**** BEGIN LOGGING AT Thu Nov 05 02:59:57 2020 Nov 05 06:32:28 ynezz: please join https://gitlab.com/openwrt/web/firmware-selector-openwrt-org/-/issues/3 Nov 05 08:32:02 svanheule[m]: ping Nov 05 08:40:14 is it normal if u do a "caldata_extract "art" 0x5000 0x844" and your openwrt device will creash? Nov 05 08:44:09 Borromini: o/ Nov 05 08:47:36 o/ Nov 05 08:49:07 aparcar[m]: BTW any decisions should be held in public, on designated and already estabilished channels, please use openwrt-adm@ Nov 05 08:52:50 aparcar[m]: Just a heads-up, I saw yesterday you were following me on a GitHub account I just deleted. I was finally able to recover my main account, yay! :) Nov 05 08:53:40 ynezz: okay Nov 05 08:53:58 rsalvaterra: I lost track. who should I follow? #follow4follow Nov 05 08:54:26 aparcar[m]: I've disabled notifications on this project explicitly, as 1) I'm not interested in this web/frontend stuff 2) I believe, that in the current CI/CD state anyone with the proper permissions should be able to accept/deploy changes Nov 05 08:54:29 This is the one: https://github.com/rsalvaterra Nov 05 08:55:18 aparcar[m]: so feel free to ping me when there are some issues with the CI/CD, otherwise not interested, good luck :) Nov 05 08:55:25 ynezz: works for me. I'll post in to adm and ask if anyone has objections Nov 05 08:56:11 ynezz: honestly I don't want to mess with web things neither, however you had some comments earlier in some MRs so I thought it's best to keep you up to date Nov 05 08:58:13 yeah, I read the openwrt-adm Nov 05 09:11:17 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_sunxi.html has been updated. (0% images and 100.0% packages reproducible in our current test framework.) Nov 05 10:06:28 hello, I'm trying to boot mainline kernel for IPQ6018 on the Xiaomi AX1800 Nov 05 10:06:56 here is the log when trying to boot own FIT image: https://pastebin.com/A6c0P87h Nov 05 10:07:08 can not get kernel output Nov 05 10:07:24 and board resets after that Nov 05 10:08:00 though I am aware I'm not using an initrd yet, I kinda expected the kernel to print something before it panics because of missing rootfs Nov 05 11:54:09 Hi all, i got a question, i have an openwrt build tree for a new hardware, how can i build all the userland and kernel modules to have them put on an http mirror ( and how can i pre-setup in the image, the URL for the downloads) ? Is there a documentation for this ? Nov 05 12:48:35 dangole: unearthed this from my link archive, which you might find useful: http://www.catb.org/esr/structure-packing/ :) Nov 05 12:57:08 tl;dr: don't use them? Nov 05 13:06:09 f00b4r0, karlp: it is useful for sending stuff over the wire which would otherwise be bloated by a lot of padding. it is not useful to actually work on that data. Nov 05 13:58:03 dangole: I'm not sure I parse your sentence. I was relating to our previous chat, the bottomline is (assuming you're not talking to hw) it's a good thing to reorder struct member to minimize slop, but it's a bad idea to force the compiler to pack a struct. Nov 05 13:58:16 Which, AIUI, matches the result you ended up with :) Nov 05 15:54:05 https://mikrotik.com/product/hex_s would this one be a good choice if i want poe passthrough with 48v (af/at) and sfp port? all look good beside that it is a mikrotik and mikrotik support tend to be pretty bad Nov 05 15:54:36 i remember recent rants about mikrotik storage patches ;p Nov 05 16:03:44 haha yes I recall someone wanting to drop support for mikrotik devices entirely or something Nov 05 16:04:18 but what does