**** BEGIN LOGGING AT Sun Nov 25 03:00:00 2018 Nov 25 04:14:32 Hi! Backporting of spi-mem is way easier than I thought and I'd like to submit it. Such a backport also means we could have the upstream spi-nand framework available in linux-v4.14 but it has a side-effect: m25p80 won't use the existing spi-flash-read operation, meaning it'll slowdown SPI reading performance on target like ar71xx/ath79 unless we wr Nov 25 04:14:32 ite a spi-mem-op for it. Shall I submit my backport or just playing with it myself because the side-effect is somehow unacceptable? Nov 25 09:46:29 pkgadd: I don't see that patch/series in patchwork? Nov 25 10:03:27 Is there an easy way to extract the text only patch from our email archive e.g. http://lists.infradead.org/pipermail/openwrt-devel/2018-August/013797.html Nov 25 10:04:07 I'm not actually interested in that patch but more the principle. Nov 25 10:06:58 similarly, if there is a patch in patchwork and I no longer have the email in my email client, is there a way to reply to that patch? Nov 25 10:30:04 you can reply by clicking the email address at the top line of the email in the web view. It has a mailto: link with the correct In-Reply-To Header(s). Nov 25 10:31:54 The "Gzip'd Text" links from http://lists.infradead.org/pipermail/openwrt-devel/ qualify as mbox files that you can load into any decent mua or process with formail & procmail. Nov 25 10:36:50 DLange: thank you - that's much closer :-) Nov 25 10:37:08 * Borromini had been looking but wasn't able to figure it out Nov 25 10:38:32 Borromini: so it wasn't just me being dumb Nov 25 10:40:35 lol ldir I feel like that most days! :-D Nov 25 10:40:49 How you doing anyway? Nov 25 10:41:04 and you Borromini: you good? Nov 25 10:42:37 I'm well - ok well enough - but busy so openwrt is taking back seat... and all the stuff I used to know has dribbled out of the ears. Nov 25 10:42:43 what's bin going on in OpenWrt land? I have spent most of the last week coffing up my lungs and feeling like I am made of ice or larver. Nov 25 10:43:14 ldir lool glad you are ok tho mate. Nov 25 10:43:51 It comes back to you it's jost slow. Nov 25 10:44:22 When I mess around with things now I make notes. Nov 25 10:44:27 Tapper: yeah. although i'll be happy when spring comes around :( :( Nov 25 10:44:43 will be off to work shortly - context switching between openwrt & work & domestic stuff & life is hard. Nov 25 10:44:57 Borromini I feel your pane! I hate the winter. Nov 25 10:45:16 ldir: that should be easy. real life isn't about bits and bytes after all ;) Nov 25 10:45:21 I love to go working out in the sticks with the ms and kids Nov 25 10:45:31 mrs Nov 25 10:46:13 walking* lol back to the old cant spell tapper hahah Nov 25 10:47:01 Yeah ldir I bet it's like when you go back to school after the summer hols hahah Nov 25 10:47:40 My god school seems like such a long time ago! Nov 25 10:57:49 Definitely Nov 25 10:58:01 Feels good :-) Nov 25 11:01:47 aha! https://patchwork.ozlabs.org/cover/959151/ so it got to v3 and was then marked superceded..well the patches did e.g. https://patchwork.ozlabs.org/patch/959150/ so that's why no longer in patchwork. Nov 25 11:08:01 rmilecki: if my dmesg says: "2 uimage-fw partitions found on MTD device firmware", what's the corresponding compatible string? Nov 25 11:08:32 russell--:[OpenWrt-Devel] Specifying "firmware" partition format on all DTS targets Nov 25 11:08:34 http://lists.infradead.org/pipermail/openwrt-devel/2018-November/014706.html Nov 25 11:11:40 compatible = "denx,uimage"; Nov 25 11:12:31 "denx" is reference to denx.de of u-boot fame? Nov 