**** BEGIN LOGGING AT Sat Jun 02 03:00:09 2018 Jun 02 04:33:40 morning Jun 02 04:35:47 morning Jun 02 04:36:20 pkgadd: I was able to upgrade using the recovery webUI (in chinese) Jun 02 04:36:36 luckily it only had one button :) Jun 02 04:47:41 abenz: ther archer c2600 has a chinese webrecovery? or are you referring to a xiaomi or zbt mt7621 device? Jun 02 04:47:54 pkgadd: no the wg3526 Jun 02 04:48:37 pkgadd: I have a usb serial I will connect it to see whats going on Jun 02 04:48:51 tried many times, sysupgrade never happens (router just restarts) Jun 02 04:49:32 ah, yep, I knew that zbt should have such a thing (xiaomi afaik doesn't) Jun 02 04:50:25 I had the wg3525 on my short list towards the end of 2016, but couldn't get it from inside .eu Jun 02 05:05:44 morning Jun 02 05:12:44 pkgadd: got serial connected, will perform a sysupgrade now and finally see Jun 02 05:13:18 brb Jun 02 06:15:12 Sending KILL to remaining processes ... hostapd hostapd hostapd hostapd hostapd hostapd hostapd hostapd hostapd hostapd Jun 02 06:15:12 Failed to kill a[ 125.891451] reboot: Restarting system Jun 02 06:15:13 ll processes. Jun 02 06:15:13 sysupgrade aborted with return code: 256 Jun 02 06:15:34 pkgadd: ^ Jun 02 06:15:57 the fix is to disable wifi before upgrade Jun 02 06:16:04 abenz: hmm, what happens if you try to kill hostapd maually before? Jun 02 06:16:08 ie wifi down, then sysupgrade works Jun 02 06:16:24 that is strange Jun 02 06:17:19 there've been a couple of hostapd changes over the last 1-2 weeks though Jun 02 06:17:28 Good morning! Jun 02 06:17:47 definately file thar in flyspray (or the mailing list) Jun 02 06:17:51 Are there any plans for a one-click "upgrade to the latest OpenWRT" thing in Luci? Jun 02 06:18:17 I noticed the new packages dir appear on the releases page and it seems like yet again I'm going to need to figure out what chipset my router has before I can upgrade :) Jun 02 06:24:55 https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=78f1974bc565d7544589a49ad8efd92c4ddec5b3 https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=c8fdd0e9c843dd483f6677dc41f7df17313aa3cd https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=69f544937f8498e856690f9809a016f0d7f5f68b in particular, all of them in just over a week's time frame Jun 02 06:33:26 pkgadd: done sent an email to list Jun 02 06:38:40 build #424 of arc770/generic is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/arc770%2Fgeneric/builds/424 Jun 02 06:48:51 glyph no mate Jun 02 06:49:11 and It's not hard to work out what chipset your router has Jun 02 06:50:37 If you cant find out what file you need to flash to then just let me know what router you have and I will find it for you kk Jun 02 06:56:19 Tapper: thanks, I always manage to figure it out somehow :) Jun 02 06:56:42 cool :-) Jun 02 06:56:49 Tapper: Mainly I just manage updates on a ridiculous number of systems and I don't want to have to think hard about each one :) Jun 02 06:58:41 Yeah I can see that being a bit of a ball ake! Jun 02 06:58:44 lol Jun 02 07:29:11 build #846 of ixp4xx/harddisk is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/ixp4xx%2Fharddisk/builds/846 Jun 02 07:37:01 build #10 of mediatek/mt7623 is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/mediatek%2Fmt7623/builds/10 Jun 02 08:07:18 blogic: you always seem to be up very early. Jun 02 08:20:11 build #251 of ar71xx/tiny is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/ar71xx%2Ftiny/builds/251 Jun 02 09:12:31 build #892 of brcm2708/bcm2709 is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/brcm2708%2Fbcm2709/builds/892 Jun 02 09:16:31 hi, why can't I choose (X) o2 Box 6431 / Arcadyan VGV7510KW22 (BRN) and o2 Box 6431 / Arcadyan VGV7510KW22 (NOR) at the same time? I think I want both images. which one makes an image, that I can boot from brnboot or uboot without flashing. with initrd? Jun 02 09:17:47 pkgadd: just tried the same thing with the "reserve" partition and I had no issues =) Jun 02 09:21:19 why are these 2 entries multple devices? Jun 02 09:29:05 blogic: I've tried removing the "netgear" and "reserve" partition on the r7800 and it doesn't seem like it causes any problems for the OEM firmware; when the OEM firmware is flashed it recreates these partitions. Here are the patches I've made in case you're interested: https://github.com/huaracheguarache/OpenWrt/blob/master/0001-Remove-netgear-partition-and-expand-ubi-partition.patch Jun 02 09:29:18 blogic: