 --direct --config-source xml:readwrite:/etc/gconf/gconf.xml.defaults --type bool --set /desktop/sugar/network/adhoc false Setting this to true, or letting it default to such will show the Ad-hoc networks by default on XO-1. It also will cause XO-1's to default to Ad-hoc if no preferred network is found. The mesh networks will still be there for manual use; but right now they seem semi-broken anyway on 12.1.0 as we attempt to connect to Mesh Network 0 and don't set a channel on the Interface. On Wed, May 2, 2012 at 10:29 AM, Martin Langhoff martin.langh...@gmail.comwrote: On Wed, May 2, 2012 at 10:18 AM, Ajay Garg ajaygargn...@gmail.com wrote: Good News. I managed to get this working (albeit via changes in sugar). The details are at :: http://git.sugarlabs.org/dextrose/mainline/commit/4ac1a5300f4c43608b0f009a23d966d404a15632 The patch seems fairly wrong to me. You are hiding the mesh icons in sugar, but the mesh is active. Packet forwarding is still happening. One of the top reasons we stopped using mesh is because it saturates the RF spectrum, which is a bad thing to do when you have many users in a small space (ie: in a school). You had the mesh disable trick working on F11, and (I assume) happy users of that feature. With this, the feature is broken, but you're making the UI look right... cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 7:59 PM, Martin Langhoff martin.langh...@gmail.comwrote: On Wed, May 2, 2012 at 10:18 AM, Ajay Garg ajaygargn...@gmail.com wrote: Good News. I managed to get this working (albeit via changes in sugar). The details are at :: http://git.sugarlabs.org/dextrose/mainline/commit/4ac1a5300f4c43608b0f009a23d966d404a15632 The patch seems fairly wrong to me. You are hiding the mesh icons in sugar, but the mesh is active. Packet forwarding is still happening. Hmm, ok (didn't know this). I believe that the number of packets being forwarded in this, would be (much) less than in the scenario when the users are actually connected to a mesh-network-channel. Kindly affirm/reject my above notion :) Regards, Ajay One of the top reasons we stopped using mesh is because it saturates the RF spectrum, which is a bad thing to do when you have many users in a small space (ie: in a school). You had the mesh disable trick working on F11, and (I assume) happy users of that feature. With this, the feature is broken, but you're making the UI look right... cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 05/02/2012 07:59 PM, Martin Langhoff wrote: On Wed, May 2, 2012 at 10:18 AM, Ajay Garg ajaygargn...@gmail.com wrote: Good News. I managed to get this working (albeit via changes in sugar). The details are at :: http://git.sugarlabs.org/dextrose/mainline/commit/4ac1a5300f4c43608b0f009a23d966d404a15632 The patch seems fairly wrong to me. You are hiding the mesh icons in sugar, but the mesh is active. Packet forwarding is still happening. One of the top reasons we stopped using mesh is because it saturates the RF spectrum, which is a bad thing to do when you have many users in a small space (ie: in a school). You had the mesh disable trick working on F11, and (I assume) happy users of that feature. With this, the feature is broken, but you're making the UI look right... I agree. The *problem* we are trying to solve is not to have a pretty mesh-icon-free-UI (which is a side effect), but disable the mesh at a hardware level. This patch *won't* solve the problem, as it will still flood the air with packet forwarding. - From the discussion, it seems to me that the kernel level switch is present in 12.x.x onwards, but not so for 11.3.x. In 10.3.x(f11) we got lucky as we were able to avoid the race condition. I suggest we keep looking for a proper solution: * Can the kernel fix be backported (might require a lot of work) * Can we tinker with the udev, postresume scripts to have it 'just' working I have reverted the commit: http://git.sugarlabs.org/dextrose/mainline/commit/20ef9a14dd55908aec6c04cf3edddc51004aabb0 cheers, m - -- Anish -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.12 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQEcBAEBAgAGBQJPoU9PAAoJEBoxUdDHDZVp+DAH/j/RUl6AarwSlz0oUGIuPa5b FxpFOwO6edA0Avd4Zv0/0x3FWlaAEHwkkyz6Vcsq3Px0lxecX0JgYrEgaXWJP4l0 YRBqROOBCzkVKxk7dEWZ003igZSGKbSmuRMlj4v4Qpv0yU9Tfi/GS3T1Q+r02B0o igF9XjmLT5lFcZ4U1e7vE/foU4f7Y5ugg/TON6u/Oh0GNF8bDdOSkY/xKhGlAkIf 72d1pSt03ypcQgUHy7mRhORj1rc1d1YCWiyZLv2iUXdR2OyR8qjukw2HjQ2Gu5ts DwTeumIy+QSF7fDIZkE83dErrmLJLUGvQ/4oqT50zhgI63o+/v8qBpa40XRo3bc= =KMS0 -END PGP SIGNATURE- ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Well, could someone point me to the kernel fix, which could solve the problem by backporting. That should be an interesting exercise. Regards, Ajay On Wed, May 2, 2012 at 8:44 PM, Anish Mangal an...@activitycentral.comwrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 05/02/2012 07:59 PM, Martin Langhoff wrote: On Wed, May 2, 2012 at 10:18 AM, Ajay Garg ajaygargn...@gmail.com wrote: Good News. I managed to get this working (albeit via changes in sugar). The details are at :: http://git.sugarlabs.org/dextrose/mainline/commit/4ac1a5300f4c43608b0f009a23d966d404a15632 The patch seems fairly wrong to me. You are hiding the mesh icons in sugar, but the mesh is active. Packet forwarding is still happening. One of the top reasons we stopped using mesh is because it saturates the RF spectrum, which is a bad thing to do when you have many users in a small space (ie: in a school). You had the mesh disable trick working on F11, and (I assume) happy users of that feature. With this, the feature is broken, but you're making the UI look right... I agree. The *problem* we are trying to solve is not to have a pretty mesh-icon-free-UI (which is a side effect), but disable the mesh at a hardware level. This patch *won't* solve the problem, as it will still flood the air with packet forwarding. - From the discussion, it seems to me that the kernel level switch is present in 12.x.x onwards, but not so for 11.3.x. In 10.3.x(f11) we got lucky as we were able to avoid the race condition. I suggest we keep looking for a proper solution: * Can the kernel fix be backported (might require a lot of work) * Can we tinker with the udev, postresume scripts to have it 'just' working I have reverted the commit: http://git.sugarlabs.org/dextrose/mainline/commit/20ef9a14dd55908aec6c04cf3edddc51004aabb0 cheers, m - -- Anish -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.12 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQEcBAEBAgAGBQJPoU9PAAoJEBoxUdDHDZVp+DAH/j/RUl6AarwSlz0oUGIuPa5b FxpFOwO6edA0Avd4Zv0/0x3FWlaAEHwkkyz6Vcsq3Px0lxecX0JgYrEgaXWJP4l0 YRBqROOBCzkVKxk7dEWZ003igZSGKbSmuRMlj4v4Qpv0yU9Tfi/GS3T1Q+r02B0o igF9XjmLT5lFcZ4U1e7vE/foU4f7Y5ugg/TON6u/Oh0GNF8bDdOSkY/xKhGlAkIf 72d1pSt03ypcQgUHy7mRhORj1rc1d1YCWiyZLv2iUXdR2OyR8qjukw2HjQ2Gu5ts DwTeumIy+QSF7fDIZkE83dErrmLJLUGvQ/4oqT50zhgI63o+/v8qBpa40XRo3bc= =KMS0 -END PGP SIGNATURE- ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 11:07 AM, Ajay Garg ajaygargn...