**** BEGIN LOGGING AT Mon Jun 22 02:59:57 2020
Jun 22 03:29:35 <owrt-snap-builds> build #406 of omap/generic is complete: Failure [failed kmodconfig]  Build details are at http://buildbot.openwrt.org/master/images/builders/omap%2Fgeneric/builds/406  blamelist: Lech Perczak <lech.perczak@gmail.com>, Adrian Schmutzler <freifunk@adrianschmutzler.de>, INAGAKI Hiroshi <musashino.open@gmail.com>, Konstantin Demin <rockdrilla@gmail.com>
Jun 22 03:31:04 <owrt-snap-builds> build #374 of bcm53xx/generic is complete: Failure [failed defconfig dltar]  Build details are at http://buildbot.openwrt.org/master/images/builders/bcm53xx%2Fgeneric/builds/374  blamelist: Lech Perczak <lech.perczak@gmail.com>, Adrian Schmutzler <freifunk@adrianschmutzler.de>, INAGAKI Hiroshi <musashino.open@gmail.com>, Konstantin Demin
Jun 22 03:31:04 <owrt-snap-builds> <rockdrilla@gmail.com>
Jun 22 09:36:18 <enyc> **** what has happened to the Chanops, please? -- topic needs updating, I'm happy to help but this is a bit silly for project....
Jun 22 15:18:02 <jow> rmilecki: ping
Jun 22 15:32:09 <dwmw2_gone> oh, my outbound ssh problem with dropbear just got weirder
Jun 22 15:32:20 <dwmw2_gone> precisely the *same* binary running on my wndr3800 works fine.
Jun 22 16:12:02 <Neighbor11111115> Hi all, does anyone know why vm.min_free_kbytes is set so high in OpenWRT sysctl init.d? I've got 128mb of memory and OpenWRT have set it to have a min_free_kbytes of 16mb. That seems a little excessive? Is there a paticular reason for this?
Jun 22 16:12:37 <Neighbor11111115> I assume that this value is for atomic kmallocs, so most likely packets on a router. Is it really going to allocate ~16mb of packets without freeing?
Jun 22 16:13:02 <Neighbor11111115> Thanks!
Jun 22 16:19:55 <jow> Neighbor11111115: driver memory buffers maybe?
Jun 22 16:20:41 <Neighbor11111115> Hrm, ok so if my system is absolutely fine booting and has over 23mb free. It is not likely to suddenly allocate 16mb of buffers is it? I assume skb is only thing that is being added often?
Jun 22 16:24:41 <rsalvaterra> "Hysterical raisins"? I don't know, I edit all that out on my builds, I let the kernel choose its defaults.
Jun 22 16:25:25 <rsalvaterra> Even on a system with 128 MiB of RAM and ath10k Wi-Fi (Archer C6 v2).
Jun 22 16:32:47 <jow> the commit message justification for the introduction of the vm_min_free_kbytes heuristic:
Jun 22 16:32:50 <jow>     Network drivers typically allocate memory in atomic context. For that to
Jun 22 16:32:53 <jow>     be reliable, there needs to be enough free memory. Set the value
Jun 22 16:32:55 <jow>     heuristically based on the total amount of system RAM.
Jun 22 16:33:04 <jow> so at some point there were network drivers that failed to allocate rx/tx buffers under memory pressure
Jun 22 16:33:19 <jow> probably not relevant for the modern arm stuff
Jun 22 16:33:59 <jow> and likely that heuristic is not tuned for the high end but rather for underpowered 16MB mips gear (logic was added in 2016)
Jun 22 16:37:04 <rsalvaterra> Yikes! Has OpenWrt ever supported devices with 16 MiB of RAM?
Jun 22 16:37:22 <karlp> you bet
Jun 22 16:37:34 <karlp> anyone seen odd make failures like this: https://paste.jvnv.net/view/JeMpA
Jun 22 16:37:40 <karlp> I've rm'd tmp, and make clean is ok,
Jun 22 16:37:44 <karlp> was hoping to avoid a dirclean.
Jun 22 16:38:01 <karlp> just updated from an ~april point on the 1907 branch, anod nothing realyl stands out in the git log
Jun 22 16:42:26 <Neighbor11111115> Thanks jow!
Jun 22 16:46:26 <plntyk> did someone change access on mailing-list ? http://lists.infradead.org/pipermail/openwrt-devel/ doesnt work anymore
Jun 22 16:46:36 <plntyk> ie the archives
Jun 22 16:48:02 <jow> rsalvaterra: yes, that used to the norm before the spoiled mvebu crow took over
Jun 22 16:48:10 <jow> *crowd
Jun 22 16:48:12 <jow> :)
Jun 22 16:48:22 <dangole> plntyk: "The machine hosting lists.infradead.org and git.infradead.org suffered
Jun 22 16:48:22 <dangole> a complete disk failure over the weekend; sadly the last backup of the
Jun 22 16:48:22 <dangole> mailman config was fairly old (May 2018) so the subscriber lists and
Jun 22 16:48:22 <dangole> archives have been reset to that date, but the lists should be running
Jun 22 16:48:22 <dangole> again."