https://github.com/huaracheguarache/OpenWrt/blob/master/0001-Remove-reserve-partition-and-expand-ubi-partition.patch Jun 02 10:24:48 Hi who is the best person to talk to about making a change in luci to make life easier for some one using a screen reader? It's a verry small change but it will save time when checking the states of a openwrt instillation. Jun 02 10:25:28 status Jun 02 10:25:52 * ldir volunteers jow Jun 02 10:26:53 Plus what's the default theme that ships with openwrt? I use so menny snapshots and community builds that I am not shure! Jun 02 10:27:12 bootstrap? Jun 02 10:27:13 Is it the bootstrap one? Jun 02 10:27:22 yeah, I think so Jun 02 10:27:28 thanks Jun 02 10:30:34 The thing is that all the info on the landing page for openwrt is just a wall of text to a screen reader. I would like for some things to be labeled as tables or heading markers put in to Ade jumping around the page with key Bord strokes. Jun 02 10:43:56 -*- luci-theme-bootstrap........................... Bootstrap Theme (default). - from make menuconfig Jun 02 11:05:08 I just realised that I'm probably a bell-end. I'm testing out whether the OEM firmware recreates two partitions after removing them with OpenWrt. Till now I've flashed an OpenWrt image without the partitions, then flashed the NETGEAR image and lastly flashed an OpenWrt vanilla snap to check whether the partitions are recreated. But even if the NETGEAR firmware doesn't create the partitions the vanilla OpenWrt snap Jun 02 11:05:09 would, right? Jun 02 11:05:53 So that would make my results invalid. Jun 02 11:07:02 ldir: is my thinking correct? Jun 02 11:19:12 huaracheguaracheHi does not the OEM fermwair reset the lay out of the flash as it is stock? Jun 02 11:19:29 Thanks ldir for looking up the thing about bootstrap mate Jun 02 11:19:54 Tapper: that's what I think, but I want to confirm it. Jun 02 11:20:49 I think it depends on the fermwair because some OEM firmware has youboot and all sorts in it when you flash Jun 02 11:21:14 in this case the firmware is over 20 mb in size Jun 02 11:21:30 Shit me side ways! Jun 02 11:21:36 that is big Jun 02 11:22:35 bloated crap, haha Jun 02 11:22:40 I just flashed a build to my wrt3200acm with a lot of packages in and that was just 11 meg Jun 02 11:24:32 hi Jun 02 11:27:24 * Tapper waves at Borromini Jun 02 11:28:51 gonna see if the netgear firmware has some sort of debug log that perhaps lists boot dmesg Jun 02 11:31:09 o/ Tapper Jun 02 11:31:18 still busy hacking stuff i see :D Jun 02 11:32:11 OMG just bin out in my back yard to chek on my 8 y old girl and the girl she was playing with was being fucking strange Jun 02 11:32:50 She was muttering under her breth and trying to touch my arm hahaha Jun 02 11:33:20 I asked her what she was doing and she said I am just joking. Jun 02 11:33:56 My kid has came in to the house now I will have to ask my wife about this when she gets back Jun 02 11:34:56 Borromini mostly breaking stuff mate Jun 02 11:35:20 join the club lol. Jun 02 11:35:34 I think that the kid was just doing it because I am blind but..... Jun 02 11:36:32 It's not the dun thing to kick a kid but man I came close! hahahahah Jun 02 11:40:43 Tapper: you have the best excuse... oh sorry didn't SEE you there ;-) Jun 02 11:42:06 Borromini: https://github.com/ldir-EDB0/openwrt/commit/68bdb486332eb1260750f5d2029292bf9d7d2397 Jun 02 11:44:00 thanks man you feel my pain. Jun 02 11:44:08 ldir haha yeah mate. I love kids I have 5 of them but that one from down the road freeks me out! Jun 02 11:47:07 Borromini: completely untested...beyond a compile for ramips at least. Jun 02 11:47:20 ramips is the only platform i have on 4.14 :^) Jun 02 11:47:22 suits me ;) Jun 02 11:47:29 what ramips stuff do you have? Jun 02 11:47:37 Hey guys, if I switch the default cpu freq on my ipq4019 kernel from ondemand to performance by using make kernel_menuconfig - would that burn out my router a whole lot faster? Or would it just boost up the performance? Jun 02 11:48:23 jow: I would like to backport the tls cache for mbedtls to 17.01 is that ok, see https://git.openwrt.org/?p=openwrt/staging/hauke.git;a=shortlog;h=refs/heads/lede-17.01 Jun 02 11:48:30 mir3g - keep trying to switch to it as my main router but something always happens to deflect me :-) Jun 02 11:49:02 nbd: It was suggested that this patch should be backported, is that ok: https://git.openwrt.org/83deb2b7386bf2ac7a8f00ff5f6f7c0ae4d993ac Jun 02 11:49:13 ldir: hehe Jun 02 