Re: [PATCH v3] drm/bridge: display-connector: implement the error path of .probe()

2025-01-30 Thread Joe Hattori
Hi Dan, Thank you for your review. On 1/31/25 14:48, Dan Carpenter wrote: Unrelated to this patch but from a naive reading of the code the conn->hpd_irq = gpiod_to_irq(conn->hpd_gpio); assignment can fail with -EPROBE_DEFER if CONFIG_GPIOLIB_IRQCHIP is enabled. The driver can functio

Re: [PATCH v3] drm/bridge: display-connector: implement the error path of .probe()

2025-01-30 Thread Dan Carpenter
Unrelated to this patch but from a naive reading of the code the conn->hpd_irq = gpiod_to_irq(conn->hpd_gpio); assignment can fail with -EPROBE_DEFER if CONFIG_GPIOLIB_IRQCHIP is enabled. The driver can function without an IRQ so we just ignore the error, but probably it would be better

[PATCH v3] drm/bridge: display-connector: implement the error path of .probe()

2025-01-30 Thread Joe Hattori
Current implementation of .probe() leaks a reference of i2c_adapter. Implement an error path and call put_device() on the obtained i2c_adapter in it to fix this refcount bug. This bug was found by an experimental static analysis tool that I am developing. Fixes: 0c275c30176b ("drm/bridge: Add bri