**** BEGIN LOGGING AT Mon Jun 18 03:00:07 2018 Jun 18 04:33:26 morning Jun 18 04:34:32 morning Jun 18 04:45:44 wigyori: could you fix/resend this one -> https://patchwork.ozlabs.org/patch/928771/ Jun 18 04:45:56 i merged 2/4 already and mkresin merged 1/4 Jun 18 04:49:49 Morning blogic Jun 18 05:01:21 Monkeh: random question - do you have an account of the same name on eevblog? Jun 18 05:20:28 blogic: any thoughts on https://github.com/openwrt/packages/pull/6270 ? Jun 18 06:22:34 mangix: what can i say, looks good Jun 18 06:22:38 did you test it ? Jun 18 06:22:44 thoroughly ? Jun 18 06:23:00 as in on real metal ? Jun 18 06:23:20 running at least the examples, the remote shell thing and had it pass a long term test ? Jun 18 06:37:21 blogic: mornin' - may I trouble you to look at https://patchwork.ozlabs.org/project/openwrt/list/?series=50457 Jun 18 06:40:46 ldir: on my list for today Jun 18 06:41:01 ldir: fighting backlog, only back in the office since couple of hours :-D Jun 18 06:42:16 excellent - thank you - please don't drown under backlog - come up for air & sanity :-) Jun 18 06:42:33 heh Jun 18 06:42:52 need to do payed work for the day then some more patches and then 18.06 Jun 18 06:43:56 ah yes, paid work, I know the priority system well :-) Jun 18 07:15:51 blogic: yep, saw lucian's comment on it, will take care of it doay Jun 18 07:15:52 today Jun 18 07:19:01 Hauke: you asked me to test the uboot patch before merging, right? :P Jun 18 07:50:38 wigyori: great ! Jun 18 08:34:37 ldir: who do we need to set fire to, to actually apply that bloody ATM patch to master and openwrt-18.06 branches? Jun 18 08:34:51 * dwmw2_gone ponders asking for commit privs Jun 18 08:35:43 stintel: did you look at that domoticz update? Jun 18 08:35:44 setting fire is a little harsh - I politely requested of blogic this morning to take a look and it's on his queue. Jun 18 08:35:57 should fix the ozwcp/ bits Jun 18 08:36:03 :) Jun 18 08:36:32 dwmw2_gone: was waiting for you Jun 18 08:36:33 in fact I note the patches have gone out of patchwork... which probably mean blogic has them in a local tree of his. Jun 18 08:36:39 so tell me what to merge and i'll do it now Jun 18 08:36:52 dwmw2_gone: sorry was on vacation and returned a few hours ago :-) Jun 18 08:36:56 slacker Jun 18 08:37:02 * ldir ROFL Jun 18 08:37:22 https://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git/commit/?id=9bbe60a67be5a1c6f79b3c9be5003481a50529ff Jun 18 08:37:24 we need that one Jun 18 08:37:47 ldir also has a patch to kill the ifx_atm_alloc_skb() hack with fire Jun 18 08:37:56 not sure why we ever had that in the first palce Jun 18 08:37:59 ldir: i removed them as dwmw2_gone said to use upstream and not those Jun 18 08:38:08 ldir: send a v3 with the real ones maybe ? Jun 18 08:38:56 ah, one *is* an upstream patch, the other is 'unfavourable' to the ifx_atm hack patch ;-) Jun 18 08:38:58 blogic: hope you enjoyed the vacation :) Jun 18 08:39:39 * ldir fiddles about and will produce a v3 shortly Jun 18 08:39:50 stintel: https://github.com/dwmw2/packages/commit/d9e2fcae57698088993fe858d703f550b368f3cd ? Jun 18 08:39:55 I am sure after doing two hours of openwrt the vacation effect will be gone :) Jun 18 08:39:59 blogic: yeah welcome back :-) Jun 18 08:39:59 hehe Jun 18 08:40:14 jow: haha Jun 18 08:40:23 jow: i was not planning on opening github Jun 18 08:40:47 ldir: thanks again for all the groundwork on this Jun 18 08:41:15 dwmw2_DRS: I looked at it but that's where it stopped Jun 18 08:42:48 ah os the gzipping is done in our Makefile Jun 18 08:42:52 yeah Jun 18 08:44:18 it reads kind of hackish but sure, lgtm Jun 18 08:50:13 dwmw2_DRS: no problem - truth be told I'm naive enough to be thrilled that I help discover & squish a kernel bug. And the fact that Eric broke it is wonderful :-) Jun 18 08:51:15 hehe yeah Jun 18 08:51:26 I didn't realise that change was so recent Jun 18 08:55:00 any idea what it may be abot? Jun 18 08:55:04 root@OpenWrt:/lib/modules/4.14.49# insmod usb-common.ko Jun 18 08:55:06 Failed to find the folder holding the modules Jun 18 08:55:20 my bootlog didn't inclue "kmodloader: loading kernel modules from /etc/modules.d/*" Jun 18 08:55:24 and no module was loaded it seems Jun 18 08:55:26 jow: ? Jun 18 08:56:28 out of curiosity, is it a known "feature" that setting TX power appears to have no effect on MT7620 devices (running 17.01)? Jun 18 09:11:03 i did rm .config, recreated it, problem still exists Jun 18 09:11:36 it's a default BCM53XX config with per CONFIG_TARGET_ALL_PROFILES and CONFIG_TARGET_PER_DEVICE_ROOTFS Jun 18 09:14:35 rmilecki: please strace it Jun 18 09:14:50 rmilecki: I suspect some uname <> folder name mismatch (CONFIG_LOCALVERSION?) Jun 18 09:15:21 jow: i just did rm -fR build_dir/target-arm_cortex-a9_musl_eabi/ Jun 18 09:15:23 rebuilding now Jun 18 09:15:30 there used to be a bug that when a kernel tree was unpacked below a git repo it considered itself dirty and appended a trailing "+" to its version Jun 18 09:15:31 if problem still exists, i'll add strace and check it Jun 18 09:16:10 this caused uname to return "x.y.z+" which didn't match the expected values, causing it to fail finding /lib/modules/x.y.z Jun 18 09:16:47 but that was years ago, maybe it has been fixed usptream by now and you're seeing another issue Jun 18 09:18:28 in any case the errors means that stat("/lib/$(uname -r)") failed Jun 18 09:19:30 blogic: new separate patches submitted. Jun 18 09:25:13 patchwork needs to handle dmarc re-written emails better. People get credited for submitted patches they've had nothing to do with. Scared the life out of me..... I haven't touched ustream-ssl :-) Jun 18 09:27:30 * ldir sets dwmw2_DRS's flamethrower to 1 candle power mode. Jun 18 09:28:35 ldir: dmarc needs to handle mailing lists better :) Jun 18 09:29:05 f00b4r0: it's known that proper TX-power management would require temperature-compensation code (which hasn't been implemented for rt2x00 yet) Jun 18 09:29:05 it kind of contradicts the point of a mailinglist Jun 18 09:29:16 yeah there's that too Jun 18 09:29:36 ldir: I still don't understand how you can remove the ifx_atm_alloc_tx symbol from the kernel, and still load the ltq-atm module Jun 18 09:29:44 you have to kill the hack in the ltq-atm module too, surely? Jun 18 09:29:49 dangole: ok so essentially the control knob is useless and the hw always use max power? Jun 18 09:30:20 I've been trying to setup AP overlaps with no luck so far ;) Jun 18 09:32:10 dwmw2_DRS: I just don't understand. Hopefully blogic will review it and tell me how I've screwed up :-) Jun 18 09:32:21 jow: root@OpenWrt:/# uname -r Jun 18 09:32:23 4.14.48 Jun 18 09:32:30 adding strace to build no Jun 18 09:35:22 jow: stat64("/lib/modules/4.14.48/", 0xbeaad988) = -1 ENOENT (No such file or directory) Jun 18 09:35:41 the correct path is /lib/modules/4.14.49 Jun 18 09:36:00 why when uname reports 4.14.48 ? Jun 18 09:36:16 riiight... Jun 18 09:36:18 what the heck Jun 18 09:36:42 f00b4r0: the code for adjusting the power is there, just the various gain values are temperature dependent and some are simply not initialized at all. keyword is 'TSSI' and the Ralink reference driver has it. Jun 18 09:37:34 dangole: color me silly but I'm not sure whether that's a "yes" or "no" answer... Jun 18 09:37:51 jow: strace uname -a https://pastebin.com/JWmRCXBn Jun 18 09:38:01 f00b4r0: yes, the knod is useless for that drivedr Jun 18 09:38:07 knob Jun 18 09:38:07 in my build tree there is build_dir/target-arm_cortex-a9_musl_eabi/linux-bcm53xx/linux-4.14.49/ Jun 18 09:38:09 ok, thanks! Jun 18 09:38:13 jow: ^ Jun 18 09:38:13 i'm pretty sure I'm using 4.14.49 Jun 18 09:38:25 waait Jun 18 09:38:27 [ 0.000000] Booting Linux on physical CPU 0x0 Jun 18 09:38:28 [ 0.000000] Linux version 4.14.48 (zajec@linux-veee) (gcc version 7.3.0 (OpenWrt GCC 7.3.0 r6427+1-ed3860c3e3)) #0 SMP Mon Jun 11 13:48:32 2018 Jun 18 09:38:53 f00b4r0: for sure the TX power knob doesn't do what you'd expect it to do. it may or may not do something, that may even be related to tx-power, but there are parts missing... Jun 18 09:39:23 dangole: ok. That's a bummer for my usecase but I'll live with it Jun 18 09:39:30 it's after I did rm -fR build_dir/target-arm_cortex-a9_musl_eabi/ bin/ .config Jun 18 09:39:37 f00b4r0: if we'd want that to be implemented properly, it could be a thermal zone in linux-hwmon-terms, some sort of kernel task to poll that and adjust the various gain and filter values and (just like the vendor driver) disable aggregation or even reduce the radio to only use one TX chain in case it really gets too hot Jun 18 09:39:44 still have to figure out why some APs are randomly rebooting anyway Jun 18 09:40:13 dangole: i see. Note that I'm not trying to increase tx-power, I want to _reduce_ it. Jun 18 09:40:28 and I was very surprised to observe that setting it to e.g. 2dBm did nothing Jun 18 09:41:17 rmilecki: keep in mind that the actual kmod packaging happens in package/linux/, not target linux/ Jun 18 09:41:33 rmilecki: maybe you just need a "make package/linux/clean world" Jun 18 09:41:45 f00b4r0: that's indeed surprising. it should do something at least, probably at a wrong offset and the actual power may change over time correlating with temperature somehow. but "no effect" is surprising to me. Jun 18 09:42:16 dangole: no noticeable effect for the purpose of reducing coverage, as "tested" via smartphone wifi scanner ap Jun 18 09:42:17 app Jun 18 09:42:46 so, not exactly ideal test conditions Jun 18 09:44:40 f00b4r0: that's still odd. i remember that changing the value had at least some measurable effect on the signal strength reported by my iwlwifi card on the other end Jun 18 09:45:23 dangole: this is with 17.01 though. Old kernel Jun 18 09:47:46 dwmw2_DRS: am speaking from position of ignorance but keen to learn. re the ifx hack, what I think my patch does is remove the dependency in the kernel atm module on the ifx_atm_alloc_tx symbol provided by the ifx driver module, because we no longer need the ifx_atm_alloc_tx call. Jun 18 09:48:08 dangole ping Jun 18 09:49:19 the dependency (well for that symbol) isn't the other way around.... ltq_atm doesn't need kernel atm (for that symbol at least) Jun 18 09:50:55 jow: i just did it Jun 18 09:50:58 didn't help... Jun 18 09:50:59 wtf Jun 18 09:59:28 jow: i tried find ./build_dir/ -name "*4.14.48*" Jun 18 09:59:33 ./build_dir/toolchain-arm_cortex-a9_gcc-7.3.0_musl_eabi/linux-4.14.48 Jun 18 09:59:51 it seems linux in toolchain dir wasn't properly refreshed or something... Jun 18 10:00:55 zajec@linux-veee:~/lede/source> ls -l build_dir/toolchain-arm_cortex-a9_gcc-7.3.0_musl_eabi/ | grep linux Jun 18 10:00:56 lrwxrwxrwx 1 zajec users 13 cze 11 17:11 linux -> linux-4.14.48 Jun 18 10:00:58 drwxr-xr-x 24 zajec users 4096 cze 11 17:12 linux-4.14.48 Jun 18 10:01:00 drwxr-xr-x 3 zajec users 4096 cze 11 17:12 linux-dev Jun 18 10:08:55 aparcar: pong Jun 18 10:22:12 anyone heard of pudn.com ? Jun 18 10:22:20 and or has a valid login Jun 18 10:22:34 there is a sketchy GPL tar ball on it that i would like to get hold of Jun 18 10:29:21 what the hell... i Just did rm -fR build_dir/toolchain-arm_cortex-a9_gcc-7.3.0_musl_eabi/ and then make Jun 18 10:29:26 it wasn't recompiled... Jun 18 10:29:45 should I remove the whole staging_dir?! Jun 18 10:29:56 i tend to remove tmp Jun 18 10:30:13 and touch toolchain/Makefile Jun 18 10:30:38 ok, trying now Jun 18 10:30:51 check your .config - is it *really* the platform you're expecting? Jun 18 10:31:43 ldir: been there done that :-D Jun 18 10:32:55 lol oh yes - don't ask me how I know or how many times ;-) Jun 18 11:06:47 how safe do you think SSDs are for tasks like OpenWrt development? Jun 18 11:06:59 compiling OpenWrt over and over means a lot of writing cycles Jun 18 11:07:14 i'd love to have SSD + HDD Jun 18 11:07:22 but new notebook I want to buy has SSD only Jun 18 11:09:14 fuck: root@OpenWrt:/# uname -r Jun 18 11:09:16 4.14.48 Jun 18 11:09:21 doesn't matter Jun 18 11:09:25 SSDs are solid now Jun 18 11:09:31 how old is that SSD? Jun 18 11:09:44 azarus: it's going to be a new one, i'm going to buy a new notebook Jun 18 11:09:50 oh, don't worry about it Jun 18 11:09:58 ok, maybe I'll give it a try... Jun 18 11:09:59 my laptop has been ssd only for years now Jun 18 11:10:06 it'll withstand anything you throw at it Jun 18 11:10:16 I don't compile as much as you do, but it's fine, and so much faster, I'd never consider going back. Jun 18 11:10:58 as long as you don't get noname device you should be fine. Jun 18 11:11:15 jow: blogic: i feel like an idiot... Jun 18 11:11:18 Either it dies in the first couple of weeks, or it'll hold out for a long, long time Jun 18 11:11:19 Could not open mtd device: firmware Jun 18 11:11:20 Can't open device for writing! Jun 18 11:11:20 and smart conveniently reports wear level, too. Jun 18 11:11:29 jow: blogic: sysupgrade never updated my kernel, just rootfs Jun 18 11:11:54 rmilecki, if you're really worried you could get a samsung PRO series SSD :P Jun 18 11:12:03 azarus: f00b4r0: I'm looking at HP EliteBook, they should put some solid SSD, I hope Jun 18 11:12:07 I have had a ssd in my PC for years now and it gets 24-7 suage including building openwrt some times Jun 18 11:12:15 rmilecki: I'd expect them to, indeed. Jun 18 11:12:22 one of those will never wear out :^) Jun 18 11:12:24 ok, you made me worry much less :) thanks Jun 18 11:12:56 rmilecki: I have an older elitebook, totally fine :) Jun 18 11:13:18 i have a 5yo EVO 850 on my main desktop (running occasional builds), it shows 99% wearability remaining Jun 18 11:13:47 f00b4r0: what SMART attribute ? Jun 18 11:13:47 the larger the SSD, the longer it'll last (for the same amount of data periodically written), btw Jun 18 11:13:47 sounds good :) Jun 18 11:14:16 f00b4r0: which smart factor are you useing for "99%" ? Jun 18 11:14:23 moments Jun 18 11:14:24 I have two 850 EVOs here, wearability is down to 96% after writing 10 TB :^) Jun 18 11:14:49 well, my machine is currently powered down so maybe Redfoxmoon can answer the SMART attribute question faster than I can ;-) Jun 18 11:15:30 my first SSDs where 2 OCZ Vertex 2, first one died after 5.5y of use Jun 18 11:15:50 karlp, at least the samsung ssds use #177 - Wear Leveling Count Jun 18 11:16:14 yeah, my old 805 evo is at "98" Jun 18 11:16:31 my intel 520s don't seem to have that Jun 18 11:16:33 mmmm, barely used then :P Jun 18 11:16:34 233 Media_Wearout_Indicator 0x0032 097 097 000 Old_age Always - 0 Jun 18 11:16:36 I actually replaced the 128gig that came int he elitebook originally, it was too small for multiple openwrt targets and some other stuff. Jun 18 11:16:56 stintel, hmhm, dang. Jun 18 11:17:17 do they track total writes somewhere? Ithought they did. Jun 18 11:17:24 Intel does Jun 18 11:17:32 249 NAND_Writes_1GiB 0x0013 100 100 000 Pre-fail Always - 66825 Jun 18 11:17:38 241 Host_Writes_32MiB 0x0032 100 100 000 Old_age Always - 1374608 Jun 18 11:17:51 lol 66TB writes Jun 18 11:17:53 karlp: on samsungs, the raw value is the average number of P/E cycles per block Jun 18 11:18:24 * karlp shrugs. I'm not concerned at alllll about reliabilty of them for desktop use, even "heavy" dev wrok Jun 18 11:18:41 yeah, you take backups anyway Jun 18 11:18:43 yeah modern SSDs are pretty solid. I wouldn't worry either Jun 18 11:18:49 also what stintel said ;^) Jun 18 11:18:59 just slightly inconvenient when your system is out of use due to a broken drive Jun 18 11:19:06 but you can have the same with HDD so Jun 18 11:19:13 I mean if you have the capability for spinning drives it's cheaper for heavy IO loads :P Jun 18 11:19:29 yup, I like SSDs, but they do sometimes fail out of the blue, like any other technical component one has Jun 18 11:19:47 (HDDs do too) Jun 18 11:19:53 yep. Jun 18 11:20:12 Had a WD caviar green mysteriously fail on me a few years back Jun 18 11:20:26 safe to say it was the last WD drive I ever bought. :^) Jun 18 11:20:38 dropped a HDD once, died instantly Jun 18 11:20:46 wouldn't have happened with a SSD Jun 18 11:20:47 I have 2x1TB WD RE with each 64GB bcache SSD cache on Intel 520, BTRFS RAID1, it's acceptable but it's almost full Jun 18 11:20:48 :P Jun 18 11:21:04 :P Jun 18 11:21:14 so I was thinking to buy 2x6TB WD RE and throw some small SSDs at it (256GB or so) for bcache Jun 18 11:21:18 *clicking head syndrome* Jun 18 11:21:33 I've got two 4 TB barracuda drives in raid 1, only 40~ gigs left of free space:^) Jun 18 11:21:47 but looking