**** BEGIN LOGGING AT Sun May 05 03:00:37 2019 May 05 03:30:19 I have a TP-Link A7. From what I can see the A7 and the C7 are similar but the firmware layout is different. I KNOW THIS IS A HACK. But... Is it possible to just hack the C7 layout for the A7 and build it? I just want a temporary release version for my A7 until it goes to release. May 05 03:31:13 I don't want to use the snapshot because of the opkg issues. But... having said that, is it enough to just run a cronjob to opkg update every few hours? May 05 03:39:16 using a snapshot is rather on the safe side May 05 03:39:21 hacking stuff is not May 05 03:40:07 and no, you can't hack it - you'd need quite meticulous work on the device (at runtime, using a serial console, from the uboot console), which would be permanent May 05 03:41:32 not liking master/ snapshots is a poor excuse for such an endeavour, as the results would be considerably more complex, dangerous and -work- (both now and long term) than just using a snapshot and maybe having to update a couple of times until the next stable release May 05 06:11:13 I guess he didn't like that answer May 05 07:07:31 blogic: I'm wondering if you've received `[PATCH v4 05/10] net: ethernet: support of_get_mac_address new ERR_PTR error` as it seems, that this patch didn't reached the patchwork yet https://patchwork.ozlabs.org/project/netdev/list/?series=105972 May 05 07:09:38 blogic: getting the same question over and over again https://patchwork.ozlabs.org/patch/1094916/#2165110 which makes me wonder if I should simply drop that DT part and call it a day May 05 07:34:37 Sanity check -- If I have run `make kernel_menuconfig` and the CONFIG lines are in `config-4.19` for the target, then they should be in `build_dir/.../linux-4.19.37/.config` after a build, yes? May 05 07:34:50 (clean build, yes) May 05 08:02:05 ynezz: the joy of upstreaming :-) May 05 08:02:38 we could always carry that detail around with us. i am worried a bit as we now need to ship mvmem, a subsystem, just to read a mac addr May 05 09:15:19 (resolved .config puzzle) May 05 10:14:01 'lo May 05 10:15:12 Hello all May 05 10:17:26 now if someone had said a year ago I'd be submitting this I'd have laughed at them https://lore.kernel.org/netdev/20190505101523.48425-1-ldir@darbyshire-bryant.me.uk/ May 05 10:20:08 why start now...or... I mean.. umm.. nice stuff ;P May 05 10:20:54 nice ascii graphics :) May 05 10:22:21 'cos a year ago I wouldn't have thought myself capable of coding such a thing. I almost understand it :-) There's some copy/paste going on of course. May 05 10:26:21 The thing I haven't yet submitted but have coded for testing/running purposes is the userspace side. That will no doubt be another task :-) May 05 10:32:32 yay I've been told off by Greg :-) May 05 10:35:06 nice work ldir May 05 10:38:12 jow: ty - there's the 'fun' other side of this that puts the DSCP into the connmark - I've an updated x_tables connmark helper that does it, but so far upstream's comments have mostly been 'we'd really like it if you could do this in nftables' - hmmmmmm, another learning cliff May 05 10:50:02 yeah, but thats only for the userspace part, right? May 05 10:50:27 I also have nftables on my list to do firewall4, but ... time and motivation May 05 10:51:13 That's allways an issue :) May 05 10:51:36 while nftables is the future in general... (well.. until someone figures out another ubertables) May 05 11:08:34 no there needs to be a kernel space tweak to netfilter xt_connmark ie. a helper to store the DSCP into the connmark (and maybe set a flag) and a user space tweak to iptables to activate it. May 05 11:13:34 jow: do you've any preference in how should we tackle per device initramfs? currently one needs to enable it in target's `FEATURES += ramdisk` so as the result its then going to produce initramfs images for every device under that target which is suboptimal **** BEGIN LOGGING AT Sun May 05 11:41:08 2019 May 05 11:41:13 Hauke: ah, because of lantiq, thanks for confirmation May 05 13:23:13 Hey May 05 13:24:01 I have a problem with trying to get the bootlog of my xiaomi mi router 4C May 05 13:24:34 I got most of it but I think my connection settings are not 100% correct because the U-Boot section is partially scrambled. May 05 13:24:43 After that everything is fine. May 05 13:24:53 Does anyone know what I did wrong? May 05 13:26:27 cperrin: the U-Boot probably uses a diffeernt baud rate as Linux May 05 13:28:48 Possibly. I tried 9600, 57600, 115200, 128000 and 256000 May 05 13:28:54 None of them made it better May 05 13:29:18 Hauke: Any suggestions? May 05 13:30:14 https://pastebin.com/jWrcAVmL This is to log BTW May 05 13:30:28 cperrin: perhaps 19200 or 38400 ? May 05 13:31:19 cperrin: hm, that doesn't look like baudrate is wrong though, then you shouldn't get anything that looks like text. May 05 13:31:33 I tried that rates May 05 13:31:38 still nothing May 05 13:31:43 That's what I thiught May 05 13:31:58 maybe it has to do with stop bits or something like that May 05 13:32:06 are you using 8N1 ? May 05 13:35:20 yes May 05 13:35:36 I disabled flow controll May 05 13:41:29 Any more ideas? May 05 13:41:41 I tinkered with differnt flow controlls ettings May 05 13:41:45 