On Thu, 20 May 2004 11:50:22 -0400 (EDT) Brendan Burns wrote:
| Hey folks,
| sorry about the last empty message... I'm still figuring out this whole
| "e-mail" thing ;-P
|
| Attached is a patch that replaces all instances of yield() in the drivers
| section of the kernel. I think my choices of schedule vs
| schedule_timeout make sense. I'd love someone to check them... Plus
| since I don't have all the appropriate hardware (read macintosh adb) it
| would be great if they were tested by those who do!
The 2-character indents are not in line with Linux coding style.
Can you explain why some changes use
schedule_timeout(1);
some use
schedule_timeout(10);
some use
schedule_timeout(5*HZ);
some just use
schedule();
They could all be correct, I'm just curious about the
differences.
+ long time=jiffies;
+ while (time-jiffies < HZ) {
+ set_current_state(TASK_INTERRUPTIBLE);
+ schedule_timeout((time+HZ)-jiffies);
use more spaces, they improve readability. e.g.:
long time = jiffies;
while (time - jiffies < HZ) {
set_current_state(TASK_INTERRUPTIBLE);
schedule_timeout((time + HZ) - jiffies);
For some of these, a new function that was just added might be
the appropriate kernel API to use. Look at msleep() [merged
2004-may-18] to see if it can be used. It's in include/linux/delay.h
in a kernel bk pull or bk patch and will be in 2.6.7.
--
~Randy
_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@xxxxxxxxxxxxxx
http://lists.osdl.org/mailman/listinfo/kernel-janitors
|
