<jow>
another quick test case is running while true; do /www/cgi-bin/luci | wc -l; done
<jow>
it should print something like 76 and eventually segfault with 0 lines
<jow>
also taking between 10 and 1000 attempts
<jow>
the wdr3600 seems to share it's dts and former machine definition code with the wdr4300 and wdr4310
<jow>
would be interesting to know if those are affected too
<schmars[m]>
and we have plenty other old ath79 devices that we can test on, if the repro case doesn't affect device stability
<schmars[m]>
ok nice, yeah
<schmars[m]>
no wdr4310 but plenty of wdr4300
<jow>
schmars[m]: well I don't think it affects stability but it will cause system load for sure
<Habbie>
any non-wdr it shares with? because from the photos, the 3600 looks familiar (but then again, tplinks often do)
<jow>
don't think so
* russell--
has a bunch of wdr3600's also
<russell-->
jow: should i worry about three of these: mips-openwrt-linux-gcc: warning: environment variable 'STAGING_DIR' not defined
<slh64>
wdr3600 and wdr4300/ wdr4310 are identical (same PCB), the only difference is the soldered-on 5 GHz PCIe wlan card (2x2 vs 3x3) and the region code for the 4310, so anything affecting the 3600 should affect 4300 & 4310 alike
<russell-->
200+ loops and counting ...
<jow>
russell--: no, that's fine for the purpose of cmpiling the test util
<jow>
russell--: also try the while true; do /www/cgi-bin/luci | wc -l; done test
<f00b4r0>
jow: I didn't realize the affected device was the WDR3600
<f00b4r0>
that device comes in two "variants" which are undistinguishable unless you open them: one has Winbond RAM the other Zentel
<f00b4r0>
the latter required a workaround for proper reboot to take effect, so the bug you're seeing might be RAM-type dependent (I wouldn't be surprised if the Zentel RAM actually needs more quirks than just the reboot one)
<schmars[m]>
here's a wdr3600 v1.1 with 2x512mbit hynix ram
<KanjiMonster>
ah, didn't expect your irc nick being different from your github handle
<Habbie>
some are known by many names
<Mangix>
Looks like I found out again that sophos_red-15w-rev1 is broken
<Mangix>
convenient that nobody uses this device.
<KanjiMonster>
Mangix: https://git.openwrt.org/?p=openwrt/svn-archive/packages.git;a=commitdiff;h=df888339a65fec311cb518bcd002d3a2a617cd30 seems to be the commit that added the symlinks
<russell-->
jow: still going
<KanjiMonster>
Mangix: seeing how 14 years ago there already was a proper .pc, I wonder if these deprecated *-config scripts are even used by anything anymore nowadays