@gmail.com wrote: I believe that the number of packets being forwarded in this, would be (much) less than in the scenario when the users are actually connected to a mesh-network-channel. Kindly affirm/reject my above notion :) I am a very pragmatic man, I would not waste your time if it was a maybe. There is no much less packet forwarding. You get 100% packet forwarding. And as Sam points out, the UI part of it can be set already with a gconf setting, via OOB. Unfortunatley, I have to agree with Anish's revert :-( Don't have the kernel patch info. Maybe look in git for changes in the libertas driver. It's a pretty low traffic driver, so you'll find it quick. cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Another comment from the unwashed: Years back, I used only ethernet to connect my XOs. Nowadays, I'm using both wired and wireless to connect between XOs. [By the way, I normally run my XOs with suspend disabled - so I have not paid much attention to problems associated with 'resume'.] For about the last year, Network Manager has become super bossy: It used to be that if I had an XO on ethernet, Network Manager might intervene after some 10+ hours - breaking that ethernet connection. I did not mind occasionally having to reconnect the ethernet (I just invoke a script that re-issues the necessary commands). But the current Network Manager intervenes in less than ten SECONDS - it breaks the connection on the ethernet (eth1) interface - presumably to set up the wireless (eth0) interface (to listen for radio signals). Bottom line - these days, in order to use ethernet on an XO, I need to __disable__ Network Manager. mikus ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Martin, just out of curiosity .. a logical query comes to my mind. Why, and to whom, are packets forwarded, even though no user has joined any channel? Please do not take this as arrogance; I just wish to clear up some logical mind-blocks :D More importantly, this would clear up some of my networking concepts as well :D Thanks in advance for being my teacher :D Thanks and Regards, Ajay On Wed, May 2, 2012 at 9:27 PM, Martin Langhoff martin.langh...@gmail.comwrote: On Wed, May 2, 2012 at 11:07 AM, Ajay Garg ajaygargn...@gmail.com wrote: I believe that the number of packets being forwarded in this, would be (much) less than in the scenario when the users are actually connected to a mesh-network-channel. Kindly affirm/reject my above notion :) I am a very pragmatic man, I would not waste your time if it was a maybe. There is no much less packet forwarding. You get 100% packet forwarding. And as Sam points out, the UI part of it can be set already with a gconf setting, via OOB. Unfortunatley, I have to agree with Anish's revert :-( Don't have the kernel patch info. Maybe look in git for changes in the libertas driver. It's a pretty low traffic driver, so you'll find it quick. cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
martin wrote: On Wed, May 2, 2012 at 11:07 AM, Ajay Garg ajaygargn...@gmail.com wrote: I believe that the number of packets being forwarded in this, would be (much) less than in the scenario when the users are actually connected to a mesh-network-channel. Kindly affirm/reject my above notion :) I am a very pragmatic man, I would not waste your time if it was a maybe. There is no much less packet forwarding. You get 100% packet forwarding. And as Sam points out, the UI part of it can be set already with a gconf setting, via OOB. Unfortunatley, I have to agree with Anish's revert :-( Don't have the kernel patch info. Maybe look in git for changes in the libertas driver. It's a pretty low traffic driver, so you'll find it quick. i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586.rpm paul cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Thanks Paul. I will test this, and get back to you once done. Thanks a ton Regards, Ajay On Thu, May 3, 2012 at 2:21 AM, Paul Fox p...@laptop.org wrote: martin wrote: On Wed, May 2, 2012 at 11:07 AM, Ajay Garg ajaygargn...@gmail.com wrote: I believe that the number of packets being forwarded in this, would be (much) less than in the scenario when the users are actually connected to a mesh-network-channel. Kindly affirm/reject my above notion :) I am a very pragmatic man, I would not waste your time if it was a maybe. There is no much less packet forwarding. You get 100% packet forwarding. And as Sam points out, the UI part of it can be set already with a gconf setting, via OOB. Unfortunatley, I have to agree with Anish's revert :-( Don't have the kernel patch info. Maybe look in git for changes in the libertas driver. It's a pretty low traffic driver, so you'll find it quick. i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586.rpm paul cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Hi all. One generic query (not necessarily related to NM crash during resume-upon-suspend) :: I cannot seem to find any grub.conf on my XO-1, wherein I could add the kernel boot parameter. So, does XO-1 have any alternative to grub.conf ? Regards, Ajay On Thu, May 3, 2012 at 2:24 AM, Ajay Garg ajaygargn...@gmail.com wrote: Thanks Paul. I will test this, and get back to you once done. Thanks a ton Regards, Ajay On Thu, May 3, 2012 at 2:21 AM, Paul Fox p...@laptop.org wrote: martin wrote: On Wed, May 2, 2012 at 11:07 AM, Ajay Garg ajaygargn...@gmail.com wrote: I believe that the number of packets being forwarded in this, would be (much) less than in the scenario when the users are actually connected to a mesh-network-channel. Kindly affirm/reject my above notion :) I am a very pragmatic man, I would not waste your time if it was a maybe. There is no much less packet forwarding. You get 100% packet forwarding. And as Sam points out, the UI part of it can be set already with a gconf setting, via OOB. Unfortunatley, I have to agree with Anish's revert :-( Don't have the kernel patch info. Maybe look in git for changes in the libertas driver. It's a pretty low traffic driver, so you'll find it quick. i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586.rpm paul cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
