**** BEGIN LOGGING AT Mon Jul 23 03:00:01 2018 Jul 23 05:15:59 guys my linksys e1200 with openwrt cant reset and access the web config, is there other solution to reset my router? Jul 23 05:29:31 cheng08: https://openwrt.org/docs/guide-user/troubleshooting/failsafe_and_factory_reset Jul 23 05:31:12 I try that sir pkgadd but not working Jul 23 05:31:46 some dev here said that linksys e1200 not supported of reset button Jul 23 05:32:03 that 'should' always work, although the actual button to spam might differ (usually the WPS or reset button) Jul 23 05:33:17 possibly, in that case you may be lucky if the OEM bootloader provides push-button tftp recovers (uploading a new firmware via tftp, when the reset button is pressed during early startup) - that will reset the configuration to its defaults as a side effect Jul 23 05:34:11 in the worst case you may need a serial console connection (usually using a TTL level (3.3V!) USB-to-serial adaptor), this will usually require opening the case and possibly soldering Jul 23 05:41:18 okay sir thank you Jul 23 05:41:57 if we use the USB-to-serial what will do next sir after we soldering Jul 23 05:43:03 in general, the serial console will give you access to an open root shell - no authentification necessary, this allows you to reset (firstboot) or reconfigure the existing configuration from the shell Jul 23 05:46:02 ok thank you sir Jul 23 05:50:26 hello Jul 23 08:00:32 blogic: hi! the qca8k commit in your staging tree only supports one cpu port, right? how much of a performance impact does that have? Jul 23 08:01:02 hello .. I am using OpenWRT CC .. It has an "fw3" package for firewall .. I want to ask that where are "core.sh" and "core_rule.sh" etc files are located in CC ?? In AA, these files ere located in /lib/firewall/ , but i don't see them in this folder in CC .. Jul 23 08:02:31 jow: Any help ?? Jul 23 08:31:27 probably don't exist anymore? Jul 23 08:31:33 what's the real question? Jul 23 08:44:00 Obaid: fw3 has been rewritten in C for CC Jul 23 08:44:07 there's no shell scripts anymore Jul 23 08:45:08 jow: Ok. Thankyou !! Jul 23 08:58:15 still curious what the real question was :) Jul 23 09:00:34 jow: hi, got my email? Jul 23 09:13:28 falstaff_: yes Jul 23 09:13:38 sorry, bad autocomplete Jul 23 09:48:45 huaracheguarache: just finished the V3 of my ath79 series Jul 23 09:49:02 doing 3 usb phy drivers just now and prepare a huge ipq401x series Jul 23 09:49:29 should be on ipqess.c (the dakota ethernet) by tonight, then onto the qca8k tag driver for that Jul 23 09:49:35 and then qca8k upstream foo Jul 23 09:49:45 and i have dual gmac support working on qca8k Jul 23 09:49:57 i am doing a upstreaming and downstream rebase week Jul 23 09:50:10 huaracheguarache: so no promises but potentially qca8k for trunk this week Jul 23 09:50:18 but we'll need to see how i get along Jul 23 09:51:49 Hi what's qca8k ? Jul 23 09:52:22 Is it a change like r71xx to ath79? Jul 23 09:53:56 Tapper: no its tha DSA driver for QCA switches Jul 23 09:54:04 supports ipq408 8075 and 8337 Jul 23 09:54:11 was told 8327 works aswell Jul 23 09:54:19 k thanks Jul 23 09:54:21 and with a bit of love, prolly all other QCA targets Jul 23 09:56:10 Any one having probs with sqm? I turned on sqm on my wrt3200acm and get a f for bufferblote Jul 23 09:56:27 I have 200 down and 12 up Jul 23 09:57:39 using cake + layer cake Jul 23 09:58:14 I will just switch back to softwair off loading Jul 23 09:58:55 sqm is broke Jul 23 10:58:21 blogic: confirming 8327 works with qca8k as well (tested in ath79). only minor diffs between 8327 and 8337 likr the cpu port exchange feature Jul 23 11:05:53 Hello Everyone Jul 23 11:06:04 Looking For Experts Helps Jul 23 11:06:08 I compile Openwrt Barrier Breaker 14.07 & I have successfully done, but when I try to install OpenVPN OpenSSL I am getting this error Collected errors: * satisfy_dependencies_for: Cannot satisfy the following dependencies for openvpn-openssl: * kernel (= 3.10.49-1-0114c71ed85677c9c1e4911437af4743) * * opkg_install_cmd: Cannot install package openvpn-openssl. Jul 23 11:11:30 Barrier Breaker is verry old! you are wasting your time. Jul 23 11:12:20 VPN-Worlds Use new fermwair Jul 23 11:14:24 does new firmware support TP-Link-MR3420 v2 router Jul 23 11:16:00 VPN-Worlds: https://wiki.openwrt.org/toh/tp-link/tl-mr3420 says it does, but it has really smal flash. Jul 23 11:16:18 yes I saw also Jul 23 11:16:21 it does Jul 23 11:42:47 stintel, o/ Jul 23 11:47:43 can anyone give me LEDE source to compile for TP-Link-MR3420 v2 Jul 23 11:50:45 VPN-Worlds, git clone https://github.com/openwrt/openwrt Jul 23 12:02:40 nitroshift > I have downloaded but target profile doesn't exist Jul 23 12:49:18 VPN-Worlds: yes it does, check target/linux/ar71xx/image/tiny-tp-link.mk Jul 23 13:20:17 i am having one of my QCA days ..... Jul 23 13:20:35 trying to figure out what the 1 trizillion line of shit code msm bus driver code do Jul 23 13:21:12 and after removing all uncalled code i end up with code that builds a bus topology and registers it without doing anything with it Jul 23 13:21:21 and it does it in the usual QCA crappy code way Jul 23 13:21:35 ah it does 1 thing, which is to call alloc/free a billion times Jul 23 13:21:56 because converting from of -> pdata -> internal struct -> bus_device is FTW Jul 23 13:22:06 blogic: So it's a memory tester in disguise Jul 23 13:22:27 for real .... Jul 23 13:24:44 blogic: While you're about - what needs doing to get the mediatek-pcie driver working on MT7620/1/8? Jul 23 13:25:03 is it not working ? Jul 23 13:25:27 Well so far as I know it's not officially supported currently, so I figured there's some slight extra magic needed Jul 23 13:25:39 I haven't tried just adding a compatible string and letting it rip. Jul 23 13:25:49 jow: there is one more regression in 18.06 & master affectting all bcm53xx devices using SPI NOR flash Jul 23 13:25:53 jow: controller problem Jul 23 13:25:58 results in fs corruption Jul 23 13:26:01 working on it Jul 23 13:26:15 Monkeh: 7620/8 work with upstream drivers Jul 23 13:26:38 blogic: Well then, I guess I'll just try it Jul 23 13:27:03 At a brief glance the MT7621 and MT7623 look basically identical as far as PCIe goes Jul 23 13:27:14 very similar for sure Jul 23 13:30:34 rmilecki: :( I'd say for the next major we stop bumping the major kernel version 3-6 months before Jul 23 13:30:43 jow: +1 Jul 23 13:30:52 the amount of fallout by 4.14 is insane Jul 23 13:30:59 jow: like bump once stable is out and then not bump till after next stable Jul 23 13:31:17 jow: but but but its bleeding edge and works on mvebu !!1! Jul 23 13:31:48 I am sure it works perfectly fine in qemu too Jul 23 13:32:20 jow: unless you use owrt, where you'll have to pad the image Jul 23 13:32:38 oh right :D Jul 23 13:37:05 Hello. Can somebody help me to choose router? I need a replacement for asus wl500w. Jul 23 13:38:12 Not more 100$, with usb and sata if it possible Jul 23 13:38:49 probably worth asking in #openwrt Jul 23 13:40:44 lantiq,xrx200 - kmod-ath9k-htc with ar9721 wifi dongle doesnt work. "ath: phy1: Mac Chip Rev 0x0f.3 is not supported by this driver" any ideas? Jul 23 13:41:19 is anyone else seeing ddns errors with hurricane electric? user.err ddns-scripts[30379]: myddns_ipv4: cURL Error: '52' Jul 23 13:41:52 oobed it Jul 23 13:41:55 noobed! Jul 23 13:43:51 ldir: only using the v6 HE stuff Jul 23 13:44:01 and that works, never setup ddns Jul 23 13:44:04 soz :( Jul 23 13:46:31 s'ok - I don't know if this is a new feature/a problem my end/a problem their end/ Jul 23 13:47:20 Newbeewrt1200ac get's my vote Jul 23 13:47:36 Newbee wrt1200ac* Jul 23 13:47:38 it was working at some point in the past(tm) now after a flash I notice it isn't. See above ^^^ :-) Jul 23 13:49:06 are the resident mt76 guru(s) around? I'm seeing something really odd on my ex3700 Jul 23 13:49:27 f00b4r0: i must follow up on the roouterboard stuff Jul 23 13:49:50 namely: can a substandard cable make the device randomly reboot? Jul 23 13:49:51 so lets put that drop support patch aside, we have an active user of the feature so dropping it is not an option Jul 23 13:50:02 but regarding the other one ... just so i understand Jul 23 13:50:15 blogic: yeah that's the beginning of the nightmare. What he's asking is to have both "devices" in tree Jul 23 13:50:26 ignore that patch Jul 23 13:50:29 that's never going to work well, unless we rename the hack in a very very obvious way Jul 23 13:50:42 ok Jul 23 13:50:56 the other one, so mikrotik has a huge partition with random chunks in it