at the prices for 2TB SSD ... I might just throw in 2x2TB in RAID1 and add more in the future Jun 18 11:22:03 going from 1TB to 6TB was based on the price/gb that was cheapest with those drives Jun 18 11:22:06 2 TB and up SSDs aren't worth it Jun 18 11:23:48 maybe they are? depends Jun 18 11:24:16 few usecases for it, but I'm sure there are people for whom it makes sense Jun 18 11:24:21 not sure what to do yet, maybe a larger bcache, a UPS (which I want to add anyway), and setting cache mode to writeback might actually be the better way to go Jun 18 11:24:21 maybe there are* Jun 18 11:24:52 but my /home is on bcache backed HDD and sometimes it's horribly slow Jun 18 11:25:41 the 3rd option is to finally build a new fileserver and throw raw video footage there, and keep workstation SSD only Jun 18 11:25:42 * karlp thinkgs bcache hacking is a waste of time. just use the ssd for /home and rust for videos/photos Jun 18 11:26:02 well, I'm actually waiting for bcachefs to go mainline Jun 18 11:26:15 and then you don't have to worry about bcache hacking, the fs will just handle it all Jun 18 11:26:20 bcache had a pretty bad bug kernel-side a short while ago, iirc? Jun 18 11:26:25 still, you only want that sort of thing if you _reallllly_ cant' fit it all on the ssd Jun 18 11:35:03 https://wikidevi.com/wiki/Netgear_Nighthawk_M1_(MR1100) Jun 18 11:36:32 CPU1: Qualcomm MDM9250 - never heard of it Jun 18 11:37:27 common phone chip Jun 18 11:37:40 given this is an LTE router, not surprising. Jun 18 11:41:20 doesn't look like it's going to be easy to run openwrt on it Jun 18 11:41:24 and pricy Jun 18 11:42:02 does openwrt run on any of the existing lte router platforms? Jun 18 11:42:07 they're kinda different worlds. Jun 18 12:30:45 rmilecki: i recall if you modified anything under package/kernel/linux you have to `touch package/kernel/linux/Makefile' before menuconfig and make, not sure if this is related to build issue you had Jun 18 12:31:01 ausjke: it's resolved now Jun 18 12:31:04 i did something really stupid Jun 18 12:31:10 sysupgrade was failingand I didn;t notice that Jun 18 12:31:14 m4t: Am I making friends again? Jun 18 12:34:36 zbt1326 dies once a day, self-soldered usb-ttl seems impact the board's stability, without serial port i'm using a remote rsyslog and a local usb-stick to hopefully log why it dies once a while. 2.4ghz performance still so-so(1MBs on average now),barely tolerable Jun 18 12:34:54 Monkeh: heh Jun 18 12:39:52 not really a panice or reboot, just suddenly all the network gone and no internet, i had to reboot it, loading vnstat today Jun 18 12:46:03 stintel: People either love me or hate me.. :) Jun 18 12:47:18 Monkeh: what have you done? Jun 18 12:47:54 Not a clue :) Jun 18 12:52:14 Monkeh, how could anyone hate you? :P Jun 18 12:52:44 * ldir burns the list of reasons :-) Jun 18 12:53:36 * ausjke thought techies only hate bugs on irc Jun 18 12:54:02 Monkeh: I think you're alright - helped me out a few times :-) Jun 18 13:24:22 i lost the last bit of faith i had in qca today Jun 18 13:28:50 blogic: ? Jun 18 13:29:37 * Tapper hugs Monkeh: Jun 18 13:30:20 fyi, my Odroid C2 was RMA'd, the new one is waiting for me in Belgium so I'll be able to finish the meson target soon Jun 18 13:30:36 stintel: nevermind Jun 18 13:30:51 stintel: do you use an SD card or the eMMC module? Jun 18 13:30:57 f00b4r0: SD for now Jun 18 13:31:01 ok Jun 18 13:31:14 I could never get the eMMC not to trash its fs under I/O load Jun 18 13:31:21 interesting Jun 18 13:31:25 I guess I should order one Jun 18 13:31:28 I should probably try the newer images now that they mainlined Jun 18 13:32:04 out of curiosity, what triggered the RMA on yours? Jun 18 13:32:28 * Tapper hugs all openwrt peoples so no body feels left out! Jun 18 13:32:38 dunno, I had repurposed it as a mini workstation at my parents' place in Belgium as I didn't have any PCs there anymore Jun 18 13:32:43 and a week later it was dead Jun 18 13:32:51 *sigh* Jun 18 13:33:01 odds are it burnt itself out Jun 18 13:33:04 yep Jun 18 13:33:16 should probably add a bigger heatsink or a small fan Jun 18 13:33:19 cooling seems lacking on this device Jun 18 13:54:54 * ldir hugs blogic - thank you! Jun 18 13:56:13 dwmw2_gone: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=d600de3ddde269bf0b324735f8f12278f82d9b37 Jun 18 13:59:01 ldir: lol Jun 18 14:00:20 time to join some fellow Belgians to watch the game Jun 18 14:00:29 laterz ;) Jun 18 14:00:32 well you committed that hard fought patch Jun 18 14:03:04 morning Jun 18 14:10:34 build #31 of mediatek/mt7622 is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/mediatek%2Fmt7622/builds/31 Jun 18 14:13:33 looks, that builds are going well Jun 18 14:18:31 please could one of the commit adults cherry-pick d600de3ddde269bf0b324735f8f12278f82d9b37 for 18.06. Ideally pppoatm ought to work on x86, (lantiq is 4.9 on 18.06 so is unaffected at present) Jun 18 14:33:59 if i changed my local feeds, how to reflect those changes into the build without ./scripts/feeds update -a? just do ./scripts/feeds update -f myfeed? Jun 18 14:34:46 ./scripts/feeds help ? Jun 18 14:35:04 -f : Force updating feeds even if there are changed, uncommitted files. Jun 18 14:35:34 this is the part confuses me, my feed has temp changes that has not git-commit yet, is -f for this? Jun 18 14:36:08 of ./scripts/feeds update myfeed will pull in all changes in myfeed anyways? Jun 18 14:36:10 ./scripts/feeds update ? Jun 18 14:36:19 why are you trying to force stuff? Jun 18 14:36:33 force is "just repull from the sources list, don't worry about things I may have changed locally" Jun 18 14:36:56 i'm unsure if 'feeds update myfeed' will update 'git uncommitted' changes, anyway i will test both Jun 18 14:37:08 karlp: i c, thanks Jun 18 14:37:20 so i should avoid -f Jun 18 14:37:28 unles that Jun 18 14:37:31 s what you want to be doing... Jun 18 14:37:43 use the hammer to hammer, use the scissors to cut... Jun 18 14:38:13 i'm making changes to local feeds and don't want to do global feeds update, just update the feeds under modification Jun 18 14:39:50 there's no "local" and "global" feeds. Jun 18 14:40:44 i read somewhere, if i modify source under build_dir/*/mypkg/myfiles, how to 'make package/.../mypkg' directly from build_dir? looks like make compile by default will clean-prepare-compile which will erase my changes under build_dir/, read it somewhere but forgot the steps now... Jun 18 14:41:06 yeah by 'local' i mean my own feeds to play with Jun 18 14:49:02 I would like to install DD-WRT's Broadcom drivers on OpenWRT. How can I do that? Jun 18 14:53:14 ldir: great, thanks Jun 18 14:53:21 logenome: you do that by using dd-wrt. Jun 18 14:53:49 blogic: for 18.06 too please Jun 18 14:53:49 I don't want to since I want to have LuCI and access to OpenWRT repos. Jun 18 14:53:52 ausjke: you can just add a line to feeds.conf with git-src and point to whatever you like. Jun 18 14:54:02 logenome: then you don't get to have dd-wrt's broadcom drivers. Jun 18 14:54:18 Are they hardcoded into the kernel? Jun 18 14:54:47 karlp: that's what i did, just to make sure i can sync-up it without update all other feeds, it works now Jun 18 14:54:48 ausjke: if you don't actually mean a feed, have you read the wiki page on "working with patches" ? it has some of the makefile targets that let you edit and rebuild Jun 18 14:55:48 i used to manually patch 100+ i.mx6 patches as I never find time to use quilt, i should Jun 18 14:56:03 that's in BB days Jun 18 14:56:15 you have a very interesting idea of "never foudn the time" Jun 18 14:56:19 logenome: dd-wrt broadcom drivers only work with dd-wrt kernels, they require a a specific abi Jun 18 14:57:07 karlp: it's about old dog learning new tricks, you know Jun 18 14:57:46 the wiki page on working with pages has been around since AA at least, if you were manually patching hundreds of files, you were defiitely on the losing side of the time equation Jun 18 14:57:54 it's about doing things harder vs smarter, and going faster as a result. Jun 18 14:58:57 actually with git-diff manually generating lots of patches are not that painful, but yes i should and will use quilt Jun 18 14:59:21 git-format-patch that is Jun 18 15:01:13 I found