**** BEGIN LOGGING AT Mon Jun 29 02:59:57 2020 Jun 29 07:03:01 anyone know if there is an effort at all to port OpenWRT to Teltonika RUTX09/10/11? Jun 29 07:08:51 gatorbeug: is there a wiki page about it? Jun 29 07:10:43 gatorbeug: they claim their RutOS is based on OpenWrt Jun 29 07:11:30 https://teltonika-networks.com/support/gpl/ Jun 29 07:26:52 It is based on OpenWRT, however what I've used of it is a bit buggy. @russell-- Jun 29 07:27:03 Ie, the OSPF settings. Though that is a bit specific. Jun 29 07:58:47 adrianschmutzler: ping Jun 29 08:44:58 Guys, does the OpenWrt patchwork pick up the patches automatically from the mailing list, or do they have to be added manually to the queue? Jun 29 09:03:50 gatorbeug: see what they changed to make it work, maybe you can port it back to openwrt Jun 29 09:04:35 rsalvaterra: if you formatted it right and your subject has [Patch] it should just work afaik Jun 29 09:05:56 It's formatted right, sure, but I still don't see it there… Jun 29 09:06:25 https://lists.openwrt.org/pipermail/openwrt-devel/2020-June/029819.html Jun 29 09:10:27 might want to ping dwmw2_gone maybe something to do with the mail server going tits up Jun 29 09:11:34 Yeah, I suspect so. Jun 29 09:12:04 perhaps patchwork is no longer subscribed? Jun 29 09:14:21 dwmw2_gone: I'm not sure, I see other patches there (sent after mine). Jun 29 09:15:19 Maybe there was some short period during which no patches were picked up. Jun 29 09:17:16 maybe related to the subject mangling? Jun 29 09:19:23 Only if something changed meanwhile… I mean, it doesn't get any more standard than [RFC PATCH]. Jun 29 09:20:58 do you see patches in patchwork *before* yours, since the outage? Did you send it *before* someone resubscribed patchwork perhaps ? Jun 29 09:23:02 Hm… let me see. Jun 29 09:23:47 let me have message-id and I can check it was delivered to patchwork Jun 29 09:25:04 dwmw2_gone: is the message-id the number of the email on pipermail? Jun 29 09:25:08 Message-Id: <20200626101458.8443-2-karlp@etactica.com> should have gone to patchwork, but didn't? Jun 29 09:25:17 no, the Message-Id: header of the message itself. Jun 29 09:25:56 Need to get into Gmail. Let me see… Jun 29 09:26:38 20200625223619.881426-1-rsalvaterra@gmail.com Jun 29 09:28:26 I resent the same patch the next day, but this is the message-id of the first email. Jun 29 09:31:19 karlp: I also don't see your Lua patches there. Jun 29 09:35:10 yeah, I know. Jun 29 09:35:24 Ill resend them tomorrow or something, there's no rush Jun 29 09:35:41 Jun 26 06:18:43 2020 (7936) post to openwrt-devel from karlp@etactica.com, size=10263, message-id=, success Jun 29 09:35:46 I think that's the one Jun 29 09:35:59 Mailman stupidly logging in localtime which is EDT (GMT-0400) there. Jun 29 09:37:43 oh, is that one of the ones that was "wrapped" by mailman due to stupid DMARC crap? Jun 29 09:38:11 do both of your domains explicitly forbid you from actually posting to mailing lists and say "any message from this person that you receive from a mailling list server is a fake" ? Jun 29 09:39:42 itnever used to, but it's office365, so who knows what reallyhappens in the background Jun 29 09:39:43 What? Not at all, I'm subscribed to debian-powerpc, I've contributed several (Linux) kernel patches and never had any issues. Jun 29 09:40:00 or maybe there was a config change which wrapped *all* messages? Jun 29 09:40:15 There was a discussion about it on the list, which I missed, and I think jow played with the settings? Jun 29 09:40:39 back to normal "no wrapping; let anyone stupid enough to reject for a DMARC 'failure' lose messages" mode now. Jun 29 09:41:17 Email is weird. :) Jun 29 09:41:24 email is really really simple Jun 29 09:41:30 not any more :) Jun 29 09:41:40 Except when it isn't. :P Jun 29 09:41:42 but then people still manage to misunderstand it and come up with stupid broken schemes like SPF and DMARC. Jun 29 09:41:49 is dmarc controlled by spf or is it different? Jun 29 09:41:57 "Something must be done" "This is something" Jun 29 09:42:18 it's similar in concept. SPF is done on the SMTP reverse-path (and is thus both pointless *and* broken). Jun 29 09:42:22 we've got spf tags on etactica.com to beable to sendmail via aws mail stuff Jun 29 09:42:31 DMARC extends it to the visible From: header, which makes it even more broken when you put in a strict policy Jun 29 09:42:34 but I don't remember doinganything for dmarc