**** BEGIN LOGGING AT Fri Mar 29 02:59:57 2019 Mar 29 05:11:55 gch981213: the ramips spi driver made it out of staging Mar 29 06:35:30 mangix: so your sysclock is at 22.5MHz and you're running flash at 80MHz? (your v2) :-) Mar 29 06:35:34 mangix: Cool. I'll do a backport later this day. Mar 29 06:36:08 mangix: anyway, thanks for it, I'll add the lost 0 Mar 29 07:47:15 mangix: https://patchwork.ozlabs.org/patch/1042418/ I would like to add more details to the commit message for this part `(with newer SPI driver)` can you please provide it? What kernel/driver version was that `dd` test made on? Mar 29 07:47:30 4.14.w/e is latest Mar 29 07:47:47 it's regular busybox dd Mar 29 07:48:29 new spi driver = you were running 4.14.y with backported driver from kernel 4.19.y ? Mar 29 07:48:49 to just make it clear Mar 29 07:48:51 the numbers given were from current master Mar 29 07:49:13 i added a comment that the upstream one is quite a bit faster though Mar 29 07:49:28 from 14 to 6 seconds Mar 29 07:49:28 BTW this is patch for for gb-pc1 Mar 29 07:49:55 oh Mar 29 07:50:03 gb-pc2 seems fine with me Mar 29 07:50:07 seems i took the wrong patch from my branch Mar 29 07:50:16 i was testing the pc2 Mar 29 07:50:23 will resend Mar 29 07:50:27 no :) Mar 29 07:50:46 I'm asking for the details for gb-pc1, gb-pc2 v2 is fine Mar 29 07:51:20 of if you would like to send v2 for gb-pc1, then go for it, no problem with me, but for me is enough if you could give me the details here and I can just edit the commit message Mar 29 07:51:26 s/of/or/ Mar 29 07:51:27 ah unfortunately I bricked my pc1. I can't provide numbers for that Mar 29 07:52:37 i sent it to the author to debrick. waiting for it to arrive Mar 29 07:53:12 no problem, I would like to just make it clear on which kernel version and spi driver version was the test done on gb-pc1, do you remember that? Mar 29 07:54:31 so for that test, I believe the initial number was with master and the latter with the upstream driver Mar 29 07:54:57 master=4.14.y upstream=5.0.y ? Mar 29 07:55:00 i would expect the number with master would be 16-17 Mar 29 07:55:50 for master, I'll just reference the dd numbers in pc2 commit message Mar 29 07:56:11 I used this btw: https://github.com/openwrt/openwrt/pull/1578 Mar 29 07:56:30 great, thanks Mar 29 07:57:43 i find it interesting that the spansion chip is slightly faster than the winbond Mar 29 08:02:33 please check it https://git.openwrt.org/?p=openwrt/staging/ynezz.git;a=summary thanks Mar 29 08:06:35 looks good Mar 29 08:06:45 actually Mar 29 08:07:06 huh nvm Mar 29 08:07:17 there was an extra 0 actually missing Mar 29 08:07:39 i don't understand the whole sysclock thing Mar 29 08:08:17 apparently unmodified devices are set to either 50 or 125MHz but the datasheet says it's supposed to be 1/4 the CPU clock Mar 29 08:53:14 quick question : does the ubnt erx (ramips, mt7530) use DSA switch drivers under 18.06? how does one figure out if a target uses DSA to begin with? Mar 29 08:57:31 movi_work: nope Mar 29 08:58:55 movi_work: here my WIP code that does use mainline kernel DSA and also phylink. https://github.com/vDorst/linux-1/tree/v5.1-rc-phylink-p5detect Mar 29 08:59:19 phylink? Mar 29 09:00:16 phylink is a new framework, needed for SFP support. Not really useful for your device but my ER-X-SFP does. Mar 29 09:00:24 ah Mar 29 09:01:12 by the way, could you explain one thing. at some point i was led to believe that DSA was required to get sw offload to work. but all platforms got it regardless if they're converted to dsa or not. Is there any actuall value to it? Mar 29 09:01:19 I code works even can use both GMACs Mar 29 09:01:45 But main issue is that I don't have a working NAND driver for +4.19 kernels. Mar 29 09:02:44 movi_work: both sw and hw offload should already works with the current driver in openwrt Mar 29 09:03:07 yes, and they do. what i'm wondering is where is the value in dsa Mar 29 09:05:20 movi_work: did you read https://www.kernel.org/doc/Documentation/networking/dsa/dsa.txt ? Mar 29 09:06:23 qyes, as in i get that it exposes every port in the switch as an interface for example Mar 29 09:08:48 DSA can make use of the hardware features in the switch. Bridging is one of those features. Mar 29 09:45:42 movi_work: Both DSA and swconfig are used for configuring switch. Upstream folks insist that each switch port should be represented as individual ethernet device and rejected swconfig at some point. Mar 29 10:18:06 blocktrron: I'm wondering if you could help test/review this https://github.com/openwrt/openwrt/pull/1857 ? I've just found out, that EX6150 is using that utility Mar 29 10:21:33 ynezz: i can test it, however it will become next week as i'm currently abroad Mar 29 10:22:35 From