Real-Time Preemption, -RT-2.6.12-rc6-V0.7.48-00
Quoted 3 lines
does -48-13 work any better?
Ingo
x86-64 needs the attached patch to compile. Some drivers also fail to compile because of SPIN_LOCK_UNLOCKED changes, e.g. softdog and four files under net/atm. Just take the x86-64 UP config from FC3's latest errata kernel.
I haven't booted it yet.
A question, though: why do you comment out the DMPS timer in drivers/char/vt.c?
Does it cause any problems? I applied the -RT tree after the multiconsole patch and I had to modify those four chunks manually.
Best regards,
Zoltán Böszörményi
x86_64/kernel/init_task.c renamed from x86_64/kernel/init_task.c.old +1 -1
@@ -11,7 +11,7 @@#include <asm/desc.h> static struct fs_struct init_fs =INIT_FS ; -static struct files_struct init_files =INIT_FILES ; +static struct files_struct init_files =INIT_FILES (init_files);static struct signal_struct init_signals =INIT_SIGNALS (init_signals);static struct sighand_struct init_sighand =INIT_SIGHAND (init_sighand);struct mm_struct init_mm =INIT_MM (init_mm);
Quoted 5 lines · Zoltan Boszormenyi
Quoted 3 lines
does -48-13 work any better?
Ingo
x86-64 needs the attached patch to compile.
(patch MIA)
Quoted 3 lines
Some drivers also fail to compile because of
SPIN_LOCK_UNLOCKEDchanges, e.g. softdog and
four files under net/atm. [...]
i'll fix those up.
Quoted 4 lines
A question, though: why do you comment out the DMPS timer in
drivers/char/vt.c? Does it cause any problems? I applied the -RT tree after the multiconsole patch and I had to modify those four chunks manually.
this is a quick hack. The timer code is fully preemptible, but for debugging one wants to have a working printk from hardirq contexts too. I.e. the mod_timer()/del_timer() would print warnings on PREEMPT_RT. We'll need some other approach to solve this cleanly.
Ingo