ajay wrote: Hi all. One generic query (not necessarily related to NM crash during resume-upon-suspend) :: I cannot seem to find any grub.conf on my XO-1, wherein I could add the kernel boot parameter. So, does XO-1 have any alternative to grub.conf ? look at olpc.fth. in /bootpart/boot/olpc.fth. if this is for libertas_disablemesh, i'd suggest adding that in modprobe.conf (or whatever the right file is under modprobe*/* these days). paul Regards, Ajay On Thu, May 3, 2012 at 2:24 AM, Ajay Garg ajaygargn...@gmail.com wrote: Thanks Paul. I will test this, and get back to you once done. Thanks a ton Regards, Ajay On Thu, May 3, 2012 at 2:21 AM, Paul Fox p...@laptop.org wrote: martin wrote: On Wed, May 2, 2012 at 11:07 AM, Ajay Garg ajaygargn...@gmail.com wrote: I believe that the number of packets being forwarded in this, would be (much) less than in the scenario when the users are actually connected to a mesh-network-channel. Kindly affirm/reject my above notion :) I am a very pragmatic man, I would not waste your time if it was a maybe. There is no much less packet forwarding. You get 100% packet forwarding. And as Sam points out, the UI part of it can be set already with a gconf setting, via OOB. Unfortunatley, I have to agree with Anish's revert :-( Don't have the kernel patch info. Maybe look in git for changes in the libertas driver. It's a pretty low traffic driver, so you'll find it quick. i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde 819f.i586.rpm paul cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff =- paul fox, p...@laptop.org =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Paul, here are the test results :: == USE-CASE 1 == a) Created file '/etc/modprobe.d/libertas.conf'. b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=1 c) Ensured that there is no echo 0 /sys/class/net/eth0/lbs_mesh script, anywhere on the XO-1. This also meant, that there was no '/etc/powerd/postresume.d/disable_mesh.sh'. d) Rebooted. e) Wifi icons were visible in Neighborhood-View, but no mesh-icons were visible. f) Upon resume-from-suspend, wifi icons were visible in Neighborhood-View, but no mesh-icons were visible. g) Observations e) and f) were observed, _every single time_. = == USE-CASE 2 == a) Created file '/etc/modprobe.d/libertas.conf'. b) Ensured that '/etc/modprobe.d/libertas.conf' contained only the following line :: options libertas libertas_disablemesh=0 c) Ensured that there is no echo 0 /sys/class/net/eth0/lbs_mesh script, anywhere on the XO-1. This also meant, that there was no '/etc/powerd/postresume.d/disable_mesh.sh'. d) Rebooted. e) Wifi icons, and mesh-icons were visible in Neighborhood-View. f) Upon resume-from-suspend, NO ICONS COULD BE SEEN IN NEIGHBORHOOD VIEW. g) Observations e) and f) were observed, _every single time_. = == USE-CASE 3 == a) Ensured that there was no such file, that contained the following line :: options libertas libertas_disablemesh b) Ensured that there is no echo 0 /sys/class/net/eth0/lbs_mesh script, anywhere on the XO-1. This also meant, that there was no '/etc/powerd/postresume.d/disable_mesh.sh'. c) Rebooted. d) Wifi icons, and mesh-icons were visible in Neighborhood-View. e) Upon resume-from-suspend, wifi-icons and mesh-icons were visible in Neighborhood-View. f) Observations d) and e) were observed, _every single time_. == SUMMARY== Barring use-case 2 (which looks a bit odd), use-cases 1 and 3 worked perfectly as expected. Thanks Paul for your prompt effort in generating the new kernel, with the patch. Thanks again. == JUST ONE LAST QUERY == Is the disable-mesh-patch the only difference between the following :: kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586 (kernel generated by you) kernel-2.6.35.13_xo1-20111005.1403.olpc.c2bd7b9.i586 (original kernel present on XO-1) Thanks again. The issue stands resolved :) The new kernel could be deployed, provided the answer to the (last, only) query is a yes. :) Regards, Ajay On Thu, May 3, 2012 at 2:21 AM, Paul Fox p...@laptop.org wrote: martin wrote: On Wed, May 2, 2012 at 11:07 AM, Ajay Garg ajaygargn...@gmail.com wrote: I believe that the number of packets being forwarded in this, would be (much) less than in the scenario when the users are actually connected to a mesh-network-channel. Kindly affirm/reject my above notion :) I am a very pragmatic man, I would not waste your time if it was a maybe. There is no much less packet forwarding. You get 100% packet forwarding. And as Sam points out, the UI part of it can be set already with a gconf setting, via OOB. Unfortunatley, I have to agree with Anish's revert :-( Don't have the kernel patch info. Maybe look in git for changes in the libertas driver. It's a pretty low traffic driver, so you'll find it quick. i've cherry-picked 65a5f2b3 onto olpc-2.6.35, and the autobuilder did the rest. this implements a new libertas_disablemesh module parameter which should keep mesh from being enabled. please test: http://rpmdropbox.laptop.org/f14-xo1/kernel-2.6.35.13_xo1-20120502.1603.olpc.bde819f.i586.rpm paul cheers, m -- martin.langh...@gmail.com mar...@laptop.org -- Software Architect - OLPC - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Any ideas please, regarding the two latest queries :) ? Regards, Ajay On Mon, Apr 30, 2012 at 1:00 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Martin and Jon for the replies. On Sun, Apr 29, 2012 at 3:04 PM, Jon Nettleton jon.nettle...@gmail.comwrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/postresume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This should be a lot cleaner solution. However, two queries :: a) Doing ifconfig on the XO-1, only shows information for eth0 and lo (no mesh device listed). So, how can the mac address for the mesh device be found? b) Are mac address for XO-1s, EXACTLY same, for every XO-1 on this planet? Looking forward to a reply. Thanks and Regards, Ajay This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