11:49:32 rmilecki: this looks like an easy fix, should I backport it: https://git.openwrt.org/adfd64ddd62c3e5d709b25e2392abe1945326566 Jun 02 11:49:46 ldir: very happy with my dir-860l myself Jun 02 11:49:51 Hauke: sure Jun 02 11:50:18 but it seems mt76 is broken in master and in 18.06 as well, despite my local 18.06 build working fine so far Jun 02 11:50:39 does the flash problem with the new revisions of the WRT3200ACM also happen in lede 17.01, if so I will add this one: https://git.openwrt.org/9dd189d4adc4158e54461b4cbac5ef4e66222a94 Jun 02 11:52:44 Borromini: there has been a lot of confusion on mt7621 recently ;-) Jun 02 11:53:37 yeah :-/ Jun 02 11:54:30 Hauke: would you say mbedtls is faster than openssl with these tweaks ? Jun 02 11:55:07 ldir: no it is still slower than openssl, but faster than before Jun 02 11:55:50 Hauke: ok thanks. I shall stick with running just openssl instead of having two ssl libs on the system :-) Jun 02 11:56:22 modern browsers will reuse the credentials from a previous session and then no asymmetric cryptographic operations are needed. Jun 02 11:56:27 * Borromini only uses OpenSSL on systems with enough storage to handle it Jun 02 11:57:05 the initial handshake will still be slow, but opening the secound page will be very fast Jun 02 12:02:32 ldir: too many differences with 18.06 already it appears, will have to take a crack at it myself once more Jun 02 12:05:54 Borromini: this release thing is overrated ;-) Jun 02 12:06:11 i bet :P Jun 02 12:06:31 * ldir enjoys the bleeding edge Jun 02 12:06:49 here i was, encouraging people to try the 18.06 snapshots Jun 02 12:07:08 next pops up a topic on the forums: 'so... i was told to give the 18.06 builds a try...' Jun 02 12:07:14 and i hid in the corner and wept. Jun 02 12:08:04 lol Jun 02 12:27:50 Hey, i'm not 100% sure how to rebase my pull request (https://github.com/openwrt/packages/pull/5563) to latest master, its a couple of months old and i don't want to mess it up, maybe anyone has the git command on hand or a maintainer can manually force a rebase on it via github ui? Jun 02 12:28:25 ldir: one clean target refresh done for 4.14, so i'm betting on a smooth ride from here on out ;) Jun 02 12:32:39 andy2244: it doesn't have any conflicts so a rebase isn't required.... unless you want to keep the change at the top of the tree? Jun 02 12:33:26 ok so against what does jenkins compile/test? Jun 02 12:33:58 i mean Travis CI Jun 02 12:34:44 hmmm, absolutely no idea! Jun 02 12:34:46 it failed on my last change, yet locally on a fresh checkout and testing against master (arm/mips) i get no errors, so i was thinking maybe a rebase can help? Jun 02 12:35:19 i mean a rebase cant hurt, so i wanted to try it :p Jun 02 12:35:31 a rebase might help provoke it into another test, but closing and re-opening the pr will give it a kick as well. Jun 02 12:35:36 andy2244: you have a local tree? Jun 02 12:37:02 ok, so git checkout master ; git pull; git checkout samba-4.8; git rebase master; assuming ok git push -f Jun 02 12:37:05 i use a separated feed not a real local tree, since i update the feed first and than just copy the changes in the branch Jun 02 12:37:35 any issues in the rebase either fix them up, or git rebase --abort to abandon Jun 02 12:38:47 ok thx will try this Jun 02 12:51:37 wee my lcd is sold <3 Jun 02 13:08:01 andy2244: jenkins performs a build test against the latest ar71xx/generic snapshot sdk at that time Jun 02 13:08:16 andy2244: to force another test aganst the current verson, force push to your feature branch Jun 02 13:08:33 andy2244: e.g. in your local clone; git checkout master; git pull --rebase Jun 02 13:09:09 andy2244: if you did a separate non-master branch for your pr: git checkout feature-branch; git rebase master; git push origin feature-branch --force Jun 02 13:09:23 @jow hey ok, so jenkins is not testing against the sdk at the time of the initial based pull request? Jun 02 13:09:32 andy2244: if you use master, then git push origin master --force Jun 02 13:09:52 andy2244: nope, it uses whatever sdk happens to be online at the time the test runs Jun 02 13:10:30 oki will do a local test against ar71xx/generic than first i guess Jun 02 13:11:14 so a rebase should do nothing for me, without any real conflicts? Jun 02 13:11:32 andy2244: can't tell without further details Jun 02 13:11:52 maybe something in buildroot or dependant libraries changed which allows the build to complete now Jun 02 13:12:05 in any