[pve-devel] [PATCH kernel] add scheduler fix for ceph on numa hosts

Fabian Grünbichler f.gruenbichler at proxmox.com
Thu Jul 14 16:00:19 CEST 2016


see https://forum.proxmox.com/threads/ceph-kernel-4-4-8-bug.28196/
and https://lkml.org/lkml/2016/3/17/570 for background
---
 Makefile                 |   1 +
 ceph-scheduler-fix.patch | 137 +++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 138 insertions(+)
 create mode 100644 ceph-scheduler-fix.patch

diff --git a/Makefile b/Makefile
index c2f9ae8..a29e062 100644
--- a/Makefile
+++ b/Makefile
@@ -258,6 +258,7 @@ ${KERNEL_SRC}/README ${KERNEL_CFG_ORG}: ${KERNELSRCTAR}
 	cd ${KERNEL_SRC}; patch -p1 < ../CVE-2016-4794-1-percpu-fix-synchronization-between-chunk-map_extend_.patch
 	cd ${KERNEL_SRC}; patch -p1 < ../CVE-2016-4794-2-percpu-fix-synchronization-between-synchronous-map-e.patch
 	cd ${KERNEL_SRC}; patch -p1 < ../CVE-2016-4470-KEYS-potential-uninitialized-variable.patch
+	cd ${KERNEL_SRC}; patch -p1 < ../ceph-scheduler-fix.patch
 	sed -i ${KERNEL_SRC}/Makefile -e 's/^EXTRAVERSION.*$$/EXTRAVERSION=${EXTRAVERSION}/'
 	touch $@
 