ajay wrote: Any ideas please, regarding the two latest queries :) ? Regards, Ajay On Mon, Apr 30, 2012 at 1:00 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Martin and Jon for the replies. On Sun, Apr 29, 2012 at 3:04 PM, Jon Nettleton jon.nettle...@gmail.comwrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/post resume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This should be a lot cleaner solution. However, two queries :: a) Doing ifconfig on the XO-1, only shows information for eth0 and lo (no mesh device listed). So, how can the mac address for the mesh device be found? it's the same as that for eth0. b) Are mac address for XO-1s, EXACTLY same, for every XO-1 on this planet? of course not. how would they tell one another apart? paul Looking forward to a reply. Thanks and Regards, Ajay This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon part 2 text/plain 129 ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Thanks Paul. On Tue, May 1, 2012 at 8:03 PM, Paul Fox p...@laptop.org wrote: ajay wrote: Any ideas please, regarding the two latest queries :) ? Regards, Ajay On Mon, Apr 30, 2012 at 1:00 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Martin and Jon for the replies. On Sun, Apr 29, 2012 at 3:04 PM, Jon Nettleton jon.nettle...@gmail.comwrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/post resume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This should be a lot cleaner solution. However, two queries :: a) Doing ifconfig on the XO-1, only shows information for eth0 and lo (no mesh device listed). So, how can the mac address for the mesh device be found? it's the same as that for eth0. Does that mean, that banning eth0-mac-address prevent the loading of wifi-hardware-device as well (in obvious addition to mesh) ? This seems very pricky. b) Are mac address for XO-1s, EXACTLY same, for every XO-1 on this planet? of course not. how would they tell one another apart? Alright. But my first doubt (same mac address for mesh-hardware and wifi-hardware) has put me in topspin :~ Regards, Ajay paul Looking forward to a reply. Thanks and Regards, Ajay This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon part 2 text/plain 129 ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
which actually brings me back to my original question :: Why is it so that putting the 'disable-mesh-script' in the 'start()' method of '/etc/init.d/Networkmanager' works (always) for (re)boot; but never works for resume-upon-suspend? Regards, Ajay On Tue, May 1, 2012 at 8:07 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Paul. On Tue, May 1, 2012 at 8:03 PM, Paul Fox p...@laptop.org wrote: ajay wrote: Any ideas please, regarding the two latest queries :) ? Regards, Ajay On Mon, Apr 30, 2012 at 1:00 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Martin and Jon for the replies. On Sun, Apr 29, 2012 at 3:04 PM, Jon Nettleton jon.nettle...@gmail.comwrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/post resume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This should be a lot cleaner solution. However, two queries :: a) Doing ifconfig on the XO-1, only shows information for eth0 and lo (no mesh device listed). So, how can the mac address for the mesh device be found? it's the same as that for eth0. Does that mean, that banning eth0-mac-address prevent the loading of wifi-hardware-device as well (in obvious addition to mesh) ? This seems very pricky. b) Are mac address for XO-1s, EXACTLY same, for every XO-1 on this planet? of course not. how would they tell one another apart? Alright. But my first doubt (same mac address for mesh-hardware and wifi-hardware) has put me in topspin :~ Regards, Ajay paul Looking forward to a reply. Thanks and Regards, Ajay This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon part 2 text/plain 129 ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
ajay wrote: Thanks Paul. On Tue, May 1, 2012 at 8:03 PM, Paul Fox p...@laptop.org wrote: ajay wrote: Any ideas please, regarding the two latest queries :) ? Regards, Ajay On Mon, Apr 30, 2012 at 1:00 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Martin and Jon for the replies. On Sun, Apr 29, 2012 at 3:04 PM, Jon Nettleton jon.nettle...@gmail.comwrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/post resume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This should be a lot cleaner solution. However, two queries :: a) Doing ifconfig on the XO-1, only shows information for eth0 and lo (no mesh device listed). So, how can the mac address for the mesh device be found? it's the same as that for eth0. Does that mean, that banning eth0-mac-address prevent the loading of wifi-hardware-device as well (in obvious addition to mesh) ? This seems very pricky. i don't know. i'm unfamiliar with NM config. it doesn't sound too hard to try. paul b) Are mac address for XO-1s, EXACTLY same, for every XO-1 on this planet? of course not. how would they tell one another apart? Alright. But my first doubt (same mac address for mesh-hardware and wifi-hardware) has put me in topspin :~ Regards, Ajay paul Looking forward to a reply. Thanks and Regards, Ajay This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon part 2 text/plain 129 ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Thanks Paul. I will give it a try myself. Just one last question :: I suppose that 'echo 0 /sys/class/net/eth0/lbs_mesh' is a hack that is olpc-customised. So, I will be really grateful if you could point me to some docs (a wiki page may be), that provide information as to how this hack affects, and is affected. Thanks for your help. Regards, Ajay On Tue, May 1, 2012 at 8:36 PM, Paul Fox p...