case it does not hurt to rebase and force-push your pr branch Jun 02 13:12:12 this will trigger a new build with new sdk Jun 02 13:12:18 if it still fails we can investigate further Jun 02 13:13:09 ok sounds good Jun 02 13:14:31 Borromini: do you remember asking whether the current temperatures in Oslo are normal at this time of the year? Jun 02 13:15:22 Borromini: well, May was the warmest May ever recorded in Norway. Jun 02 13:18:38 Borromini: right now it's 30+ here, lol Jun 02 13:23:26 o_O Jun 02 13:23:51 we had a sweltering heat for the past few days:-/ Jun 02 13:24:44 luckily it's cooler now but it will pick up again later this week. it seems we can't have any normal warm days anymore >_> always very humid Jun 02 13:26:07 high humidity is unbearable Jun 02 13:26:33 but that luckily doesn't happen very often here in Norway Jun 02 13:28:34 funny thing is that Northern Norway is set to get some sleet and snow from monday Jun 02 13:30:53 heh Jun 02 13:36:05 huaracheguarache, lucky bastard ;-; Jun 02 13:36:29 deathly heat here at the bottom of norway :^) Jun 02 13:40:29 Hi Jun 02 13:40:30 Redfoxmoon: I live near Oslo, but I don't mind the heat too much. Jun 02 13:40:45 dang. Jun 02 13:40:51 25-30 celcius here :^) Jun 02 13:41:07 Sleeping in the basement solves a lot of things =) Jun 02 13:42:33 no basement here... Jun 02 13:42:59 okay my own .47 bump is breaking in the ugliest way i've ever seen >_> Jun 02 13:59:43 ldir: i put up my bump here static inline void sb800_prefetch(struct device *dev, int on) {} Jun 02 13:59:46 woopsie Jun 02 13:59:49 + static inline bool usb_amd_pt_check_port(struct device *device, int port) Jun 02 13:59:51 +static inline void usb_enable_intel_xhci_ports(struct pci_dev *xhci_pdev) {} Jun 02 13:59:57 https://gist.github.com/Borromini/df50456db77920b6f7b1c58628496fd3 Jun 02 14:00:14 but it breaks badly. Jun 02 14:00:28 as indicated at the top Jun 02 14:00:51 you're the adult here so if you know how to fix this let me know :) Jun 02 14:30:16 @jow oki the rebase worked fine, yet travis fails with "The job exceeded the maximum time limit for jobs, and has been terminated." .... Jun 02 15:46:06 andy2244: that just means that the build take too long for the allocated time budget, it is not any indication of a mistake on your side Jun 02 15:46:12 *took Jun 02 15:46:33 this can happen for packages having a huge dependency chain for example Jun 02 15:49:43 @jow yeah i guessed so much, so i will leave it as is, since a local compile of mips/arm/mipsel all worked fine against latest master. So i guess i'm back waiting for the final review. Jun 02 16:47:19 the mt76 git-revert was in openwrt-18.06 and just tested with zbt1326, seems working. Jun 02 16:47:40 my one-day effort of fixing it did not work but happy to see the branch is good for zbt1326 now Jun 02 18:14:44 oh cool, ramips data corruption made it into the kernel Jun 02 18:14:50 *fix Jun 02 18:22:31 why do I get "LZMA: uncompress or overwrite error 7 - must RESET b" ? Jun 02 18:34:17 jospezial: when do you get that error? Jun 02 18:50:59 new release on monday? Jun 02 19:01:10 jow: should I always generate a ECDSA certificate for uhttpd, or should this be hidden behind some config option and what should be the default value? Jun 02 19:01:52 jow: an EC key also has a curve name instead of a key size Jun 02 19:02:23 most users probably want to have both a rsa and an ecdsa key in parallel Jun 02 19:08:27 Hauke: out of curiousity, do you know a place where ECDSA does not work? Jun 02 19:10:24 mangix: old browsers Jun 02 19:12:19 mangix: hmm cloudflare says that only very old browsers are unsupported: https://support.cloudflare.com/hc/en-us/articles/203041594-What-browsers-work-with-Cloudflare-s-SSL-certificates- Jun 02 19:12:37 a quick ssllabs look shows, iOS 4 and below, XP, and Android less than 4 Jun 02 19:13:59 hmm then we can probably also completly switch to ECDSA by default Jun 02 19:14:51 as these versions also do not support SNI https will not work good with them in general Jun 02 19:19:26 I think so. I don't think many users are still on XP. Jun 02 19:27:45 huaracheguarache: send the patch to the ML please Jun 02 19:32:26 Hello all Jun 02 19:32:51 Is it possible to use adblock simple to block custom urls? Jun 02 19:33:00 Like Microsoft. Com? Jun 02 19:57:47 blogic: ok, will do Jun 02 20:25:59 Hauke, I try to boot from uboot in RAM openwrt-lantiq-xrx200-arcadyan_vgv7510kw22-brn-initramfs-kernel.bin or