Resolve inconsistent cell name for IRQ type in device trees #79356
Description
Introduction
Let's consolidate over a single name to describe the IRQ type in interrupt controller bindings.
Problem description
There are currently 7 interrupt controller bindings using "sense" as cell name to describe the IRQ type and 6 using "flags". Drivers that shall be used with different controllers may be forced to look after both cell names - or fail to do that, see e.g.
zephyr/drivers/serial/uart_ns16550.c
Line 1791 in c710f88
This is a source of subtle problems.
Proposed change
I have no strong preference about a unified name. We can either pick one of the two or maybe even rename them all to type so that using IRQ_TYPE_* constants for it feels more natural. I'm happy to write a PR for any of such renamings.
Future bindings should then be forced to use the new name for the flags/sense/type cell.
Dependencies
None.
Concerns and Unresolved Questions
See above: decision about the name is needed.
Alternatives
Applying lots of changes like this:
diff --git a/drivers/serial/uart_ns16550.c b/drivers/serial/uart_ns16550.c
index a46dc34aad7..8e1b39e436e 100644
--- a/drivers/serial/uart_ns16550.c
+++ b/drivers/serial/uart_ns16550.c
@@ -1787,9 +1787,12 @@ static const struct uart_driver_api uart_ns16550_driver_api = {
};
#define UART_NS16550_IRQ_FLAGS(n) \
- COND_CODE_1(DT_INST_IRQ_HAS_CELL(n, sense), \
- (DT_INST_IRQ(n, sense)), \
- (0))
+ (COND_CODE_1(DT_INST_IRQ_HAS_CELL(n, flags), \
+ (DT_INST_IRQ(n, flags)), \
+ (0)) | \
+ COND_CODE_1(DT_INST_IRQ_HAS_CELL(n, sense), \
+ (DT_INST_IRQ(n, sense)), \
+ (0)))
/* IO-port or MMIO based UART */
#define UART_NS16550_IRQ_CONFIG(n) \
One could also provide some common macro for that, of course, but all that would look strange.