and gabor at one point rev-enged it and split it manually to smaller chunks ? Jul 23 13:51:08 he did that on ar71xx, yes Jul 23 13:51:28 and there is a tool to read the blob in mikrotik format called robo-cfg ? Jul 23 13:51:31 or similar ? Jul 23 13:51:37 more or less, yes Jul 23 13:51:49 rbcfg Jul 23 13:51:57 it reads the "soft_config" partition Jul 23 13:52:07 the hard_config partition is processed by boot routines (on ar71xx) Jul 23 13:52:10 and on ramips based targets the format changed Jul 23 13:52:16 the format appears to be the same Jul 23 13:52:20 ok Jul 23 13:52:22 offsets are different, so are sizes Jul 23 13:52:27 so ideally we use a fw splitter Jul 23 13:52:30 ideally yes Jul 23 13:52:34 that's my ultimate goal Jul 23 13:52:36 ok, got it Jul 23 13:52:56 and we support rbcfg ? Jul 23 13:53:01 on ar71xx we do Jul 23 13:53:03 and whats inside the soft config ? Jul 23 13:53:21 Tapper: it is too expensive. Only one usb... Jul 23 13:53:23 on ramips we don't yet because it's hardcoded for the endianness of ar71xx Jul 23 13:54:07 f00b4r0: and whats inside that blob ? Jul 23 13:54:12 macc addr and so on ? Jul 23 13:54:15 blogic: boot delay, boot device, boot key (when serial enabled), boot protocol (dhcp/bootp, for netboot), bootloader choice (regular or backup), cpumode, cpufreq scaling, uart speed Jul 23 13:54:23 possibly more we haven't yet rev-eng'd Jul 23 13:54:27 ah its the uboot-env style thing Jul 23 13:54:38 macaddr, serial and device name are in hard_config Jul 23 13:54:41 which is read-only Jul 23 13:54:41 like the mikrotik boot env Jul 23 13:54:53 indeed Jul 23 13:55:07 Newbee TP-Link c7 then Jul 23 13:55:38 fine, thanks ! Jul 23 13:55:54 yw Jul 23 13:56:12 i'm sorry this wasn't clear enough in my emails, I'll try to be clearer next time ;) Jul 23 13:56:16 f00b4r0: When you say substandard cable.. what cable? Jul 23 13:56:19 c7 is a bit old now but it's better than the asus Jul 23 13:56:58 Monkeh: it's a cable that's more or less a cat5, but some wires are shorter than other. At the plug, out of ~40m cable some of the wires are shorter by ~1m Jul 23 13:57:16 Monkeh: reason is hard to explain. Retrofit of an 80s analog telcom 32-pair cable Jul 23 13:57:21 I.. eee. Jul 23 13:57:24 ;) Jul 23 13:57:30 I wouldn't expect reboots Jul 23 13:57:35 But that's a big difference. Jul 23 13:57:35 i wouldn't either Jul 23 13:57:41 something is out of whack Jul 23 13:57:48 disconnect I'd get, but the device resets Jul 23 13:58:06 it's a custom built image, maybe I missed something important Jul 23 13:58:09 I don't know what the requirements for matching between pairs are. Jul 23 13:58:16 I assume the pairs at least are intact. Jul 23 13:58:19 they are Jul 23 13:58:22 no crosstalk Jul 23 13:58:41 And no metre length differences. Jul 23 13:58:51 .. or centimetre, for that matter. Jul 23 13:59:15 well, that's where I have a problem: there is a metre difference in-pair Jul 23 13:59:37 .. between conductors in a single pair? Jul 23 14:00:02 basically, the previous system was an Ericsson analog PBX where the phones were hooked in series. So the 32 pairs were interrupted in a few places (where the phones were) and I had to reconnect them to reassemble a "suitable" network cable Jul 23 14:00:04 yes Jul 23 14:00:12 Burn in hell. :P Jul 23 14:00:15 You can't use that Jul 23 14:00:31 I suppose I could try to improve that situation. I also suspect the measurement isn't very accurate (using cheap chinese electronic cable tester) Jul 23 14:00:43 well, the fun part is: it works no trouble with everything else but these APs Jul 23 14:01:06 Tapper: tp-link much better. If I connect 1Tb hdd with samba server will be enough performance? Jul 23 14:01:10 and I still don't understand why the AP would reboot on a bad cable Jul 23 14:01:13 Maybe their switch silicon is a little special needs Jul 23 14:01:18 But that doesn't excuse the cable Jul 23 14:01:25 it doesn't, we agree on that ;) Jul 23 14:02:39 i'll do more tests see if I can figure the root cause (i.e. if it's really a cable problem) Jul 23 14:03:16 And then go kill that cable. Jul 23 14:03:17 :P Jul 23 14:03:24 it's in the walls Jul 23 14:03:32 which is why I went through the pain Jul 23 14:03:38 Kill the walls, they're harbouring it. Jul 23 14:03:45 No mercy. Jul 23 14:03:46 