openwrt-lantiq-xrx200-arcadyan_vgv7510kw22-nor-initramfs-kernel.bin made from git source. I used loadx to copy it. tftpboot crashes the router Jun 02 20:29:51 Hi, an issue with Lede on a Tp-link tl-wr841n v9.1, I can t access it after trying 192.168.2.1 lan static ip to make a wifi repeater out of it ... Any idea to fix ? Jun 02 20:31:00 spikemcc, Are you sure, you are in itsnew net? Jun 02 20:31:44 Is there any obvious reason why mvebu is not neon but vfpv3? https://github.com/openwrt/openwrt/commit/586ece71a66327f48e83ba67287892517e8dcba1 Jun 02 20:32:00 What do you mean ? Sorry I m a french speaker first ... Jun 02 20:33:43 Using lede-17.01.4-ar71xx-generic-tl-wr841-v9-squashfs-factory(dot)bin Jun 02 20:33:53 spikemcc, Is your computer in the same net as the 192.168.2.1 ? Jun 02 20:35:12 spikemcc, if you use dhcp than you can make it the easy way. Jun 02 20:36:17 not sure if its on the same net sadly Jun 02 20:37:30 your computer IP schould be in 192.168.2.x Jun 02 20:39:50 blogic: nvm, need to do some more testing first Jun 02 20:40:37 spikemcc_, the dhcp server from that ruoter could still give 192.168.1 or other.x adresses Jun 02 20:48:05 jospezial: when uboot uncompressed the binary it probably overwrote the original data it was uncompressing Jun 02 20:48:16 jospezial: try to increase the load address Jun 02 20:53:51 almost there ;) Did everything in the documentation but I don t see where to create the new network that the repeater mode does ... Jun 02 20:54:17 dissent1: wrt1900v1 is the reason Jun 02 20:54:22 no neon on there Jun 02 20:55:15 mangix: I see, thanks! Jun 02 20:55:37 i have neon on my local builds. no idea if it makes a difference Jun 02 20:56:16 diizzy tried making a subtarget for neon capable mvebu devices. didn't fly. Jun 02 20:56:53 http://i.imgur.com/WbrRz5I.png http://i.imgur.com/uT8t0cX.png Jun 02 20:58:44 Followed this : Jun 02 20:58:47 https://openwrt.org/docs/guide-user/network/wifi/relay_configuration?s[]=repeater Jun 02 20:59:39 you making a bridge or a repeater? Jun 02 21:00:02 wifi repeater, what differ for a bridge ? Jun 02 21:01:25 I aim to add a voip router in a charity office without router, the charity office next to it has the main router ... Jun 02 21:01:51 wow, this looks hard to set up Jun 02 21:02:42 Main router to Tpl-link tl-wr841n v9.1 on Lede to Linksys PAP2T Voip router on voip.ms Jun 02 21:03:50 Hauke, loadaddr set in uboot env or load address in kernel image ? Jun 02 21:04:03 Would have avoided this mess by using Linux on a computer but they are pussies so I need to workaround and waste my time on this ... Jun 02 21:05:04 spikemcc: there's a section on accessing the LEDE device Jun 02 21:06:03 also, if you're not using any LEDE specific feature, I'd recommend DD-WRT for this. This seems fairly error prone on OpenWrt Jun 02 21:06:43 jospezial: loadaddr set in uboot env Jun 02 21:07:04 you can also just cahnge it with "loadx 0xaabbccdd" and then bootm 0xaabbccdd Jun 02 21:07:16 then it is not persitent Jun 02 21:07:38 That s fixed, I now need to turn it into a bridge but it don t work ... Jun 02 21:08:25 are you configuring with LuCI or CLI? Jun 02 21:09:37 Luci, not sure if it support DD-wrt Jun 02 21:11:02 Hauke, so both commands with the same adress value or better different for more load place? Jun 02 21:11:32 spikemcc: ftp://ftp.dd-wrt.com/betas/2018/06-01-2018-r36079/tplink_tl-wr841ndv9/ Jun 02 21:13:11 factory or webflash Jun 02 21:13:15 ? Jun 02 21:15:37 probably web. Jun 02 21:16:34 turns out zbt1326 is not a solid product per se Jun 02 21:16:43 ausjke: elaborate Jun 02 21:17:08 because i doubt it is Jun 02 21:17:26 during the wifi debug I probably rebooted 10-20 times, I open the box soldered the uart wires, now it can not boot up most of the time Jun 02 21:17:52 ...interesting... Jun 02 21:18:09 release PCIe RST: RALINK_RSTCTRL = 7000000 is where it stops at serial output, basically, one of the problem is that it no longer finds pcie1 and pcie2 links Jun 02 21:18:21 i have a different mt7621 box that has a similar problem Jun 02 21:18:21 so wifi2 and wifi5 both are not found Jun 02 21:18:52 it has a 3.5mm serial plug. if it is plugged in while booting, booting fails. no output from uboot Jun 02 21:19:03 if it's not, it boots. Jun 02 21:19:13 at which point i can replug Jun 02 21:19:17 horrible design Jun 02 21:19:18 to make things worse, even its recovery image can not be booted up sometimes, i.e. the serial port has nothing come out sometimes, seems