https://raphaelhertzog.com/2012/08/08/how-to-use-quilt-to-manage-patches-in-debian-packages/ as another useful page re quilt and filled in a couple of questions I had after the openwrt quilt page. Jun 18 15:02:00 Does anyone here use NFS root for development? Jun 18 15:04:29 RichardD: no more since 2008 Jun 18 15:05:14 ausjke: Oh I see. What's your build/run/debug workflow look like? Jun 18 15:06:54 oh, fuck me backwards with a chainsaw Jun 18 15:07:02 ldir: I worked out why ltq-atm still builds Jun 18 15:07:09 because it was more fucked than I ever thought Jun 18 15:07:22 that ifx_atm_alloc_tx hack which is exported from the kernel.... was never used *anyway* Jun 18 15:07:31 because ltq_atm.c defines its OWN Jun 18 15:07:36 static struct sk_buff* (*ifx_atm_alloc_tx)(struct atm_vcc *, unsigned int) = NULL; Jun 18 15:07:39 not 'extern' Jun 18 15:07:40 'static' Jun 18 15:07:50 so it's not setting the pointer that the kernel was exporting. Jun 18 15:07:53 never did, afaict Jun 18 15:08:01 and that's why it doesn't break now that the kernel isn't exporting that symbol Jun 18 15:08:09 I... words... not.... Jun 18 15:08:10 while cable-modem is dominant in the US, looks like europe still has more DSL Jun 18 15:09:05 and here ATM means money machine, is ATM gear still used in backbone network these days? Jun 18 15:09:26 hi all - how hard is it to compile packages from source natively in OpenWRT? Jun 18 15:09:27 mostly not Jun 18 15:09:42 I see chaos chalmer includes gcc and make Jun 18 15:09:52 dwmw2_gone: lol - thanks for solving the mysteries :-D Jun 18 15:10:20 My target platform is a RavPower FileHub Plus RP-WD03 (MIPS CPU, 32 MB RAM, 8 MB Flash, SD card slot) Jun 18 15:10:36 ausjke: ATM was very nailed down to that 622 mbits/sec rate. Pretty much everybody and their pet llama went to something that behaved like gigabit ethernet (though it may have been over dwdm or whatever) Jun 18 15:10:56 atm not backbone... but last mile quite popular Jun 18 15:10:58 allotrope94: that's pretty tight for any native build Jun 18 15:11:04 allotrope94: i mean, it's impossible Jun 18 15:11:22 I want to compile bsdgames, nethack, lynx, and gforth, among other things Jun 18 15:11:29 why compile natively? Jun 18 15:11:42 ausjke: there is a snapshot for the rp-wd03 Jun 18 15:12:13 I guess for simplicity. Is the OpenWRT toolchain difficult to set up for a newcomer? Jun 18 15:12:14 allotrope94: 32MB/8MB is just not enough, try it inside a qemu instead for native compiling? Jun 18 15:12:33 ausjke: not enough for native compiling? Jun 18 15:12:42 adding stuff as a package in openwrt is fairly simple Jun 18 15:12:45 Or not enough for OpenWRT altogether? Jun 18 15:12:50 I still haven't worked out how to invoke the toolchain directly :) Jun 18 15:13:01 allotrope94: absoulyte not ,just add them to openwrt Makefile and cross-compile might be the easiet route Jun 18 15:13:13 dwmw2_gone: You want to build natively on the router target rather than cross-compile? Jun 18 15:13:25 valdisk: yes Jun 18 15:13:34 valdisk: $DEITY no, but allotrope seems to be a masochist Jun 18 15:13:52 Unless cross-compiling on a netbook (Lubuntu x86) is much easier :) Jun 18 15:13:56 Which it seems to be :) Jun 18 15:14:28 Can I use old package makefiles from a place like here? https://github.com/jameshilliard/openwrt-packages Jun 18 15:15:26 I guess I would follow these steps for cross-compilation: https://wiki.openwrt.org/doc/devel/crosscompile Jun 18 15:15:50 yesterday read some news: worldwide, 80+% routers are now produced from China Jun 18 15:16:09 Wow. Where for the other 20-%? Jun 18 15:16:25 Taiwan? :D Jun 18 15:16:31 don't know, maybe latvia? Jun 18 15:16:45 Yeah. Thinking that too. Maybe Malaysia also? Jun 18 15:17:14 f00b4r0: politically corrently speaking taiwan is part of china, but i guess that 80% does not include taiwan, if it does, it might be 95%, leave 5% to Latvia Jun 18 15:17:29 ausjke: hence smiley Jun 18 15:18:42 * ausjke hears latvia is the most introvert nation in europe and has the highest IQ on average, thus Routerboard Jun 18 15:19:13 dwmw2_gone: I think jow had a script for doing what you want.... summit like https://pastebin.com/wenkR992 Jun 18 15:20:33 Does Broadcom compile their drivers for DD-WRT only, with no way of adapting them to OpenWRT? Jun 18 15:20:51 No way to write a wrapper? Jun 18 15:23:19 I doubt you'll find anyone here very interested in helping you on this task. Jun 18 15:23:29 you might have better luck asking the dd-wrt people how to install luci? Jun 18 15:24:04 I think I already asked about installing OpenWRT packages on DD-WRT. Jun 18 15:24:30 have not checked dd-wrt for years, is it still 'compatible' with new opewnrt devel? Jun 18 15:26:15 I think OpenWRT and DD-WRT diverged too much. Jun 18 15:26:44 logenome: unfortunately there is no way to write a wrapper as the precompiled object files directly link into kernel internals Jun 18 15:26:52 has kernel 4.14.50 been tested on ADSL with lantiq xrx200/HH5a? I tried 4.14.48, and WANIP disonnects after a small amount of internet traffic Jun 18 15:27:18 the driver is not structured like for example the proprietaty nvidia one where a clean glue layer is present between kernel and driver core Jun 18 15:27:37 DDJ2: have you ever checked bugs.openwrt.org? Jun 18 15:27:49 yes a few weeks ago Jun 18 15:28:10 DDJ2: https://bugs.openwrt.org/index.php?do=details&task_id=1567 Jun 18 15:30:25 yes I saw that, had no WAN IP at all, after 4.14.40 (?)+, on another of same router WAN IP does show but does not last more than 90 seconds, after soon as internet traffic, IP lost, Jun 18 15:32:43 To connect to a client WiFi network and allow SSH over ethernet simultaneously, should I change the default IP from 192.168.1.1 to 192.168.2.1 in /etc/config/network before connecting to WiFi? Jun 18 15:33:10 The client WiFi network being the one generated by my other router Jun 18 15:33:40 ok, i'll try current later, Jun 18 15:34:58 DDJ2: wait for a new build or build your own image. a fresh build of xrx200 is pending Jun 18 15:37:52 jow: what happened to /var becoming writeable? Jun 18 15:38:26 isn't /var still just a link to /tmp? Jun 18 15:40:35 yes Jun 18 15:40:39 mangix: ? Jun 18 15:40:57 i remember talks about making /var writeable Jun 18 15:40:57 * jow didn't touch /var Jun 18 15:41:25 or maybe moving /var/run to /run Jun 18 15:41:25 mangix: talks yes, actions some, results none, plans none Jun 18 15:41:43 got it Jun 18 15:42:13 if that ever comes to be please make it optional. I don't plan on replacing flash chips every now and then ;P Jun 18 15:44:00 what's the reliability of NOR fkash? Jun 18 15:44:17 not great. Jun 18 15:44:19 NAND was 50k writes i think Jun 18 15:44:40 mangix: Let's just say you want to take backups of the device. :) Jun 18 15:44:46 NOR should be more I would think Jun 18 15:44:49 especially when you hit the same spot on small storage devices Jun 18 15:45:09 f00b4r0: Is there still crap out there that doesn't do proper wear leveling? Jun 18 15:45:15 oh because of no wear leveling? Jun 18 15:45:49 valdisk: wear levelling? On raw mtd? Jun 18 15:55:15 Is this the correct channel to ask for device recommendations? Jun 18 15:55:56 one without broadcom drivers only :) Jun 18 15:59:05 mangix: I suspect 50k is being very generous for some of the cheap nand they use in routers heh Jun 18 15:59:28 hi I want to run a custom wifi channel scanning script right after boot, but I cannot get it to work. tried with init.d and rc.local Jun 18 15:59:35 OK, Asus, Netgear or Zyxel, with OpenWRT support DD-WRT optional, up to $300, gigabit LAN, 802.11ac WiFi. Jun 18 15:59:50 is there a recommended way to do it? I guess I have to wait for wlan0 to boot properly up first Jun 18 16:01:22 * blogic gets chainsaw Jun 18 16:01:28 dwmw2_gone: are you sure ? Jun 18 16:02:26 about the ifx_atm_alloc_tx thing? Jun 18 16:02:34 * ldir dons hazmat suit Jun 18 16:02:49 1d0a9d0c047 package/platform/lantiq/ltq-atm/src/ltq_atm.c (John Crispin 2012-12-15 02:00:33 +0000 264) static struct sk_buff* (*ifx_atm_alloc_tx)(struct atm_vcc *, unsigned int) = NULL; Jun 18 16:03:15 i see my name Jun 18 16:03:19 let me have a look Jun 18 16:03:47 I was just really confused by the fact that it built at all Jun 18 16:04:11 Hey everyone. Is