especially Jun 29 09:42:46 http://david.woodhou.se/why-not-spf.html Jun 29 09:43:20 anyway, mystery solved. Patchwork missed those messages because they were wrapped due to DMARC insanity; that's stopped now. Jun 29 09:43:28 if you resend it should see them. Jun 29 09:43:42 Hmm… should I resend? Jun 29 09:43:49 you might be technicallycorrect, (about spf) but when people like goog and yahoo/office all require it, it doesn' Jun 29 09:43:54 t matter if it's wrong, it's required Jun 29 09:47:13 Resent. Third time's the charm…? :P Jun 29 09:50:23 karlp: sure, but they don't require a record that ends with -all Jun 29 09:50:42 Meanwhile, I think I'm just going to remove tmp-on-zram support from procd. It just doesn't make sense when you have swap on zram. Jun 29 09:50:44 SPF can say "yes, this message IS valid" Jun 29 09:50:49 or it can say "I don't know" Jun 29 09:50:54 that's useful enough, if that's what it's used for Jun 29 09:51:01 SPF *cannot* say "No, this message isn't valid." Jun 29 09:51:14 which is what people abuse it for, too frequently Jun 29 10:08:15 Hmm… is /tmp/shm used for anything? I see procd creating the directory (and trying to mount it as tmpfs, lol), but it's empty. Jun 29 10:11:54 I'll just leave it be for now. **** BEGIN LOGGING AT Mon Jun 29 10:19:14 2020 Jun 29 10:42:01 dwmw2_gone: I see the patch on patchwork already, looks like mailman is working properly again, thanks! Jun 29 10:45:57 hmm.. so extroot on non mtd devices is broken? Jun 29 11:14:26 Oh, FFS…! Jun 29 11:14:31 if (mount_zram_on_tmp()) { Jun 29 11:14:31 mount("tmpfs", "/tmp", "tmpfs", MS_NOSUID | MS_NODEV | MS_NOATIME, "mode=01777"); Jun 29 11:14:31 mkdir("/tmp/shm", 01777); Jun 29 11:14:31 } else { Jun 29 11:14:31 mkdir("/tmp/shm", 01777); Jun 29 11:14:31 mount("tmpfs", "/tmp/shm", "tmpfs", MS_NOSUID | MS_NODEV | MS_NOATIME, Jun 29 11:14:32 "mode=01777"); Jun 29 11:14:32 } Jun 29 11:15:11 You'd think mount_zram_on_tmp() would return true if we were mounting over zram, right? Wrong, it's exactly the other way around. *facepalm* Jun 29 11:15:27 *sigh* Jun 29 11:19:02 rsalvaterra: typically if its an int return, its 0 = okay, -1 = fail. If its a bool return function then 1 (true) = okay, 0 (false) = fail Jun 29 11:19:54 jow: Sure, but in that case the name of the function is misleading. Jun 29 11:21:42 not for me. Its a function with side effects (mount something) that might fail with -1 Jun 29 11:22:00 it would be misleading if it'd be called is_zram_mounted_on_tmp() Jun 29 11:22:08 while having the same return value semantics Jun 29 11:23:22 I do agree that an if (mount_zram_on_tmp() != 0) { ... } would be clearer Jun 29 11:23:57 Meh, no biggie. I'm killing that function altogether. Jun 29 11:24:51 I had to fall back to my Archer C6, since the Omnia didn't like the change, but that's development. :P Jun 29 11:26:14 why are you removing that function? Jun 29 11:26:36 I'm removing support for tmp on zram from procd. Jun 29 11:26:44 It just doesn't make sense. Jun 29 11:27:35 dwmw2_gone: karlp: I mistakingly enabled "from_is_list" which wraps all mailing list messages (basically produces one-message-digests). That might have provented patchwork from picking up the patch embedded in the inner mime message attachment Jun 29 11:28:17 this settign is now back to its normal non-wrapping setting. Wrapping is now only performed for DMARC quarantine/reject domains Jun 29 11:30:24 better to turn off wrapping completely Jun 29 11:30:35 few recipients will make the mistake of *rejecting* unwrapped messages. Jun 29 11:30:42 that will result in mass-bounces and disabled subscriptions Jun 29 11:30:55 which was the reason to enable it in the first place Jun 29 11:31:10 only for people with broken receive setups, and they can fix those. Jun 29 11:31:21 if wrapping can be done only for broken recipients, that might be useful. Jun 29 11:31:29 or let them subscribe to the digest instead Jun 29 11:31:40 Gmail has a "broken receive setup" and it cannot be fixed Jun 29 11:32:43 not by its users anyway Jun 29 11:37:58 jow: the rationale for removing zram support from procd is simple: it's redundant and more limited than zram-swap. Jun 29 11:50:11 In other words, if you think you need to mount tmp on zram, what you really need is zram-swap. ;) Jun 29 11:51:21 uhm why? Jun 29 11:51:27 is tmpfs using swap? Jun 29 12:19:00 jow: ok, as an example, let's imagine we only use procd's tmp on zram. Jun 29 12:20:24 