and we're not talking modern concrete walls. Stone walls Jul 23 14:03:47 why is it so arm Jul 23 14:03:48 warm Jul 23 14:03:48 heh Jul 23 14:03:52 Purge the unclean. Jul 23 14:03:55 lol Jul 23 14:04:03 burn the house! Jul 23 14:04:05 death to all xenos ... Jul 23 14:05:05 burn the house is what you do if you ever have a macerating toilet thing go faulty. Jul 23 14:05:26 I just pre-emptively burn when macerators are involved. Jul 23 14:05:52 blogic: btw, one important point: the RB M33G is in 18.06: whatever is decided on those two patches, at least one will apply there Jul 23 14:06:47 blogic: also, since you're looking at routerboard stuff, https://github.com/openwrt/openwrt/pull/1181 also has patches for 18.06 Jul 23 14:06:51 * f00b4r0 hides Jul 23 14:06:59 http://www.diyfaq.org.uk/humour.html with a bit of casual racism that f00b4r0 will appreciate Jul 23 14:07:10 since we're talking about routerboards Jul 23 14:07:13 who wants one Jul 23 14:07:23 (gotta fix support though, only condition) Jul 23 14:07:24 :D Jul 23 14:07:25 f00b4r0: start using the ML please :-) Jul 23 14:07:32 stuff wont get lost that way Jul 23 14:07:59 blogic: I will from now on. I'm hoping you'll forgive this last perusal of github :) Jul 23 14:08:10 i did not even notice Jul 23 14:08:15 that is the problem ... :-) Jul 23 14:08:25 blogic: the way I understood it, the two disagreements are 1) keep existing board support or not Jul 23 14:08:34 blogic: 2) allow nested partitions in the dts Jul 23 14:08:42 blogic: I get that. I promise this was my last submission on github :) Jul 23 14:09:14 jow: 1) yes, one active user of the (broken) support voiced himself so we'll keep it till there is an alternative Jul 23 14:09:51 jow: 2) removing apples and adding bannans is not an option, the solution would be to add a FW splitter and then a big rbcfg part Jul 23 14:10:05 jow: 2) is not actually possible with DTC using fixed partitions Jul 23 14:10:07 for 2) there are diverging opinions. mkresin strongly objects to have nested partitions in the dts syntax proposed as it is either out of style for the target or even deprecated / discouraged / only working by change upstream Jul 23 14:10:19 *by chance Jul 23 14:10:24 blogic: alternative WIP here: https://github.com/f00b4r0/openwrt/commits/rb750gr3routerboot Jul 23 14:10:27 i don't have the hw Jul 23 14:10:35 jow: i am making the call and saying no nested parts but a splitter Jul 23 14:10:57 blogic: that is fine but f00b4r0 already pointed out that he's not planning to invest work in a splitter right now Jul 23 14:11:01 jow: then remove the manually split parts and add one big one that userland and kernel space can work with Jul 23 14:11:08 jow: err what? I'm not using the out of style syntax, quite the opposite actually Jul 23 14:11:23 and not having the sub partitions was apparently no option (I suppose to keep rbcfg working) Jul 23 14:11:41 jow: well either he cares so much that he will help fix it the proper way or he does not care in which case i dont understand the flamery Jul 23 14:11:46 either way i made the call Jul 23 14:11:50 jow: rmilecki's patch in fact build exactly on the syntax I'm using. Jul 23 14:11:51 I quite like the idea of properly implemented nesting, and IMO it can help clarify the otherwise magic numbers in the dts Jul 23 14:12:14 Monkeh: fixed-partitions does not allow it Jul 23 14:12:21 blogic: how so? Jul 23 14:12:29 Monkeh: but you could patch it and get it approved upstream, we can use it once that happened Jul 23 14:12:36 blogic: Subject to this patch: https://patchwork.ozlabs.org/patch/943613/ Jul 23 14:12:43 f00b4r0: if i am not mistaken it'll spew an error Jul 23 14:12:50 blogic: nowhere in docs have I found anything claiming partitions can't overlap Jul 23 14:12:53 If that gets upstream I think it's a very logical way to organise things. Jul 23 14:13:01 blogic: absolutely not. No error neither at compile or load time Jul 23 14:13:16 f00b4r0: well once rob accepts it fine until then - no Jul 23 14:13:16 DT docs say: For backwards compatibility partitions as direct subnodes of the flash device are Jul 23 14:13:19 supported. This use is discouraged. Jul 23 14:13:28 jow: that's exactly why I'm not using that Jul 23 14:13:35 jow: basically, what we have no is not optimal Jul 