this a development-only channel, or may I ask support related questions here? Jun 18 16:04:24 at least, I was confused that ltq-atm.ko loaded and worked, after commit 0276e1f7608 stopped exporting the ifx_atm_alloc_tx symbol Jun 18 16:04:32 dwmw2_gone: That's the sort of line that's legal in C, and occasionally called for, but usually does something other than what you thought it did. ;) Jun 18 16:04:35 so I stared and stared at it until I realised it never ever worked in the first place Jun 18 16:04:42 dwmw2_gone: ah ok Jun 18 16:04:42 @Axtroz You can ask anything OpenWRT-related Jun 18 16:04:54 dwmw2_gone: do you know what the ppe32 is ? Jun 18 16:05:06 its the packet processing engine 32bit fuck Yeah ! Jun 18 16:05:13 :) Jun 18 16:05:19 its the flow offloading HW inside lantiq HW Jun 18 16:05:33 dwmw2_gone: YOu don't need to export a symbol if something in the same compilation unit takes the address and passes the pointerout to elsewhere.. Jun 18 16:05:34 there is this amazing shit nugget piece of driver with only 80k lines Jun 18 16:05:47 the driver uses that callback Jun 18 16:05:57 valdisk: I know. Jun 18 16:06:05 Thanks. I'm building for x86-64 but the combined image gets stuck on waiting for root partition. I tried booting a ramfs image, and while that boots, keyboard input doesn't work. Any hints what could be wrong? Jun 18 16:06:22 and at the time i made the atm driver work by simply defining the callback pointers inside the driver rather than the kernel hackery that is used to hook atm drivers and/or shit nugget drivers into it Jun 18 16:06:37 blogic: I deliberately didn't look that hard. I was only establishing that the ifx_atm_alloc_tx hook was redundant Jun 18 16:06:38 i could tell you a story but would get into legal issues Jun 18 16:06:50 that particular hook was never used Jun 18 16:07:13 afair no Jun 18 16:07:19 even if the hooking had worked, it would never have been invoked since we don't end up in vcc_sendmsg() for the PPPoATM case via pppoatm_send() Jun 18 16:07:28 as it told a in kernel pointer about a function that the offloading layer used Jun 18 16:07:30 we can kill it all with fire Jun 18 16:07:42 dwmw2_gone: no Jun 18 16:07:51 be more dominant Jun 18 16:07:52 right. But nothing ever changed that pointer from NULL. Not even ltq-atm, which *tried* Jun 18 16:08:01 acid, then napalm, then nuke and then LOIC Jun 18 16:08:14 and then get a gog to poop on the place it once was Jun 18 16:08:20 What really confused me was ldir killing the *hook* with fire and then the ltq-atm module still working Jun 18 16:08:23 that'll do it what it deserves Jun 18 16:08:25 but I understnad that now :) Jun 18 16:09:16 sadly, LTQ does write pretty good software compared to the other vendors Jun 18 16:09:28 which sheds a horrible light on them Jun 18 16:10:01 any experience running ash scripts on boot? Jun 18 16:10:11 right, that being sorted i think its time to look at some more github PRs Jun 18 16:10:12 :) Jun 18 16:10:16 hey, at least we have code! Jun 18 16:10:26 dwmw2_gone: hurra Jun 18 16:10:28 we should actually kill the ltq-atm side too Jun 18 16:10:40 my concern is "Have I done the right thing(tm)" or have I done an Eric and left a ticking timebomb? Jun 18 16:10:43 alx__: boot is async, rc.local might run before the network even started, better register hotplug handlers in /etc/hotplug.d/ to do stuff Jun 18 16:10:47 i lost faith in the industry liberating that dsl codebase Jun 18 16:10:54 :) Jun 18 16:10:56 dwmw2_gone: that was 3 years hard work Jun 18 16:12:48 thanks jow ill try that :) Jun 18 16:17:59 @karlp I would like a device recommendation. Asus, Netgear or Zyxel, up to $300, 802.11ac. Jun 18 16:19:03 I have been considering Netgear R8000. Jun 18 16:19:12 Any idea why USB Support is empty in my x86-64 configuration? Jun 18 16:25:43 blogic: You had faith in QCA at some point? Silly boy.. :) Jun 18 16:27:56 Monkeh: i like to think good of people Jun 18 16:28:13 i can see the beauty in an ugly dog with 2 legs Jun 18 16:28:14 People maybe, big companies.. :) Jun 18 16:28:24 Axtroz: Umm.. because you didn't tell it to include USB? :) Jun 18 16:28:35 Monkeh: i was talking about the developers Jun 18 16:28:45 they are only people Jun 18 16:29:17 i had the joy of snooping inside qsdk 5.3 from fall 2017 today Jun 18 16:29:28 * Monkeh passes blogic the spoon Jun 18 16:29:32 and they managed to outcrap themselves Jun 18 16:29:46 Let's get those eyeballs out Jun 18 16:29:52 <[mbm]> sounds about right Jun 18 16:29:57 they are gone already Jun 18 16:30:07 [mbm]: aloha Jun 18 16:30:11 Did they burst into flames? Jun 18 16:30:15 <[mbm]> hola Jun 18 16:30:24 [mbm]: while you are here, nss is a dual core ubicom32 ? Jun 18 16:31:06 <[mbm]> nnss = network subsystem; it's on the qcom krait cores and is a modified ubicom core Jun 18 16:31:17 ok Jun 18 16:31:26 and it runs the ubicom rtos inside Jun 18 16:31:34 <[mbm]> sorta Jun 18 16:31:37 well a modified version Jun 18 16:31:38 ok Jun 18 16:31:46 and long shot, maybe you know ... Jun 18 16:32:04 when i bring upt he nss the ethernet irqs go away and a nss irq comes up that fires endlessly Jun 18 16:32:13 low cpu load but lots of irqs Jun 18 16:32:20 <[mbm]> originally ubicom was BE but qcom changed it to LE and rewrote a bunch of stuff Jun 18 16:32:23 do you happen to know if that is expected ? Jun 18 16:32:45 i would have expected it to fire only upon events Jun 18 16:33:00 but even in idle state i am seeing ~10khz irqs Jun 18 16:33:12 <[mbm]> idea behind the nss is sort of a programmable ethernet filter Jun 18 16:33:24 all works, able to pass tcp with wirespeed at 128 byte frames with 200 flow Jun 18 16:33:27 all well Jun 18 16:33:34 logenome: you already asked, I have no recommendation for you, tagging me explicitly won't make me have one all of a sudden :) Jun 18 16:33:37 and i'd expect mailbox irqs to be high Jun 18 16:33:53 but i am puzzled why they fly so high in idle state Jun 18 16:34:16 <[mbm]> so that when you set the equiv of an iptables reject rule that can happen directly at the lowlevel without needing to go all the way through the linux network stack Jun 18 16:34:30 correct Jun 18 16:34:43 i saw that you can even do dpi and all sorts of corner case stuff Jun 18 16:34:48 seems very feature complete Jun 18 16:34:59 still black box magic Jun 18 16:34:59 hallelujah: 4.14.50 works on xrx200 Jun 18 16:35:06 Why do I read that as 'very full of holes'? :) Jun 18 16:35:11 * [mbm] did work for ubicom and qcom but can't really remember all the details Jun 18 16:35:38 [mbm]: no worries Jun 18 16:36:03 i am trying to figure out how the wifi offloading works and evaluating if we can combine this with mac80211 Jun 18 16:36:11 what's everyone's source browser tool these days? used source-navigator that can track function calls but it's not updated for years, not really like vim/ctags/emacs for that Jun 18 16:36:25 @karlp You said you can't recommend anything in private. Jun 18 16:36:26 <[mbm]> ubicom rtos is ipOS and is actually rather minimal Jun 18 16:36:35 That's all you said. Jun 18 16:36:36 no, I said I won't even talk to you in private :) Jun 18 16:36:42 Oh yes, blogic, you were going to email me about a serial adapter project? Jun 18 16:36:58 if I had anything to answer with, I would have just answered in the channel. Jun 18 16:37:01 Monkeh: and then i went on vacation and am fighting backlog :-) Jun 18 16:37:15 Well, just making sure you remembered as well :) Jun 18 16:37:26 _I_ have a zbt we1326, that I'm happy with, but that doesn't match your requirements Jun 18 16:37:53 karlp: how on earth you can get 300mbps on zbt1326 assuming we're having the same hw and sw? Jun 18 16:38:09 on the 5ghz. Jun 18 16:38:20 I just report what the speed test apps say on my phone. Jun 18 16:38:32 2.4gig is slowwww but works. 