platform hook for timer_interrupt()
Quoted 5 lines · Martin Hicks
SGI has a need to run platform specific code during the timer interrupt. Attached are patches for both 2.4 and 2.6 that add a hook inside the timer interrupt that will run platform specific code if required.
Any opinions on this?
Why not use add_timer()?
Signature · 3 lines
"It's not Hollywood. War is real, war is primarily not about defeat or victory, it is about death. I've seen thousands and thousands of dead bodies. Do you think I want to have an academic debate on this subject?" -- Robert Fisk
Quoted 7 lines · Martin Hicks
Hello,
SGI has a need to run platform specific code during the timer interrupt. Attached are patches for both 2.4 and 2.6 that add a hook inside the timer interrupt that will run platform specific code if required.
Any opinions on this?
Can't say that I'm fond of increasing the number of runtime platform
hooks. Is there a need for this to be changed runtime? If not, you could make it a real machvec so it will get compiled out for non-generic, non-sn2 kernel targets. A platform hook in time_init() and an sn2 wrapper around timer_interrupt() might do the job too.
Alex
Signature · 1 line
Alex Williamson HP Linux & Open Source Lab
Quoted 8 lines · Matthew Wilcox
Quoted 5 lines · Martin Hicks
SGI has a need to run platform specific code during the timer interrupt. Attached are patches for both 2.4 and 2.6 that add a hook inside the timer interrupt that will run platform specific code if required.
Any opinions on this?
Why not use
add_timer()?
This is code that must be run on each CPU, so the generic timer interface doesn't help us. I'm not certain, but some of this code may actually only have to run on each node (not every CPU).
mh
Signature · 1 line
Wild Open Source Inc. [email protected]
Quoted 5 lines · Martin Hicks
Quoted 1 line
Why not use
add_timer()?This is code that must be run on each CPU, so the generic timer interface doesn't help us. I'm not certain, but some of this code may actually only have to run on each node (not every CPU).
How about smp_call_function() in a regular timer then?
Signature · 3 lines
"It's not Hollywood. War is real, war is primarily not about defeat or victory, it is about death. I've seen thousands and thousands of dead bodies. Do you think I want to have an academic debate on this subject?" -- Robert Fisk
Quoted 1 line · Matthew Wilcox
How about
smp_call_function()in a regular timer then?
smp_call_function() is kind of messy since it uses IPI to call on each CPU. Not sure that's a good solution since it won't complete until all the CPUs have executed the given function.
If kernel threads or tasklets don't cut it, I could understand why using a timers machvec would be "lighter weight".
grant