Luke-Jr has quit [Read error: Connection reset by peer]
ekathva has joined #openwrt-devel
amaumene has joined #openwrt-devel
Luke-Jr has joined #openwrt-devel
test has joined #openwrt-devel
csharper2005 has joined #openwrt-devel
hanetzer1 has joined #openwrt-devel
hanetzer has quit [Ping timeout: 480 seconds]
Lechu has joined #openwrt-devel
Misanthropos has quit [Ping timeout: 480 seconds]
rua has quit [Ping timeout: 480 seconds]
Grommish has joined #openwrt-devel
rua has joined #openwrt-devel
test has quit [Remote host closed the connection]
<csharper2005>
Hi guys! My new device pr on GitHub has no feedback from the reviewer (adschm) since sep 2021. I guess that opening the new one is the worst possible idea. Are there any alternatives from your experience?
GNUmoon2 has quit [Remote host closed the connection]
test has joined #openwrt-devel
Lechu has joined #openwrt-devel
floof58 has quit [Ping timeout: 480 seconds]
floof58 has joined #openwrt-devel
ekathva has quit [Remote host closed the connection]
ekathva has joined #openwrt-devel
<ynezz>
csharper2005: I think, that for that mtd parser review, you can try to ping rmilecki, but I assume, he's going to ask you to upstream it as well
<ynezz>
we don't want to maintain that indefinitely
<ynezz>
other then that it looks ok from the quick glance
<ynezz>
that being said, it's usually enough to start the upstreaming process for the patch, it's not strictly necessary, that the patch lands in the upstream tree in order to have it accepted here
dangole has joined #openwrt-devel
m has joined #openwrt-devel
<csharper2005>
ynezz: thanks for your answer! This point about upstreaming wasn't clear for me before your explanation. How can I contact rmilecki? Is it ok just to mention him here? I haven't yet familiar with the best practices in this community.
kar200 has joined #openwrt-devel
kar200 has quit [Quit: Page closed]
kar200 has joined #openwrt-devel
kar200 has left #openwrt-devel [#openwrt-devel]
kar200 has joined #openwrt-devel
kopijahe has joined #openwrt-devel
kar200 has quit [Remote host closed the connection]
<rsalvaterra>
Grommish: pong
GNUmoon2 has joined #openwrt-devel
<Grommish>
rsalvaterra: I switched to your zstd swap a while back.. What's the best way to test load it to see if it's actually working? I've not seen anything in swap yet
<Grommish>
rsalvaterra: Don't get me wrong, it reports correctly, I've just never seen anything actually put into it and was wondering if you knew a decent way to force it
<rsalvaterra>
Grommish: If you have a swap device with the size you've chosen, it's working. :)
<Grommish>
rsalvaterra: Good enough :)
<rsalvaterra>
As far as testing is concerned, I think I've seen one of my 128 MiB devices swapping a couple of times, but it's rather unusual for my setup.
<rsalvaterra>
My 2 GiB laptop, however, does it all the time and works perfectly (zram swap only, no swap partition/file at all).
<Grommish>
Gotcha.. I'll just have to generate some traffic and see what happens. I'm still cramming suricata into places it doesn't belong..
<Grommish>
Yah, I'm using the /dev/zram0 @ 15%
<rsalvaterra>
You can always dd if=/dev/zero of=/tmp/hog… :)
<Grommish>
That sounds spicy and dangerous.. It's on the list!
<rsalvaterra>
Since tmpfs is backed by anonymous memory, if you have a big enough /tmp, it will surely start swapping.
<Grommish>
BTW, So far I see no reason a medium-powered router couldn't run IDS/IPS/NMS.. I've got 25k signature rules and 1400 IP rules loaded.. Takes about 530Mb RAM in footprint size overall and I lose about 50MBit/sec throughput thru the engine
<rsalvaterra>
Oh, so you did got it working on MIPS64!