@laptop.org wrote: ajay wrote: Thanks Paul. On Tue, May 1, 2012 at 8:03 PM, Paul Fox p...@laptop.org wrote: ajay wrote: Any ideas please, regarding the two latest queries :) ? Regards, Ajay On Mon, Apr 30, 2012 at 1:00 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Martin and Jon for the replies. On Sun, Apr 29, 2012 at 3:04 PM, Jon Nettleton jon.nettle...@gmail.comwrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/post resume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This should be a lot cleaner solution. However, two queries :: a) Doing ifconfig on the XO-1, only shows information for eth0 and lo (no mesh device listed). So, how can the mac address for the mesh device be found? it's the same as that for eth0. Does that mean, that banning eth0-mac-address prevent the loading of wifi-hardware-device as well (in obvious addition to mesh) ? This seems very pricky. i don't know. i'm unfamiliar with NM config. it doesn't sound too hard to try. paul b) Are mac address for XO-1s, EXACTLY same, for every XO-1 on this planet? of course not. how would they tell one another apart? Alright. But my first doubt (same mac address for mesh-hardware and wifi-hardware) has put me in topspin :~ Regards, Ajay paul Looking forward to a reply. Thanks and Regards, Ajay This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon part 2 text/plain 129 ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Tue, 2012-05-01 at 20:50 +0530, Ajay Garg wrote: Thanks Paul. I will give it a try myself. Just one last question :: I suppose that 'echo 0 /sys/class/net/eth0/lbs_mesh' is a hack that is olpc-customised. So, I will be really grateful if you could point me to some docs (a wiki page may be), that provide information as to how this hack affects, and is affected. Ajay: http://wiki.laptop.org/go/Mesh_Network_Details http://wiki.laptop.org/go/Wireless_Driver_README Are installing the dextrose-platform rpm? http://download.sugarlabs.org/dextrose/testing/dx3/rpms/source/dextrose-platform-2-2.fc14.src.rpm Jerry Thanks for your help. Regards, Ajay On Tue, May 1, 2012 at 8:36 PM, Paul Fox p...@laptop.org wrote: ajay wrote: Thanks Paul. On Tue, May 1, 2012 at 8:03 PM, Paul Fox p...@laptop.org wrote: ajay wrote: Any ideas please, regarding the two latest queries :) ? Regards, Ajay On Mon, Apr 30, 2012 at 1:00 PM, Ajay Garg a...@activitycentral.com wrote: Thanks Martin and Jon for the replies. On Sun, Apr 29, 2012 at 3:04 PM, Jon Nettleton jon.nettle...@gmail.comwrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/post resume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This should be a lot cleaner solution. However, two queries :: a) Doing ifconfig on the XO-1, only shows information for eth0 and lo (no mesh device listed). So, how can the mac address for the mesh device be found? it's the same as that for eth0. Does that mean, that banning eth0-mac-address prevent the loading of wifi-hardware-device as well (in obvious addition to mesh) ? This seems very pricky. i don't know. i'm unfamiliar with NM config. it doesn't sound too hard to try. paul b) Are mac address for XO-1s, EXACTLY same, for every XO-1 on this planet? of course not. how would they tell one another apart? Alright. But my first doubt (same mac address for mesh-hardware and wifi-hardware) has put me in topspin :~ Regards, Ajay paul Looking forward to a reply. Thanks and Regards, Ajay This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore)
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Tue, May 1, 2012 at 3:55 PM, Ajay Garg a...@activitycentral.com wrote: which actually brings me back to my original question :: Why is it so that putting the 'disable-mesh-script' in the 'start()' method of '/etc/init.d/Networkmanager' works (always) for (re)boot; but never works for resume-upon-suspend? Because on suspend/resume the NetworkManager service isn't started/restarted so the script in /etc/init.d isn't run where as on reboot it is. There's means though to run specific scripts on suspend/resume if that sort of thing is necessary. Peter ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Hi all. I'll ask this straight :: Is there an already tested-cum-recommended method, that could disable mesh-network, both during reboots and resume-from-suspend? Possible options (not tested by me) :: a) A driver, encapsulating the patch at http://lists.infradead.org/pipermail/libertas-dev/2007-June/000448.html (as per http://dev.laptop.org/pub/firmware/libertas/RELEASE_NOTES) Query : How can I ascertain if this patch is present on a particular XO-1? b) (For Reboot) :: Add the bootstrap script 'echo 0 /sys/class/net/eth0/lbs_mesh' in '/etc/init.d/NetworkMaanager'. (For Resume-Upon-Suspend) :: Add the disable_mesh.sh script (at http://download.sugarlabs.org/dextrose/testing/dx3/rpms/source/dextrose-platform-2-2.fc14.src.rpm) in /etc/powerd/postresume.d. Query : While the (For Reboot) case is guaranteed to work stably every single time, can the same be said about (For Resume-Upon-Suspend) ? c) Any other unknown in-the-dark hack. Regards, Ajay On Tue, May 1, 2012 at 9:36 PM, Peter Robinson pbrobin...@gmail.com wrote: On Tue, May 1, 2012 at 3:55 PM, Ajay Garg a...@activitycentral.com wrote: which actually brings me back to my original question :: Why is it so that putting the 'disable-mesh-script' in the 'start()' method of '/etc/init.d/Networkmanager' works (always) for (re)boot; but never works for resume-upon-suspend? Because on suspend/resume the NetworkManager service isn't started/restarted so the script in /etc/init.d isn't run where as on reboot it is. There's means though to run specific scripts on suspend/resume if that sort of thing is necessary. Peter ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Just an updated version :: Hi all. I'll ask this straight :: Is there an already tested-cum-recommended method, that could disable mesh-network, both during reboots and resume-from-suspend? Possible options (not tested by me) :: a) A driver, encapsulating the patch at http://lists.infradead.org/pipermail/libertas-dev/2007-June/000448.html (as per http://dev.laptop.org/pub/firmware/libertas/RELEASE_NOTES) Query : How can I ascertain if this patch is present on a particular XO-1? b) (For Reboot) :: Add the bootstrap script 'echo 0 /sys/class/net/eth0/lbs_mesh' in '/etc/init.d/NetworkMaanager'. (For Resume-Upon-Suspend) :: Add the disable_mesh.sh script (at http://download.sugarlabs.org/dextrose/testing/dx3/rpms/source/dextrose-platform-2-2.fc14.src.rpm) in /etc/powerd/postresume.d. Query : While the (For Reboot) case is guaranteed to work stably every single time, can the same be said about (For Resume-Upon-Suspend) ? Answer : No. The (Resume-Upon-Suspend) case does not work. So, this solution is rejected. c) Any other unknown in-the-dark hack. Regards, Ajay On Wed, May 2, 2012 at 9:10 AM, Ajay Garg a...