5gig is great. Jun 18 16:38:46 in that case i hooked a laptop to its lan port, use phone speed test, the best is under 80mbps Jun 18 16:39:02 * karlp shrugs Jun 18 16:39:11 do you actually have > 80Mbps upstream? :) Jun 18 16:39:19 no just download Jun 18 16:39:31 curl -o /dev/null 100MB.file Jun 18 16:39:46 blogic: is qca8k coming on the r7800 for the release? Jun 18 16:39:58 mangix: nope Jun 18 16:40:04 we will tag rc1 tonight or tomorrow Jun 18 16:40:13 i'll do a last round of cherry picks today Jun 18 16:42:21 * dangole quickly squash-picks the oxnas reboot into 18.06 before blogic creates rc tag Jun 18 16:42:48 dangole: hah Jun 18 16:42:56 hope i did not catch you of guard there Jun 18 16:43:53 i ask since iperf results with qca8k sewm horrible Jun 18 16:44:29 around 400mbps with offloading enabled Jun 18 16:44:39 hw or sw offloading Jun 18 16:46:11 need a usb dongle to test 11ac Jun 18 16:46:46 blogic: sw Jun 18 16:46:55 qca8k doesn't do hw? Jun 18 16:46:57 quite a lot usb3 wifi AC1200 adapter at amazon, need make sure linux supports them Jun 18 16:49:48 https://www.wirelesshack.org/top-linux-compatible-usb-wireless-adapters.html best usb wifi for pentesting Jun 18 16:52:54 why does space rat keep comming in and then out again Jun 18 16:53:10 logenome wrt1200 wrt1900ac or wrt3200acm Jun 18 16:55:12 Tapper: /ignore #openwrt-devel MODES JOINS PARTS QUITS Jun 18 16:58:02 blogic: May I trouble you to cherry-pick d600de3ddde269bf0b324735f8f12278f82d9b37 for 18.06. Ideally pppoatm ought to work on x86, (lantiq is 4.9 on 18.06 so is unaffected at present) :-) Jun 18 16:58:24 ldir: sh0re Jun 18 16:58:45 Fanx Jun 18 16:59:02 no barny Jun 18 16:59:38 hey guys, since you're talking about 18.06, are there any know issues with 17.01.4 about wireless clients not having connectivity? Jun 18 17:00:23 dangole: is this true for oxnas ? https://github.com/openwrt/openwrt/pull/1062 Jun 18 17:00:24 i mean, they do concect to the wireless network, it's just they can't "reach" anywhere... Jun 18 17:00:40 do they get dhcp ? Jun 18 17:02:07 blogic: no, nothing Jun 18 17:02:31 dhcp is from another router tho Jun 18 17:03:00 and the wifi interface is inside the bridge Jun 18 17:03:02 ? Jun 18 17:03:05 yep Jun 18 17:03:06 brctl show Jun 18 17:03:08 odd Jun 18 17:03:21 and you are using what HW ? Jun 18 17:03:40 # brctl show Jun 18 17:03:42 bridge name bridge id STP enabled interfaces Jun 18 17:03:43 br-lan 7fff.e894f6c41586 no eth0.1 Jun 18 17:03:45 wlan0 Jun 18 17:03:46 wlan1 Jun 18 17:03:53 TP-Link TL-WDR4300 v1 Jun 18 17:04:09 Tapper it's $340 where I live. Jun 18 17:10:03 <[mbm]> blogic: forgot to mention earlier but a lot of the ubicom code got dumped on code aurora when qcom bought ubicom if you're really interested in the ubicom internals Jun 18 17:11:05 [mbm]: yeah i already found some of the stuff Jun 18 17:12:28 <[mbm]> nss happened while ubicom was workong on the ip8k Jun 18 17:15:11 karlp I'm wary of buying a router from Aliexpress because it may contain a unremovable Chinese backdoor to be honest. Jun 18 17:16:43 sorry blogic was on the phone. What does that meen? Jun 18 17:17:52 logenome what about the 1900ac or the wrt1200? Jun 18 17:18:00 Tapper: to remove the join/leave messages on irssi Jun 18 17:18:12 logenome: yes, the mt7621 in _my_ device is irrecoverablly backdoored, but the mt7621 in _your_ "freedom filled" device, is somehow clean.... Jun 18 17:18:41 karlp: /man ignore Jun 18 17:18:51 what's irssi? Jun 18 17:18:56 What about other parts of the device, like the pre-flashed firmware? Jun 18 17:19:00 Tapper: ok, you use a different irc client Jun 18 17:19:15 How can you replace that without relying on it for replacement? Jun 18 17:19:18 karlp: the most misunderevaluated feature of the interwebz Jun 18 17:19:22 I use instabird Jun 18 17:19:26 Directly connecting to the flash pins? Jun 18 17:22:42 Tapper Still expensive. And ugly. Jun 18 17:22:43 You give them too much credit. Jun 18 17:43:21 seems like SpaceRat has some problem with connection to this channel :-) Jun 18 17:52:05 Why are Linksys routers so ugly? Jun 18 17:52:38 logenome: Some industrial designer thought that making them "distinctive" would help market share. Jun 18 17:53:03 giant spiders Jun 18 17:54:16 I remember there were some issues with stability on Marvell chips. Jun 18 17:54:29 Not sure if I want to get Linksys because of that. Jun 18 17:57:04 logenome: Hmm.. dunno. My Linksys wrt1200ac seems reasonably stable except when I stuff new lede-snapshot images on it ;) Jun 18 18:00:58 [mbm]: hello! Jun 18 18:03:52 Hi all, I just noticed 18.06rc1 was pushed back to 24th of this month - is the 18.06 branch now being built with LuCI in prep for rc1 or are the snapshots still minimal? Jun 18 18:07:45 dwmw2_gone: & blogic https://patchwork.ozlabs.org/patch/931120/ Jun 18 18:10:21 who takes care about look of router? It is device hidden somewhere running 24/7, where look is las item in the priority list :-) Jun 18 18:10:53 It sits on my desk. Jun 18 18:11:17 I'm not sure I want to hide it. Jun 18 18:12:19 you are the first one who takes care about look of hist router. My god... :-) Jun 18 18:13:11 And if I can't find Linksys or ZBT? Jun 18 18:14:04 asus or tp-link :-) Jun 18 18:14:21 <[mbm]> hello agb Jun 18 18:15:24 ultito Netgear too maybe? Jun 18 18:17:13 it is up to you, but don't forget to choose model where you can run openwrt :-) Jun 18 18:17:38 That's obvious. Jun 18 18:17:46 But WHAT model? Jun 18 18:18:13 I wanted AC68U or Netgear R7000, but they don't have Wi-Fi with OpenWRT. Jun 18 18:49:35 oh nice, didn't know the gl-ar150 had already been ported to ar79/devicetree Jun 18 18:49:54 bleep bloop bailing on ar71xx Jun 18 18:55:16 ultito Any ideas? Jun 18 19:08:03 gzc__: :-) Jun 18 19:08:11 gzc__: one of the first devices i ported Jun 18 19:08:44 neat! thank you :D trying to figure out how to Contribute more and trying to catch up Jun 18 19:11:17 worste case contribute test coverage Jun 18 19:12:45 build #27 of oxnas/generic is complete: Failure [failed signpack] Build details are at http://release-builds.openwrt.org/18.06/images/builders/oxnas%2Fgeneric/builds/27 blamelist: Koen Vandeputte , Daniel Golle Jun 18 19:13:48 neat, I'll start looking at how tests work here Jun 18 19:14:59 any chance there's a donation bin? i'd love to get some maintainer a gl-ar750, kinda my favorite thing right now Jun 18 19:21:04 wigyori: Ok I just saw that blogic merged the sunxi patches Jun 18 19:21:22 I tested them on the boards I have, but I do not have the old sunxi boards Jun 18 19:21:49 hopefully it works, otherwise people will complain ;-) Jun 18 19:22:44 Hauke: prepend stuff with RFC or mark them as under review or assign them to yourself Jun 18 19:22:58 then i'll know in future Jun 18 19:23:16 ldir: what shalt i cherry-pick Jun 18 19:23:33 blogic: no problem Jun 18 19:23:37 I will do that next time Jun 18 19:23:46 sorry did not realize Jun 18 19:23:50 was on a merge frenzy Jun 18 19:24:10 Merge it all and let the users sort it out! :D Jun 18 19:24:57 blogic: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=d600de3ddde269bf0b324735f8f12278f82d9b37 if you would kind Sir Jun 18 19:26:02 if you're on a merge frenzy, mind checking out https://github.com/openwrt/openwrt/pull/957 ? I think I got an irc-based ack from jow and nbd. Jun 18 19:28:46 agb: and i guess both to 18.06 aswell ? Jun 18 19:30:36 that would be great! it's a minimal enough change that it shouldn't introduce any 18.06 regressions and means I can drop my local patch-tree to build on busybox-based distros Jun 18 19:31:11 ldir: done Jun 18 19:31:57 agb: done Jun 18 19:32:19 blogic: magic - that'll make dwmw2 happy :-) Jun 18 19:32:37 ldir: great, mission of the day accomplished Jun 18 19:32:58 ldir: admitedly, i am getting a vdsl/atm line next month Jun 18 19:33:10 otherwise i'd not give a donkeys ... :P Jun 18 19:33:27 vdsl with atm? Jun 18 19:33:37 that is very uncommon in germany Jun 18 19:33:38 ah no, that would be ptm Jun 18 19:33:53 blogic: danke. Jun 18 19:33:55 all the merging effort for nothing, darn :P Jun 18 19:34:03 you getting adsl/atm then Jun 18 19:34:12 ldir: no vdsl/ptm Jun 18 19:34:19 loosing my isdn unfortunatley Jun 18 19:34:21 you should swithc to ADSL 16MBit/s ;-) Jun 18 19:34:28 Hauke: i have 16 mbit Jun 18 19:34:30 then you get ATM Jun 18 19:34:34 well 7/1 in reality Jun 18 19:34:46 and i'll get vdsl with vectoring Jun 18 19:34:51 so in theory 100.20 i think Jun 18 19:34:55 100/20 Jun 18 19:35:04 but we'll see Jun 18 19:35:13 nice Jun 18 19:35:17 i dont trust german ISPs Jun 18 19:35:17 my parents have 2/.5 on a good day Jun 18 19:35:31 ldir: must be a first world country Jun 18 19:35:43 two different