<Grommish>
Yah :) In theory it should work on anything I can get it to compile on as long as it can physically run it.. They've got issues with their SC_ATOMIC macros they are working to fix still, but seems to work
<Grommish>
alerts on my EICAR test at least
<rsalvaterra>
So, 530 MiB of RAM on MIPS64… The memory footprint on 32-bit systems will be smaller (and MIPS isn't exactly known for its efficient instruction encoding).
<Grommish>
x86_64 should be able to include hyperscan assuming I bother to try to put it in
danitool has joined #openwrt-devel
<rmilecki>
csharper2005: why choosing a python for creating binary formats, it doesn't fit that task well ;/
<rsalvaterra>
rmilecki: Why choosing Python at all? :P
* rsalvaterra
runs
<Grommish>
Althought the reason I started this is because I figured the ASICs in actual router hardware would be much better at packet handling than a standard x86_64 PC running it
<rsalvaterra>
aparcar[m]: What does the "test formalities" CI failure in the dnsmasq pull request mean?
<rsalvaterra>
Grommish: Are those ASICs supported by the kernel? Or do the work transparently?
<Grommish>
rsalvaterra: Well, at least on my device, my throughput loss is less than when I tried on a standalone box with a lot more RAM and more than a dual-core 1Ghz MIPS64
<Grommish>
and aside from the fact I had to enable the MIPS FP in the kernel, it's the default 5.4 Openwrt kernel
<rsalvaterra>
That's sad… trap-and-emulate is transparent to the applications, yes, but it's atrociously slow… :(
<rsalvaterra>
We need proper soft-float for MIPS64. :P
<Grommish>
rsalvaterra: The tree still supports the ER-Lit, which has no FPU
<Grommish>
So, it's that or nothing :D
<rsalvaterra>
Oh, I'm not suggesting we should only support hardware float, that would basically obsolete 90 % of our targets. :P
<neggles>
Grommish: hehehe, found that one weeks ago ;)
<Grommish>
rsalvaterra: Well, I'm wondering if I can turn it on just for my device :D and see what happens
<Grommish>
Octeon3 has an FPU
<rsalvaterra>
Grommish: Not an IEEE 754-compliant one, I think. :P
<Grommish>
That would be a question for neggles :)
<rsalvaterra>
Wasn't it missing sqrt, for example?
<Grommish>
Because I have absolutely no idea
<neggles>
rsalvaterra: WELL YOU SEE THE THING ABOUT THAT IS
<neggles>
*MAYBE!*
<neggles>
that was not entirely intentional caps but i'll roll with it
<neggles>
not all octeons are created equal, some have FPUs that are fully compliant, some have no FPU at all, and some have FPUs which only have some of the features
<rsalvaterra>
Caps-lock == cruise-control for cool :P
<neggles>
this is a #truefact
<neggles>
but I was going for more of a "hahahahahaAHAHAHAHAhahahahahahahahaa please kill me" kind of vibe :P
<rsalvaterra>
That deep in the rabbit hole, huh? :P
<neggles>
friends don't let friends do Octeon
<neggles>
I've come across octeon 3 dual-core chips that have fully compliant FPUs and also ones which have no FPU at all
rua has quit [Ping timeout: 480 seconds]
<rsalvaterra>
Oh, my… the cynic in me would think they've silicon issues with the FP units and just shipped them stating they weren't fully compliant.
<neggles>
I suspect you're right. That, or the ones which claim to be FPU-less just have it disabled in firmware/with eFuses.
<neggles>
e.g. the Juniper SRX300/SRX320 use an Octeon III CN7020 which, during boot, claims to be FPU-less
<neggles>
but the SDK doco says all octeon IIIs have an FPU...
<Grommish>
See, Iv'e got a CN7020AAP1.2.. I should check to see what it says
<neggles>
I should really ramboot one of these SRXes into openwrt
<csharper2005>
rmilecki: the first version was a shell script. It looked more creepy. :) Then I found and reuse a few python scripts for the recipe those were already merged in OpenWrt. And I decided to use python too... I'm ready to rewrite them in another language if it necessary for merging the pr. What do you think about the mtd parser patch?