@activitycentral.com wrote: Hi all. I'll ask this straight :: Is there an already tested-cum-recommended method, that could disable mesh-network, both during reboots and resume-from-suspend? Possible options (not tested by me) :: a) A driver, encapsulating the patch at http://lists.infradead.org/pipermail/libertas-dev/2007-June/000448.html (as per http://dev.laptop.org/pub/firmware/libertas/RELEASE_NOTES) Query : How can I ascertain if this patch is present on a particular XO-1? b) (For Reboot) :: Add the bootstrap script 'echo 0 /sys/class/net/eth0/lbs_mesh' in '/etc/init.d/NetworkMaanager'. (For Resume-Upon-Suspend) :: Add the disable_mesh.sh script (at http://download.sugarlabs.org/dextrose/testing/dx3/rpms/source/dextrose-platform-2-2.fc14.src.rpm) in /etc/powerd/postresume.d. Query : While the (For Reboot) case is guaranteed to work stably every single time, can the same be said about (For Resume-Upon-Suspend) ? c) Any other unknown in-the-dark hack. Regards, Ajay On Tue, May 1, 2012 at 9:36 PM, Peter Robinson pbrobin...@gmail.comwrote: On Tue, May 1, 2012 at 3:55 PM, Ajay Garg a...@activitycentral.com wrote: which actually brings me back to my original question :: Why is it so that putting the 'disable-mesh-script' in the 'start()' method of '/etc/init.d/Networkmanager' works (always) for (re)boot; but never works for resume-upon-suspend? Because on suspend/resume the NetworkManager service isn't started/restarted so the script in /etc/init.d isn't run where as on reboot it is. There's means though to run specific scripts on suspend/resume if that sort of thing is necessary. Peter ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 02, 2012 at 09:14:26AM +0530, Ajay Garg wrote: Is there an already tested-cum-recommended method, that could disable mesh-network, both during reboots and resume-from-suspend? Not that I know of. But I'm curious, why do you need to do this? Is there some other problem you think you will solve with this? There might be an alternate solution. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Thanks James for the reply. On Wed, May 2, 2012 at 9:21 AM, James Cameron qu...@laptop.org wrote: On Wed, May 02, 2012 at 09:14:26AM +0530, Ajay Garg wrote: Is there an already tested-cum-recommended method, that could disable mesh-network, both during reboots and resume-from-suspend? Not that I know of. But I'm curious, why do you need to do this? Is there some other problem you think you will solve with this? There might be an alternate solution. No, nothing of that sort. Just wish to remove the mesh-icons from Neighborhood-View. Right now, we were trying the disable-mesh-script (''echo 0 /sys/class/net/eth0/lbs_mesh) for our purpose. Now, I am thinking of removing all references to devices of type DEVICE_TYPE_802_11_OLPC_MESH from sugar-code. I think that should do it. Thanks to all for your quick responses. Thanks and Regards, Ajay -- James Cameron http://quozl.linux.org.au/ ___ Sugar-devel mailing list sugar-de...@lists.sugarlabs.org http://lists.sugarlabs.org/listinfo/sugar-devel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 02, 2012 at 09:33:36AM +0530, Ajay Garg wrote: Thanks James for the reply. On Wed, May 2, 2012 at 9:21 AM, James Cameron qu...@laptop.org wrote: On Wed, May 02, 2012 at 09:14:26AM +0530, Ajay Garg wrote: Is there an already tested-cum-recommended method, that could disable mesh-network, both during reboots and resume-from-suspend? Not that I know of. But I'm curious, why do you need to do this? Is there some other problem you think you will solve with this? There might be an alternate solution. No, nothing of that sort. Just wish to remove the mesh-icons from Neighborhood-View. But why? Right now, we were trying the disable-mesh-script (''echo 0 /sys/class/net/ eth0/lbs_mesh) for our purpose. Now, I am thinking of removing all references to devices of type DEVICE_TYPE_802_11_OLPC_MESH from sugar-code. I think that should do it. Yes, that should do it. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Wed, May 2, 2012 at 9:39 AM, James Cameron qu...@laptop.org wrote: On Wed, May 02, 2012 at 09:33:36AM +0530, Ajay Garg wrote: Thanks James for the reply. On Wed, May 2, 2012 at 9:21 AM, James Cameron qu...@laptop.org wrote: On Wed, May 02, 2012 at 09:14:26AM +0530, Ajay Garg wrote: Is there an already tested-cum-recommended method, that could disable mesh-network, both during reboots and resume-from-suspend? Not that I know of. But I'm curious, why do you need to do this? Is there some other problem you think you will solve with this? There might be an alternate solution. No, nothing of that sort. Just wish to remove the mesh-icons from Neighborhood-View. But why? Just that, we do not wish to set up a mesh-network, as per say. By removing all references to the device, we could provide a soft solution :) Thanks again. Regards, Ajay Right now, we were trying the disable-mesh-script (''echo 0 /sys/class/net/ eth0/lbs_mesh) for our purpose. Now, I am thinking of removing all references to devices of type DEVICE_TYPE_802_11_OLPC_MESH from sugar-code. I think that should do it. Yes, that should do it. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [Sugar-devel] Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Wed 02 May 2012 09:39:50 AM IST, James Cameron wrote: On Wed, May 02, 2012 at 09:33:36AM +0530, Ajay Garg wrote: Thanks James for the reply. On Wed, May 2, 2012 at 9:21 AM, James Cameron qu...