the "Passive PoE up to 57V" mean in the PoE-out specs. to me it reads as non-standard so useless Nov 05 16:06:55 [stintel](https://matrix.to/#/@freenode_stintel:matrix.org): Likely exactly that Nov 05 16:08:06 oh matrix is at it again ¯\_(ツ)_/¯ Nov 05 16:09:22 how so? or I mean how is the bridge failing this time Nov 05 16:09:48 Honest curiosity Nov 05 16:10:09 05|17:06:55 < olmari> [stintel](https://matrix.to/#/@freenode_stintel:matrix.org): Likely exactly that Nov 05 16:10:20 a lot of cruft in your reply Nov 05 16:10:47 stintel: it outputs same as gets on input Nov 05 16:11:00 so if you get 48v - it outputs 48v Nov 05 16:11:17 raspberrypi 4 can be powered by poe and i think about powering by that thing Nov 05 16:11:41 too bad the pi3/4 need a HAT for PoE and that makes it less compatible with other HATs :( Nov 05 16:11:57 [damex](https://matrix.to/#/@freenode_damex:matrix.org): so it's only not passive, but "passthrough" as well Nov 05 16:11:59 it does not specifically require af/at so 48v should be fine (chips permit so) if pinout is fine Nov 05 16:12:46 stintel: oh that is completely fine by me. i am using them as a tiny compute nodes :D Nov 05 16:13:08 issue with generally "passive poe" is that they won't generally work with "active" or standard stuff Nov 05 16:13:26 https://www.cisco.com/c/en/us/support/switches/catalyst-3560cx-8xpd-s-switch/model.html Nov 05 16:13:36 though you prolly know all this Nov 05 16:13:36 this has also SFP and PoE-PD + PT Nov 05 16:13:42 but different price range :P Nov 05 16:14:40 ah no, the WS-C3560CX-8PT-S Nov 05 16:14:42 well, raspberrypi works. spec of chips on poe hat permit so and i tested with 48v output on 'unifi switch 8' (af/at on input and it 'passthrough' 48v) Nov 05 16:15:04 it passes passive 48v on port8 Nov 05 16:17:00 i would be glad to get that tiny device that can have af/at poe on all ports Nov 05 16:17:51 https://mikrotik.com/product/RB960PGS this one is twice bigger (deeper?) and have poe on all ports, although i can not power it by poe to redistribute power ;( Nov 05 16:19:16 yeah PoE-PD+PT devices are very rare Nov 05 16:19:30 I was generally suprised switches are nowadays, well, an thing at all, but not expensive either.. and even 24(12) managed ones not THAT expensive, like just over 300€ instead 4figures. Nov 05 16:19:33 hmm... or it can o_O rechecked specs - seems like we can feed passive 48-54v and get 802.3af/at on all ports? Nov 05 16:20:24 umm...missed part...8 port where 4 poe at/af at beginning part Nov 05 16:22:10 olmari: well, i would like to get a tiny device, not that gigantic 12~24 port managed switches :) Nov 05 16:22:44 more switches - more segmentation in a tiny racks for tiny single board computers ;) Nov 05 16:22:52 yeah I see your needs :) Nov 05 16:23:51 for "1" device or generally smaller stuff everything is always more Nov 05 16:24:29 i currently have ubnt edgeswitch 8 150w that could power some other tiny switch that would redistribute 30w of power (enough for 4 boards) Nov 05 16:29:06 for snything I am responsible for, there nowadayss exist tiny or big "active" poe switch, then any device needing power is bette to be "active" one Nov 05 16:30:43 olmari: sure, that would be great. huge pro for 'active' power - you could monitor power consumption on a port. never seen it on passive/passthrough ports. Nov 05 16:30:43 in soho context etc, corporate plannings are own figures then Nov 05 16:32:25 because passive is quite literally what it says.. there just exit some arbitary powwer brick somewhere Nov 05 16:33:10 exists* Nov 