25 11:12:38 yes Nov 25 11:12:51 i was looking for a vendor who introduced uimage Nov 25 11:12:59 and it seems uimage was introruced by u-boot Nov 25 11:13:04 and u-boot was developed by denx Nov 25 11:13:11 that's why I came up with "denx" Nov 25 11:13:15 seems plausible Nov 25 11:15:35 I pull latest git and tried denx,uimage. Seems to work. Don't get any errors. Nov 25 11:16:14 *pulled Nov 25 11:18:32 micmac1: great :) Nov 25 11:53:06 Any idea what the equivalent of ath79_gpio_output_select(DIR825C1_GPIO_LED_BLUE_USB,AR934X_GPIO_OUT_GPIO); could be on ath79? Nov 25 11:59:19 micmac1: try git grep DIR825C1_GPIO_LED_BLUE_USB Nov 25 12:00:47 micmac1: Somehow write a subnode for pinmux in dts. Nov 25 12:01:38 But I haven't figure out the usage of pinctrl-single yet :( Nov 25 12:04:20 Argh Nov 25 12:04:26 Sounds complicated Nov 25 12:04:28 Thanks :) Nov 25 12:05:50 russell--: that returns some hits in the ar71xx mach file? Nov 25 12:06:20 yeah, provides the corresponding gpio and active_lowness Nov 25 12:07:10 Mmh yes, but if I define the led in dts it's there but doesn't work. In mach file the led was defined and then Nov 25 12:07:14 Sorry Nov 25 12:07:34 And then *on top* this ath79_gpio_output_select was set to make it work Nov 25 12:08:36 micmac1: you probably already checked other ar71xx mach files that use something similar but that do have an ath79 equivalent already? Nov 25 12:08:49 I have the feeling that I need pinctrl also to make the 5GHz led blink (ap9x_pci_setup_wmac_led_pin(0, 0); Nov 25 12:09:43 Borromini: yes, for instance cr5000. But there is nothing in the DTS file that would suggest they do something similar like was done on ar71xx. But I'm still comparing Nov 25 12:10:31 micmac1: have you looked at /sys/kernel/debug/gpio to see the direction? Nov 25 12:12:09 yes, it's gpio-11 ( |d-link:blue:usb ) out hi Nov 25 12:12:31 But when I change the brightness it doesn't go to low Nov 25 12:13:12 change it how? Nov 25 12:14:00 echo 255 > /sys/class/leds/d-link\:blue\:usb/brightness Nov 25 12:14:27 what's the value of trigger? Nov 25 12:14:48 usbport Nov 25 12:15:07 try echo timer > trigger Nov 25 12:15:17 does it blink? Nov 25 12:15:44 No Nov 25 12:15:46 :) Nov 25 12:15:54 It's the weirdest led I've ever seen Nov 25 12:16:00 I haven't seen many, though :) Nov 25 12:16:21 it does sound like a pinmux'ing problem Nov 25 12:16:27 oh noooooo Nov 25 12:16:36 but, just a guess Nov 25 12:25:36 russell--: ath79_gpio_output_select is actually a function that sets pinmux register :) Nov 25 12:28:27 I found a board that was converted, qca9558_openmesh_om5p-ac-v2.dts Nov 25 12:31:25 magic numbers-- Nov 25 12:31:43 It went from ath79_gpio_output_select(2, 0); to pinctrl-single,bits = <0x0 0xff00 0x0>; Nov 25 12:32:01 Kind of tough to see the relation in the numbers :) Nov 25 12:33:53 There's also ath79_gpio_output_select(16, 0); to pinctrl-single,bits = <0x10 0xff 0x0>; Nov 25 12:34:37 At least 0x10 fits 16 Nov 25 12:35:05 Maybe 0x0 for the first one is a typo Nov 25 12:37:57 pinctrl-single,bits is documented in Documentation/devicetree/bindings/pinctrl/pinctrl-single.txt in the kernel tree Nov 25 12:38:29 Yep, looking at that currently Nov 25 12:39:16 micmac1: That two nodes aren't actually enabled. To enable them you'll need to define pinctrl-X=<&name> somewhere else. Nov 25 12:42:36 gch981213[w]: you mean the pinctrol in qca9558_openmesh_om5p-ac-v2.dts isn't actually used? Nov 25 12:42:42 *pinctrl Nov 25 12:59:34 micmac1: yes. ar9341_tplink_tl-wr841n-v2.dts has an example of enabling jtag_disable_pins defined in ar934x.dtsi Nov 25 15:36:34 I