@laptop.org wrote: On Wed, May 02, 2012 at 09:14:26AM +0530, Ajay Garg wrote: Is there an already tested-cum-recommended method, that could disable mesh-network, both during reboots and resume-from-suspend? Not that I know of. But I'm curious, why do you need to do this? Is there some other problem you think you will solve with this? There might be an alternate solution. No, nothing of that sort. Just wish to remove the mesh-icons from Neighborhood-View. But why? Because it isn't really used much. Sugar supports adhoc networking (since 0.88, IIRC) and that is used instead (atleast in dextrose builds). So, its better to make the mesh functionality disabled/invisible to the user. Right now, we were trying the disable-mesh-script (''echo 0 /sys/class/net/ eth0/lbs_mesh) for our purpose. Now, I am thinking of removing all references to devices of type DEVICE_TYPE_802_11_OLPC_MESH from sugar-code. I think that should do it. Yes, that should do it. - -- Anish -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.12 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQEcBAEBAgAGBQJPoLT8AAoJEBoxUdDHDZVpC1QH/0ukS8tfT0IejDCczQkT7SCp fiAuifIh8JCayK7j3CAY5BMUHoGlL8IRxw/xKqAxF+OZLKklb9GiFIok+5UJHC2o S2HAHVe/MCHbmcL1npV2qvlSnzVpEIXFtbQvjasX0IGJx6rC5RBDO5cKOXbbKAII QLwCwBgJmEbRLa5G+2ji++g21Y6OD6WL34ilSwLArcyEwgfozZlP1en0BAStCw1i 543yfN91f72KGNU4hNfa6LlR4GLhcjemLN2TLFB/rgliPF7VEFc77O25Y+YvKnZ8 7A8ZtBvbTt8F5eKGfbzJhfUVYCatTU9gD0H9w3DX/fdQ76+ySxGDrmBBnCsJwbY= =GKXZ -END PGP SIGNATURE- ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Hi, On Wed, May 02 2012, Ajay Garg wrote: Just wish to remove the mesh-icons from Neighborhood-View. Have you considered just removing the icons directly? diff --git a/src/jarabe/desktop/meshbox.py b/src/jarabe/desktop/meshbox.py index 20dc413..0aa8c7f 100644 --- a/src/jarabe/desktop/meshbox.py +++ b/src/jarabe/desktop/meshbox.py @@ -635,9 +635,9 @@ class MeshBox(gtk.VBox): def enable_olpc_mesh(self, mesh_device): mesh_mgr = OlpcMeshManager(mesh_device) -self._add_olpc_mesh_icon(mesh_mgr, 1) -self._add_olpc_mesh_icon(mesh_mgr, 6) -self._add_olpc_mesh_icon(mesh_mgr, 11) +#self._add_olpc_mesh_icon(mesh_mgr, 1) +#self._add_olpc_mesh_icon(mesh_mgr, 6) +#self._add_olpc_mesh_icon(mesh_mgr, 11) # the OLPC mesh can be recognised as a normal wifi network. remove # any such normal networks if they have been created -- Chris Ball c...@laptop.org http://printf.net/ One Laptop Per Child ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Yes Chris, that certainly is an option. But removing the references to the device itself, would clean up both the model and the view; whereas this (as I think) only removes the view. Regards, Ajay On Wed, May 2, 2012 at 10:18 AM, Chris Ball c...@laptop.org wrote: Hi, On Wed, May 02 2012, Ajay Garg wrote: Just wish to remove the mesh-icons from Neighborhood-View. Have you considered just removing the icons directly? diff --git a/src/jarabe/desktop/meshbox.py b/src/jarabe/desktop/meshbox.py index 20dc413..0aa8c7f 100644 --- a/src/jarabe/desktop/meshbox.py +++ b/src/jarabe/desktop/meshbox.py @@ -635,9 +635,9 @@ class MeshBox(gtk.VBox): def enable_olpc_mesh(self, mesh_device): mesh_mgr = OlpcMeshManager(mesh_device) -self._add_olpc_mesh_icon(mesh_mgr, 1) -self._add_olpc_mesh_icon(mesh_mgr, 6) -self._add_olpc_mesh_icon(mesh_mgr, 11) +#self._add_olpc_mesh_icon(mesh_mgr, 1) +#self._add_olpc_mesh_icon(mesh_mgr, 6) +#self._add_olpc_mesh_icon(mesh_mgr, 11) # the OLPC mesh can be recognised as a normal wifi network. remove # any such normal networks if they have been created -- Chris Ball c...@laptop.org http://printf.net/ One Laptop Per Child ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
If all you wish to do is remove the mesh icons from the network neighbourhood, why would you also want to clean up the model? On Wed, May 02, 2012 at 10:29:45AM +0530, Ajay Garg wrote: Yes Chris, that certainly is an option. But removing the references to the device itself, would clean up both the model and the view; whereas this (as I think) only removes the view. Regards, Ajay On Wed, May 2, 2012 at 10:18 AM, Chris Ball c...@laptop.org wrote: Hi, On Wed, May 02 2012, Ajay Garg wrote: Just wish to remove the mesh-icons from Neighborhood-View. Have you considered just removing the icons directly? diff --git a/src/jarabe/desktop/meshbox.py b/src/jarabe/desktop/meshbox.py index 20dc413..0aa8c7f 100644 --- a/src/jarabe/desktop/meshbox.py +++ b/src/jarabe/desktop/meshbox.py @@ -635,9 +635,9 @@ class MeshBox(gtk.VBox): def enable_olpc_mesh(self, mesh_device): mesh_mgr = OlpcMeshManager(mesh_device) -self._add_olpc_mesh_icon(mesh_mgr, 1) -self._add_olpc_mesh_icon(mesh_mgr, 6) -self._add_olpc_mesh_icon(mesh_mgr, 11) +#self._add_olpc_mesh_icon(mesh_mgr, 1) +#self._add_olpc_mesh_icon(mesh_mgr, 6) +#self._add_olpc_mesh_icon(mesh_mgr, 11) # the OLPC mesh can be recognised as a normal wifi network. remove # any such normal networks if they have been created -- Chris Ball c...@laptop.org http://printf.net/ One Laptop Per Child -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Well.. Hmm.. So that it could mean less (leftover-half-bits) bugs to play with :P Regards, Ajay On Wed, May 2, 2012 at 10:42 AM, James Cameron qu...@laptop.org wrote: If all you wish to do is remove the mesh icons from the network neighbourhood, why would you also want to clean up the model? On Wed, May 02, 2012 at 10:29:45AM +0530, Ajay Garg wrote: Yes Chris, that certainly is an option. But removing the references to the device itself, would clean up both the model and the view; whereas this (as I think) only removes the view. Regards, Ajay On Wed, May 2, 2012 at 10:18 AM, Chris Ball c...@laptop.org wrote: Hi, On Wed, May 02 2012, Ajay Garg wrote: Just wish to remove the mesh-icons from Neighborhood-View. Have you considered just removing the icons directly? diff --git a/src/jarabe/desktop/meshbox.py b/src/jarabe/desktop/meshbox.py index 20dc413..0aa8c7f 100644 --- a/src/jarabe/desktop/meshbox.py +++ b/src/jarabe/desktop/meshbox.py @@ -635,9 +635,9 @@ class MeshBox(gtk.VBox): def enable_olpc_mesh(self, mesh_device): mesh_mgr = OlpcMeshManager(mesh_device) -self._add_olpc_mesh_icon(mesh_mgr, 1) -self._add_olpc_mesh_icon(mesh_mgr, 6) -self._add_olpc_mesh_icon(mesh_mgr, 11) +#self._add_olpc_mesh_icon(mesh_mgr, 1) +#self._add_olpc_mesh_icon(mesh_mgr, 6) +#self._add_olpc_mesh_icon(mesh_mgr, 11) # the OLPC mesh can be recognised as a normal wifi network. remove # any such normal networks if they have been created -- Chris Ball c...