2011-04-14 21:23:20

by Dan Magenheimer

[permalink] [raw]
Subject: [PATCH] xen: cleancache shim to Xen Transcendent Memory

[PATCH] xen: cleancache shim to Xen Transcendent Memory

This patch provides a shim between the kernel-internal cleancache
API (see Documentation/mm/cleancache.txt) and the Xen Transcendent
Memory ABI (see http://oss.oracle.com/projects/tmem).

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]>
Reviewed-by: Jeremy Fitzhardinge <[email protected]>
Cc: Konrad Rzeszutek Wilk <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Nick Piggin <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Rik Van Riel <[email protected]>
Cc: Jan Beulich <[email protected]>
Cc: Chris Mason <[email protected]>
Cc: Andreas Dilger <[email protected]>
Cc: Ted Ts'o <[email protected]>
Cc: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Cc: Nitin Gupta <[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(+)

diff -Napur -X linux-2.6.39-rc3/Documentation/dontdiff linux-2.6.39-rc3-cleancache/arch/x86/include/asm/xen/hypercall.h linux-2.6.39-rc3-cleancache-xen/arch/x86/include/asm/xen/hypercall.h
--- linux-2.6.39-rc3-cleancache/arch/x86/include/asm/xen/hypercall.h 2011-04-11 18:21:51.000000000 -0600
+++ linux-2.6.39-rc3-cleancache-xen/arch/x86/include/asm/xen/hypercall.h 2011-04-14 09:54:19.210857901 -0600
@@ -447,6 +447,13 @@ 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)
{
diff -Napur -X linux-2.6.39-rc3/Documentation/dontdiff linux-2.6.39-rc3-cleancache/drivers/xen/Makefile linux-2.6.39-rc3-cleancache-xen/drivers/xen/Makefile
--- linux-2.6.39-rc3-cleancache/drivers/xen/Makefile 2011-04-11 18:21:51.000000000 -0600
+++ linux-2.6.39-rc3-cleancache-xen/drivers/xen/Makefile 2011-04-14 09:54:19.236096051 -0600
@@ -1,5 +1,6 @@
obj-y += grant-table.o features.o events.o manage.o balloon.o
obj-y += xenbus/
+obj-y += tmem.o

