<Christophe[m]1>
huh, urngd fails to build with glibc? /staging_dir/toolchain-aarch64_cortex-a53_gcc-12.3.0_glibc/include/features.h:413:4: error: #warning _FORTIFY_SOURCE requires compiling with optimization (-O) [-Werror=cpp] https://pastebin.mozilla.org/5zp32OjH/raw
xavifr has quit [Remote host closed the connection]
<Christophe[m]1>
And an image with urngd excluded fails with a "Fatal glibc error: cannot get entropy for arc4random" kernel panic. https://pastebin.mozilla.org/s2pNWinh/raw
xavifr has joined #openwrt-devel
<KanjiMonster>
Christophe[m]1: at first I was confused because there is a -Os at the begining, but later on there is a -O0 which probably overrides the -Os and causes the warning upgraded to error
<Christophe[m]1>
KanjiMonster: Oh, I didn't see that either. I was wondering if it doesn't like -Os
<KanjiMonster>
-Os should be fine, at least in my local tests. So the next question would be where the -O0 comes from
<Christophe[m]1>
I see. Can switch off fortification via PKG_BUILD_FLAGS?
<KanjiMonster>
according to include/hardening.mk, PKG_FORTIFY_SOURCE
<KanjiMonster>
there are a few packages in feeds that set this
<KanjiMonster>
(to 0)
<KanjiMonster>
Christophe[m]1: looking at the musl includes, looks like it also has that requirement, just doesn't warn about it. So it might make sense to just change urngd to also pass -D_FORTIFY_SOURCE=0 in https://git.openwrt.org/?p=project/urngd.git;a=blob;f=CMakeLists.txt;h=7a3be27c4189d95bf6ab7c176e07549adbd66abb;hb=319df8a19f16586f1cce445059490494e897dedf#l25
<Christophe[m]1>
yup, now it compiles
<KanjiMonster>
or rather -U_FORTIFY_SOURCE, to avoid a redefined warning
<Christophe[m]1>
No, but pretty likely the issue I ran into.
<KanjiMonster>
ynezz: https://git.openwrt.org/?p=project/urngd.git;a=commit;h=c7f7b6b65b82eda4675b42d8cd28d76ea7aebf1a - I think you forgot to push the tag
mcbridematt has quit [Quit: Leaving]
mcbridematt has joined #openwrt-devel
mcbridematt has quit [Remote host closed the connection]
mcbridematt has joined #openwrt-devel
<Christophe[m]1>
I found out that "Fatal glibc error: cannot get entropy for arc4random" doesn't appear when I enter failsafe mode. It pauses "Waiting for kernel randomness to be initialised..." for some seconds "random: crng init done"
cmonroe has quit [Ping timeout: 480 seconds]
clandmeter has joined #openwrt-devel
robimarko has joined #openwrt-devel
<ynezz>
KanjiMonster: oops, fixed, thanks
<KanjiMonster>
the v1.0.1 tag is also weird, doesn't point to a commit in the ma* branch but one outside of any branches
<KanjiMonster>
ynezz: the commit you tagged isn't part of the master branch
<KanjiMonster>
not sure what's happening there
<ynezz>
well, who knows what I was doing 3 years ago :P
<ynezz>
but you're right, its leading nowhere, I don't have those commits in my local repository anymore as well
<ynezz>
probably tagging on the wrong/development branch and removed that?
<ynezz>
anyways, that "tagging" is mostly for the versioning in CMake to set human URNGD_VERSION
<ynezz>
should be fixed now
<Christophe[m]1>
On successful boots I'm seeing "random: crng init done" right after"mtk_rng 1020f000.rng: registered RNG driver". on unsuccessful boots it's not even showing up after "urandom-seed: Seeding with /etc/urandom.seed" a couple seconds later.
<KanjiMonster>
ynezz: oooh, even signed ones. nice!
<Christophe[m]1>
unfortunately my usb-ttl cable gave up so I'm doing the same
<Christophe[m]1>
I found "random: procd: uninitialized urandom read (4 bytes read)" in logs of a musl build.
philipp64 has quit [Ping timeout: 480 seconds]
dangole has joined #openwrt-devel
shibboleth has joined #openwrt-devel
schwicht has joined #openwrt-devel
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
dgcampea is now known as Guest7133
dgcampea has joined #openwrt-devel
Guest7133 has quit [Remote host closed the connection]
schwicht has joined #openwrt-devel
aiyion has joined #openwrt-devel
minimal has joined #openwrt-devel
xavifr has joined #openwrt-devel
hurricos has joined #openwrt-devel
<f00b4r0>
just in case: suggestions as to the tiniest serial client that can be installed on openwrt (to access some other device serial console)?
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
schwicht has joined #openwrt-devel
<f00b4r0>
seems to be the aptly named "serialconsole" :)
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<BWhitten>
hey all, looking at a PR at the moment for the rockchip and the bump in uboot requires use of binman, which has dependencies on pyelftools and swig. The build workers have swig but not pyelftools, can I run a standalone to test easily if I PR adding that package?
<BWhitten>
ahhhh we'll see what the workflow reckons when its run is approved, being a one liner :)