**** BEGIN LOGGING AT Thu Aug 22 02:59:56 2019
**** BEGIN LOGGING AT Thu Aug 22 06:45:36 2019
Aug 22 09:10:33 <julieng> embden_new[m]: The implementation of crossbar_route_irq_to_xen() and crossbar_route_irq_to_guest looks fairly suspicious.
Aug 22 09:12:04 <julieng> I don't understand why you check against > 191.
Aug 22 09:14:04 <embden_new[m]> Ok ,I will check it
Aug 22 09:22:54 <embden_new[m]> julieng:  could you explain in few words the overall picture of usage route_irq_to_guest, route_irq_to_xen and map_irq_to_domain?
Aug 22 09:34:08 <julieng> embden_new[m]: Leaving aside the crossbar for a moment, route_irq_to_guest will configure a given hardware interrupt to be routed to the guest.
Aug 22 09:34:23 <julieng> route_irq_to_xen will configure a given hardware interrupt to be routed to Xen.
Aug 22 09:34:58 <julieng> map_irq_to_domain() is just a wrapper for routing an IRQ to the guest and giving permission to the guest to manage the interrupt.
Aug 22 09:35:59 <julieng> As we discussed in the past, the crossbar is here to overcome the GIC limitations (for this platform).
Aug 22 09:37:27 <julieng> So the crossbar will have more interrupts than the GIC. This means that you can't configure the crossbar when translating for all the interrupts.
Aug 22 09:37:36 <julieng> This can only be done for Xen interrupts.
Aug 22 09:38:46 <julieng> For interrupts routed to the guest, we need to wait on the guest to use the crossbar.
Aug 22 09:40:26 <julieng> From my understanding, the use case between x15 and Tegra X1/K1 is slightly different.
Aug 22 09:42:44 <julieng> There secondary interrupt controller is using the same Interrupt ID as the GIC. So they don't have an interrupt controller with more interrupts than the other.
Aug 22 09:51:20 <embden_new[m]> Ok, I understand
**** BEGIN LOGGING AT Thu Aug 22 19:11:52 2019
**** BEGIN LOGGING AT Thu Aug 22 19:36:37 2019
Aug 22 20:35:17 <embden_new[m]> julieng:  I call a function of irq translation from handle_device
Aug 22 20:35:39 <embden_new[m]> how can I allow data writes into the crossbar registers?
Aug 22 20:43:22 <embden_new[m]> oh, everything allright, I found my mistake
Aug 22 20:55:14 <embden_new[m]> julieng: sstabellini  drhunter  I have a peace of ok news - I make my code with a better distinction between platform and general code work. So, tomorrow, I will prepare documentation and start cleaning up the improved version. It is still messy but it is a platform-aware kind of messy: https://github.com/embeddedden/xen/commit/7936452bd86f06581bffbe60d0a669e21a19805f
**** ENDING LOGGING AT Fri Aug 23 03:00:40 2019