nothing helped May 05 13:43:43 It's not an impossible idea that the baud rate divider value set in uboot for 115200 is incorrect, and it is then set correctly once kernel boots. May 05 13:47:05 I would also check that the serial earth pin has good continuity as I've seen strange things related to earths May 05 13:49:45 ldir: all othe baud rates I tried only made it worse. Maybe I wired it wrong: I have three wires RX,TR,GND. There is a fourth pin VCC but I don't use it. Should I? May 05 13:52:22 It's unusual to need VCC - if used at all it is used to power a lever shifter on the serial adapter. It depends on the serial adapter. TX, RX, GND are the normal ones. May 05 13:53:37 The fact it comes good suggests you have them in the correct order. I have seen issues where grounds weren't quite soldered correctly, poor continuity meant the whole thing shifted up & down, so mostly it worked ok but some characters were iffy. May 05 13:54:24 is there a sample log where someone has got all the characters correctly or are you blazing the trail? May 05 14:03:31 if the adapter plugs into usb, do NOT use vcc May 05 14:03:57 jow: that would probably need a lot of refactoring, I've found something bit easier with `FEATURES := selective-ramdisk` and NEEDS_INITRAMFS=1 https://git.openwrt.org/06f88b818823 May 05 14:04:26 jow: working example https://git.openwrt.org/01db74ced8859b May 05 14:05:23 ldir: I seem to be the first one that is looking at this device. I have found nothing about this model. May 05 14:05:55 So no clean log yet May 05 14:06:18 VCC one needs pretty much only if one has isolated serial adapter, which you would know if you have...and like said here, you already have some data flowing trough which means TTL-serial hardware itself works May 05 14:06:56 I have a USB TTL adapter btw May 05 14:09:20 I don't have a magic answer - it looks to me like the rate/format is subtly different from when the kernel boots. May 05 14:09:37 ynezz: not opposed to that appraoch, it just feels and looks like a kludge bypassing the image buld code May 05 14:14:13 yeah, its exactly that May 05 16:14:53 Is it just "habit" that tar-based sysupgrade files have a .bin extension rather than .tar, or a specific choice? May 05 16:17:16 My theory: back from where everthing kinda started was those linksys wrt54g(s/l) systems and they took .bin May 05 16:17:23 "stuck ever since" May 05 16:20:07 Confusing as can be when you're working with both the NOR and NAND versions at the same time! May 05 16:21:38 Also most of the OEM files are also bin May 05 16:36:40 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html May 05 16:42:59 jeffsf: I remember it's because users keep trying to extract tarballs before flashing and OpenWrt decided to use .bin instead then. May 05 16:47:21 hah May 05 16:54:55 gch981213: Bah, ugh, OK -- I'll change it back before I finish with this little beast May 05 16:56:47 * jeffsf sort of misses the rush of the fans on the laptop telling him that it is working hard compiling May 05 17:27:14 jeffsf: I think it also has to do with the metadata that the images have. So they are tar + something May 05 19:41:38 and can somebody tell me where to look for these files cdc_ether.ko rndis_host.ko usbnet.ko May 05 19:41:39 maybe some kind of shared repository. just such drivers. May 05 19:41:39 it's hard for me to find the source. everything is scattered round. I'm interested in these files on different kernels. in particular, currently on Linux 3.10.108 May 05 19:42:47 don't cross post May 05 21:51:43 We find ourselves in need to inject per-router secrets @work, and really, doing it at image creation time is... hell. Did anyone do something like extending the sysupgrade process to runtime-append to the rootfs at flash time? May 05 21:52:29 (for the record, this is to provide non-joke initial randomness *and* equipment-specific wifi password, etc). May 05 21:53:57 -- it needs to survive a factory reset unchanged, btw, which is why we thought of adding it to the rootfs. May 05 21:59:51 I find it quite questionable to add secrets to images May 05 22:02:02 I put per device wifi passwords in the uboot env when we flash the image. May 05 22:10:20 ynezz: indeed. May 05 22:10:41 karlp: can that be done in a generic way per arch (i.e. something that works on everything ar71xx, ath79...) ? May 05 22:11:03 karlp: I tried to look into a few tp-links we had around, and some looked like they didn't even *have* uboot env areas. May 06 00:01:06 What is the best way to report snapshot bugs for a specific device. I have an Archer A7 and I found a bug that when you delete all the VLAN entries, the add button will not add a blank vlan entry. I also have a question about performance. It seems that I can't get above 150mbps from WIFI to WAN but I can WIFI to LAN. May 06 00:03:32 I can get the full bandwidth of my ISP WIFI > LAN > router to ISP but WIFI > WAN > router to ISP only 150mbps. May 06 00:04:46 The OEM firmware works fine in this configuration. I'm ultimately going to replace my router with the A7 once I get the speed up. I tried it WIFI > WAN > ISP modem with the same slow results. May 06 00:13:26 hmh: how many archs do you have?! **** ENDING LOGGING AT Mon May 06 02:59:57 2019