05 16:34:30 olmari: managed 10GbE 802.3bt is still pricy though :P Nov 05 16:35:16 although compared to some Cisco L3 GbE switches ... the price is ok'ish Nov 05 16:35:40 the 1440W PoE budget I have is a bit overkill though 😂 Nov 05 16:35:51 haha 1440w :D Nov 05 16:35:53 10g still are yeah.. so are likely bt too :) Nov 05 16:36:01 what are you gonna do with it? ;p Nov 05 16:37:20 I was suprised that managed 24(12) at/af 1gbit was just over 300 euro and not even bad UI, from HP (at least branding) Nov 05 16:37:20 in theory I didn't need the extra PSU, this was a miscommunication between supplier and me, but now I do have redundant PSU which is also nice. especially since I'm using several PoE-PD devices Nov 05 16:39:59 now if a less noisy alternative comes out in the future I'll still be looking at replacing it :P Nov 05 16:50:32 so choosing new device (switch with openwrt) - would having IMAGE_SIZE := 6976k be a problem ? Nov 05 16:50:54 there is a recommendation to not buy new devices with 8M of flash ;/ Nov 05 16:51:54 zyxel_gs1900-10hp looks almost perfect. so far only downside is just tiny flash ;/ Nov 05 16:57:10 damex: Why would you need more than 6976 kiB of flash… in a switch…? O_o Nov 05 16:57:53 I can think of a few things Nov 05 16:58:03 lldpd, net-snmp, just to name a few Nov 05 16:58:03 rsalvaterra: i don't need that much. i am just worried that it might not be supported for a long time if there is too little storage space available Nov 05 16:58:18 stintel: that should not take that much... Nov 05 16:58:21 stintel: Yeah, but you're even crazier than me. :P Nov 05 16:58:40 i would definitely need snmpd+mibs+lldpd there Nov 05 16:58:50 damex: oh I managed to grow my image for my Unifi AP AC Pro so that it didn't fit in the 8MB anymore Nov 05 16:59:13 8MB would be a dealbreaker for me Nov 05 16:59:22 I'm not buying anything <16MB anymore Nov 05 16:59:53 yeah, even that cheap ass ubnt aircube ac has 16MB storage Nov 05 17:00:27 I'd also argue a switch, which isn't exposed to the internet, wouldn't need the hardening setting a router does… and those also increase the image size *a lot*. Nov 05 17:00:36 *settings Nov 05 17:03:03 https://github.com/bkobl/openwrt/blob/rtl83xx_dev/target/linux/rtl838x/image/Makefile of that list only zyxel_gs1900-10hp has small physical footprint (for its amount of ports) and supports poe Nov 05 17:03:32 so in that case maybe no rtl838x switch ;/ Nov 05 17:06:10 rsalvaterra: sure, but i'd highly prefer that device is supported by openwrt so it have regular release support and such Nov 05 17:06:57 oh well, storage just put https://mikrotik.com/product/hex_s offline and it is not available for purchase anymore ;/ Nov 05 17:07:10 s/storage/store/ Nov 05 17:10:07 Granted, most people don't like to tweak stuff… Nov 05 17:14:35 Speaking of tweaking, we might want to this for x86(-64)… https://github.com/rsalvaterra/linux/commit/e565c13fb081f18dd9d1f2ecdca6f142a6d7cb1e Nov 05 17:14:45 … we have it for ARM and MIPS already. Nov 05 17:18:15 I'd so love for OEMs to be more.. well.. open.. if nothing else, for peoples wanting to.. I kinda hate that for example some Inteno device that is very closed (drivers), broadcom 11ac, based on whatever openwrt heavy mod anyhow is o god damn good on wlan Nov 05 17:20:56 olmari: there is also lots of switches that based on broadcom SoCs and locked down so you can't boot anything from uboot (no flash and no ethernet initialized at that stage or just no support present) Nov 05 17:21:02 olmari: Only laws will fix that. The OEM's