23 14:13:39 jow: Yes, that refers to the use of a partitions subnode Jul 23 14:13:45 and i wont merge another sub optimal solution Jul 23 14:13:59 it'll stay as is until there is a real clean and upstream compliant solution Jul 23 14:14:08 and now i think i spent enough time on this Jul 23 14:14:31 blogic: I have yet to be pointed at anything that says that partitions cannot overlap. Jul 23 14:14:32 it'll be hard to change my mind until there is an upstream patch so lets not waste time trying to do so Jul 23 14:14:46 * blogic points at himself Jul 23 14:14:50 heh Jul 23 14:15:02 so in the meantime we keep garbage in 18.06? Jul 23 14:15:08 correct Jul 23 14:15:14 its better than 17.01 Jul 23 14:15:16 *sigh* Jul 23 14:15:21 which is better than 15.0x Jul 23 14:15:29 which is better than CC bb aa, backfire .... Jul 23 14:15:31 jow: As far as that goes, I have an unreviewed patch to deal with it: https://github.com/eamaclean/source/commit/fe5f8d52fb141278c91bb201d0b76bb032027c33 Jul 23 14:15:32 small steps at a time Jul 23 14:15:44 it works and has a user and 18.06 is near final Jul 23 14:16:09 and partitions { compatible = "fixed-partition"; } works? Jul 23 14:16:13 f00b4r0: i am more worried about the ath10k issue Jul 23 14:16:14 i'm still confused why people want 50K worth of interactive shell auto-package-suggestion-thingy in an embedded world. Jul 23 14:16:24 from the DT docs it lookedl ike only a binding with compatible = "brcm,bcm947xx-cfe-partitions"; exists Jul 23 14:16:32 salcedo: they do ? Jul 23 14:16:44 jow: the patches I submitted work. They have been tested. Jul 23 14:17:30 jow: https://elixir.bootlin.com/linux/latest/source/Documentation/devicetree/bindings/mtd/partition.txt Jul 23 14:17:35 ^ Jul 23 14:18:26 Monkeh: yeah, but for fixed partitions the doc states "For backwards compatibility partitions as direct subnodes of the flash device are supported. This use is discouraged." Jul 23 14:18:37 Yes.. Jul 23 14:18:38 ah nvm, wrong context Jul 23 14:18:50 jow: yes, that's what most of the DTS files in ramips use, but many other targets use the correct implementation Jul 23 14:19:02 including what I submitted Jul 23 14:19:05 f00b4r0: correct, we'll update those shortly Jul 23 14:19:11 together with the pinctrl carnage Jul 23 14:19:16 lol Jul 23 14:19:20 As far as overlapping partitions I don't see that they're explicitly disallowed, but without clear syntax I don't approve Jul 23 14:21:57 blogic: luckily, there's nothing controversial in #1181. If you don't want the generic spinor patch just ignore it, it's not strictly necessary Jul 23 14:22:02 ;P Jul 23 14:22:25 Monkeh: I agree Jul 23 14:23:08 jow: can i resubmit with only the sub parts, leaving a comment that the top part should be added back when the new syntax goes upstream? Jul 23 14:23:14 so the reamining issues seems to be the principal question on whether it makes sense to declare fixed (for now) sub partitions for something that should be provided by a splitter Jul 23 14:23:23 have a feeling this is the only way to move forward Jul 23 14:23:48 as the layout could theoreticly change etc. pp. Jul 23 14:23:53 f00b4r0: are you using this or is this a religious thing ? Jul 23 14:24:00 or let me rephrase Jul 23 14:24:01 using what? Jul 23 14:24:10 what does the patch make work that is currently not working Jul 23 14:24:13 i don't have religious opinions :P Jul 23 14:24:21 blogic: support on a non-modded device Jul 23 14:24:27 blogic: See above github link for the ~17,500 line horror adding the partitions subnode for ramips. It does need review rather than pushing. :) Jul 23 14:24:30 means a rb which doesn ot need replacement of the bootloader Jul 23 14:24:30 blogic: what are we talking about now? Jul 23 14:24:32 i'm conused Jul 23 14:24:36 confused Jul 23 14:24:54 well resubmit what you want Jul 23 14:24:59 on the ML Jul 23 14:25:52 f00b4r0: for my own understanding... these sub partitions are needed to make rbcfg happy? (which does not yet work due to endianess issues?) Jul 23 14:26:34 seems everyone is confused Jul 23 14:26:48 jow: i think that we used small chunks on ar71xx Jul 23 14:27:06 and we also want the SDK/vendor way Jul 23 14:27:08 jow: these subpartitions are needed: - to make rbcfg happy ("soft_config"); - to easily reference the correct