06:50 <krzk> arnd: thanks, indeed I forgot about this one. It's coming via thermal tree. I'll take care about it, thanks!

09:11 <Xogium> hmm one cannot use usb mass storage gadget with nand flash, right ?

09:11 <ukleinek> [florian]: I don't understand the specification for brcmstb pwm. The quoted part assumes PWM_ON = 0?

09:11 <ukleinek> the specification should have s/</<=/ in the first line?

09:12 <Xogium> I suppose an alternative could be to setup a loop device as the mass storage backend ? Not sure though

09:12 <ukleinek> Xogium: AFAIK you need a block device for mass storage

09:12 <ukleinek> ubiblock might work

09:13 <Xogium> yeah that was my understanding as well

09:13 <Xogium> nup ubiblock is read-only

09:14 <Xogium> only alternative might be usb gadget for ethernet… So the system can download updates and install them like that

09:14 <Xogium> yeah good idea, I'll go bother him :D

12:44 <ajb-linaro> is there a debug option that can get me better kernel backtraces in gdb?

13:45 <arnd> ajb-linaro: on 32-bit arm and on x86, you can pick between multiple unwinders, on 64-bit you only get the frame pointer based unwinding IIRC. Regardless of the unwinder, you probably want CONFIG_KALLSYMS_ALL=y to get symbolic names (this is usually enabled already), and you can set CONFIG_READABLE_ASM=y and CONFIG_LTO_NONE=y to get less aggressive inlining

13:51 <ukleinek> ..ooOO(CONFIG_DEBUG_INFO)

14:00 <arnd> ah right, I misread ajb-linaro's question as being about kgdb, which I assume needs the KALLSYMS option, while of course for qemu+gdb you want DEBUG_INFO

14:05 <apritzel> arnd: do you have any idea where I should start with the multi-v5-defconfig journey? I see that there is no change since 5.14, which means that some options were broken already before that

14:06 <apritzel> arnd: do you know what's a known-good version is I could use as a baseline? v5.4, maybe?

14:10 <arnd> apritzel: in v5.4, I see differences for FB_XGI=y and SOC_CAMERA=y , in v5.7 the savedefconfig output has the same contents as the defconfig file

14:11 <arnd> make O=obj-arm multi_v5_defconfig ARCH=arm savedefconfig ; sort obj-arm/defconfig arch/arm/configs/multi_v5_defconfig | uniq -c | grep -w 1

14:15 <apritzel> arnd: yeah, I created something similar for the last 10 v5.x.0 releases, and see the changes now, but just was curious where to start

14:16 <arnd> the zboot_rom changes and v4l_platform_drivers stuff came in linux-5.8, so no need to bisect those any more

14:16 <ukleinek> arnd: in case you want to sort out entries like CONFIG_SOMETHING=12, too, I suggest to use uniq -u

14:17 <arnd> ah right, that's a bit easier

14:42 <ajb-linaro> arnd: yeah I already had DEBUG_INFO - maybe it's agressive inlining that is flattening the backtrace?

16:45 <abelloni> mripard: do you plan to review https://lore.kernel.org/all/20220211122643.1343315-1-andre.przywara@arm.com/ ? :)

16:46 <robher> geertu: Are you going to respin that shmem warning fix patch?

16:51 <mripard> abelloni: I'm no longer maintainer, so it's very likely that I won't :)

16:59 <arnd> all pull requests taken care of, emails sent out and marked in patchwork as necessary

16:59 <arnd> if anyone sent something for soc@kernel.org and did not get a reply, please let me know

17:05 <arnd> Hmm, I only merged 625 commits for 5.18 so far, it feels like I'm missing something

17:16 <geertu> robher: Sorry, I don't get what you mean

17:17 <robher> geertu: the patch to fix scpi 'shmem' binding warnings

17:18 <robher> setting maxItems...

17:18 <geertu> robher: Oh right. Perhaps.

17:20 <robher> geertu: I've got a few other warning fixes to send to Linus.

17:38 <geertu> robher: OK, working on it...

17:49 <geertu> robher: v2 sent

17:50 <robher> geertu: thanks!