the latest irqlock patch, against 2.5.27-BK-current can be found at:
http://redhat.com/~mingo/remove-irqlock-patches/remove-irqlock-2.5.27-F8
the big change since -F3 is the introduction of a separate preemption,
hardirq and softirq bitrange within preempt_count:
* PREEMPT_MASK: 0x000000ff
* HARDIRQ_MASK: 0x0000ff00
* SOFTIRQ_MASK: 0x00ff0000
* IRQ_MASK: 0x00ffff00
with this it was possible to have a functional in_irq() again, so the
skbuff.c hack could be reverted.
The definition of the bitrange depends on the following base defines:
#define NR_PREEMPT 256
#define NR_HARDIRQ 256
#define NR_SOFTIRQ 256
NOTE: other, non-x86 architectures might want to define a tighter
bitrange, to make the preempt-only check cheaper. On x86 the above
bitrange is the most optimal.
compiles, boots, works just fine.
Changes in -F8:
- preempt/hardirq/softirq count separation, cleanups.
- skbuff.c fix.
- use irq_count() in scheduler_tick()
Changes in -F3:
- the entry.S cleanups/speedups by Oleg Nesterov.
- a rather critical synchronize_irq() bugfix: if a driver frees an
interrupt that is still being probed then synchronize_irq() locks up.
This bug has caused a spurious boot-lockup on one of my testsystems,
ifconfig would lock up trying to close eth0.
- remove duplicate definitions from asm-i386/system.h, this fixes
compiler warnings.
Ingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Jul 23 2002 - 22:00:42 EST