**** BEGIN LOGGING AT Sat May 25 02:59:58 2019 **** BEGIN LOGGING AT Thu May 30 09:28:10 2019 May 30 09:38:29 sounds like a plan! May 30 09:41:20 BTW has anynone tested the urngd series yet? It would be nice to get at least one Acked-by or Reviewed-by as I'm quite hesitant to push this by myself, I would like to share possible anger which it might introduce with someone :) May 30 10:29:11 jow: could you please adapt the build bot the at91 target now has two subtargets sam9x and sama5 May 30 10:37:17 tmn505_: this patch should continue with v2 right? https://patchwork.ozlabs.org/patch/1100700/ May 30 10:38:16 tmn505_: if so I'm going to change the state from New to Changes Requested May 30 10:50:41 Hi what does the value "interface" mean in network config? e.g. the default has network.wan='interface' and and network.wan6='interface'? May 30 10:52:44 buggrit - just seen a minor size optimisation I could do in the act_ctinfo module just been accepted upstream May 30 11:30:05 Neighbor11111115: that means that the section named 'wan' in the 'network' uci config has the type 'interface' May 30 11:31:35 Ahh of course, sorry that was silly of me. Thanks KanjiMonster **** BEGIN LOGGING AT Thu May 30 11:38:25 2019 May 30 14:19:20 ynezz: yes May 30 14:30:48 kab-el: Hi. How is Your progress with mvebu kernel 4.19 bump? May 30 15:36:14 For (GNU) screen, I have vague memories of it being able to record and replay timing of the content. Does this ring a bell for anyone? May 30 15:37:48 NM -- `script`, not `screen` May 30 16:45:19 <_abc_> Hi. Does anyone have a link to the manpage or wiki page of the command /sbin/block ? I'm at the point where I am using 5 arguments it takes, one of which is not in the self-printed help... namely hotplug May 30 16:45:39 <_abc_> (the next step is to read sources) May 30 16:54:27 https://git.openwrt.org/?p=project/fstools.git;a=blob;f=block.c;h=39212d247fe46e07b4ea4113edf4687b59b7d4e7;hb=HEAD#l1805 May 30 16:55:41 which then dispatches to the appropriate s/r, such as main_hotplug(argc,argv) May 30 16:57:39 * jeffsf scratches hiss head as to why he bought another router he doesn't need **** ENDING LOGGING AT Thu May 30 17:08:18 2019 **** BEGIN LOGGING AT Thu May 30 19:16:09 2019 May 30 19:16:37 ye it requires a rebuild May 30 19:17:13 The driver will detect the flash based on reading its manufacturer ID and device ID then doing a table lookup. If you want to *use* more of the space, yes, you need to rebuild the entire image May 30 19:18:03 If you do so, you need to make sure you don't overwrite critical parts (like boot loader) or device-specific information, like ART May 30 19:18:26 <_abc_> I want to put in a much larger flash. Is 4MB now, will be 32 May 30 19:18:51 TP-Link has a "nasty habit" of putting the ART at the top of the address space, so you can't easily extend to larger flash May 30 19:19:11 <_abc_> I know. I will deal with that. I can always add another mtd partition and custom mount it, no? May 30 19:19:21 <_abc_> Above the 4MB "wall" with ART at the top. May 30 19:19:22 i saw a tutorial to mod the wr841 for twice the flash May 30 19:19:23 32 MB SPI NOR generally has a different way of being driven than 16 MB and below May 30 19:19:35 maybe that could be of any use for you May 30 19:19:38 <_abc_> I know, I'll probably go for 16 1st May 30 19:20:02 <_abc_> stealthplayerx: I saw that, he had to copy the spi flash start and edited some header parts in it and did not need to recompile, iirc. May 30 19:20:20 So will need to make sure the driver supports it, as well as completely repartitioning the device, including manually moving any needed partitions, and making sure that things like ART extraction are updated May 30 19:21:03 <_abc_> Yes, except I would like to leave ART where it is, since it's address is cast in stone in the dtsi, and add a 5th mtd partition from 4MB to the top. May 30 19:21:03 In my opinion, a relatively underpowered device, both in terms of CPU and RAM, isn't worth the time May 30 19:21:23 <_abc_> jeffsf: It's an experiment for now, I am adding hardware to it, it is no longer a router. May 30 19:21:41 Nothing is "cast in stone" -- that's why there is generally a DTS per board May 30 19:21:44 <_abc_> It's junk anyway, I did the USB mod on it May 30 19:22:00 <_abc_> jeffsf: I would like to not recompile at all and deal with the extra mtd partition in user mode. May 30 19:22:27 <_abc_> perhaps pivot the overlay mount onto it. May 30 19:22:28 <_abc_> Later May 30 19:22:49 <_abc_> I wonder if the kernel will get confused by the larger device. May 30 19:22:57 <_abc_> SPI flash. Without recompilation. May 30 19:23:06 No, it just ignores the "extra" space May 30 19:23:24 <_abc_> Where is the table of supported flash spi devices usually? May 30 19:23:32 tmn505_: Hi. Unfortunately I have exams now, so I took vacation from work :-( Maybe I'll find some time to look into this sometime this week but not promising anything. Sorry May 30 19:23:53 * _abc_ needs to index the github openwrt repo May 30 19:24:16 linux: drivers/mtd/spi-nor/ and the associated header files May 30 19:24:52 tmn505_: it took me some 2 hours to walk through all the patches from 4.14 and select those still needed for 4.19. Maybe it won't take too long to try to merge. I think I'll try tomorrow May 30 19:26:49 https://elixir.bootlin.com/linux/v4.19.46/source/drivers/mtd/spi-nor May 30 19:27:12 clickable cross references May 30 19:28:30 <_abc_> Normally none of the quad speeds will be turned off by default, right? I think I'll have spansion or atmel spi flash. Or winbond. May 30 19:30:09 I would *guess* that it is auto-detected by the driver -- there may be SoC or board limitations May 30 19:31:01 <_abc_> The older 17.x version running on this old 741 with mods says, in dmesg: May 30 19:31:03 <_abc_> [ 1.700000] ath79-spi ath79-spi: master is unqueued, this is deprecated May 30 19:31:11 <_abc_> [ 1.710000] m25p80 spi0.0: found s25sl032p, expected m25p80 May 30 19:31:17 <_abc_> [ 1.720000] m25p80 spi0.0: s25sl032p (4096 Kbytes) May 30 19:31:33 <_abc_> So it deals with expecting a 2MB one and finds a 4MB one May 30 19:31:50 "expected m25p80" is pretty common -- just means that the mach file (predecessor to DTS) used the "generic" name May 30 19:31:59 <_abc_> oh May 30 19:32:05 <_abc_> But the size is much smaller May 30 19:32:09 <_abc_> then it goes on: May 30 19:32:10 <_abc_> [ 1.720000] 5 tp-link partitions found on MTD device spi0.0 May 30 19:32:13 <_abc_> why 5? May 30 19:32:34 <_abc_> Ah ok it lists them May 30 19:33:11 Probably four from the mach/DTS, then one from auto-split of ROM from overlay (OpenWrt patches to the kernel) May 30 19:33:19 <_abc_> I'll need to get the sources and read them. May 30 19:34:03 I'd forgo the ar71xx ones and just deal with ath79 -- better to learn the current way if you're doing this as a learnind experience May 30 19:34:10 <_abc_> "u-boot" "kernel" "rootfs" "rootfs_data" "art" "firmware" May 30 19:34:17 <_abc_> What is firmware? Radio fw? May 30 19:34:38 <_abc_> the ar71xx what? May 30 19:34:51 <_abc_> ar71xx is the old name or such? May 30 19:35:19 ar71xx was a mach-based platform. Deprecated in favor of the upstream-supported ath79 May 30 19:35:42 <_abc_> Wait, so what I am reading is not what is in a 18.06.2 741 image? May 30 19:35:46 Basically if your device is still on ar71xx, it's about to be left in the dus May 30 19:35:56 <_abc_> ok, no problem May 30 19:36:16 ath79 is what is being used on `master` -- effectively no further work on ar71xx May 30 19:36:23 <_abc_> ok May 30 19:39:58 and I guess ath79 needs more than the 32/4 several ar71xx devices have? May 30 19:40:32 4/32 devices are already on life-support May 30 19:41:05 Hence the "tiny" target -- the 32 MB of RAM is a problem even with virtually nothing istalled May 30 19:41:48 ath79 isn't that much bigger, but it has been a problem for a few years already May 30 19:42:23 hmm, I wonder if I can just replace the small chips on the ar71xx devices I have with bigger ones May 30 19:42:36 The core dev team has been doing great things too keep 4/32 running, but the writing has been on the wall for some time now May 30 19:43:12 You can buy a complete router with 16/128 starting at under US$20 May 30 19:43:35 kab-el: np and thx. I have enough time after this weekend, so if that's ok with You I'll merge our work and post it in Your stead. May 30 19:46:30 <_abc_> jeffsf: depends where, some places those are marked up as premium for no good reason. May 30 19:46:51 <_abc_> I'm in Eastern Europe, the tail edge of $everything. May 30 19:53:23 <_abc_> jeffsf: say a type of $20 router with 16/128? May 30 19:54:05 GL.iNet AR300M-Lite, if still available, VIXMINI is its replacement May 30 19:54:34 If you need two Ethernet ports, the MT300N-V2 May 30 19:55:23 <_abc_> That is more like $45 with wifi May 30 19:55:53 <_abc_> And that's the aliexpress price, not a shop price, not fair. May 30 19:56:44 GL.iNET GL-MT300N-V2 -- US$20.49 then less 5% on Amazon US May 30 19:56:55 <_abc_> qualcomm has good open source support? May 30 19:57:14 <_abc_> jeffsf: I saw it on amazon, slowly no longer available May 30 19:57:37 "Good" is always relative -- their data sheets are still proprietary, but the upstream and OpenWrt support are both very good May 30 19:58:10 The AR300M-Lite was $15 or so, but they're gone here in the US now May 30 19:58:14 <_abc_> Better than Broadcom then :) May 30 19:58:19 Much, yes May 30 19:58:47 <_abc_> I had to manually compile some ralink usb wifi stick drivers and I have scars from that. How bad can it get. May 30 19:58:47 I've heard good things about the MTK-based devices, but don't own any May 30 19:59:07 <_abc_> I have a lot of mtk based low end stuff, including my phone. May 30 19:59:31 <_abc_> I did not see the sdk but the built in stuff is okay. Even engineer mode included in the phone from factory etc. May 30 19:59:48 <_abc_> Everything the savvy user needs to brick it :) May 30 20:00:26 i had a phone with a mtk in it, you cna even erase the damn imei May 30 20:00:32 <_abc_> yep May 30 20:00:43 Other options and opinions on "cheap" devices at https://forum.openwrt.org/t/whats-your-favourite-cheap-lede-openwrt-device/1362 May 30 20:00:46 <_abc_> 90% of the lower cost phones and smart watches are mtk here May 30 20:01:17 I own several GL.iNet devices, so I can speak from experience on those, but not most of the others May 30 20:01:23 <_abc_> I caught a lot of mac 112233445566 in my wifi logs in the past, from the bus stop in front ... May 30 20:01:53 <_abc_> It's interesting the mtk device have a GPL leaflet included in the box here in the EU. May 30 20:01:59 <_abc_> With a link on it, which works. May 30 20:02:30 my most powerful devices are lantiq May 30 20:02:55 ynezz: the proper way for describing your changes is in above your SoB, not behind it ;p (see https://www.kernel.org/doc/html/latest/process/submitting-patches.html#sign-your-work-the-developer-s-certificate-of-origin ) May 30 20:08:00 KanjiMonster: ok, noted, thanks May 30 20:08:34 but I swear, I've seen it many times in many places already May 30 20:14:56 Just to double check, `ucidef_set_led_netdev` can't be replaced by a DTS property, at least at this time, correct? May 30 20:29:36 <_abc_> re: uci set system.@button[-1].button=wps :: index -1 means "at next available"? May 30 20:30:56 <_abc_> hm? May 30 20:31:54 ldir:nice! May 30 20:34:23 <_abc_> https://forum.openwrt.org/t/wireless-router/36380/3 the things people pull off May 30 20:34:35 KanjiMonster: just saying http://paste.ubuntu.com/p/DKJ33Kr9f9/ :) May 30 20:46:08 _abc_: https://openwrt.org/docs/guide-user/base-system/uci May 30 20:46:17 You can also use negative indexes, such as system.@timeserver[-1]. “-1” means the last one, “-2” means the second-to-last one, and so on. May 30 20:47:56 You perhaps want `add_list` May 30 20:48:16 <_abc_> Ah ok, I get it. May 30 20:48:49 <_abc_> Do you consider the syntax system in uci better than cisco ios clisjh? May 30 20:48:54 <_abc_> *clish? May 30 20:49:41 I think it is appropriate for resource-constrained devices that are often managed by non-experts May 30 20:50:00 as well as those of us who hand-edit the config files May 30 20:50:06 <_abc_> As long as they read the great docs online. After they find them. May 30 20:50:14 Hence LuCI May 30 20:50:33 <_abc_> cisco also have a web gui. I avoid it like the plague. May 30 20:51:30 Yep, too slow on my SG300-series switches to do anything but a minor change -- LuCI is pretty solid, especially for "roughing out" config I haven't seen before May 30 20:52:05 Even better now that nginx and TLS is supported by packages May 30 20:53:36 <_abc_> Why are package sizes not shown in the luci software selector for installation? I've asked this since 12.x times. May 30 20:54:13 <_abc_> Is that a design decision? I had to poke and peek at various Packages files manually to find out what I needed, more than once. And the info is not in one place. Why not add it to the web gui? May 30 20:55:35 `opkg` is evolving -- at least for me, getting ABI compatibility information is a higher priority. May 30 20:56:22 I'm not a core dev by any stretch, but many struggle with devices that barely have enough resources to run, not to mention manage generally unneeded metadata May 30 20:56:23 <_abc_> jeffsf: yes, and the evolver should be on this channel? May 30 20:57:02 <_abc_> Adding an average or 10 bytes in a table per row will not break the bank, the table stays in ram normally. May 30 21:09:41 "break the bank" -- You've got a 32 MB device -- tmpfs on my running Archer C7v2 shows 19 MB of RAM May 30 21:10:27 <_abc_> Hmm? Yes it is tight but the 100 packages adding sizes in the table are just another 10 bytes per package, 1kbyte. May 30 21:11:11 <_abc_> And it is quite critical. I had to back out of bricked luci due to exhausted flash manually a number of times. May 30 21:12:22 3,823 packages in openwrt_packages alone May 30 21:13:45 <_abc_> okay 38 kbytes more then May 30 21:13:54 <_abc_> in ram **** ENDING LOGGING AT Fri May 31 02:59:58 2019