nostackp := $(call cc-option, -fno-stack-protector)
CFLAGS_features.o := $(nostackp)
diff -Napur -X linux-2.6.39-rc3/Documentation/dontdiff linux-2.6.39-rc3-cleancache/drivers/xen/tmem.c linux-2.6.39-rc3-cleancache-xen/drivers/xen/tmem.c
--- linux-2.6.39-rc3-cleancache/drivers/xen/tmem.c 1969-12-31 17:00:00.000000000 -0700
+++ linux-2.6.39-rc3-cleancache-xen/drivers/xen/tmem.c 2011-04-14 09:54:19.236917913 -0600
@@ -0,0 +1,264 @@
+/*
+ * Xen implementation for transcendent memory (tmem)
+ *
+ * Copyright (C) 2009-2010 Oracle Corp. All rights reserved.
+ * Author: Dan Magenheimer
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/pagemap.h>
+#include <linux/cleancache.h>
+
+#include <xen/xen.h>
+#include <xen/interface/xen.h>
+#include <asm/xen/hypercall.h>
+#include <asm/xen/page.h>
+#include <asm/xen/hypervisor.h>
+
+#define TMEM_CONTROL 0
+#define TMEM_NEW_POOL 1
+#define TMEM_DESTROY_POOL 2
+#define TMEM_NEW_PAGE 3
+#define TMEM_PUT_PAGE 4
+#define TMEM_GET_PAGE 5
+#define TMEM_FLUSH_PAGE 6
+#define TMEM_FLUSH_OBJECT 7
+#define TMEM_READ 8
+#define TMEM_WRITE 9
+#define TMEM_XCHG 10
+
+/* Bits for HYPERVISOR_tmem_op(TMEM_NEW_POOL) */
+#define TMEM_POOL_PERSIST 1
+#define TMEM_POOL_SHARED 2
+#define TMEM_POOL_PAGESIZE_SHIFT 4
+#define TMEM_VERSION_SHIFT 24
+
+
+struct tmem_pool_uuid {
+ u64 uuid_lo;
+ u64 uuid_hi;
+};
+
+struct tmem_oid {
+ u64 oid[3];
+};
+
+#define TMEM_POOL_PRIVATE_UUID { 0, 0 }
+
+/* flags for tmem_ops.new_pool */
+#define TMEM_POOL_PERSIST 1
+#define TMEM_POOL_SHARED 2
+
+/* xen tmem foundation ops/hypercalls */
+
+static inline int xen_tmem_op(u32 tmem_cmd, u32 tmem_pool, struct tmem_oid oid,
+ u32 index, unsigned long gmfn, u32 tmem_offset, u32 pfn_offset, u32 len)
+{
+ struct tmem_op op;
+ int rc = 0;
+
+ op.cmd = tmem_cmd;
+ op.pool_id = tmem_pool;
+ op.u.gen.oid[0] = oid.oid[0];
+ op.u.gen.oid[1] = oid.oid[1];
+ op.u.gen.oid[2] = oid.oid[2];
+ op.u.gen.index = index;
+ op.u.gen.tmem_offset = tmem_offset;
+ op.u.gen.pfn_offset = pfn_offset;
+ op.u.gen.len = len;
+ set_xen_guest_handle(op.u.gen.gmfn, (void *)gmfn);
+ rc = HYPERVISOR_tmem_op(&op);
+ return rc;
+}
+
+static int xen_tmem_new_pool(struct tmem_pool_uuid uuid,
+ u32 flags, unsigned long pagesize)
+{
+ struct tmem_op op;
+ int rc = 0, pageshift;
+
+ for (pageshift = 0; pagesize != 1; pageshift++)
+ pagesize >>= 1;
+ flags |= (pageshift - 12) << TMEM_POOL_PAGESIZE_SHIFT;
+ flags |= TMEM_SPEC_VERSION << TMEM_VERSION_SHIFT;
+ op.cmd = TMEM_NEW_POOL;
+ op.u.new.uuid[0] = uuid.uuid_lo;
+ op.u.new.uuid[1] = uuid.uuid_hi;
+ op.u.new.flags = flags;
+ rc = HYPERVISOR_tmem_op(&op);
+ return rc;
+}
+
+/* xen generic tmem ops */
+
+static int xen_tmem_put_page(u32 pool_id, struct tmem_oid oid,
+ u32 index, unsigned long pfn)
+{
+ unsigned long gmfn = xen_pv_domain() ? pfn_to_mfn(pfn) : pfn;
+
+ return xen_tmem_op(TMEM_PUT_PAGE, pool_id, oid, index,
+ gmfn, 0, 0, 0);
+}
+
+static int xen_tmem_get_page(u32 pool_id, struct tmem_oid oid,
+ u32 index, unsigned long pfn)
+{
+ unsigned long gmfn = xen_pv_domain() ? pfn_to_mfn(pfn) : pfn;
+
+ return xen_tmem_op(TMEM_GET_PAGE, pool_id, oid, index,
+ gmfn, 0, 0, 0);
+}
+
+static int xen_tmem_flush_page(u32 pool_id, struct tmem_oid oid, u32 index)
+{
+ return xen_tmem_op(TMEM_FLUSH_PAGE, pool_id, oid, index,
+ 0, 0, 0, 0);
+}
+
+static int xen_tmem_flush_object(u32 pool_id, struct tmem_oid oid)
+{
+ return xen_tmem_op(TMEM_FLUSH_OBJECT, pool_id, oid, 0, 0, 0, 0, 0);
+}
+
+static int xen_tmem_destroy_pool(u32 pool_id)
+{
+ struct tmem_oid oid = { { 0 } };
+
+ return xen_tmem_op(TMEM_DESTROY_POOL, pool_id, oid, 0, 0, 0, 0, 0);
+}
+
