10:21 <CounterPillow> C macro basilisk of the day: https://elixir.bootlin.com/linux/v6.13.7/source/include/linux/compiler.h#L247

10:24 <CounterPillow> ("basilisk" in the sense that if you look directly at it, it will kill you :D)

10:54 <mmind00> 1-line macro with 40-lines comment :-)

11:14 <CounterPillow> I first found a copy of it in the tools subdirectory that had no explanation, imagine how I felt

11:51 <qschulz> TIL https://www.kernel.org/doc/html/v6.14-rc7/core-api/printk-formats.html#error-pointers

12:18 <CounterPillow> Yeah I highly recommend using those for any errno, it's way nicer than just throwing an int at people, even if it means doing a ERR_PTR(ret)

12:18 <robmur01> qschulz: yup, very handy - only downside is it looks a bit weird when you use it unconditionally so success is "(NULL)" :)

12:20 <CounterPillow> I've been thinking about writing a cocci thing to find instances of if (ret)/if (ret < 0) { dev_err with ret passed to %d } to fix up every user-facing error that relies on users knowing the Linux errno headers by heart

12:21 <CounterPillow> (I did write a short weggli query for it but weggli isn't powerful enough to get both if (ret) and if (ret < 0) in just one query)

13:27 <qschulz> TIL https://elixir.bootlin.com/linux/v6.13.7/source/include/linux/string_choices.h

14:02 <CounterPillow> qschulz: neat

15:16 <montjoie> hello my rk2588-rock-5b reboot in middle of init() with recent kernel, does someone tried a recent kernel on it ?

15:24 <CounterPillow> if that's with a recent linux-next then that's because of the iommu breakage

15:25 <CounterPillow> but that should be a panic

15:25 <montjoie> no I try ony stable

15:25 <CounterPillow> might be u-boot not initialising fusb302 then, but kernel initialising it and then cutting off its own power

15:26 <montjoie> it is a dumb power supply (not PD)

15:26 <montjoie> 6.6.83 just booted fine

15:27 <montjoie> lets go for a bisect

15:27 <CounterPillow> try disabling the fusb nodes anyway, easier than bisect ;)

16:05 <montjoie> at least the error is repeatable

19:33 <montjoie> automatic bisect to the rescue

21:28 <sre> CounterPillow: fusb302 has not yet been added in the upstream kernel DT for Rock 5B.

21:28 <CounterPillow> Oh, good to know, sorry for the red herring

21:32 <montjoie> arm64_defconfig is so long to compile, omg

21:34 <CounterPillow> throw out all the nvidia and mellanox stuff and it goes somewhat faster :D

21:57 <montjoie> probably will do a rockchip_defconfig

21:57 <montjoie> did it for meson64/sunxi64