**** BEGIN LOGGING AT Sat Jan 16 02:59:58 2016 Jan 16 09:14:43 nbd r48258 trunk/package/libs/ustream-ssl/Makefile * ustream-ssl: fix copy&paste mistake in mbedtls variant title Jan 16 10:24:54 nbd r48259 trunk/scripts/download.pl * download.pl: fix openssl sha256 digest output Jan 16 19:04:28 emailed in a patch to fix sysupgrade with -c generating a proper file list Jan 16 19:39:10 I submitted a tiny patch yesterday - is it OK? Jan 16 19:39:12 Allow UCI dhcp classifier to accept a MAC list Jan 16 19:48:25 any interest in a sysupgrade patch that takes a config option to restore the partition table from the boot disk after flashing? Jan 16 19:49:25 handy for x86 Jan 16 19:51:13 kaloz nbd jow ? Jan 16 19:52:24 does it have sanity checks for when the image partition layout changes? Jan 16 19:52:34 if so, then yes, i'm interested Jan 16 19:52:40 brb Jan 16 19:55:11 not sure how to implement that exactly Jan 16 19:55:23 but ill look into it Jan 16 19:55:26 and request feedback Jan 16 21:05:11 ugh this hfsc bug is annoying Jan 16 21:09:16 nyt: https://bugzilla.kernel.org/show_bug.cgi?id=109581 ? Jan 16 21:09:23 yea Jan 16 21:10:06 Yeah, that hit me on trunk as well. Luckily I tried running it in a VM before actually upgrading my router. Jan 16 21:15:27 seems its in all of the 4.x kernels, were any recent changes mdae to it? Jan 16 21:16:32 I checked sch_hfsc.c itself and there didn't seem to be anything new that was relevant. nbd said it was probably in the fq_codel.c file though. Jan 16 21:25:50 https://github.com/torvalds/linux/commit/6ac644a8ae2dabf884a1b01e82e32d96ffe6eee5 Jan 16 21:25:51 ? Jan 16 21:26:40 hm no Jan 16 21:35:12 nbd: so i tried witiboards eeprom from http://ftp.mqmaker.com/WiTi/Users/Benn/eeprom/, didn't help, now i have a copy of eeprom from a friend who too has a witiboard (but with the original firmware), and the wifi is still unusuable :( Jan 16 21:35:51 nbd: looks like i found a way to sanity check Jan 16 21:46:10 using sfdisk to read the partition table before do_upgrade, and again after Jan 16 21:47:26 then using grep to subtract all the lines from the original table from the after table and ensure its empty Jan 16 21:47:31 if so, can restore the original table Jan 16 22:04:59 nbd, this is what I've got so far... haven't tested yet, thoughts? http://pastebin.com/z6Q87iDw Jan 16 22:08:30 aside from obv mistakes like before and after reading from the same file Jan 16 22:11:53 need to exclude the 0 size partitions earlier, missed that from my testing to adding it to the script Jan 16 22:25:31 http://pastebin.com/wDGMP6zt Jan 16 22:26:10 removed the cuts, unnecessary Jan 16 22:28:21 http://pastebin.com/WZB67BkG shoudl do the trick, going to add it to sysupgrade now Jan 16 22:31:32 kab-el: which openwrt version were you trying? Jan 16 22:33:14 nyt: well, if you simply check if they match completely, restoring the partition table is going to be useless Jan 16 22:33:27 im not checking if they match completely Jan 16 22:33:42 just checking to make sure the partitions after exist exactly in the partitions before Jan 16 22:33:46 ah Jan 16 22:33:50 ok, i misread that part Jan 16 22:33:55 i tested all cases, it works so far Jan 16 22:34:00 nice Jan 16 22:34:14 this shit drives me crazy lol Jan 16 22:34:28 another thing: you then also need to make sure that sfdisk gets built into the image Jan 16 22:34:32 i did Jan 16 22:34:34 or at least that it's enabled by default Jan 16 22:34:45 at least i think i did Jan 16 22:34:46 like this, right? Jan 16 22:34:56 [nyt@cc:~/archive/openwrt/apu2/target/linux/x86]: grep sfdisk Makefile Jan 16 22:34:56 FEATURES:=squashfs ext4 vdi vmdk pcmcia targz sfdisk Jan 16 22:35:04 or is that not right Jan 16 22:35:10 nope, sfdisk is not a feature flag Jan 16 22:35:12 it's a package Jan 16 22:35:22 how do I add Jan 16 22:35:45 nbd: yes, that is a good idea, to actually try openwrt, iam going to compile trunk Jan 16 22:35:52 nbd: i was just using openwrt kernel Jan 16 22:35:59 nyt: DEFAULT_PACKAGES += sfdisk Jan 16 22:36:01 with gentoo Jan 16 22:36:02 yeah just saw Jan 16 22:36:02 ty Jan 16 22:36:10 kab-el: which version of mt76 were you using? Jan 16 22:36:21 and that goes in the makefile right? Jan 16 22:36:27 