<neggles>
but i don't think there's DSA support for this broadcom hurricane chip yet...
<Grommish>
neggles: [ 0.000000] FPU revision is: 00739600
goliath has joined #openwrt-devel
<rsalvaterra>
neggles: Actually, I have a soft spot for exotic achitectures, and Octeon (MIPS64) is one of them. No hardware, unfortunately. :P
<Grommish>
Actually, i did make one more kernel change from default 5.4 - CVMSEG size: 3 cache lines (384 bytes)
<Grommish>
because default is 0
<neggles>
yeah you don't/shouldn't need that with 5.10
<neggles>
since your device can't pciboot
<rsalvaterra>
Another possibility… could the FP units be fused-out?
<neggles>
and rsalvaterra don't get me wrong, I love the stupid weirdos, it'd just be *really* nice if marvell would hand over a functional / vaguely-up-to-date SDK
<neggles>
I own... Many... and for some reason keep acquiring more?
<neggles>
they're the turbo subaru of SoCs
<Grommish>
neggles: PCIe: Link timeout on port 0, probably the slot is empty
<Grommish>
and Port 1/2 Not in PCIe mode
rua has joined #openwrt-devel
<neggles>
Grommish: oh no no, pciboot = booting over pcie, as a device
<neggles>
which is what the SNIC10e/LiquidIO lineup do
<Grommish>
Ah
<Grommish>
5.10 was still memleaking last time I checked, but it's been about a month
<neggles>
rsalvaterra: they may be fused off yes
<neggles>
see, Grommish here's got a CN7020-AAP pass1.2
<neggles>
and the SRX320 I have here? OCTEON CN7020-AAP pass 1.2, Core clock: 1200 MHz, IO clock: 600 MHz, DDR clock: 667 MHz (1334 Mhz DDR)
<neggles>
and yet.... CAVIUM's OCTEON 70XX/71XX CPU Rev. 0.2 with no FPU implemented
<neggles>
maybe it's just that FreeBSD 6 doesn't have octeon FPU support?
<stintel>
I don't understand how something can start breaking in different ways from a patch release
<karlp>
because semver is a cake is a lie... at least for many projects, no matter wht they like to believe
<stintel>
le wtf, worked around the zlib host lib leakage with --without-zlib, now it's complaining about missing zstd dep again, which I previously fixed locally by --without-zstd
<stintel>
and it's not the usual missing \
* stintel
stabs gnutls again and again and again
<stintel>
autohell being autohell obviously not helping
csharper2005 has quit [Read error: Connection reset by peer]
<nbd>
rsalvaterra: i think it should at least be possible to enable the interpreter via build config
csharper2005 has joined #openwrt-devel
<nbd>
sometimes some architectures have JIT limitations where some programs can only be run by the interpreter
<nbd>
always-on might be a good idea to detect those (and fix them to be JIT-able or fix the JIT)
<nbd>
but sometimes falling back to the interpreter might be a reasonable workaround
<stintel>
apparently it doesn't work yet, but like your thoughts on the idea
<stintel>
needs to be STAGING_DIR_HOSTPKG rather than STAGING_DIR_HOST
<aiyion>
lynxis: I saw you closed #7071 a few days ago. Im building gluon on OpenWrt 21.02 and just encountered the same problem as the author.
<aiyion>
As you wrote on github, the 19.07 branch produced fine results. Mine did as well, as tI built the image two years ago. the device knows its revision is 1.
<aiyion>
The image Im building know appears to become revision 0.