interest is to keep everything closed. Planned obsolescence… Nov 05 17:21:05 so bad ;( Nov 05 17:21:45 https://www.broadcom.com/products/ethernet-connectivity/switching/strataxgs/bcm5334x in mine - soc is supposed to be 'open' but no way to flash anything over there beside using SOIC clip ;p Nov 05 17:22:45 rsalvaterra: well, another one - people might stop buying (well, they don't) Nov 05 17:23:17 damex: Good luck with that. Nov 05 17:23:17 they could just push that closed devices at the lower price brackes and make them 'work' Nov 05 17:23:24 mm... still it irks me that the dmaned device is quite ace as 11ac AP Nov 05 17:23:27 Joe User doesn't care. Nov 05 17:23:37 rsalvaterra: yeah, that's true ;( Nov 05 17:24:18 We're talking about people who buy a new PC because the one they have is getting slower. Nov 05 17:24:55 And I mean hdd-full-of-crap slower. Nov 05 17:25:50 (And here I am, building OpenWrt and Linux on a Pentium D 950 with 3.2 GiB of RAM, from 2006… :P) Nov 05 17:27:27 300KiB o_O Nov 05 17:27:44 stintel: Ah, you saw the patch. :P Nov 05 17:29:03 My /boot, at the moment… Nov 05 17:29:03 -rw-r--r-- 1 root root 4460720 nov 2 09:18 vmlinuz-5.10.0-rc2+ Nov 05 17:29:03 -rw-r--r-- 1 root root 4264480 nov 4 17:12 vmlinuz-5.10.0-rc2-dcde+ Nov 05 17:29:31 Not too shabby, heh? :) Nov 05 17:30:01 yeah fancy although for x86 I don't really care :P Nov 05 17:30:16 Actually it's more like 200 kiB. I guess I fat-fingered the commit message. Nov 05 17:30:44 Yeah… x86 is not exactly space constrained. :P Nov 05 17:32:01 This is basically the poor man's LTO. Nov 05 17:32:14 there might be x86 boards that might love that reduction Nov 05 17:37:36 Are there build issues with procd right now? Nov 05 17:39:09 aparcar[m]: Not that I know of, at least with my config. Nov 05 17:39:38 Okay thanks Nov 05 17:41:01 hmm... is there support for 98DX3236's MII in 5.4 ? Nov 05 17:41:13 i wonder if someone brought that devices to openwrt ;p Nov 05 17:41:51 (looking at CRS305-1G-4S+IN with 1Gbe + 4xSFP+) Nov 05 17:42:20 found attempt to bring it up two years ago https://github.com/TobleMiner/openwrt/tree/mikrotik-crs305 Nov 05 20:35:09 ah, screwed by a new feature.... Nov 05 20:36:49 jow: just ran into a new feature that requires warning in the directions.... Nov 05 21:13:34 jg_: hi. can you elaborate? I am a bit out of the loop Nov 05 21:18:50 anyone knows the people from NSLU2-Linux? I'd tell them to stop logging this channel. Nov 05 21:19:43 Also why is the channel topic back to NSLU2 and not whitequark.org Nov 05 21:22:09 aparcar[m]: topic points to whitequark here? Nov 05 21:22:46 I got : Forum: https://forum.openwrt.org/ | Wiki: http://openwrt.org/ | Release: https://downloads.openwrt.org/releases/19.07.3 | Notes: https://openwrt.org/releases/19.07/notes-19.07.3 | Logs: http://logs.nslu2-linux.org/livelogs/openwrt-devel/ Nov 05 21:23:15 all right, matrix bug :) Nov 05 21:40:01 :D Nov 05 22:33:08 Borromini: There. Now you can buy the ER4. XD Nov 05 22:50:18 damex: congratz :) Nov 05 23:08:23 meh still tempted to buy one of those :P Nov 05 23:15:49 Uh, guys… https://dnsflagday.net/2020/#dns-flag-day-2020 Nov 05 23:16:54 … shouldn't dnsmasq default to an EDNS buffer size of 1232 bytes too? Nov 06 01:57:22 heya guys, I was wondering if any of you had experience with loading uboot to run in RAM via JTAG ? Nov 06 02:26:12 looks like i need to figure out why APK has a smaller package than OpenWrt... Nov 06 02:37:41 oh I see. zlib Nov 06 02:40:11 oh no. i see lzw in the source **** ENDING LOGGING AT Fri Nov 06 02:59:57 2020