07:32 <lucastanure> Hi, there is an server to download Ubuntu/Debian images for vim3? Or the only way is to build myself using https://docs.khadas.com/vim3/FenixScript.html ?

08:10 <narmstrong> lucastanure: https://docs.khadas.com/firmware/Vim3Firmware.html

08:10 <narmstrong> https://docs.khadas.com/firmware/Vim3Firmware.html#mainline

08:10 <narmstrong> for mainline based image

08:13 <lucastanure> yes, I saw that, whats the way to get the board to aways boot from SD Card, it seems that I ways need to press some buttons to get the board into the SD card

08:33 <narmstrong> lucastanure: https://docs.khadas.com/vim3/HowtoEraseEMMC.html

14:28 <emOne> would it make sense that I am getting input output errors because of bent pins inside a usb port?

14:33 <chewitt> depends whether the bent pins result in breaking or partial contacts or not .. if yes, expect errors

15:40 <brads> jbrunet: have a question if you don't mind about the mmc driver, I applied -rt patches to 5.9-rc3 but had the system stall using the mmc. Had a chat to the preempt_rt developers and they were able to resolve it by changing the IRQ type used here https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/mmc/host/meson-gx-mmc.c?h=v5.9-rc6#n1142 from IRQF_ONESHOT to

15:40 <brads> 0

15:42 <brads> The deadlock was caused as it runs in hard interrupt context & was seeing this lock checking report on startup https://pastebin.com/W7VHy1V8 but with the IRQ type set to 0 no further mmc issue and no deadlock reports

15:43 <brads> Do you know if it would be safe to make this change?

15:44 <brads> I know the mmc driver is a bit of a complex beast

16:11 <narmstrong> brads: never understood what IRQF_ONESHOT is used for... but seems it breaks RT

16:12 <narmstrong> brads: can you send a patch to the linux-amlogic _+linux-mmc mailing-list ?

16:20 <narmstrong> brads: seems bogus to have ONESHOT if we provide both handlers, I find a few "threaded irqs must provide a primary handler or set the IRQF_ONESHOT flag", so yes it should be removed

16:21 <narmstrong> `Specifying this flag is mandatory if the primary handler is set to NULL. The default primary handler does nothing more than to return IRQWAKETHREAD to wake up a kernel thread to execute the thread_fn IRQ handler.`

16:32 <brads> narmstrong: sure will do, thanks for the confirmation