just installed (OpenWrt SNAPSHOT r8466-251c350727 / LuCI Master (git-18.329.25235-26e903c)) on a TP-Link Archer C20 v4 (https://openwrt.org/toh/hwdata/tp-link/tp-link_archer_c20_ac750_v4) Nov 25 15:36:43 well upgraded from a previous snapshot really, and now it won't show me any memory graphs. using the material theme if that matters Nov 25 15:36:51 where do I start to troubleshoot this? Nov 25 17:59:07 this commit looks intresting: https://git.kernel.org/linus/a9911937e7d332761e8c4fcbc7ba0426bdc3956f Nov 25 18:25:47 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html Nov 25 20:02:21 I'm hoping somebody in here might have a clue; I was building a customized version of the current snapshot for a openwrt-ar71xx-tiny-device-tl-wr841-v9, but it doesn't produce a factory.bin file, and there's no sysupgrade.bin file either. How could I get these? Nov 25 20:04:56 There should definitely be a way to produce them, as they're available for that router in the downloads section. I just don't know what I'm doing wrong; I was following the guide closely. Nov 25 20:39:11 monx: are you building with BUILD_LOG=1? if so, check build logs. Nov 25 20:40:30 russel--: Will do, thanks. Is that an env variable? Nov 25 20:41:40 you can either add it to the command line or via menuconfig (or equivalent0 Nov 25 20:41:56 the usual reason for image not being created, and especially for tiny devices, is that your config results in an oversized image Nov 25 20:42:41 Even if I just used the defaults? The only thing I added in menuconfig was luci Nov 25 20:42:50 from memory, you should find the relevant info in logs/target/linux/install.txt Nov 25 20:43:21 well, kernel 4.14 and luci is very, very tight - probably not going to fit into 4 MB Nov 25 20:44:10 russell--: Thanks, let me re-compile and check Nov 25 20:45:06 pkgadd: Does openwrt-18.06.1-ar71xx-wr841-v9-squashfs-factory.bin from the downloads section come with luci? Nov 25 20:45:32 release images, like 18.06.1, do - daily snapshot images don't Nov 25 20:46:32 Then I suppose somebody was able to cram luci on that router before Nov 25 20:47:20 Indeed, the logs read [mktplinkfw] *** error: images are too big by 38105 bytes Nov 25 20:47:34 if you don't need pppoe, that's something you can shed Nov 25 20:48:24 18.06.1 came with kernel 4.9, 4.14 combined with the newer gcc version needs 200-300 KB more space Nov 25 20:49:27 # CONFIG_PACKAGE_kmod-ppp is not set Nov 25 20:49:40 # CONFIG_PACKAGE_ppp is not set Nov 25 20:49:50 So disabling pppoe is the way to go? Nov 25 20:50:00 Not entirely sure what pppoe is used for. Assuming I remove it, what would be the implications? Nov 25 20:50:08 Other than "you can't use pppoe anymore" :p Nov 25 20:50:23 if your isp doesn't need it, then you don't either Nov 25 20:50:24 ADSL/ VDSL connections usually requires PPPoE Nov 25 20:50:34 cable or ftth usually don't Nov 25 20:50:54 pkgadd: my fiber provider does, sigh Nov 25 20:51:15 ouch Nov 25 20:51:46 especially at those speeds, PPPoE must be a quite significant performance drop Nov 25 20:52:21 not really, it's mostly just connection negotiation Nov 25 20:52:38 although it does eat a few bytes out of the mtu Nov 25 20:53:11 speed test says 900-ish Mbps Nov 25 20:54:35 Is there a way I can find out if pppoe is required? I live in a dorm and don't have control over the router that's directly hooked with the street Nov 25 20:54:54 look at your /etc/config/network Nov 25 20:55:35 if it doesn't say proto pppoe anywhere, you should be good Nov 25 20:55:51 That's assuming I already have a working OpenWrt installation, isn't it? :p Nov 25 