07:24 <sicelo> tmlind: freemangordon: remember the l3 interrupt issue i posted a couple of weeks ago? https://paste.debian.net/1284876/ for ease of reference

07:26 <sicelo> and fmg asked if it could be possible to find address or device that's at fault ... TRM seems to suggest it's possible even on OMAP3, see page 643, 5.2.4.3

07:29 <sicelo> we probably can't tell here because the kernel code calls BUG_ON() without attempting to decode the source of the error, https://github.com/torvalds/linux/blob/master/drivers/bus/omap_l3_smx.c#L171-L177

07:30 <sicelo> i added a printk just above that line 177, printk("omap_l3_smx.c: omap3_l3_app_irq: L3_SI_FLAG_STATUS_0: %llu \n", status);

07:34 <sicelo> got the value 524288, (0x80000) ... using table 5-29, I suppose that's flag bit 19, pointing to sDMA Rd IA

07:36 <sicelo> just thought i'd give you guys an update ... maybe you will have some ideas on how to proceed further

09:24 <freemangordon> sicelo: thanks, I'll read the TRM later on to see if we can extract some more info

09:25 <freemangordon> what the ?!? https://github.com/torvalds/linux/blob/master/drivers/bus/omap_l3_smx.c#L168

09:27 <freemangordon> sicelo: what about moving the bug down?

09:28 <freemangordon> BUG_ON() I mean

09:28 <freemangordon> and see if you can get any more info from https://github.com/torvalds/linux/blob/master/drivers/bus/omap_l3_smx.c#L183 and down

09:34 <sicelo> yes, thanks. that's what i was going to try next :-)

09:35 <sicelo> i was yet to see what the __ffs does (line 184) so i can know if line 186 will contain the correct address

12:59 <sicelo> freemangordon: so with our timeout error, `int_type` will be 0, and err_source will be 19. looks good. what i can't grok yet is how `base` will be evaluated on line 186. it appears to expect omap3_l3_bases to be a multidimensional array, something like omap3_l3_bases[0][19] in our case.

13:00 <sicelo> ah ... sorry i get it now. always good to ask :P

13:01 <sicelo> so `bases` will be calculated correctly ... i'll make the change soon

13:07 <freemangordon> :)

13:34 <sicelo> i'll also try what tmlind suggested at https://lore.kernel.org/lkml/YmkFvmE+2CD3Bjs+@atomide.com/ ... but for now, already compiling such that we can see what's going on with the L3 error itself

14:22 <freemangordon> right