diff --git a/ceph-scheduler-fix.patch b/ceph-scheduler-fix.patch
new file mode 100644
index 0000000..2466f82
--- /dev/null
+++ b/ceph-scheduler-fix.patch
@@ -0,0 +1,137 @@
+commit 8974189222159154c55f24ddad33e3613960521a
+Author: Peter Zijlstra <peterz at infradead.org>
+Date:   Thu Jun 16 10:50:40 2016 +0200
+
+    sched/fair: Fix cfs_rq avg tracking underflow
+    
+    As per commit:
+    
+      b7fa30c9cc48 ("sched/fair: Fix post_init_entity_util_avg() serialization")
+    
+    > the code generated from update_cfs_rq_load_avg():
+    >
+    > 	if (atomic_long_read(&cfs_rq->removed_load_avg)) {
+    > 		s64 r = atomic_long_xchg(&cfs_rq->removed_load_avg, 0);
+    > 		sa->load_avg = max_t(long, sa->load_avg - r, 0);
+    > 		sa->load_sum = max_t(s64, sa->load_sum - r * LOAD_AVG_MAX, 0);
+    > 		removed_load = 1;
+    > 	}
+    >
+    > turns into:
+    >
+    > ffffffff81087064:       49 8b 85 98 00 00 00    mov    0x98(%r13),%rax
+    > ffffffff8108706b:       48 85 c0                test   %rax,%rax
+    > ffffffff8108706e:       74 40                   je     ffffffff810870b0 <update_blocked_averages+0xc0>
+    > ffffffff81087070:       4c 89 f8                mov    %r15,%rax
+    > ffffffff81087073:       49 87 85 98 00 00 00    xchg   %rax,0x98(%r13)
+    > ffffffff8108707a:       49 29 45 70             sub    %rax,0x70(%r13)
+    > ffffffff8108707e:       4c 89 f9                mov    %r15,%rcx
+    > ffffffff81087081:       bb 01 00 00 00          mov    $0x1,%ebx
+    > ffffffff81087086:       49 83 7d 70 00          cmpq   $0x0,0x70(%r13)
+    > ffffffff8108708b:       49 0f 49 4d 70          cmovns 0x70(%r13),%rcx
+    >
+    > Which you'll note ends up with sa->load_avg -= r in memory at
+    > ffffffff8108707a.
+    
+    So I _should_ have looked at other unserialized users of ->load_avg,
+    but alas. Luckily nikbor reported a similar /0 from task_h_load() which
+    instantly triggered recollection of this here problem.
+    
+    Aside from the intermediate value hitting memory and causing problems,
+    there's another problem: the underflow detection relies on the signed
+    bit. This reduces the effective width of the variables, IOW its
+    effectively the same as having these variables be of signed type.
+    
+    This patch changes to a different means of unsigned underflow
+    detection to not rely on the signed bit. This allows the variables to
+    use the 'full' unsigned range. And it does so with explicit LOAD -
+    STORE to ensure any intermediate value will never be visible in
+    memory, allowing these unserialized loads.
+    
+    Note: GCC generates crap code for this, might warrant a look later.
+    
+    Note2: I say 'full' above, if we end up at U*_MAX we'll still explode;
+           maybe we should do clamping on add too.
+    
+    Signed-off-by: Peter Zijlstra (Intel) <peterz at infradead.org>
+    Cc: Andrey Ryabinin <aryabinin at virtuozzo.com>
+    Cc: Chris Wilson <chris at chris-wilson.co.uk>
+    Cc: Linus Torvalds <torvalds at linux-foundation.org>
+    Cc: Mike Galbraith <efault at gmx.de>
+    Cc: Peter Zijlstra <peterz at infradead.org>
+    Cc: Thomas Gleixner <tglx at linutronix.de>
+    Cc: Yuyang Du <yuyang.du at intel.com>
+    Cc: bsegall at google.com
+    Cc: kernel at kyup.com
+    Cc: morten.rasmussen at arm.com
+    Cc: pjt at google.com
+    Cc: steve.muckle at linaro.org
+    Fixes: 9d89c257dfb9 ("sched/fair: Rewrite runnable load and utilization average tracking")
+    Link: http://lkml.kernel.org/r/20160617091948.GJ30927@twins.programming.kicks-ass.net
+    Signed-off-by: Ingo Molnar <mingo at kernel.org>
+
+---
+ kernel/sched/fair.c |   33 +++++++++++++++++++++++++--------
+ 1 file changed, 25 insertions(+), 8 deletions(-)
+
+--- a/kernel/sched/fair.c
++++ b/kernel/sched/fair.c
+@@ -2682,6 +2682,23 @@ static inline void update_tg_load_avg(st
+ 
+ static inline u64 cfs_rq_clock_task(struct cfs_rq *cfs_rq);
+ 
++/*
++ * Unsigned subtract and clamp on underflow.
++ *
++ * Explicitly do a load-store to ensure the intermediate value never hits
++ * memory. This allows lockless observations without ever seeing the negative
++ * values.
++ */
++#define sub_positive(_ptr, _val) do {				\
++	typeof(_ptr) ptr = (_ptr);				\
++	typeof(*ptr) val = (_val);				\
++	typeof(*ptr) res, var = READ_ONCE(*ptr);		\
++	res = var - val;					\
++	if (res > var)						\
++		res = 0;					\
++	WRITE_ONCE(*ptr, res);					\
++} while (0)
++
+ /* Group cfs_rq's load_avg is used for task_h_load and update_cfs_share */
+ static inline int update_cfs_rq_load_avg(u64 now, struct cfs_rq *cfs_rq)
+ {
+@@ -2690,15 +2707,15 @@ static inline int update_cfs_rq_load_avg
+ 
+ 	if (atomic_long_read(&cfs_rq->removed_load_avg)) {
+ 		s64 r = atomic_long_xchg(&cfs_rq->removed_load_avg, 0);
+-		sa->load_avg = max_t(long, sa->load_avg - r, 0);
+-		sa->load_sum = max_t(s64, sa->load_sum - r * LOAD_AVG_MAX, 0);
++		sub_positive(&sa->load_avg, r);
++		sub_positive(&sa->load_sum, r * LOAD_AVG_MAX);
+ 		removed = 1;
+ 	}
+ 
+ 	if (atomic_long_read(&cfs_rq->removed_util_avg)) {
+ 		long r = atomic_long_xchg(&cfs_rq->removed_util_avg, 0);
+-		sa->util_avg = max_t(long, sa->util_avg - r, 0);
+-		sa->util_sum = max_t(s32, sa->util_sum - r * LOAD_AVG_MAX, 0);
++		sub_positive(&sa->util_avg, r);
++		sub_positive(&sa->util_sum, r * LOAD_AVG_MAX);
+ 	}
+ 
+ 	decayed = __update_load_avg(now, cpu_of(rq_of(cfs_rq)), sa,
+@@ -2764,10 +2781,10 @@ static void detach_entity_load_avg(struc
+ 			  &se->avg, se->on_rq * scale_load_down(se->load.weight),
+ 			  cfs_rq->curr == se, NULL);
+ 
+-	cfs_rq->avg.load_avg = max_t(long, cfs_rq->avg.load_avg - se->avg.load_avg, 0);
+-	cfs_rq->avg.load_sum = max_t(s64,  cfs_rq->avg.load_sum - se->avg.load_sum, 0);
+-	cfs_rq->avg.util_avg = max_t(long, cfs_rq->avg.util_avg - se->avg.util_avg, 0);
+-	cfs_rq->avg.util_sum = max_t(s32,  cfs_rq->avg.util_sum - se->avg.util_sum, 0);
++	sub_positive(&cfs_rq->avg.load_avg, se->avg.load_avg);
++	sub_positive(&cfs_rq->avg.load_sum, se->avg.load_sum);
++	sub_positive(&cfs_rq->avg.util_avg, se->avg.util_avg);
++	sub_positive(&cfs_rq->avg.util_sum, se->avg.util_sum);
+ }
+ 
+ /* Add the load generated by se into cfs_rq's load average */
-- 
2.1.4




More information about the pve-devel mailing list