area that stores hw config ("hard_config"), - to clearly state where the bootloaders live ("routerboot" and "routerboot2"), and "bios" I don't remember what it is, you'd have to ask gabor Jul 23 14:27:13 coz the userland tool wont work otherwise Jul 23 14:27:15 and these sub partitions are also meant as some kind of "extra write protection" ? Jul 23 14:27:37 jow: correct Jul 23 14:27:43 jow: the current setup is good Jul 23 14:27:48 well that's the role I think the top partition should bear since there are holes between the sub ones Jul 23 14:27:49 adding a huge master blob is bad Jul 23 14:28:02 jow: i'll nak anything that is not a target agnostic splitter Jul 23 14:28:10 i really thought I explained myself in the cover letter :P Jul 23 14:28:20 The top level partition is meant to ensure nobody thinks the gaps are free to use (although you have to be a little nuts to go poking into random 'empty' bits of vendor firmware) Jul 23 14:28:52 Monkeh: you expect all owrt users to be _sane_? Have you looked at some submissions lately? ;) Jul 23 14:29:21 the top level part is also necessary if we want to have a runtime splitter Jul 23 14:29:38 f00b4r0: sure but it can be ro :-) Jul 23 14:29:39 of course until we do the sub parts can live on their own, as they do on ar71xx Jul 23 14:29:44 f00b4r0: I sat down and manually edited 256 dts files. No, I don't expect sanity, I lack it myself. :) Jul 23 14:30:11 the gap between soft_config and bios? Jul 23 14:30:14 the key difference between ar71xx and ramips is that on ar71xx, there are no holes. so the risk of "leaving the door open" is much lower Jul 23 14:30:51 jow: and more importantly the gap between bios and firmware Jul 23 14:30:58 that's the one gap I'd be very worried to see reclaimed Jul 23 14:31:07 recovering from that would be nearly impossible Jul 23 14:31:57 0x31000..0x40000 ? Jul 23 14:32:01 correct Jul 23 14:32:28 sooo Jul 23 14:32:34 can't we simply enlarge it? Jul 23 14:32:38 again, in a perfect world we'd have a runtime splitter and we wouldn't be arguing now ;P Jul 23 14:32:42 this the holes only exist on the ramips boards ? Jul 23 14:32:44 I means its too long in the original proposal as well Jul 23 14:32:55 if that is the case, you just finally killed your patch Jul 23 14:32:56 blogic: ramips, very probably ipq as well Jul 23 14:33:08 as the splitter will simply register them as reserved Jul 23 14:33:08 have to redig into mikrotik code Jul 23 14:33:10 at least judging from "//·valid·data·only·extends·up·to·0x4f·but·make·it·0x1000·to·match·ar71xx " Jul 23 14:33:26 ok, a splitter or no ptch unless it fixes a regression Jul 23 14:33:31 jow: 0x1000 is the smallest addressable mtd block, AIUI Jul 23 14:33:35 right Jul 23 14:33:40 hence that Jul 23 14:33:42 writing a splitter would have taken less time that the aggregate we spent discussing Jul 23 14:33:52 blogic: ;P Jul 23 14:33:59 f00b4r0: i am serious Jul 23 14:34:08 blogic: I don't have data to test the splitter Jul 23 14:34:14 that was the other reason for the top partition Jul 23 14:34:25 "please send me your blob" Jul 23 14:34:47 i know i may look like an ass but I do try to anticipate everything Jul 23 14:34:55 (sometimes i try too hard i guess) Jul 23 14:36:14 cause I know getting a splitter in will also requires serious proof that it works. So, chicken-and-egg problem ahead, right? Jul 23 14:36:34 hence my insistance. Maybe it makes more sense now? Jul 23 14:36:59 no Jul 23 14:37:11 :/ Jul 23 14:37:22 personally i find your way of dealing with this issue very aggressive Jul 23 14:37:35 i also now that i understand the whole issue consider it a brain dead patch Jul 23 14:37:42 heh Jul 23 14:37:52 and i think it simply wasted time and the solution will be exactly what was proposed at the start Jul 23 14:38:13 well good then. I can save everybody's time and do something else. Jul 23 14:38:15 and now, that 3 devs spent several hours trying to understand the full details and all conclude the same, i no longer want to talk about it Jul 23 14:38:21 kthx bye Jul 23 14:39:18 jow: mind installing ccache on the bots please? That'd be the last step. Jul 23 14:39:44 f00b4r0: will do in a minute Jul 23 14:39:47 thx Jul 23 14:45:54 sometimes I wonder how nl80211 apis are designed Jul 23 14:46:09 they seem to love