companies want to deploy fiber at my parents Jun 18 19:35:56 i heard that civilisation has an inverse relation to absolute bandwidth Jun 18 19:35:59 with 1.5s bufferbloat in the downstream direction. Jun 18 19:36:18 currently they only have VDSL and will soon get vectoring, they are living in a small village Jun 18 19:36:25 ldir: admit it, its a gprs link Jun 18 19:36:43 so if you actually *use* the line to download anything it's unusable. Jun 18 19:36:53 * blogic looks out the window, cow fields ... Jun 18 19:36:58 Hauke: yep, small village here too Jun 18 19:37:23 right, i was wondering Jun 18 19:37:29 Now that pppoatm on a hh5a works again..... maybe I can actually install it and put in some bufferbloat mitigation. :-) Jun 18 19:37:33 we have ath10k and ath10k-ct Jun 18 19:37:46 we might want to consider a mac80211 and a mac80211-nbd Jun 18 19:37:54 lol Jun 18 19:38:02 so that people can choose to just use vanilla compat-wireless Jun 18 19:38:35 also goes for some other features like the mc->uc bridge code Jun 18 19:38:48 I was wondering if we should make ath10k-ct the default Jun 18 19:38:53 at least we get support from ben Jun 18 19:39:01 basically allowing us to enable the "magic features" no one knows about what they do how toey work and that they even exist Jun 18 19:39:12 Hauke: correct Jun 18 19:39:43 this qca FW always crashes for me Jun 18 19:40:19 i was about to start a qca rant Jun 18 19:40:22 dont trigger me Jun 18 19:40:53 I read it in the IRC history but haven't found the context Jun 18 19:41:13 conetxt is that i tried to do something with qsdk5.3 today Jun 18 19:41:21 and i nearly lost all faith in humanity Jun 18 19:42:21 seriously the people that thought up that openwrt integration design should be banned from any electronic equipment for live Jun 18 19:42:29 and all their spouses for 10 gens to come Jun 18 19:42:56 see what you did ? now you got me ranting Jun 18 19:43:14 blogic: I'd appreciate a sanity check of https://patchwork.ozlabs.org/patch/931120/ if it diverts from a full on rant :-) Jun 18 19:43:33 sorry ;-) Jun 18 19:44:47 blogic: dwmw2 thinks the commit message is too nice lol :-) Jun 18 19:45:56 It is not not worth to rant about shitty software, just try to stay as far away as possible from it, at least that is my current strategie Jun 18 19:46:12 I tried to improve the software but that is very hard Jun 18 19:46:41 when you are not officially in charge Jun 18 19:47:41 ldir: done Jun 18 19:47:52 Hauke: correct Jun 18 19:48:05 uh ... can we have gptfdisk for 18.06, pretty please? https://github.com/openwrt/packages/pull/5621 has been waiting for love since february. Jun 18 19:48:27 Hauke: I have a similar situation at a broadcast facility at which I work. When I'm there I'm responsible for it but not in charge, so my ability to actually apply sensible changes is challenging and ultimately very stressful. Jun 18 19:48:28 drmr: that is the packages feed Jun 18 19:48:45 I am aware. so ... noone cares, then? Jun 18 19:49:02 i did nto say that Jun 18 19:49:29 sorry, that came across as snide. Jun 18 19:49:58 blogic: oh I didn't expect you to commit it !! :-) Brilliant! Thank you! Jun 18 19:50:07 imho, gdisk is rather important for anyone using openwrt on a NAS or as a NAS. Jun 18 19:50:15 ldir: fixed the commit message Jun 18 19:50:40 yeah I noticed :-) Kept the commit title - laughing. Jun 18 19:50:53 sooo Jun 18 19:50:55 github Jun 18 19:50:58 "base-files: add new backup options" https://patchwork.ozlabs.org/project/openwrt/list/?series=37466 is a series I'd like to see merged (not necessarily for 18.06), while I'm not the author, I've been using it since April and it does help a lot to keep my overlay and config backups sane Jun 18 19:51:04 (fdisk can do gpt, but it's a major discomfort in the posterior to use. Jun 18 19:51:10 the big green button Jun 18 19:51:20 i use "rebase and merge" right ? Jun 18 19:51:39 and packages master is on github and i dont need to merge this to a server and have it mirrored Jun 18 19:52:02 oh shoot, I have no idea honestly. (it's not my pull request, btw.) Jun 18 19:52:39 * ldir has no idea - they don't give me commit rights (*everyone* sighs in relief) Jun 18 19:52:59 rebase and merge will yield a linear history with the individual commits of the PR stacked on top of the current tip. Jun 18 19:54:07 ^^^ that sounds good... no infuriating merge commit. Jun 18 19:54:43 But more commits looks more active! Jun 18 19:54:45 you don't like the commit graphs that look like Los Angeles freeways? Jun 18 19:55:02 Monkeh: and you wonder why people hate you ;-) Jun 18 19:55:03 agb: what do those freeways look like Jun 18 19:55:21 drmr: try now, i think it might have worked Jun 18 19:55:38 pkgadd: ahh yes Jun 18 19:55:44 * Monkeh sprinkles a few DIP packages on ldir's floor Jun 18 19:55:44 pkgadd: its on my list Jun 18 19:55:46 looks legit Jun 18 19:55:51 :) Jun 18 19:55:53 scared to merge it to 18.06 Jun 18 19:56:09 potentially breaking sysupgrade this late .... Jun 18 19:56:10 blogic: freeways of 3-6 lanes in each direction, criss-crossing all over each-other, often with large fly-over junctions. https://en.wikipedia.org/wiki/Southern_California_freeways#/media/File:SocalfreewaysystemWIKI.jpg Jun 18 19:56:23 I'd skip it for 18.06, but master would be great Jun 18 19:56:26 Monkeh: and I'm in bare feet!!!! :-) Jun 18 19:57:00 agb: I was going to say 'a flattened ball of spaghetti' Jun 18 19:57:49 pkgadd: yep, i'll ask jow about ti tomorrow Jun 18 19:58:00 thanks a lot Jun 18 19:58:05 pkgadd: i've had the series open various times and had it in my local tree already Jun 18 19:58:38 Speaking of sysupgrade Jun 18 19:59:04 Is there a reason the metadata doesn't include a checksum? Jun 18 19:59:14 yes Jun 18 19:59:28 no one thought of doing so yet, i believe that is the reason Jun 18 19:59:34 Oh, okay. Jun 18 19:59:42 nah prolly because of the signing Jun 18 20:00:05 signature will fail i guess Jun 18 20:00:19 Monkeh: pretty much: https://www.google.com/maps/@34.032311,-118.2192822,1810m/data=!3m1!1e3 Jun 18 20:01:31 there's at least a dozen interchanges like that across the LA area. https://www.google.com/maps/@33.9297663,-118.2818415,1279m/data=!3m1!1e3 Jun 18 20:02:00 blogic: I didn't think the images themselves were signed? Jun 18 20:02:27 agb: i cant help to be reminded of sim city Jun 18 20:02:34 all looks to rasterized and neat Jun 18 20:03:05 Monkeh: that is what ucert is for i believe Jun 18 20:03:19 but i might be totally off Jun 18 20:03:37 Ah, that was just added the other day, I recall spotting that Jun 18 20:04:38 Anyway, back to figuring out why sysupgrade is making a mess of saving configs in this branch.. Jun 18 20:05:18 Or more likely mtd Jun 18 20:06:28 blogic: this is what happens when cities are built after the invention of the horseless cart and a nation needs to keep industry going in the void of post-WWII peace. :-/ Jun 18 20:08:39 agb: we need a war Jun 18 20:08:43 that is the solution Jun 18 20:09:17 blogic: could you set http://patchwork.ozlabs.org/patch/930522/ to accepted ? Jun 18 20:09:19 funny places Jun 18 20:09:33 glamour glow shopping Jun 18 20:09:38 free walking jesus Jun 18 20:09:42 panda express Jun 18 20:09:51 seaside laundromat Jun 18 20:09:58 anarchy motor repairs Jun 18 20:10:16 trailer terrace mobile park ... Jun 18 20:10:31 we found the qca crack supply place Jun 18 20:10:46 blogic: can you convince Merkel to take over Britain once they Brexit? maybe then Trump will call her a bad hombre and get Putin and Kim Jong Un to join him in fighting the bad hombres. Jun 18 20:11:00 merkel is about o loose her job Jun 18 20:11:10 germany plotocs are in turmoil Jun 18 20:11:26 one one side we have the "we are not nazis and to prrof it we take all refugees Jun 18 20:11:30 but who will keep Germans safe from themselves? Jun 18 20:11:51 and on the other side you have the "we are not nazis but we dont want refugees" crowd Jun 18 20:12:06 agb: horst will i guess Jun 18 20:12:13 crap knows Jun 18 20:12:18 i am not even bloody german Jun 18 20:12:53 anyhow, german politics is in the process of imploding, its funny to watch Jun 18 20:13:14 right, back to work .... i know i should not have eaten that clown earlier today Jun 18 20:13:30 tasted