Skip to contentSkip to conversation

2.5.44-mm4

DISCUSSION Andrew Morton 1 participant 1 message Oct 24, 2002 Memory management

lore
Share & tools
lore
#0 lore raw

url: http://www.zip.com.au/~akpm/linux/patches/2.5/2.5.44/2.5.44-mm4/

Having a few stability problems so most of the new things have been removed. Once this thing is working properly we can start moving forward again.

If the people who had problem with -mm4 could please retest? If problems remain, please try popping off shpte-ng.patch.

Also be suspicious of CONFIG_PREEMPT=y. For me, with preempt, smp and spinlock debugging the kernel dies immediately doing an unlock of already-unlocked kernel_flag when bringing up the first migration thread. That's on base 2.5.44. May not be a preempt problem; could be that preempt is simply exposing it.

+read-barrier-depends.patch

RCU fix

+deferred-lru-add-fix.patch

lru_cache_add fix

-for-each-cpu.patch

Dropped. Rusty has a different cpu iterator patch

-task-unmapped-base-fix.patch

Folded into ingo-mmap-speedup

-larger-cpu-masks.patch
-adam-loop.patch
-rcu-stats.patch
-generic-nonlinear-mappings-D0.patch

Over in experimental/

+md-01-driverfs-core.patch
+md-02-driverfs-topology.patch
+md-03-numa-meminfo.patch
+md-04-memblk_online_map.patch
+md-05-node_online_map.patch

The NUMA driverfs interfaces from Matt Dobson. Queued up in experimental/ too. It all adds only a few hundred bytes to a non-NUMA build.