providing setters for stuff but no getters Jul 23 14:47:01 can't seem to find a way to query the effective mesh id of a given meshpoint interface Jul 23 14:47:07 seems to be exposed nowhere Jul 23 14:47:35 Well they know what they put in, right? Why would they need to query anything? ;) Jul 23 14:48:40 I know, I know Jul 23 14:48:53 real men make no mistakes, hence no need to take a 2nd look Jul 23 14:49:03 its also why real man only ever need "cat" to write programs Jul 23 14:49:27 (without readline history!) Jul 23 14:49:34 lol Jul 23 15:10:47 jow: no obvious breakage spotted, by the way? Jul 23 15:11:58 f00b4r0: didn't notice some yet Jul 23 15:13:47 jow: cool then Jul 23 16:04:46 aww, SpaceRat is back Jul 23 16:07:27 obvious bot is obvious Jul 23 16:32:57 is anyone having issue compibling on xubuntu 18.04 keep getting this error /openwrt/include/toplevel.mk:216: recipe for target 'world' failed Jul 23 16:33:55 and yes updated and tryed make distclean but same error too Jul 23 16:35:22 pkgadd: ping Jul 23 16:42:24 Martini Jul 23 16:48:07 In file included from efiemu/runtime/efiemu.c:33: Jul 23 16:48:07 ../include/grub/efiemu/runtime.h:36:1: error: alignment 1 of 'struct efi_variable' is less than 8 [-Werror=packed-not-aligned] Jul 23 16:48:10 after switching to gcc8 Jul 23 19:17:12 blogic: If the upstream driver is working on mt7620/28, are things like a clock driver already written somewhere? Jul 23 19:32:16 MrFantastik: not without any context whatsoever Jul 23 19:32:26 oh he left Jul 23 19:32:44 MrFantastik: sry, ignore Jul 23 19:57:03 blogic: pong Jul 23 19:58:17 pkgadd: cant find the mail with that eep dump you mailed me Jul 23 19:58:23 pkgadd: mind resending it ? Jul 23 19:58:31 Monkeh: upstream works with 2 mionor patches in our tree Jul 23 19:58:43 Monkeh: there is no common_clock support on ramips yet Jul 23 20:05:50 sure, just a minute Jul 23 20:06:42 you're looking for APPSBLENV? Jul 23 20:09:29 blogic: sent Jul 23 20:10:48 pkgadd: thx Jul 23 20:19:55 blogic: The patches I guess replace the current OF logic for things like clock controls? Jul 23 20:21:00 Monkeh: ? Jul 23 20:21:04 dont get the problem Jul 23 20:21:45 blogic: Was fiddling with using the upstream driver on MT7621 - currently due to lack of clock driver etc there's no way to make it work unmodified. I guess I'm going to need to write that. ;) Jul 23 20:22:37 huh ? Jul 23 20:22:55 ok so some folks started upstreaming stuff in an un ccordinated way Jul 23 20:23:11 cant really judge, i know i'll have to clean up the mess after them eventually Jul 23 20:23:20 I don't mean the mess in the staging dir - I mean the upstream mediatek-pcie driver which currently supports mt7623 and friends Jul 23 20:24:17 It looks to be essentially the same as MT7621, there's just some glue missing Jul 23 20:24:42 (and, well, that's also essentially the same as 7620 and 7628) Jul 23 20:25:03 ahhhh Jul 23 20:25:05 ok ok Jul 23 20:25:08 soo ... Jul 23 20:25:17 gimme 2 sec, need to get a new bottle of water Jul 23 20:25:29 its blasting hot in my office Jul 23 20:26:41 I have a growing number of MT MIPS boards, so I'm trying to figure out how to clean things up and get away from the hardcoded magic Jul 23 20:32:55 sure Jul 23 20:33:06 so 7623 is arm and 7621 is mips Jul 23 20:33:10 Yep Jul 23 20:33:19 But it seems like the peripherals are mostly c+p Jul 23 20:33:22 the mips targets have no power regulators or complex clock trees Jul 23 20:33:46 so drivers generally work once the whole clk/regulator stuff gets removed Jul 23 20:33:46 Just some clock enables and a little PLL block I can't find any docs on. Jul 23 20:34:16 Which is always fun. Jul 23 20:36:29 blogic: But basically so far I need to write simple clock and reset drivers, I don't think there's much else (except the mystery PLL, if you know anything about it) Jul 23 20:39:19 ? Jul 23 20:40:33 The existing driver sets up some PLL registers (RALINK_PCIE_CLK_GEN and GEN1) which.. well, the MT7621 datasheet loosely suggests they exist. And that's all I know about them. Jul 23 20:41:13 The 20 and 28 datasheets don't appear to acknowledge their existence. Jul 23 20:43:27 uff Jul 23 20:43:34 i'll need to dig into that Jul 23 20:43:42 added to my upstream todo list Jul 23 20:44:10 Well, I'll do a little reading and