weird, should have been a dead give away Jun 18 20:14:03 blogic: Has their government ragequit yet? :P Jun 18 20:14:21 anxiously awaiting it to be honest Jun 18 20:14:32 Brexit was funny. "Err.. hmm.. I didn't expect that result." *quit* Jun 18 20:15:24 it did not even took 24h for them to say that they do not want to fund NHS with the money they "save" with Brexit Jun 18 20:15:35 in Germany this normally takes longer Jun 18 20:16:16 Tapper: ENGLAND Jun 18 20:20:31 PANAMA! Jun 18 20:20:56 did I win? Jun 18 20:21:25 panama is an easy win Jun 18 20:21:53 yeah, it's a good Van Halen song. Jun 18 20:23:31 jwh yes mate good game there and I was all so eating a big fat stake my wife made for me so tapper is one happy camper lol Jun 18 20:33:42 Tapper: lol, pub was mental Jun 18 20:33:57 build #867 of ramips/rt305x is complete: Failure [failed pkgbuild] Build details are at http://phase1.builds.lede-project.org/builders/ramips%2Frt305x/builds/867 blamelist: Vincent Wiemann , John Crispin , Koen Vandeputte , Linus Walleij , Magnus Fr?hling Jun 18 20:33:57 , Lucian Cristian , Kevin Darbyshire-Bryant , Hannu Nyman , Daniel Golle , Paul Spooren , Jeremiah McConnell , Rafa? Mi?ecki , Daniel Engberg , Hauke Jun 18 20:33:57 Mehrtens , Sven Roederer , Weijie Gao , Andy Boyett , Chen Minqiang , Felix Fietkau , Thomas Nixon , Tomasz Maciej Nowak , Krystian Kozak , David Bauer , Marty Jun 18 20:33:57 E. Plummer , Rosy Song , Zoltan HERPAI , Mirko Parthey Jun 18 20:34:07 Tapper: panama easy win Jun 18 20:34:25 may be Jun 18 20:34:37 who added kmod-chaoskey ? Jun 18 20:34:44 it is broken ;-) Jun 18 20:35:06 Hauke: hannu did Jun 18 20:35:08 i merged it Jun 18 20:35:47 blogic: will you fix it or should I look into this? Jun 18 20:35:56 Hauke: i'd revert it Jun 18 20:36:02 if you want to fix it go for it Jun 18 20:36:21 I will fix it Jun 18 20:36:31 ok, looks like deps are missing Jun 18 20:41:56 Tapper: ah well either way, get some fun coz they won :D Jun 18 20:42:15 well i hope so, shes being a bit cagey Jun 18 20:45:41 blogic: one of the issues with german politics is that they are so damn obsessed with that "schwarze null" Jun 18 20:52:08 blogic: fixed Jun 18 20:52:17 Hauke: cheers Jun 18 20:52:26 lets wait for the next problem Jun 18 20:52:36 ath79 blows up aswell Jun 18 20:52:41 building a tree just now Jun 18 20:56:53 build #47 of gemini/generic is complete: Failure [failed images] Build details are at http://phase1.builds.lede-project.org/builders/gemini%2Fgeneric/builds/47 blamelist: Vincent Wiemann , Magnus Kroken , John Crispin , Koen Vandeputte , Linus Walleij Jun 18 20:56:53 , Magnus Fr?hling , Lucian Cristian , Kevin Darbyshire-Bryant , Hannu Nyman , Daniel Golle , Paul Spooren , Jeremiah McConnell , Rafa? Mi?ecki , Daniel Jun 18 20:56:53 Engberg , Hauke Mehrtens , Sven Roederer , Weijie Gao , Andy Boyett , Chen Minqiang , Felix Fietkau , Thomas Nixon , Tomasz Maciej Nowak , Krystian Kozak Jun 18 20:56:53 , David Bauer , Marty E. Plummer , Rosy Song , Zoltan HERPAI , Mirko Parthey Jun 18 20:56:55 I installed openwrt on my Ubiquity ER-X-SFP. I get these errors in dmesg "Data buffer not 16 bytes aligned: 8ee4c178 Jun 18 20:57:36 OpenWrt SNAPSHOT, r7263-b1a173d7c3 Jun 18 20:58:33 this was probably caused by one of Linus patches Jun 18 20:59:06 Hauke: yep Jun 18 20:59:09 one at a time Jun 18 20:59:20 was to be expected by that pile of patches Jun 18 20:59:47 yes Jun 18 21:00:21 triggered with "opkg update, opkg install luci" Jun 18 21:12:32 Monkeh: lol i just recognized the name searching for some rigol 1052e scope stuff iirc Jun 18 21:21:53 Rene__: this looks to be coming from target/linux/ramips/patches-4.14/0039-mtd-add-mt7621-nand-support.patch and is only an info message, the data is copied to a 16 bytes aligned buffer if required alignment isn't met and then written to NAND Jun 18 21:33:12 Rene__ hi Jun 18 21:57:08 ldir: Ok. Jun 18 21:57:23 Tapper: Hi Jun 18 22:14:25 Hauke: i am too stupid to figure out the gemini issue Jun 18 22:14:30 its parallel build related Jun 19 00:04:41 hey just wondering if anyone has the same problem: with lantiq dsl and vdsl2 vectoring the snr magin just slowly counts down along with the sync rate Jun 19 00:05:43 the snr margin can go back up if it goes down sometimes Jun 19 00:06:13 like if there's genuine interference, but there also seems to be a constant eating of snr Jun 19 00:07:34 and it never goes back up. Like I leave a conection going for 10 days, the sync rate will go from 95mbit downstream to 73 and if I reboot the modem it will be 95 again Jun 19 00:08:29 a broadcom based modem doesn't have the same problem Jun 19 00:08:47 I cannot confirm this on my vectoring line with lantiq, sorry. Jun 19 00:09:24 you in australia ? Jun 19 00:09:32 germany. Jun 19 00:09:46 what is your max sync rate ? Jun 19 00:10:26 nvr mind if u dont want to answer for privacy purposes Jun 19 00:10:28 Max. Attainable Data Rate (ATTNDR): Down: 178.491 Mb/s / Up: 89.492 Mb/s Jun 19 00:10:55 Data Rate: Down: 99.998 Mb/s / Up: 43.993 Mb/s Jun 19 00:11:06 so, capped by my profile. Jun 19 00:11:12 is that on a xrx200 ? Jun 19 00:11:18 yup. Jun 19 00:11:43 a TP-Link W8970. Jun 19 00:13:53 yeah it's strange with my line if I use the openwrt with vectoring firmware the speed is capped at 95,767 exactly, every time on first sync. But with the broadcom device it will go over 100mbit and stay there Jun 19 00:14:20 i'm just wondering if there's some pipe dsl command I can enter to turn on sra or something Jun 19 00:30:01 95,767 may not be the exact number, but it's close Jun 19 02:31:44 Does anyone @here know-how I can get openwrt running on a NetApp FAS3020? I have no idea how to Jun 19 02:32:13 Build a OS image for this CFE bootloader they use. Jun 19 02:33:30 Any help getting openwrt running on a BetApp FAS 3020 GREATLY appreciated!! Jun 19 02:34:09 Rhcev6: have patience. not everyone who might be familiar with the hardware is going to always be at their computer Jun 19 02:34:14 My username here is also my Gmail username for email. Jun 19 02:34:30 Rhcev6: you should start by looking at the table og hardware (ToH) on the wiki Jun 19 02:34:39 s/og/of/ Jun 19 02:37:06 Well it is basically a xeon x86 board. What is the wing me is the bootloader. If I could just figure out how to create an OS image for it to load from CFor even just figure out how to build a grub2 binary I'd be set. Jun 19 02:37:36 then i would definitely look at the wiki. Jun 19 02:37:59 there is a lot of documentation readily available, and you can even search it all Jun 19 02:38:12 NetApp FAS3020 uses a modded Broadcom CFE 1.x bootloader. Jun 19 02:38:46 ...broadcrap. Jun 19 02:39:31 @fishman Do you happen to have a link to anything related to working with or creating things for the Broadcom CFE? Jun 19 02:39:39 i personally have no idea what their CFE is like, but their wifi sucks for anything open source Jun 19 02:39:55 Rhcev6: again, no, you should search the wiki. Jun 19 02:41:02 Rhcev6: and to clarify my position on broadcrap: if it just works or takes less than 15 minutes, great. if not, i get rid of it and move on with my life Jun 19 02:41:46 I saw some bits scattered here and there that mention the CFE which I think might be for SOC routers. Maybe building an image for those is the same tools and procedures as any Broadcom based CFE? Jun 19 02:42:39 i have no idea. search the wiki and forums while you hang out here to see if anyone else can offer insight Jun 19 02:44:45 The hardware isn't Broadcom at all. NetApp just grabbed source code for the CFE from Broadcom and use it in their FAS3020 BUT it is a dual Intel Xeon 400mhz FSB board with 4Gb RAM. Jun 19 02:46:15 I signed up for dev mail list access and was just going to see is anyone there might know but figured I'd check in here too. Jun 19 02:50:20 On a side note sad to see no support for wrt54g v3. I don't need the full router bits. All I need is bare OS and any add on's to be able to create a wifi mesh network so ony cell I can go anywhere on my property and seamlessly switch to the strongest signal with highest through put without the hugecisco aironet investment Jun 19 02:51:35 Someone just gave me 8 wrt54g v3 units with 15 ram and 4 flash. **** ENDING LOGGING AT Tue Jun 19 03:00:03 2018