+int tmem_enabled;
+
+static int __init enable_tmem(char *s)
+{
+ tmem_enabled = 1;
+ return 1;
+}
+
+__setup("tmem", enable_tmem);
+
+/* cleancache ops */
+
+static void tmem_cleancache_put_page(int pool, struct cleancache_filekey key,
+ pgoff_t index, struct page *page)
+{
+ u32 ind = (u32) index;
+ struct tmem_oid oid = *(struct tmem_oid *)&key;
+ unsigned long pfn = page_to_pfn(page);
+
+ if (pool < 0)
+ return;
+ if (ind != index)
+ return;
+ mb(); /* ensure page is quiescent; tmem may address it with an alias */
+ (void)xen_tmem_put_page((u32)pool, oid, ind, pfn);
+}
+
+static int tmem_cleancache_get_page(int pool, struct cleancache_filekey key,
+ pgoff_t index, struct page *page)
+{
+ u32 ind = (u32) index;
+ struct tmem_oid oid = *(struct tmem_oid *)&key;
+ unsigned long pfn = page_to_pfn(page);
+ int ret;
+
+ /* translate return values to linux semantics */
+ if (pool < 0)
+ return -1;
+ if (ind != index)
+ return -1;
+ ret = xen_tmem_get_page((u32)pool, oid, ind, pfn);
+ if (ret == 1)
+ return 0;
+ else
+ return -1;
+}
+
+static void tmem_cleancache_flush_page(int pool, struct cleancache_filekey key,
+ pgoff_t index)
+{
+ u32 ind = (u32) index;
+ struct tmem_oid oid = *(struct tmem_oid *)&key;
+
+ if (pool < 0)
+ return;
+ if (ind != index)
+ return;
+ (void)xen_tmem_flush_page((u32)pool, oid, ind);
+}
+
+static void tmem_cleancache_flush_inode(int pool, struct cleancache_filekey key)
+{
+ struct tmem_oid oid = *(struct tmem_oid *)&key;
+
+ if (pool < 0)
+ return;
+ (void)xen_tmem_flush_object((u32)pool, oid);
+}
+
+static void tmem_cleancache_flush_fs(int pool)
+{
+ if (pool < 0)
+ return;
+ (void)xen_tmem_destroy_pool((u32)pool);
+}
+
+static int tmem_cleancache_init_fs(size_t pagesize)
+{
+ struct tmem_pool_uuid uuid_private = TMEM_POOL_PRIVATE_UUID;
+
+ return xen_tmem_new_pool(uuid_private, 0, pagesize);
+}
+
+static int tmem_cleancache_init_shared_fs(char *uuid, size_t pagesize)
+{
+ struct tmem_pool_uuid shared_uuid;
+
+ shared_uuid.uuid_lo = *(u64 *)uuid;
+ shared_uuid.uuid_hi = *(u64 *)(&uuid[8]);
+ return xen_tmem_new_pool(shared_uuid, TMEM_POOL_SHARED, pagesize);
+}
+
+static int use_cleancache = 1;
+
+static int __init no_cleancache(char *s)
+{
+ use_cleancache = 0;
+ return 1;
+}
+
+__setup("nocleancache", no_cleancache);
+
+static struct cleancache_ops tmem_cleancache_ops = {
+ .put_page = tmem_cleancache_put_page,
+ .get_page = tmem_cleancache_get_page,
+ .flush_page = tmem_cleancache_flush_page,
+ .flush_inode = tmem_cleancache_flush_inode,
+ .flush_fs = tmem_cleancache_flush_fs,
+ .init_shared_fs = tmem_cleancache_init_shared_fs,
+ .init_fs = tmem_cleancache_init_fs
+};
+
+static int __init xen_tmem_init(void)
+{
+ struct cleancache_ops old_ops;
+
+ if (!xen_domain())
+ return 0;
+#ifdef CONFIG_CLEANCACHE
+ BUG_ON(sizeof(struct cleancache_filekey) != sizeof(struct tmem_oid));
+ if (tmem_enabled && use_cleancache) {
+ char *s = "";
+ old_ops = cleancache_register_ops(&tmem_cleancache_ops);
+ if (old_ops.init_fs != NULL)
+ s = " (WARNING: cleancache_ops overridden)";
+ printk(KERN_INFO "cleancache enabled, RAM provided by "
+ "Xen Transcendent Memory%s\n", s);
+ }
+#endif
+ return 0;
+}
+
+module_init(xen_tmem_init)
diff -Napur -X linux-2.6.39-rc3/Documentation/dontdiff linux-2.6.39-rc3-cleancache/include/xen/interface/xen.h linux-2.6.39-rc3-cleancache-xen/include/xen/interface/xen.h
--- linux-2.6.39-rc3-cleancache/include/xen/interface/xen.h 2011-04-11 18:21:51.000000000 -0600
+++ linux-2.6.39-rc3-cleancache-xen/include/xen/interface/xen.h 2011-04-14 09:54:19.237875256 -0600
@@ -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. */