see if I can't magic up something to handle the clock enables and resets properly. Jul 23 20:44:27 so basically we should add common_clock to ramips Jul 23 20:44:41 can you recommend some link to read about the lede/openwrt re-merge? Jul 23 20:44:44 there are no real regulators tha t we can add Jul 23 20:44:59 ring0: not really what you wanna know ? Jul 23 20:47:19 blogic, I'm mainly interested in why lede is not continuing on its own. if I remember correctly the split/fork happened due to openwrt's not easy to maintain/old/cluttered codebase? Jul 23 20:47:36 ring0: not really Jul 23 20:47:48 is that kinda magically solved? Jul 23 20:47:53 it was just a layer8 problem and not a codebase issue Jul 23 20:49:13 so, after everybody calmed down it was decided to continue together, again? Jul 23 20:49:31 that is a very simple way of putting it Jul 23 20:50:18 ring0: so owrt as we have it now is LEDE but with owrt branding Jul 23 20:50:29 we kept our new inclusive dev pattern Jul 23 20:50:39 the rules .... Jul 23 20:51:58 well, I'm just curious. have been travelling for some time and suddenly lede was gone but openwrt back ;) Jul 23 20:52:28 ring0: consider owrt a lede with owrt branding Jul 23 20:53:04 so the old name was of some major importance, i understand Jul 23 20:55:07 looking forward to more new talks of nbd at congresses :) Jul 23 20:55:25 ring0: lulz Jul 23 20:55:46 i prefered the hamburg one Jul 23 20:55:55 let me check Jul 23 20:55:58 was a 15 min drive to get there :-) Jul 23 20:58:58 blogic, yes, 2013 in hh Jul 23 21:08:14 Hi has any one found out what is with spacerat? Jul 23 21:15:48 blogic: sounds good =) sorry for the late reply Jul 23 21:16:27 huaracheguarache: :-) Jul 23 21:16:41 Tapper: mayor tom ? Jul 23 21:17:43 blogic what do you meen Jul 23 21:22:47 Tapper: space, mayor tom, ... Jul 23 21:22:52 u know ... Jul 23 21:22:55 pun intended Jul 23 21:23:31 No dude sorry. Jul 23 21:23:40 david bowie ? Jul 23 21:23:52 ok, forget it, bad joke Jul 23 21:32:57 blogic: It's uh, major. ;) Jul 23 21:33:16 Monkeh: :P Jul 23 21:33:33 the bigger question would be Bowie or Schilling (and I sincerely hope for the former ;) Jul 23 21:33:50 pkgadd: ouch ,,, Jul 23 21:45:34 Package mtr is missing dependencies for the following libraries: Jul 23 21:45:34 libcap.so.2 Jul 23 21:45:39 even with the makefile adjust Jul 23 21:50:57 mm, are there some examples in the tree of ubus (C) consumers? need to write a nice lightweight app that periodically posts info, rather than using cron+curl+whatever Jul 23 22:19:21 hm, I find myself wanting to use server type packages on some of my openwrt devices Jul 23 22:19:25 when did I become one of those people Jul 23 22:20:20 When your openwrt devices started having gigs of RAM and 10s of gigs of storage... Jul 23 22:20:28 ah, they do not :D Jul 23 22:22:38 should probably put something a bit more generic on them, but theres a lack of mips64 disributions Jul 23 22:43:28 in that case you should have picked ARMv7 or ARMv8 instead ;) Jul 23 22:45:10 heh Jul 23 23:00:30 nyt: configure.ac needs a patch. was working on it but gave up Jul 23 23:00:51 OTOH, upstream recommends libcap, which is why you can't disable it Jul 23 23:26:38 bleh, device has disappeared anyway Jul 23 23:26:45 getting sick of these broken bastard serial/uart cables Jul 23 23:27:09 * jwh orders a 3620 Jul 23 23:28:42 never had a 3620 fail, ever Jul 23 23:28:58 ~20 years old, still solid Jul 23 23:38:48 weird since i havr libcap package sekected Jul 23 23:56:21 nyt: about to send a patch disabling libcap support Jul 24 00:00:51 i decided against trying to make it configurable Jul 24 00:01:05 since i failed the first time :) Jul 24 00:13:31 ugh, wondering why my package wasn't showing up Jul 24 00:13:49 feeds script doesn't install it but I can't figure out why Jul 24 01:02:16 rmilecki: I did a build from the 18.06 branch with early printk turned on for the E3000 and opened a new bugreport with the serial console output: https://bugs.openwrt.org/index.php?do=details&task_id=1688 Jul 24 01:15:07 mmm Jul 24 01:15:42 so disabling failsafe doesn't actually stop someone just hitting enter and bypassing login, probably because the uci check fails before firstboot/boot is finished **** ENDING LOGGING AT Tue Jul 24 03:00:00 2018