like a hardware quality issue to me Jun 02 21:20:43 some kind of short somewhere Jun 02 21:22:05 as it has the recovery image i thought i probably don't really need a serial connection Jun 02 21:22:36 until it failed to boot on known good image, i had to solder serial, and saw the problem Jun 02 21:23:22 serial port should not really impact pcie link in any way, they're usually not related on either pcb board or the silicon itself, its load should not impact pcie... Jun 02 21:24:07 i'm pretty certain the board is already half dead before I open the box and solder the uart Jun 02 21:24:40 jospezial: you have to use the same value Jun 02 21:24:47 loadx loads the binray into ram Jun 02 21:24:59 and bootm starts the binary from that address Jun 02 21:26:42 not only it said pcie1/pcie2 disabled, it can not even finish booting most of the time, what a piece of junk Jun 02 21:27:30 * ausjke is going to write a auto-reboot script to reboot another zbt1326 200 times and see if it will die in the middle Jun 02 21:27:33 ausjke: i recommend getting rid of your serial wires. fun fact: the dts file is totally wrong regarding its pcie ports Jun 02 21:28:00 mangix: yes i fought with the dts the whole day yesterday to no avail Jun 02 21:28:23 it's a mess and it's delicate to get a good fix Jun 02 21:28:56 mangix: problem is that before i solder the uart it already is refusing to boot up sometimes, which forced me to do the serial Jun 02 21:29:23 junk, got it Jun 02 21:30:17 to the point if you press&hold the reset button, plug in power, it won't boot at all Jun 02 21:31:03 the way the antenna is soldered to the PCB also concerns me, not impressed Jun 02 21:32:20 in short, pcie on cheap consumer board might not be solid, unlike pci that could be more robust, pcie is either you have it all, or you have nothing Jun 02 21:33:55 are the recent issues with mt7621 wifi fixed? Jun 02 21:34:02 paulius: yes Jun 02 21:34:08 yay :) Jun 02 21:34:23 Hauke, the default is set to 0x81000000 . My images are about 7Mbyte. Jun 02 21:34:24 openwrt dts files are messed up apparently Jun 02 21:34:41 need to be fixed before fixing the pci drivert Jun 02 21:34:41 anf hw offloading works? Jun 02 21:34:43 *driver Jun 02 21:34:53 and* Jun 02 21:35:02 what's broken with hw offloading? Jun 02 21:35:09 jospezial: and waht is the load address in the image? Jun 02 21:35:14 how much ram do you have? Jun 02 21:35:21 dunno, read something somewhere Jun 02 21:35:45 jospezial: this is u-boot isn't it? Jun 02 21:36:12 Hauke, 80002000 Jun 02 21:36:26 yes Jun 02 21:37:28 jospezial: if your compressed image is 7MB big it could be that it does not fit Jun 02 21:38:14 jospezial: if you have 64MB ram just try to load it to 0x82000000 Jun 02 21:38:19 at ram offset of 32 MB Jun 02 21:38:37 then you have more space to uncompress Jun 02 21:39:51 jospezial: I will leave now Jun 02 21:40:00 ok will try. Jun 02 22:08:51 interesting, with openwrt-18.06 I now only have eth0(wan), no more eth1(lan) Jun 02 22:09:03 on zbt1326 that it Jun 02 22:25:31 is it possible to dump the flash with brnboot or uboot? Jun 02 22:26:32 serial connection is running Jun 02 22:29:49 can I disable lzma compression of the kernel image? Jun 02 22:35:10 jospezial: you could "make kernel_menuconfig" in your source tree and change the compression method, but there isn't a 'none' setting Jun 02 22:37:05 even if you managed to boot uncompressed vmlinux it'd be too big for most kernel flash partitions i think Jun 02 22:38:40 https://unix.stackexchange.com/a/128025 might work? Jun 02 22:38:41 the strange thing is I can boot openwrt-lantiq-xrx200-arcadyan_vgv7510kw22-nor-squashfs-sysupgrade.bin but that does panic because of no root fs Jun 02 22:40:46 I want to dump the flash before flashing. Jun 02 22:41:28 oh Jun 02 22:41:39 if flash is memory mapped you can dump it as hex from u-boot Jun 02 22:41:43 slow but it works Jun 02 22:43:38 best bet is to make a minimal initramfs image though Jun 02 22:43:54 m4t, can you tell me the command? I find the startadress myself Jun 02 22:44:26 md i think Jun 02 22:44:43 md.b ? Jun 02 22:45:35 i dumped flash on an octeon target like this md.w B7C00000 4000000 Jun 02 22:45:48 where b7c is the start of mmap'd flash and 40... is the size in hex Jun 02 22:46:13 it'd be much easier and quicker to just boot an initramfs Jun 02 22:46:38 serial is kind of unreliable (with the absence of flow control, etc.) Jun 02 22:48:08 package mtd-utils in the initramfs and nanddump