On 8/28/25 6:54 PM, chengkaitao wrote:
diff --git a/block/mq-deadline.c b/block/mq-deadline.c
index b9b7cdf1d3c9..1a031922c447 100644
--- a/block/mq-deadline.c
+++ b/block/mq-deadline.c
@@ -41,19 +41,16 @@ static const int fifo_batch = 16; /* # of sequential requests treated as o
enum dd_data_dir {
DD_READ = READ,
DD_WRITE = WRITE,
+ DD_DIR_COUNT = 2
};
-enum { DD_DIR_COUNT = 2 };
-
This change is not an improvement in my opinion because it makes it
less clear what the role of DD_DIR_COUNT is.
enum dd_prio {
- DD_RT_PRIO = 0,
- DD_BE_PRIO = 1,
- DD_IDLE_PRIO = 2,
- DD_PRIO_MAX = 2,
+ DD_RT_PRIO,
+ DD_BE_PRIO,
+ DD_IDLE_PRIO,
There is code that depends on DD_RT_PRIO < DD_BE_PRIO < DD_IDLE_PRIO so
I'd like to keep the explicit enum values.
+ DD_PRIO_COUNT
};
-enum { DD_PRIO_COUNT = 3 };
I see the above change as a step backwards because it makes the role of
DD_PRIO_COUNT less clear.
- for (prio = DD_BE_PRIO; prio <= DD_PRIO_MAX; prio++) {
+ for (prio = DD_BE_PRIO; prio < DD_PRIO_COUNT; prio++) {
The current code is easier to read IMHO than the new code.
Thanks,
Bart.