20:55:59 yes Nov 25 20:57:03 if there is a router between your router and the ISP then you almost certainly don't need it Nov 25 20:57:24 but, check your admin interface wan configuration Nov 25 20:57:37 probably just dhcp Nov 25 20:57:45 Perfect, I know for a fact that this is the case Nov 25 20:57:49 I'll go ahead and drop it Nov 25 20:58:57 And maybe also strip the binaries Nov 25 20:59:26 opkg is also an easy removal, you won't be able to use it on a 4/32 system anyways Nov 25 21:00:23 I won't? :s Nov 25 21:00:29 also write on your arm in sharpie, "need a new router with more ram and flash" Nov 25 21:00:41 :D Nov 25 21:01:16 After all this pain, this might actually be worth considering Nov 25 21:01:36 4/32 might have been o.k. in 2008-2010, but not afterwards Nov 25 21:01:42 you can probably find something better on ebay for the cost of a few sandwiches Nov 25 21:03:17 Do you always convert into sandwiches? Nov 25 21:03:21 people jumping to 802.11ac is great, because they are selling off their old 802.11n dualband device *with better driver support* for cheap Nov 25 21:04:11 even 802.11ac/ wave1 gear can be surprisingly cheap, with e.g. the BT Home Hub 5 Type A for under 20 EUR in total Nov 25 21:05:03 is there something wrong with sandwiches as a universal currency? Nov 25 21:05:09 not the fastest kid around the block, but a very solid device with amazing capabilities - and 2*500 MHz mips, 128 MB NAND and 128 MB RAM - and 802.11ac/wave1 dual-band wlan Nov 25 21:06:00 Apparently, there's also CONFIG_PACKAGE_kmod-pppoe, not juts CONFIG_PACKAGE_kmod-ppp. Should I throw out both? Nov 25 21:06:10 yes Nov 25 21:06:12 well, I was going for EUR as you can only (sensibly) get the device in question in europe - and because 20 quid is slightly above sandwich territory ;) Nov 25 21:06:45 128 MB NAND sounds pretty usable, could actually run oh-my-zsh on there Nov 25 21:06:51 or rather 8-9 GBP plus shipping Nov 25 21:08:02 Guess I'll set aside a few EUR next month, sounds compelling Nov 25 21:08:35 check the wiki first, initial flashing isn't for the faint of heart, but it's a great device Nov 25 21:09:16 need to get a second one before the end of march Nov 25 21:10:27 Hm, ppp, pppe and pppox are all marked with -*-, and pressing "N" or toggling the selections doesn't do anything Nov 25 21:10:45 something depends on them Nov 25 21:10:49 the "luci" metapackage depends on ppp Nov 25 21:11:10 Ah :p Nov 25 21:11:54 ou can just add luci-base and hte parts yourself if you don't need ppp Nov 25 21:12:10 Hm, there's still something that depends on it (after removing Collections/luci) Nov 25 21:12:16 press ? Nov 25 21:12:27 it will say ... selected by... xxxx=y or something Nov 25 21:12:50 Ah, perfect, thanks Nov 25 21:13:45 I suppose I won't need PACKAGE_kmod-isdn4linux, lol Nov 25 21:20:51 3.8M, woohoo! Nov 25 21:21:06 Thanks, everyone :) Nov 25 21:21:37 (including luci-base and some luci-related modules) Nov 25 23:21:44 Openwrt has issues with the devicetree files, but I don't see the problem. Build output: https://paste.debian.net/1053227/ Nov 25 23:25:33 the three dts files. http://paste.ubuntu.com/p/yTdDcNHKCg/ http://paste.ubuntu.com/p/sgG6wd955z/ http://paste.ubuntu.com/p/7rYbjwkSyd/ Nov 26 01:08:04 Rene__: You need to apply the owrtdtb patch. Nov 26 01:14:50 There was a patch for MIPS similar to this one but I can't find it now...https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=target/linux/generic/pending-4.14/332-arc-add-OWRTDTB-section.patch;h=58f14b6ca5493f97f33bce2df7d4162be77e214d;hb=HEAD **** ENDING LOGGING AT Mon Nov 26 02:59:59 2018