or mtd_debug read depending if nand/nor respectively iirc Jun 02 22:49:06 your command dumps the flash to ram? but how can I get it to my pc? I send my images with sx to the router ram. Jun 02 22:49:14 no it dumps it to stdout Jun 02 22:49:33 its a raw memory read. if flash is memory mapped then a raw memory read of the correct address space == flash contents Jun 02 22:50:09 thats really a last ditch method i think, even a few MB can take hours depending on baud rate Jun 02 22:50:49 considering that every byte becomes 2 bytes, plus whitespace, addresses Jun 02 22:51:00 sending 7 MB to the router takes 10 minutes Jun 02 22:51:04 (due to hex conversion and formatting of md output) Jun 02 22:51:19 over which method? zmodem? Jun 02 22:51:26 xmodem Jun 02 22:51:43 this would take probably 3x that or more Jun 02 22:51:50 just boot an initramfs... Jun 02 22:52:15 plus given the number of bytes there's a good chance some bits will get flipped Jun 02 22:53:08 I wrote before: I get "LZMA: uncompress or overwrite error 7 - must RESET b" Jun 02 22:54:31 can do you tftpboot in u-boot? Jun 02 22:54:35 -do Jun 02 22:55:05 no it crashes the router after first * Jun 02 22:55:24 did you check printenv to see which addresses stock is booting with? Jun 02 22:55:30 e.g. bootcmd Jun 02 22:56:38 ipaddr and serverip? Jun 02 22:56:51 if you're in u-boot you could probably change bootargs to do init=/bin/sh to get a root prompt Jun 02 22:56:57 no... Jun 02 22:57:02 the memory addresses it's using Jun 02 23:03:27 blogic: i'm fairly sure the newest ramips git-revert broke eth0.1 and eth0.2 on zbt1326 at least Jun 02 23:03:42 wifi are up, but LAN wire ports are gone Jun 02 23:03:51 do the snapshot or release images have an initramfs/initrd or other inbuild root fs? Jun 02 23:04:17 i dont think so Jun 02 23:04:18 to boot without flash Jun 02 23:04:43 if one of your issues is not having enough ram to uncompress you'd want to make it as small as possible Jun 02 23:04:55 stock images are too big for memory on a lot of targets Jun 02 23:07:56 it has 64Mbyte ram. my images are 7Mbyte. the booting openwrt-lantiq-xrx200-arcadyan_vgv7510kw22-nor-squashfs-sysupgrade.bin is 9 Mbyte. Jun 02 23:28:12 ausjke: no way Jun 02 23:28:43 mangix: i'm make-distclean and make-j now to verify Jun 02 23:29:05 in case something messed up as I did too many 'make's these days Jun 02 23:29:06 don't save settings either Jun 02 23:30:48 i realized it's after lunchtime i could not have network access after i used the newest openwrt-1806-with-wifi-fixes, which led to serial-soldering, which led to an even worse situation, now with a backup zbt1326 i still have no eth0.0/eth0.1 access, but if i program an old image it works fine Jun 02 23:31:23 by 'old image' i mean an image before 5/14 when the bug was introduced Jun 02 23:32:40 i have a patch that slightly fixes mt7621. i doubt it would get taken though Jun 02 23:33:13 what does it do, if you pastebin i can play with it Jun 02 23:33:19 jospezial: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=3e6d303d61861734e8d7da731971f46e004cfbde Jun 02 23:33:31 fixes the pcie entries Jun 02 23:33:43 in mt7621.dtsi. not relevant to ethernet Jun 02 23:34:27 ethernet connects through mdio or something Jun 02 23:34:41 mt7621.dtsi impacts a lot of boards...and pci interrupt-map is there, are we sure the pcie-interrupt-mapping are the same for all mt7621 boards? Jun 02 23:34:58 not talking about that Jun 02 23:35:17 dtb currently errors on mt7621.dtsi. my patch just fixes the warnings Jun 02 23:35:21 what if they layout pcie differently, say, some boards have a pcie-to-4G modem on bus 1, another does not have it and put wifi on bus1 instead... Jun 02 23:35:41 oh that is broken in openwrt currently Jun 02 23:35:52 the workaround is to ignore the dts files Jun 02 23:35:59 and hardcode it in the pci driver Jun 02 23:36:14 maybe the pcie interrupt mapping should be in board dts, e.g. zb1326.dts instead of mt7621.dtsi Jun 02 23:36:43 no zb1326.dts should only enable pci1 and 2 Jun 02 23:36:52 pci0 is bogus IIRC Jun 02 23:37:02 what does the INterrupt 4 do? Jun 02 23:37:12 anywhere I can find a schematics on zbt1326... Jun 02 23:37:40 doubt it. look at the circuit the board. better yet, look at /proc/interrupts Jun 02 23:37:41 to write a correct pcie-interrupt-mapping you have to know where those pcie devices are IDSEL-ed Jun 02 23:38:03 /proc/interrupts has the numbers on column 8 Jun 02 23:38:21 but if the mapping is not