On Friday, May 19, 2023 9:13 PM, Andrew Lunn wrote: > > I have one MSI-X interrupt for all general MAC interrupt (see TXGBE_PX_MISC_IEN_MASK). > > It has 32 bits to indicate various interrupts, GPIOs are the one of them. When GPIO > > interrupt is determined, GPIO_INT_STATUS register should be read to determine > > which GPIO line has changed state. > > So you have another interrupt controller above the GPIO interrupt > controller. regmap-gpio is pushing you towards describing this > interrupt controller as a Linux interrupt controller. > > When you look at drivers handling interrupts, most leaf interrupt > controllers are not described as Linux interrupt controllers. The > driver interrupt handler reads the interrupt status register and > internally dispatches to the needed handler. This works well when > everything is internal to one driver. > > However, here, you have two drivers involved, your MAC driver and a > GPIO driver instantiated by the MAC driver. So i think you are going > to need to described the MAC interrupt controller as a Linux interrupt > controller. > > Take a look at the mv88e6xxx driver, which does this. It has two > interrupt controller embedded within it, and they are chained. Now I add two interrupt controllers, the first one for the MAC interrupt, and the second one for regmap-gpio. In the second adding flow, irq = irq_find_mapping(txgbe->misc.domain, TXGBE_PX_MISC_GPIO_OFFSET); err = regmap_add_irq_chip_fwnode(fwnode, regmap, irq, 0, 0, chip, &chip_data); and then, config.irq_domain = regmap_irq_get_domain(chip_data); gpio_regmap = gpio_regmap_register(&config); "txgbe->misc.domain" is the MAC interrupt domain. I think this flow should be correct, but still failed to get gpio_irq from gpio_desc with err -517. And I still have doubts about what I said earlier: https://lore.kernel.org/netdev/20230515063200.301026-1-jiawenwu@xxxxxxxxxxxxxx/T/#me1be68e1a1e44426ecc0dd8edf0f6b224e50630d There really is nothing wrong with gpiochip_to_irq()?? > > > If you are getting errors when removing the driver it means you are > > > missing some level of undoing what us done in probe. Are you sure > > > regmap_del_irq_chip() is being called on unload? > > > > I used devm_* all when I registered them. > > Look at the ordering. Is regmap_del_irq_chip() being called too late? > I've had problems like this with the mv88e6xxx driver and its > interrupt controllers. I ended up not using devm_ so i had full > control over the order things got undone. In that case, the external > devices was PHYs, with the PHY interrupt being inside the Ethernet > switch, which i exposed using a Linux interrupt controller. I use no devm_ functions to add regmap irq chip, register gpio regmap, and call their del/unregister functions at the position corresponding to release. irq_domain_remove() call trace still exist. [ 104.553182] Call Trace: [ 104.553184] <TASK> [ 104.553185] irq_domain_remove+0x2b/0xe0 [ 104.553190] regmap_del_irq_chip.part.0+0x8a/0x160 [ 104.553196] txgbe_remove_phy+0x57/0x80 [txgbe] [ 104.553201] txgbe_remove+0x2a/0x90 [txgbe] [ 104.553205] pci_device_remove+0x36/0xa0 [ 104.553208] device_release_driver_internal+0xaa/0x140 [ 104.553213] driver_detach+0x44/0x90 [ 104.553215] bus_remove_driver+0x69/0xf0 [ 104.553217] pci_unregister_driver+0x29/0xb0 [ 104.553220] __x64_sys_delete_module+0x145/0x240 [ 104.553223] ? exit_to_user_mode_prepare+0x3c/0x1a0 [ 104.553226] do_syscall_64+0x3b/0x90 [ 104.553230] entry_SYSCALL_64_after_hwframe+0x72/0xdc