2011-04-15 22:02:15

by Ian Campbell

[permalink] [raw]
Subject: Re: [PATCH] xen: cleancache shim to Xen Transcendent Memory

On Thu, 2011-04-14 at 14:20 -0700, Dan Magenheimer wrote:
> [PATCH] xen: cleancache shim to Xen Transcendent Memory
>
> This patch provides a shim between the kernel-internal cleancache
> API (see Documentation/mm/cleancache.txt) and the Xen Transcendent
> Memory ABI (see http://oss.oracle.com/projects/tmem).

There's no need to build this into a kernel which doesn't have
cleancache (or one of the other frontends), is there? I think there
should be a Kconfig option (even if its not a user visible one) with the
appropriate depends/selects.

Ian.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

2011-04-15 22:53:47

by Dan Magenheimer

[permalink] [raw]
Subject: RE: [PATCH] xen: cleancache shim to Xen Transcendent Memory

> From: Ian Campbell [mailto:[email protected]]
> Sent: Friday, April 15, 2011 4:02 PM
> Subject: Re: [PATCH] xen: cleancache shim to Xen Transcendent Memory
>
> On Thu, 2011-04-14 at 14:20 -0700, Dan Magenheimer wrote:
> > [PATCH] xen: cleancache shim to Xen Transcendent Memory
> >
> > This patch provides a shim between the kernel-internal cleancache
> > API (see Documentation/mm/cleancache.txt) and the Xen Transcendent
> > Memory ABI (see http://oss.oracle.com/projects/tmem).
>
> There's no need to build this into a kernel which doesn't have
> cleancache (or one of the other frontends), is there? I think there
> should be a Kconfig option (even if its not a user visible one) with
> the appropriate depends/selects.

Yes, you're right. It should eventually depend on

CONFIG_CLEANCACHE || CONFIG_FRONTSWAP

though there's no sense merging this xen cleancache
shim at all unless/until Linus merges cleancache
(and hopefully later some evolution of frontswap).

And once cleancache (and/or frontswap) is merged,
there's very little reason NOT to enable one or
both on a Xen guest kernel.

Thanks,
Dan

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href
ilto:"[email protected]"> [email protected] </a>

2011-04-18 08:47:21

by Ian Campbell

[permalink] [raw]
Subject: RE: [PATCH] xen: cleancache shim to Xen Transcendent Memory

On Fri, 2011-04-15 at 23:53 +0100, Dan Magenheimer wrote:
> > From: Ian Campbell [mailto:[email protected]]
> > Sent: Friday, April 15, 2011 4:02 PM
> > Subject: Re: [PATCH] xen: cleancache shim to Xen Transcendent Memory
> >
> > On Thu, 2011-04-14 at 14:20 -0700, Dan Magenheimer wrote:
> > > [PATCH] xen: cleancache shim to Xen Transcendent Memory
> > >
> > > This patch provides a shim between the kernel-internal cleancache
> > > API (see Documentation/mm/cleancache.txt) and the Xen Transcendent
> > > Memory ABI (see http://oss.oracle.com/projects/tmem).
> >
> > There's no need to build this into a kernel which doesn't have
> > cleancache (or one of the other frontends), is there? I think there
> > should be a Kconfig option (even if its not a user visible one) with
> > the appropriate depends/selects.
>
> Yes, you're right. It should eventually depend on
>
> CONFIG_CLEANCACHE || CONFIG_FRONTSWAP
>
> though there's no sense merging this xen cleancache
> shim at all unless/until Linus merges cleancache
> (and hopefully later some evolution of frontswap).