right then /proc/interrupts are wrong too Jun 02 23:38:35 not exactly Jun 02 23:38:51 zbt1326.dts specifies pci0 and 1 Jun 02 23:38:58 yes Jun 02 23:39:06 therefore the interrupts should show up as 11 and 31 Jun 02 23:39:15 but last i remember, they don't Jun 02 23:39:30 they show up as 31 and 32 Jun 02 23:39:41 which is 24,25 + 7 Jun 02 23:39:55 in my case it's 24 and 25 , or 31 and 32, add 7 that is Jun 02 23:40:04 there is no 11 whatsoever Jun 02 23:40:16 so like i said, the board is using pcie1 and 2 Jun 02 23:40:20 not pci0 Jun 02 23:40:28 which is why the driver broke the board Jun 02 23:40:31 the dts file is wrong Jun 02 23:40:51 i agree, except that i have not figured out how to fix the dts myself Jun 02 23:41:08 oh go to the pci section of the dts file Jun 02 23:41:19 change pcie0 and pcie1 to pcie1 and pcie2 Jun 02 23:41:32 wait a minute.... Jun 02 23:41:51 it could also be reversed with the wifi interfaces. won't really matter with the old driver Jun 02 23:42:26 it's reversed, totally insane Jun 02 23:42:47 that...makes some sense Jun 02 23:43:04 so I guess you need to reverse mt76@0,0 and mt76@1,0 Jun 02 23:43:24 you're getting somewhere... Jun 02 23:43:27 the chinese developers looked at the board the opposite way i guess Jun 02 23:43:40 erm, the chipset Jun 02 23:44:03 in the current dts, the 0x10000, 0x20000, 0x30000 are supposed to be left-shifted 11 bits Jun 02 23:44:17 mangix: Which board is it you have which fails to boot with serial attached? Jun 02 23:44:50 Monkeh: gnubee gb-pc1 Jun 02 23:45:04 i think it broke over time Jun 02 23:45:08 was not originally broken Jun 02 23:45:15 so they should have IDSEL as 0x20, 0x40, 0x60, but I could not find schematic/whatever to verify Jun 02 23:45:33 and wondering where did they get the 0x10000, 0x20000, 0x30000 for 4/24/25 interrupt-map Jun 02 23:46:06 pci driver Jun 02 23:46:17 they're hardcoded values Jun 02 23:46:29 i tried changing them, broke everything Jun 02 23:47:02 mangix: You're aware you'll deselect SPI boot if your serial adapter pulls the TX pin down? Jun 02 23:47:36 huh? My serial adapter is a 3.5mm port built-in to the board Jun 02 23:47:55 If whatever you're plugging into that happens to pull the board's TX pin down.. Jun 02 23:49:02 it's an adapter that currently works with the GB-PC2 Jun 02 23:49:31 We've established it exists Jun 02 23:49:32 That's a start. Jun 02 23:51:13 dunno, maybe i bent or shorted something Jun 02 23:51:17 too lazy to look Jun 02 23:51:34 especially since it's in production Jun 02 23:51:57 That's unfortunate Jun 02 23:55:55 mangix: eth0.0/1 is fine, a fresh built from openwrt-1806 fixed it Jun 02 23:56:13 i probably built too many times and something caught in the middle Jun 02 23:56:29 blogic: sorry, ethernet on ramips is fine, it's my build that is wrong Jun 02 23:58:43 ausjke: do you wish to be a guinea pig? Jun 03 00:01:37 mangix: don't get it, anyway i need zbt1326 to be solid for years to come, if possible Jun 03 00:02:01 and it's summer time, many are in the house to use it daily Jun 03 00:02:13 never mind then Jun 03 00:28:40 is it wise to delete everything in the dl folder when doing a make clean? Jun 03 00:38:56 biangbiangmian: no, it's pretty much the only thing that makes sense and is safe to keep under basically all circumstances (unless your dl has grown to much and contains lots of ancient stuff) Jun 03 01:08:36 biangbia_: it's advised to put dl somewhere else and do a symlink under openwrt, so make-distclean does not erase the content, as download is the most time consuming part for a new build Jun 03 01:10:44 pkgadd, I tried that with tftpput but it seems every tftp action reboots the router Jun 03 01:15:50 jospezial: I've only used dumping the flash via serial on the o2 box 6431 and from an initramfs image on the BT HH5a Jun 03 01:16:13 both worked well (serial takes its time though) Jun 03 01:17:44 never played with danube though (and besides xrx200, I only have (sadly unsupported) AMAZON-ME hardware) Jun 03 01:27:17 mangix: after fresh new build the 'broken' zbt1326 is back alive, i'm rebooting it every 3 minutes for the next 24 hours and see how robust it is, before I load iperf and cpubenchmark onto it after this test Jun 03 01:35:29 lol Jun 03 01:40:48 can't explain why i could not get into recovery mode sometimes still though Jun 03 01:41:09 but none need that constantly, so if a few tries can get it to work it should be fine **** ENDING LOGGING AT Sun Jun 03 03:00:01 2018