Skip to contentSkip to conversation

Cleancache: shim to Xen Transcendent Memory

PATCH Superseded Dan Magenheimer 1 participant 1 message Sep 3, 2010

Subsystems (5)
Memory management x86 OCFS2 Xen hypervisor interface +1 more subsystems
lore
Share & tools
lore

Dan Magenheimer

author
#0 lore raw

PATCH v5 1/1 Cleancache: shim to Xen Transcendent Memory

[PATCH V5] Cleancache: shim to Xen Transcendent Memory

Changes from V0 to V5:
- uses new in-kernel cleancache API to match cleancache V5 patch
- uses new Xen tmem ABI (v1) just pushed upstream to Xen (skipped V1-V4 to match cleancache version numbers)

This companion patch to the cleancache V5 patchset (see
http://lkml.org/lkml/2010/9/3/383) provides one consumer for the proposed cleancache internal kernel API. This user is Xen Transcendent Memory ("tmem"), supported in Xen 4.0+. (A second user, zmem, was separately posted by Nitin Gupta and is not yet updated to cleancache V5.)

This patch affects Xen directories only; while broader review is welcome, the patch is provided primarily to answer concerns expressed about the apparent non-existence of consumers for the proposed cleancache patch.

Xen tmem provides "hypervisor RAM" as an ephemeral page-oriented pseudo-RAM store for cleancache pages, shared cleancache pages, and frontswap pages. Tmem provides enterprise-quality concurrency, full save/restore and live migration support, compression and deduplication.

A presentation showing up to 8% faster performance and up to 52% reduction in sectors read on a kernel compile workload, despite aggressive in-kernel page reclamation ("self-ballooning") can be found at:

http://oss.oracle.com/projects/tmem/dist/documentation/presentations/TranscendentMemoryXenSummit2010.pdf

Signed-off-by Dan Magenheimer <[email protected]>

Diffstat:

 arch/x86/include/asm/xen/hypercall.h     |    7 
 drivers/xen/Makefile                     |    1 
 drivers/xen/tmem.c                       |  264 +++++++++++++++++++++
 include/xen/interface/xen.h              |   22 +
 4 files changed, 294 insertions(+)
arch/x86/include/asm/xen/hypercall.h +7 -0
diff -Napur linux-2.6.36-rc3/arch/x86/include/asm/xen/hypercall.h linux-2.6.36-rc3-cleancache_tmem_only/arch/x86/include/asm/xen/hypercall.h
@@ -423,6 +423,13 @@ HYPERVISOR_hvm_op(int op, void *arg)
        return _hypercall2(unsigned long, hvm_op, op, arg);
 }
 
+static inline int
+HYPERVISOR_tmem_op(
+	struct tmem_op *op)
+{
+	return _hypercall1(int, tmem_op, op);
+}
+
 static inline void
 MULTI_fpu_taskswitch(struct multicall_entry *mcl, int set)
 {
drivers/xen/Makefile +1 -0
diff -Napur linux-2.6.36-rc3/drivers/xen/Makefile linux-2.6.36-rc3-cleancache_tmem_only/drivers/xen/Makefile
@@ -1,5 +1,6 @@
 obj-y	+= grant-table.o features.o events.o manage.o
 obj-y	+= xenbus/
+obj-y	+= tmem.o
 
 nostackp := $(call cc-option, -fno-stack-protector)
 CFLAGS_features.o			:= $(nostackp)
drivers/xen/tmem.c +264 -0

264 lines not on this page · raw message

include/xen/interface/xen.h +22 -0
diff -Napur linux-2.6.36-rc3/include/xen/interface/xen.h linux-2.6.36-rc3-cleancache_tmem_only/include/xen/interface/xen.h
@@ -58,6 +58,7 @@
 #define __HYPERVISOR_event_channel_op     32
 #define __HYPERVISOR_physdev_op           33
 #define __HYPERVISOR_hvm_op               34
+#define __HYPERVISOR_tmem_op              38
 
 /* Architecture-specific hypercall definitions. */
 #define __HYPERVISOR_arch_0               48

@@ -461,6 +462,27 @@ typedef uint8_t xen_domain_handle_t[16];
 #define __mk_unsigned_long(x) x ## UL
 #define mk_unsigned_long(x) __mk_unsigned_long(x)
 
+#define TMEM_SPEC_VERSION 1
+
+struct tmem_op {
+	uint32_t cmd;
+	int32_t pool_id;
+	union {
+		struct {  /* for cmd == TMEM_NEW_POOL */
+			uint64_t uuid[2];
+			uint32_t flags;
+		} new;
+		struct {
+			uint64_t oid[3];
+			uint32_t index;
+			uint32_t tmem_offset;
+			uint32_t pfn_offset;
+			uint32_t len;
+			GUEST_HANDLE(void) gmfn; /* guest machine page frame */
+		} gen;
+	} u;
+};
+
 #else /* __ASSEMBLY__ */
 
 /* In assembly code we cannot use C numeric constant suffixes. */
fs/ocfs2/super.c +3 -0
@@ -42,6 +42,7 @@
 #include <linux/seq_file.h>
 #include <linux/quotaops.h>
 #include <linux/smp_lock.h>
+#include <linux/cleancache.h>
 
 #define MLOG_MASK_PREFIX ML_SUPER
 #include <cluster/masklog.h>

@@ -2284,6 +2285,8 @@ static int ocfs2_initialize_super(struct
 		mlog_errno(status);
 		goto bail;
 	}
+	sb->cleancache_poolid =
+		cleancache_init_shared_fs((char *)&uuid_net_key, PAGE_SIZE);
 
 bail:
 	mlog_exit(status);