Cleancache isn't in already? I thought I saw references to it in
drivers/staging?

> And once cleancache (and/or frontswap) is merged,
> there's very little reason NOT to enable one or
> both on a Xen guest kernel.

There are software knobs to allow the host- and guest-admin to opt in or
out as they desire though, right?

Ian.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

2011-04-18 14:12:46

by Dan Magenheimer

[permalink] [raw]
Subject: RE: [PATCH] xen: cleancache shim to Xen Transcendent Memory

> From: Ian Campbell [mailto:[email protected]]
> > >
> > > On Thu, 2011-04-14 at 14:20 -0700, Dan Magenheimer wrote:
> > >
> > > There's no need to build this into a kernel which doesn't have
> > > cleancache (or one of the other frontends), is there? I think there
> > > should be a Kconfig option (even if its not a user visible one)
> with
> > > the appropriate depends/selects.
> >
> > Yes, you're right. It should eventually depend on
> >
> > CONFIG_CLEANCACHE || CONFIG_FRONTSWAP
> >
> > though there's no sense merging this xen cleancache
> > shim at all unless/until Linus merges cleancache
> > (and hopefully later some evolution of frontswap).
>
> Cleancache isn't in already? I thought I saw references to it in
> drivers/staging?

Linus said he would review it after 2.6.39-rc1 was released,
but has neither given thumbs up nor thumbs down so I'm
assuming he didn't have time and it will be reconsidered
for 2.6.40. This latest patchset (V8) is updated in linux-next.

Yes, zcache is in driver/staging and has references to it.
I guess that proves the chicken comes before the egg...
or was it vice versa? :-)

> > And once cleancache (and/or frontswap) is merged,
> > there's very little reason NOT to enable one or
> > both on a Xen guest kernel.
>
> There are software knobs to allow the host- and guest-admin to opt in
> or out as they desire though, right?

Definitely. Both Xen and a Linux guest have runtime
options, which currently default to off.

Dan

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href
ilto:"[email protected]"> [email protected] </a>

2011-04-18 16:03:47

by Ian Campbell

[permalink] [raw]
Subject: RE: [PATCH] xen: cleancache shim to Xen Transcendent Memory

On Mon, 2011-04-18 at 07:12 -0700, Dan Magenheimer wrote:
> > From: Ian Campbell [mailto:[email protected]]
> > > >
> > > > On Thu, 2011-04-14 at 14:20 -0700, Dan Magenheimer wrote:
> > > >
> > > > There's no need to build this into a kernel which doesn't have
> > > > cleancache (or one of the other frontends), is there? I think there
> > > > should be a Kconfig option (even if its not a user visible one)
> > with
> > > > the appropriate depends/selects.
> > >
> > > Yes, you're right. It should eventually depend on
> > >
> > > CONFIG_CLEANCACHE || CONFIG_FRONTSWAP
> > >
> > > though there's no sense merging this xen cleancache
> > > shim at all unless/until Linus merges cleancache
> > > (and hopefully later some evolution of frontswap).
> >
> > Cleancache isn't in already? I thought I saw references to it in
> > drivers/staging?
>
> Linus said he would review it after 2.6.39-rc1 was released,
> but has neither given thumbs up nor thumbs down so I'm
> assuming he didn't have time and it will be reconsidered
> for 2.6.40. This latest patchset (V8) is updated in linux-next.
>
> Yes, zcache is in driver/staging and has references to it.
> I guess that proves the chicken comes before the egg...
> or was it vice versa? :-)

Ah, I didn't realise that the relaxations associated with staging
allowed for unsatisfiable (due to the other half being out of tree)
Kconfig items as well.

> > > And once cleancache (and/or frontswap) is merged,
> > > there's very little reason NOT to enable one or
> > > both on a Xen guest kernel.
> >
> > There are software knobs to allow the host- and guest-admin to opt in
> > or out as they desire though, right?
>
> Definitely. Both Xen and a Linux guest have runtime
> options, which currently default to off.

Good to know, thanks!

Ian.