<aiyion>
Is that something a dirclean is expected to get rid of?
owrt-2102-builds has quit [Quit: buildmaster reconfigured: bot disconnecting]
owrt-2102-builds has joined #openwrt-devel
owrt-snap-builds has quit [Quit: buildmaster reconfigured: bot disconnecting]
owrt-snap-builds has joined #openwrt-devel
owrt-1907-builds has quit [Quit: buildmaster reconfigured: bot disconnecting]
owrt-1907-builds has joined #openwrt-devel
owrt-snap-builds has quit [Remote host closed the connection]
owrt-1907-builds has quit [Remote host closed the connection]
owrt-2102-builds has quit [Remote host closed the connection]
owrt-2102-builds has joined #openwrt-devel
Luke-Jr has joined #openwrt-devel
ekathva_ has joined #openwrt-devel
ekathva has quit [Remote host closed the connection]
ekathva_ is now known as ekathva
owrt-2102-builds has quit [Quit: buildmaster reconfigured: bot disconnecting]
ekathva_ has joined #openwrt-devel
ekathva has quit [Ping timeout: 480 seconds]
csharper2005 has joined #openwrt-devel
foxtrot has quit [Quit: Goodbye]
foxtrot has joined #openwrt-devel
foxtrot is now known as Guest325
csharper2005 has quit [Read error: Connection reset by peer]
csharper2005 has joined #openwrt-devel
csharper2005 has quit [Read error: Connection reset by peer]
csharper2005 has joined #openwrt-devel
philipp64 has quit [Quit: philipp64]
Misanthropos has joined #openwrt-devel
test has quit [Remote host closed the connection]
csharper2005 has quit [Read error: Connection reset by peer]
mattytap has quit [Remote host closed the connection]
mattytap has joined #openwrt-devel
hanetzer3 has joined #openwrt-devel
<lynxis>
aiyion: strange. dirclean might fix it. not sure how the image/Makefile are processed down to the image. so if there is an intermediate step which uses an artifact from an earlier build. not sure. wouldn't expect it.
<lynxis>
aiyion: before dirclean. grep -R for the tplink hwid in ./tmp/ (and maybe ./build_dir/ ./staging_dir/.)
<stintel>
ok so gnutls 3.7.4 introduced certificate compression (RFC8879), this adds optional zlib and zstd dependencies
<stintel>
explains part of the mess
hanetzer2 has quit [Ping timeout: 480 seconds]
<lynxis>
aiyion: would be good to know if it's 100% reproducible produce such wrong image.
<aiyion>
lynxis: I just reproduced it on commit f65edc9b990c2bcc10c9e9fca29253adc6fe316d
<aiyion>
(after make distclean in order to get around a bugyy stub symlink)
<aiyion>
Im just building master in order to verify the bugs gone; and will bisect afterwords.
<aiyion>
I feel like something just did not get backported to 21.02.
<lynxis>
aiyion: but target/linux/../image/*.mk does contain the right value?
csharper2005 has joined #openwrt-devel
<lynxis>
aiyion: found it.
minimal has joined #openwrt-devel
<lynxis>
aiyion: image/common-tp-link.mk: Device/tplink-safeloader sets hwrev =0x0, Device/tplink_tl-wr1043nd-v4 uses it as template.
<aiyion>
And what changed in master?
<lynxis>
master is also broken.
<lynxis>
neoraider: updated the issues
<aiyion>
nice thanks.
owrt-2102-builds has joined #openwrt-devel
srslypascal is now known as Guest330
srslypascal has joined #openwrt-devel
<lynxis>
will check it later. go out to catch the last sunshine.
<aiyion>
no problem, you helped me a lot already. thanks!
Guest330 has quit [Ping timeout: 480 seconds]
mattytap_ has joined #openwrt-devel
mattytap has quit [Ping timeout: 480 seconds]
philipp64 has joined #openwrt-devel
ekathva__ has joined #openwrt-devel
gnustomp[m] has quit [Server closed connection]
gnustomp[m] has joined #openwrt-devel
goliath has quit [Quit: SIGSEGV]
ekathva_ has quit [Ping timeout: 480 seconds]
csharper2005 has quit [Ping timeout: 480 seconds]
csharper2005 has joined #openwrt-devel
csharper2005 has quit [Read error: Connection reset by peer]