nyt: yes Jan 16 22:38:21 nbd: the newest, from the mt7603 branch Jan 16 22:39:34 newest from when? Jan 16 22:39:42 and what openwrt version did you use the kernel from? Jan 16 22:39:50 also whats with e1000e and realtek stuff being built by default for x86? just because its common? Jan 16 22:40:07 nyt: i guess so Jan 16 22:40:30 nbd: from the day before yesterday Jan 16 22:41:02 nbd: newest, vanilla 4.3.3 with openwrt patches Jan 16 22:41:15 both kernel and mt76 from the day before yesterday? Jan 16 22:41:20 yes Jan 16 22:41:24 no Jan 16 22:41:35 kernel 4.3.2, sorry, the 4.3.3 is probelmatic on witi Jan 16 22:41:50 ah, so old kernel which doesn't have some dts fixes Jan 16 22:41:54 it freezes the system, sometimes at boot, sometimes later Jan 16 22:42:17 i think it is maybe because of the changes of interrupt contoller api Jan 16 22:42:25 i'm going to stop guessing what's in there or not Jan 16 22:42:31 this is getting too confusing for me Jan 16 22:42:38 i'll simply wait for a test with plain current openwrt Jan 16 22:42:43 ok Jan 16 22:42:58 4.3.3 worked just fine for me on the witi btw. Jan 16 22:43:30 nbd: btw, i posted in into bugs, too, that include/prereq-build.mk checks for openssl, but it does not work for libressl Jan 16 22:43:52 https://dev.openwrt.org/ticket/21534 Jan 16 22:44:22 (with patch attached) Jan 16 22:44:43 ok, please post that as described in the ticket Jan 16 22:44:50 to the mailing list Jan 16 22:48:11 btw, do you know if blogic plans to add support for the IP1001 GE2 interface? if not, maybe i would try to do it, i was reading the code of raeth last month... Jan 16 22:48:45 i think he's mainly working on other stuff at the moment Jan 16 22:48:51 not sure if he plans to work on this himself Jan 16 22:51:54 ok, so maybe next month ill give it a try, thanks Jan 16 22:52:29 nbd, should i make this avail for all archs or just x86? Jan 16 22:54:06 nyt: x86 only for now Jan 16 23:09:06 so install_bin for the ramfs stuff Jan 16 23:09:17 shouldn't fail if the file doesnt exist Jan 16 23:09:22 yeah nm i just read the method Jan 16 23:09:29 bit all over the place sorry, lack of sleep :/ Jan 16 23:19:17 well Jan 16 23:19:19 i have the patch done Jan 16 23:19:21 just need to test it Jan 16 23:19:27 but want some eyes on it first to make sure im doing it properly Jan 16 23:19:30 but pastebin is eating dirt Jan 16 23:20:12 there's also bpaste, dpaste, pastebin.ca, and on and on :-) Jan 16 23:23:43 http://pastebin.com/8v3DQuTd Jan 16 23:40:21 nbd ;) Jan 16 23:40:26 or someone, plz to review Jan 16 23:50:10 can I give you a suggestion? Jan 16 23:50:28 yea Jan 16 23:51:17 send it as RFC to the mailing list, if everything is fine, you will only need to replace [RFC] with [PATCH] Jan 16 23:51:37 will do Jan 16 23:51:45 but just wanted some eyes on it before then Jan 16 23:51:49 going to test it out properly in a min Jan 16 23:57:16 i also have to figure out how to submit a patch for flashrom lol Jan 16 23:57:43 which kind? Jan 16 23:59:45 adding an include for sys/io.h and removing -lnsl in the build Jan 17 00:00:00 also also a fix for cryptodev building Jan 17 00:00:15 just not enough time atm :/ Jan 17 00:01:39 oh -lnsl removal was for tcp wrappers Jan 17 00:01:48 new file: utils/cryptodev-linux/patches/002-aed_request_set_assoc-fix.patch Jan 17 00:01:49 new file: utils/flashrom/patches/0003-fix-io-h-include.patch Jan 17 00:01:49 modified: libs/tcp_wrappers/Makefile Jan 17 00:02:03 will get that stuff mailed out tomorrow hoepfully Jan 17 00:07:07 nyt: ML + patchwork + RFC is much better for getting a review Jan 17 00:07:15 than posting link to pastebin on IRC Jan 17 00:18:17 fixed a bug or two, but it totally works Jan 17 00:18:27 thats sweet Jan 17 00:44:04 just changed my root partition size, and it worked good detecting the change Jan 17 00:44:04 Partition layout changed, not restoring. Jan 17 01:23:53 sent to list Jan 17 01:24:03 at least now i can actually not be annoyed using this device ;) Jan 17 01:24:12 /dev/sdb3 28.0G 151.0M 26.4G 1% /backup Jan 17 01:24:17 kept blowing out my other partition Jan 17 01:24:36 between this and the other sysupgrade bug i found / submitted a patch for **** ENDING LOGGING AT Sun Jan 17 02:59:58 2016