Jun 22 16:49:00 <rsalvaterra> jow: It's not my fault I have a 2 GiB Omnia! XD
Jun 22 16:49:05 <rsalvaterra> Wait, actually, it is.
Jun 22 16:49:17 <jow> karlp: I think this happens if CONFIG_TARGET is unset / pointing to something not existing anymore
Jun 22 16:49:36 <jow> rsalvaterra: :P
Jun 22 16:51:29 <rsalvaterra> Speaking of the Omnia, does anyone around here have a first gen Omnia running master using the WAN interface (not SFP)?
Jun 22 16:52:14 <rsalvaterra> I still have that nasty issue with the GPIO IRQ…
Jun 22 16:54:40 <karlp> jow: make menuconfig and exit seems tohave worked? weird though.
Jun 22 16:54:55 <jow> maybe your .config was screwed?
Jun 22 16:55:08 <karlp> I guess, not sure what would have touched it though.
Jun 22 16:55:32 * karlp chalks it up to happenstance
Jun 22 17:05:29 <rsalvaterra> jow: speaking of mvebu, what about my Thumb-2 patches? O:3
Jun 22 17:05:48 <rsalvaterra> (Kidding, I know you guys are crazy busy with other stuff.)
Jun 22 17:36:44 <owrt-snap-builds> build #417 of x86/generic is complete: Failure [failed pkgbuild]  Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2Fgeneric/builds/417  blamelist: Petr ?tetiar <ynezz@true.cz>, Daniel Gonz?lez Cabanelas <dgcbueu@gmail.com>
Jun 22 18:28:41 <owrt-snap-builds> build #411 of x86/geode is complete: Failure [failed pkgbuild]  Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2Fgeode/builds/411  blamelist: Petr ?tetiar <ynezz@true.cz>, Daniel Gonz?lez Cabanelas <dgcbueu@gmail.com>
Jun 22 18:29:53 <rmilecki> jow: pong
Jun 22 18:39:58 <rmilecki> jow: blogic: thanks for your comments on (not) procd patch
Jun 22 18:40:33 <rmilecki> jow: blogic: what do you think about rpcd internal module with "rc" object" providing "list" and "action" methods?
Jun 22 18:41:04 <rmilecki> or maybe replace "action" method name with "call" or "exec" as it also takes "action" argument - to avoid duplicating "action" name?
Jun 22 18:57:17 <owrt-snap-builds> build #377 of x86/legacy is complete: Failure [failed pkgbuild]  Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2Flegacy/builds/377  blamelist: Petr ?tetiar <ynezz@true.cz>, Daniel Gonz?lez Cabanelas <dgcbueu@gmail.com>
Jun 22 19:11:37 <owrt-snap-builds> build #407 of omap/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/omap%2Fgeneric/builds/407
Jun 22 19:26:15 <owrt-snap-builds> build #375 of bcm53xx/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/bcm53xx%2Fgeneric/builds/375
Jun 22 19:48:11 <rsalvaterra> Guys, I asked this a while back (at the forum, I think), but I didn't get a definitive answer: would a patch to remove zram support (mounting /tmp as zram) from procd be acceptable?
Jun 22 19:50:41 <rsalvaterra> The rationale is that it doesn't make a lot of sense. /tmp is mounted as tmpfs, and tmpfs is backed by anonymous pages, which can be swapped out at will. What people actually want is swap on zram, and that's already provided by the zram-swap package.
Jun 22 19:53:41 <rsalvaterra> So, procd zram is both redundant and inefficient, since it actually formats the zram block device as ext4.
Jun 22 19:54:45 <rsalvaterra> And speaking of zram, another thing we really want is to swap on zram *with discard*.
Jun 22 19:57:21 <rsalvaterra> (The semantics is that discarded blocks are returned as free memory to the kernel immediately.)
Jun 22 21:10:21 <ldir> lynxis: ping
Jun 22 21:25:05 <owrt-snap-builds> build #350 of x86/64 is complete: Failure [failed pkgbuild]  Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2F64/builds/350  blamelist: Petr ?tetiar <ynezz@true.cz>, Daniel Gonz?lez Cabanelas <dgcbueu@gmail.com>
Jun 22 22:52:39 <owrt-snap-builds> build #354 of armvirt/64 is complete: Failure [failed pkgbuild]  Build details are at http://buildbot.openwrt.org/master/images/builders/armvirt%2F64/builds/354  blamelist: Petr ?tetiar <ynezz@true.cz>, Daniel Gonz?lez Cabanelas <dgcbueu@gmail.com>
**** ENDING LOGGING AT Tue Jun 23 03:03:14 2020