my experience, the Netgear wizard asks you "Are you sure to downgrade to a lower version" instead of failing hard. At least this is the case for the atheros-delta and ramips-sercomm units. Mar 29 10:31:40 blocktrron: thanks, no worries, enjoy your stay Mar 29 10:57:45 is cotequerioz here on irc? Mar 29 10:58:05 had we formally agreed in the past that we were merging on packages via rebase to avoid the merge commits or had that been left as personal prefernce? Mar 29 10:58:45 eh, looks like mangix/stintle keep them too. watevs then I guess :) Mar 29 11:00:13 doesn't ring a bell Mar 29 11:02:36 yeah, if it's already prevalent it doesn't matter Mar 29 11:44:50 karlp: yes I am Mar 29 11:49:47 I was not aware of that. Mar 29 11:49:58 apparently it's nto a thing, don't worry about it :) Mar 29 11:52:49 OK Mar 29 11:59:42 cotequeiroz: thanks for that suggestion for strongswan in submenu btw! Mar 29 12:11:13 You're welcome. Mar 29 12:15:51 gah, please don't do this Mar 29 12:16:32 mixing selecting depends with nonselecting depends will cause recursinve dependency troubles and end up introducing issues for other packages like https://github.com/openwrt/packages/issues/8535 Mar 29 12:46:38 jow: Are you talking about the collapsing menu in strongwan? Mar 29 14:03:10 ynezz: please let's ask DTS authors to speify license, see https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=3f019bf96cdf52668750f67cb139fa1b5c538f0b ("ramips: add Netgear EX6150") Mar 29 14:03:42 ynezz: you can just link to the https://openwrt.org/submitting-patches#dts_checklist Mar 29 14:10:56 rmilecki: damn, sorry, I've missed this one Mar 29 14:11:02 I'll fix it Mar 29 14:11:25 no problem, it happens, you can't notice everything and I didn't review it so I'm to blame as well ;) Mar 29 14:11:37 ynezz: if you can contact author & request for specifying license, that's great Mar 29 14:11:43 blocktrron: ^ Mar 29 14:25:10 someone reports ath10k-ct FW crash on 18.06.2, and not in recent stuff. Should we consider backporting ath10k-ct driver/firmware to a stable release, or just tell them to run new beta stuff? https://github.com/greearb/ath10k-ct/issues/82 Mar 29 14:29:45 I'm not against backporting Mar 29 14:30:39 stintel: jow also considered backporting the current wifi drivers from master to 18.06 Mar 29 14:43:14 what mac80211 kernel version is that release using? Or more specifically, what ct driver version is it using? Maybe 4.13? Mar 29 14:43:51 or would you refresh the entire mac80211 package as part of the backport? Mar 29 14:44:09 ynezz: sorry, i missed it too. If you want to fix it do GPL2+ or MIT Mar 29 14:44:28 Alternatively i can send a patch this evening Mar 29 14:47:41 greearb_: openwrt 18.06 uses ath10k-ct 4.13 Mar 29 14:48:00 the plan would be to replacve mac80211 and ath10k-ct with the version from master Mar 29 14:49:50 ok Mar 29 15:19:35 q Mar 29 16:23:07 nice. got a build failure, because of "curl: (35) Cannot communicate securely with peer: no common encryption Mar 29 16:23:10 | algorithm(s)" Mar 29 16:23:14 damnit, that was meant to be one line. Mar 29 16:23:47 downloads ok from curl on the command line and via a browser: https://gitorious.digitalstrom.org/dss-misc/upnpbeacon/-/archive/v0.1.1/upnpbeacon-v0.1.1.tar.bz2 Mar 29 16:57:55 karlp: not sure what you mean? Mar 29 17:01:15 it's not an upstream package (yet) but the curl in the buildroot refuses to download the file from that url, with that "no common encryption algorithms" Mar 29 17:01:25 I have to download it manually and put it in the dl file. Mar 29 17:01:42 and my regular host curl with the same args "-f --insecure" works ok. Mar 29 17:01:48 not an error I've ever seen. Mar 29 17:27:28 not that Mar 29 17:27:39 something about merging packages Mar 29 17:28:20 * mangix is amused curl gets used by the build system but wget is the dependency Mar 29 18:39:37 Maybe its no use anymore, but i recently succeeded in unlocking an locked bootloader of BCM3380 cable modems. With this unlock, any can use TFTP to upload images and boot them from RAM instead of flashing it to SPI flash. I also found a way how to build bootloader upgrade files that can be used with the standard eCos /ip_hal/bootloader command. It passed all checks including the signature Mar 29 18:39:37 check Mar 29 18:42:03 i posted a bootlog at: https://openwrt.org/toh/cisco/epc3925#unlocked_bootloader Mar 29 20:44:58 my 4.19 ramips bump compiled Mar 29 20:45:04 this will not go well... Mar 29 20:46:19 🧱 Mar 29 20:50:10 aaaaaannnd dts changes are needed Mar 29 20:50:13 unfortunate Mar 29 20:52:49 but not unexpected Mar 30 02:24:23 mangix: download.pl checks if curl is present. If it's not available it falls back to wget. **** ENDING LOGGING AT Sat Mar 30 02:59:57 2019