This means all files written in /tmp are written on an ext4 filesystem which zram compresses with lzo (or lzo-rle, on Linux 5.4), since we can't even choose a different algo. Jun 29 12:23:50 But the people who may actually need to compress memory in order to have a usable system will surely also enable zram-swap, since inactive anonymous memory can be swapped-out and compressed as needed by the kernel. Jun 29 12:24:38 So, now we have a zram device for /tmp, and a zram device for swap. Jun 29 12:25:47 But, the thing is… by design, tmpfs is backed by anonymous memory. The memory pages allocated to tmpfs can be swapped out at any time, if the kernel sees fit. Jun 29 12:26:25 So, why have a zram device dedicated to /tmp, when a zram swap device fills the need of both use cases? Jun 29 12:31:57 I have no ideia why procd's tmp on zram was implemented in the first place, but it doesn't make sense, due to the inherent way the kernel's memory management subsystem works. Jun 29 12:32:01 *idea Jun 29 13:22:04 I'm trying to get understand 802.11 better. Are the standards behind a paywall? It seems so just wanted to verify here. And if so are there any alternatives? Jun 29 13:26:04 Neighbor11111115: you can download the latest draft Jun 29 13:26:35 Sorry if I'm being stupid, but where from? It all seems behind a paywall for me Jun 29 13:32:19 Neighbor11111115: query Jun 29 13:34:06 Indeed, it doesn't seem to be accessible now :( Jun 29 13:40:15 https://ieeexplore.ieee.org/browse/standards/get-program/page/series?id=68 seem to tell that one can create an IEEE account for free and download after that. Jun 29 14:03:31 rsalvaterra: sometimes it helps to use sth like err = mount_zram_on_tmp(); Jun 29 14:03:43 if (err) is easy to read then Jun 29 14:10:00 Thanks PaulFertser Jun 29 14:10:11 sorry what do you mean by query Jow? Jun 29 14:10:20 jow * Jun 29 14:10:27 Neighbor11111115: it means I've sent you a private message Jun 29 14:10:36 Thanks Jun 29 14:29:36 sorry I asked yesterday, but d/c so don't know if someone responded. But are 802.11 phys able to measure the noise floor? And if so how do they do that? Jun 29 14:31:39 Neighbor11111115: somehow certain devices (e.g. ath10k) return SNR of the received frames. Also, if you have ath9k you can see what parameters it reports as part of ANI (adaptive noise immunity) debugfs files. Jun 29 15:04:40 rmilecki: indeed, that's why the construct threw me a bit off. But it's fine, I've got it fixed on my Omnia already. :) Jun 29 15:05:26 In any case, does anyone disagree with my reasoning for removing procd's tmp on zram feature? Jun 29 15:05:50 If not, I'll send patched for it. Just after taking a nap. :P Jun 29 15:05:55 rsalvaterra: i don't know, you may try sending question to ML, or just a patch if you have one Jun 29 15:05:59 *patches Jun 29 15:07:12 rmilecki: I don't have any questions, my reasoning is just based on the way the Linux mm subsystem (swap, in particular) works. Jun 29 15:08:38 I do agree, patches speak louder than words. I'll just send them and wait for the fallout. ;) Jun 29 16:26:31 How do I ensure a kernel gets build with CONFIG_SWAP. From kernel_menuconfig it looks activated: [*] Support for paging of anonymous memory (swap), but from ../linux/.config.set it is not set. What causes this inconsistency? Jun 29 17:33:08 jmv2009: that's really strange. Is swap also enabled in the kernel config? Jun 29 17:50:38 Thanks PaulFertser. Do you know they calculate noise? I guess not since you said somehow. Are they able to measure signals inbetween frames to guess the floor noise? Jun 29 20:20:36 Neighbor11111115: I know what should help you if you're cool enough Jun 29 20:21:43 Neighbor11111115: https://github.com/open-sdr/openwifi Jun 29 20:36:32 Neighbor11111115: I would expect wifi card recievers to use some kind of AGC and most probably it's listening (and adjusting) to everything, including noise. Jun 29 21:46:08 hey, i'm looking for a way to detect/check kernel config options at runtime any ideas? I want to check for 'IO_URING' kernel support. Jun 29 22:59:43 if it was compiled in, (it's off by default) you can get a config.gz in /proc somewhere Jun 29 23:02:26 yeah i would need something that works by default. Jun 29 23:03:13 i also just did setup a PR to add it as config option, this way i can at least check at buildtime and exclude/include the module i want to build it for. **** ENDING LOGGING AT Tue Jun 30 02:59:57 2020