@laptop.org http://printf.net/ One Laptop Per Child -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Okay. Take care not to add any when you change all those lines of code. ;-) On Wed, May 02, 2012 at 10:47:41AM +0530, Ajay Garg wrote: Well.. Hmm.. So that it could mean less (leftover-half-bits) bugs to play with :P Regards, Ajay On Wed, May 2, 2012 at 10:42 AM, James Cameron qu...@laptop.org wrote: If all you wish to do is remove the mesh icons from the network neighbourhood, why would you also want to clean up the model? On Wed, May 02, 2012 at 10:29:45AM +0530, Ajay Garg wrote: Yes Chris, that certainly is an option. But removing the references to the device itself, would clean up both the model and the view; whereas this (as I think) only removes the view. Regards, Ajay On Wed, May 2, 2012 at 10:18 AM, Chris Ball c...@laptop.org wrote: Hi, On Wed, May 02 2012, Ajay Garg wrote: Just wish to remove the mesh-icons from Neighborhood-View. Have you considered just removing the icons directly? diff --git a/src/jarabe/desktop/meshbox.py b/src/jarabe/desktop/ meshbox.py index 20dc413..0aa8c7f 100644 --- a/src/jarabe/desktop/meshbox.py +++ b/src/jarabe/desktop/meshbox.py @@ -635,9 +635,9 @@ class MeshBox(gtk.VBox): def enable_olpc_mesh(self, mesh_device): mesh_mgr = OlpcMeshManager(mesh_device) -self._add_olpc_mesh_icon(mesh_mgr, 1) -self._add_olpc_mesh_icon(mesh_mgr, 6) -self._add_olpc_mesh_icon(mesh_mgr, 11) +#self._add_olpc_mesh_icon(mesh_mgr, 1) +#self._add_olpc_mesh_icon(mesh_mgr, 6) +#self._add_olpc_mesh_icon(mesh_mgr, 11) # the OLPC mesh can be recognised as a normal wifi network. remove # any such normal networks if they have been created -- Chris Ball c...@laptop.org http://printf.net/ One Laptop Per Child -- James Cameron http://quozl.linux.org.au/ -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Haha.. Okies.. :) :) Regards, Ajay On Wed, May 2, 2012 at 10:55 AM, James Cameron qu...@laptop.org wrote: Okay. Take care not to add any when you change all those lines of code. ;-) On Wed, May 02, 2012 at 10:47:41AM +0530, Ajay Garg wrote: Well.. Hmm.. So that it could mean less (leftover-half-bits) bugs to play with :P Regards, Ajay On Wed, May 2, 2012 at 10:42 AM, James Cameron qu...@laptop.org wrote: If all you wish to do is remove the mesh icons from the network neighbourhood, why would you also want to clean up the model? On Wed, May 02, 2012 at 10:29:45AM +0530, Ajay Garg wrote: Yes Chris, that certainly is an option. But removing the references to the device itself, would clean up both the model and the view; whereas this (as I think) only removes the view. Regards, Ajay On Wed, May 2, 2012 at 10:18 AM, Chris Ball c...@laptop.org wrote: Hi, On Wed, May 02 2012, Ajay Garg wrote: Just wish to remove the mesh-icons from Neighborhood-View. Have you considered just removing the icons directly? diff --git a/src/jarabe/desktop/meshbox.py b/src/jarabe/desktop/ meshbox.py index 20dc413..0aa8c7f 100644 --- a/src/jarabe/desktop/meshbox.py +++ b/src/jarabe/desktop/meshbox.py @@ -635,9 +635,9 @@ class MeshBox(gtk.VBox): def enable_olpc_mesh(self, mesh_device): mesh_mgr = OlpcMeshManager(mesh_device) -self._add_olpc_mesh_icon(mesh_mgr, 1) -self._add_olpc_mesh_icon(mesh_mgr, 6) -self._add_olpc_mesh_icon(mesh_mgr, 11) +#self._add_olpc_mesh_icon(mesh_mgr, 1) +#self._add_olpc_mesh_icon(mesh_mgr, 6) +#self._add_olpc_mesh_icon(mesh_mgr, 11) # the OLPC mesh can be recognised as a normal wifi network. remove # any such normal networks if they have been created -- Chris Ball c...@laptop.org http://printf.net/ One Laptop Per Child -- James Cameron http://quozl.linux.org.au/ -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
Thanks Martin and Jon for the replies. On Sun, Apr 29, 2012 at 3:04 PM, Jon Nettleton jon.nettle...@gmail.comwrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/postresume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This should be a lot cleaner solution. However, two queries :: a) Doing ifconfig on the XO-1, only shows information for eth0 and lo (no mesh device listed). So, how can the mac address for the mesh device be found? b) Are mac address for XO-1s, EXACTLY same, for every XO-1 on this planet? Looking forward to a reply. Thanks and Regards, Ajay This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/postresume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: Wanting to know a bit of (NetworkManager) workflow upon resume-from-suspend
One thing I forgot to add. You can use the standard udi format to specify devices. i.e /org/freedesktop/NetworkManager/Devices/0 On Sun, Apr 29, 2012 at 11:34 AM, Jon Nettleton jon.nettle...@gmail.com wrote: On Sun, Apr 29, 2012 at 10:04 AM, Martin Abente martin.abente.lah...@gmail.com wrote: Are you guys still using this? http://git.sugarlabs.org/dextrose-platform/mainline/blobs/master/etc/powerd/postresume.d/disable_mesh.sh If so, you should remove it IF there is no way to guarantee that it will run before NM picks up the device. At least it will avoid the crash... I would ask in the NM community if there is a better way to disable a particular device, like banning a device(?). Edit /etc/NetworkManager/NetworkManager.conf Add a line to the [main] section like no-auto-default=xx:xx:xx:xx:xx (obviously replacing the x's with the mac-address of your mesh device.) This does not stop NM from managing your device, but does stop it from auto-connecting the device. You would still be able to go into NM and manually enable the mesh network. If you want NM to completely leave the device alone you can go one more step. Also in /etc/NetworkManager/NetworkManager.conf change the plugins line to plugins=ifcfg-rh,keyfile Then add a section that looks like this. [keyfile] unmanaged-devices=mac:xx:xx:xx:xx:xx:xx (Where X's are the mac address of the device you want to ignore) Hope that helps, let me know if you have further questions. -Jon ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel