summaryrefslogtreecommitdiffstats
path: root/libglusterfs/src
diff options
context:
space:
mode:
authorM. Mohan Kumar <mohan@in.ibm.com>2013-11-09 14:51:53 +0530
committerVijay Bellur <vbellur@redhat.com>2013-11-10 21:25:49 -0800
commitc8fef37c5d566c906728b5f6f27baaa9a8d2a20d (patch)
tree03c833446bc73bfa3da6c621315b590c0d65c748 /libglusterfs/src
parentd5335f9e40f6e9533f7812d153b9727bcc04aa4e (diff)
glusterfs: zerofill support
Add support for a new ZEROFILL fop. Zerofill writes zeroes to a file in the specified range. This fop will be useful when a whole file needs to be initialized with zero (could be useful for zero filled VM disk image provisioning or during scrubbing of VM disk images). Client/application can issue this FOP for zeroing out. Gluster server will zero out required range of bytes ie server offloaded zeroing. In the absence of this fop, client/application has to repetitively issue write (zero) fop to the server, which is very inefficient method because of the overheads involved in RPC calls and acknowledgements. WRITESAME is a SCSI T10 command that takes a block of data as input and writes the same data to other blocks and this write is handled completely within the storage and hence is known as offload . Linux ,now has support for SCSI WRITESAME command which is exposed to the user in the form of BLKZEROOUT ioctl. BD Xlator can exploit BLKZEROOUT ioctl to implement this fop. Thus zeroing out operations can be completely offloaded to the storage device , making it highly efficient. The fop takes two arguments offset and size. It zeroes out 'size' number of bytes in an opened file starting from 'offset' position. This patch adds zerofill support to the following areas: - libglusterfs - io-stats - performance/md-cache,open-behind - quota - cluster/afr,dht,stripe - rpc/xdr - protocol/client,server - io-threads - marker - storage/posix - libgfapi Client applications can exloit this fop by using glfs_zerofill introduced in libgfapi.FUSE support to this fop has not been added as there is no system call for this fop. Changes from previous version 3: * Removed redundant memory failure log messages Changes from previous version 2: * Rebased and fixed build error Changes from previous version 1: * Rebased for latest master TODO : * Add zerofill support to trace xlator * Expose zerofill capability as part of gluster volume info Here is a performance comparison of server offloaded zeofill vs zeroing out using repeated writes. [root@llmvm02 remote]# time ./offloaded aakash-test log 20 real 3m34.155s user 0m0.018s sys 0m0.040s [root@llmvm02 remote]# time ./manually aakash-test log 20 real 4m23.043s user 0m2.197s sys 0m14.457s [root@llmvm02 remote]# time ./offloaded aakash-test log 25; real 4m28.363s user 0m0.021s sys 0m0.025s [root@llmvm02 remote]# time ./manually aakash-test log 25 real 5m34.278s user 0m2.957s sys 0m18.808s The argument log is a file which we want to set for logging purpose and the third argument is size in GB . As we can see there is a performance improvement of around 20% with this fop. Change-Id: I081159f5f7edde0ddb78169fb4c21c776ec91a18 BUG: 1028673 Signed-off-by: Aakash Lal Das <aakash@linux.vnet.ibm.com> Signed-off-by: M. Mohan Kumar <mohan@in.ibm.com> Reviewed-on: http://review.gluster.org/5327 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Vijay Bellur <vbellur@redhat.com>
Diffstat (limited to 'libglusterfs/src')
-rw-r--r--libglusterfs/src/call-stub.c67
-rw-r--r--libglusterfs/src/call-stub.h16
-rw-r--r--libglusterfs/src/defaults.c33
-rw-r--r--libglusterfs/src/defaults.h18
-rw-r--r--libglusterfs/src/globals.c1
-rw-r--r--libglusterfs/src/glusterfs.h1
-rw-r--r--libglusterfs/src/syncop.c29
-rw-r--r--libglusterfs/src/syncop.h2
-rw-r--r--libglusterfs/src/xlator.c1
-rw-r--r--libglusterfs/src/xlator.h16
10 files changed, 184 insertions, 0 deletions
diff --git a/libglusterfs/src/call-stub.c b/libglusterfs/src/call-stub.c
index 2f07a0074c5..ac79cf0711c 100644
--- a/libglusterfs/src/call-stub.c
+++ b/libglusterfs/src/call-stub.c
@@ -2241,6 +2241,62 @@ out:
}
+call_stub_t *
+fop_zerofill_cbk_stub(call_frame_t *frame, fop_zerofill_cbk_t fn,
+ int32_t op_ret, int32_t op_errno,
+ struct iatt *statpre, struct iatt *statpost,
+ dict_t *xdata)
+{
+ call_stub_t *stub = NULL;
+
+ GF_VALIDATE_OR_GOTO ("call-stub", frame, out);
+
+ stub = stub_new (frame, 0, GF_FOP_ZEROFILL);
+ GF_VALIDATE_OR_GOTO ("call-stub", stub, out);
+
+ stub->fn_cbk.zerofill = fn;
+
+ stub->args_cbk.op_ret = op_ret;
+ stub->args_cbk.op_errno = op_errno;
+
+ if (statpre)
+ stub->args_cbk.prestat = *statpre;
+ if (statpost)
+ stub->args_cbk.poststat = *statpost;
+ if (xdata)
+ stub->args_cbk.xdata = dict_ref (xdata);
+out:
+ return stub;
+}
+
+call_stub_t *
+fop_zerofill_stub(call_frame_t *frame, fop_zerofill_t fn, fd_t *fd,
+ off_t offset, size_t len, dict_t *xdata)
+{
+ call_stub_t *stub = NULL;
+
+ GF_VALIDATE_OR_GOTO ("call-stub", frame, out);
+ GF_VALIDATE_OR_GOTO ("call-stub", fn, out);
+
+ stub = stub_new (frame, 1, GF_FOP_ZEROFILL);
+ GF_VALIDATE_OR_GOTO ("call-stub", stub, out);
+
+ stub->fn.zerofill = fn;
+
+ if (fd)
+ stub->args.fd = fd_ref (fd);
+
+ stub->args.offset = offset;
+ stub->args.size = len;
+
+ if (xdata)
+ stub->args.xdata = dict_ref (xdata);
+out:
+ return stub;
+
+}
+
+
static void
call_resume_wind (call_stub_t *stub)
{
@@ -2468,6 +2524,12 @@ call_resume_wind (call_stub_t *stub)
stub->args.fd, stub->args.offset,
stub->args.size, stub->args.xdata);
break;
+ case GF_FOP_ZEROFILL:
+ stub->fn.zerofill(stub->frame, stub->frame->this,
+ stub->args.fd, stub->args.offset,
+ stub->args.size, stub->args.xdata);
+ break;
+
default:
gf_log_callingfn ("call-stub", GF_LOG_ERROR,
"Invalid value of FOP (%d)",
@@ -2670,6 +2732,11 @@ call_resume_unwind (call_stub_t *stub)
STUB_UNWIND(stub, discard, &stub->args_cbk.prestat,
&stub->args_cbk.poststat, stub->args_cbk.xdata);
break;
+ case GF_FOP_ZEROFILL:
+ STUB_UNWIND(stub, zerofill, &stub->args_cbk.prestat,
+ &stub->args_cbk.poststat, stub->args_cbk.xdata);
+ break;
+
default:
gf_log_callingfn ("call-stub", GF_LOG_ERROR,
"Invalid value of FOP (%d)",
diff --git a/libglusterfs/src/call-stub.h b/libglusterfs/src/call-stub.h
index f0872b1219a..45bef80443f 100644
--- a/libglusterfs/src/call-stub.h
+++ b/libglusterfs/src/call-stub.h
@@ -71,6 +71,7 @@ typedef struct {
fop_fsetattr_t fsetattr;
fop_fallocate_t fallocate;
fop_discard_t discard;
+ fop_zerofill_t zerofill;
} fn;
union {
@@ -117,6 +118,7 @@ typedef struct {
fop_fsetattr_cbk_t fsetattr;
fop_fallocate_cbk_t fallocate;
fop_discard_cbk_t discard;
+ fop_zerofill_cbk_t zerofill;
} fn_cbk;
struct {
@@ -745,6 +747,20 @@ fop_discard_cbk_stub(call_frame_t *frame,
struct iatt *statpre, struct iatt *statpost,
dict_t *xdata);
+call_stub_t *
+fop_zerofill_stub(call_frame_t *frame,
+ fop_zerofill_t fn,
+ fd_t *fd,
+ off_t offset,
+ size_t len, dict_t *xdata);
+
+call_stub_t *
+fop_zerofill_cbk_stub(call_frame_t *frame,
+ fop_zerofill_cbk_t fn,
+ int32_t op_ret, int32_t op_errno,
+ struct iatt *statpre, struct iatt *statpost,
+ dict_t *xdata);
+
void call_resume (call_stub_t *stub);
void call_stub_destroy (call_stub_t *stub);
void call_unwind_error (call_stub_t *stub, int op_ret, int op_errno);
diff --git a/libglusterfs/src/defaults.c b/libglusterfs/src/defaults.c
index a3c8d97f112..2ebb251504c 100644
--- a/libglusterfs/src/defaults.c
+++ b/libglusterfs/src/defaults.c
@@ -473,6 +473,17 @@ default_discard_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
}
int32_t
+default_zerofill_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+ int32_t op_ret, int32_t op_errno, struct iatt *pre,
+ struct iatt *post, dict_t *xdata)
+{
+ STACK_UNWIND_STRICT(zerofill, frame, op_ret, op_errno, pre,
+ post, xdata);
+ return 0;
+}
+
+
+int32_t
default_getspec_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
int32_t op_ret, int32_t op_errno, char *spec_data)
{
@@ -900,6 +911,17 @@ default_discard_resume(call_frame_t *frame, xlator_t *this, fd_t *fd,
return 0;
}
+int32_t
+default_zerofill_resume(call_frame_t *frame, xlator_t *this, fd_t *fd,
+ off_t offset, size_t len, dict_t *xdata)
+{
+ STACK_WIND(frame, default_zerofill_cbk, FIRST_CHILD(this),
+ FIRST_CHILD(this)->fops->zerofill, fd, offset, len,
+ xdata);
+ return 0;
+}
+
+
/* FOPS */
int32_t
@@ -1325,6 +1347,17 @@ default_discard(call_frame_t *frame, xlator_t *this, fd_t *fd,
}
int32_t
+default_zerofill(call_frame_t *frame, xlator_t *this, fd_t *fd,
+ off_t offset, size_t len, dict_t *xdata)
+{
+ STACK_WIND_TAIL(frame, FIRST_CHILD(this),
+ FIRST_CHILD(this)->fops->zerofill, fd, offset, len,
+ xdata);
+ return 0;
+}
+
+
+int32_t
default_forget (xlator_t *this, inode_t *inode)
{
gf_log_callingfn (this->name, GF_LOG_WARNING, "xlator does not "
diff --git a/libglusterfs/src/defaults.h b/libglusterfs/src/defaults.h
index f3cbb3a4bb7..0747027bc35 100644
--- a/libglusterfs/src/defaults.h
+++ b/libglusterfs/src/defaults.h
@@ -255,6 +255,13 @@ int32_t default_discard(call_frame_t *frame,
off_t offset,
size_t len, dict_t *xdata);
+int32_t default_zerofill(call_frame_t *frame,
+ xlator_t *this,
+ fd_t *fd,
+ off_t offset,
+ size_t len, dict_t *xdata);
+
+
/* Resume */
int32_t default_getspec_resume (call_frame_t *frame,
xlator_t *this,
@@ -477,6 +484,13 @@ int32_t default_discard_resume(call_frame_t *frame,
off_t offset,
size_t len, dict_t *xdata);
+int32_t default_zerofill_resume(call_frame_t *frame,
+ xlator_t *this,
+ fd_t *fd,
+ off_t offset,
+ size_t len, dict_t *xdata);
+
+
/* _cbk */
int32_t
@@ -695,6 +709,10 @@ int32_t default_discard_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
int32_t op_ret, int32_t op_errno, struct iatt *pre,
struct iatt *post, dict_t *xdata);
+int32_t default_zerofill_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+ int32_t op_ret, int32_t op_errno, struct iatt *pre,
+ struct iatt *post, dict_t *xdata);
+
int32_t
default_getspec_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
int32_t op_ret, int32_t op_errno, char *spec_data);
diff --git a/libglusterfs/src/globals.c b/libglusterfs/src/globals.c
index 1e4cbf0edeb..259c5c885f7 100644
--- a/libglusterfs/src/globals.c
+++ b/libglusterfs/src/globals.c
@@ -70,6 +70,7 @@ const char *gf_fop_list[GF_FOP_MAXVALUE] = {
[GF_FOP_FREMOVEXATTR]= "FREMOVEXATTR",
[GF_FOP_FALLOCATE] = "FALLOCATE",
[GF_FOP_DISCARD] = "DISCARD",
+ [GF_FOP_ZEROFILL] = "ZEROFILL",
};
/* THIS */
diff --git a/libglusterfs/src/glusterfs.h b/libglusterfs/src/glusterfs.h
index b38d6d53e2c..474631e3ff9 100644
--- a/libglusterfs/src/glusterfs.h
+++ b/libglusterfs/src/glusterfs.h
@@ -212,6 +212,7 @@ typedef enum {
GF_FOP_FREMOVEXATTR,
GF_FOP_FALLOCATE,
GF_FOP_DISCARD,
+ GF_FOP_ZEROFILL,
GF_FOP_MAXVALUE,
} glusterfs_fop_t;
diff --git a/libglusterfs/src/syncop.c b/libglusterfs/src/syncop.c
index 866dca50ea4..c1620bb7088 100644
--- a/libglusterfs/src/syncop.c
+++ b/libglusterfs/src/syncop.c
@@ -2173,6 +2173,35 @@ syncop_discard(xlator_t *subvol, fd_t *fd, off_t offset, size_t len)
return args.op_ret;
}
+int
+syncop_zerofill_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
+ int op_ret, int op_errno, struct iatt *prebuf,
+ struct iatt *postbuf, dict_t *xdata)
+{
+ struct syncargs *args = NULL;
+
+ args = cookie;
+
+ args->op_ret = op_ret;
+ args->op_errno = op_errno;
+
+ __wake (args);
+
+ return 0;
+}
+
+int
+syncop_zerofill(xlator_t *subvol, fd_t *fd, off_t offset, size_t len)
+{
+ struct syncargs args = {0, };
+
+ SYNCOP (subvol, (&args), syncop_zerofill_cbk, subvol->fops->zerofill,
+ fd, offset, len, NULL);
+
+ errno = args.op_errno;
+ return args.op_ret;
+}
+
int
syncop_lk_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
diff --git a/libglusterfs/src/syncop.h b/libglusterfs/src/syncop.h
index 0fc9e14ba0f..f790981f0bc 100644
--- a/libglusterfs/src/syncop.h
+++ b/libglusterfs/src/syncop.h
@@ -403,6 +403,8 @@ int syncop_fallocate(xlator_t *subvol, fd_t *fd, int32_t keep_size, off_t offset
size_t len);
int syncop_discard(xlator_t *subvol, fd_t *fd, off_t offset, size_t len);
+int syncop_zerofill(xlator_t *subvol, fd_t *fd, off_t offset, size_t len);
+
int syncop_rename (xlator_t *subvol, loc_t *oldloc, loc_t *newloc);
int syncop_lk (xlator_t *subvol, fd_t *fd, int cmd, struct gf_flock *flock);
diff --git a/libglusterfs/src/xlator.c b/libglusterfs/src/xlator.c
index d029475504a..483c588a9f1 100644
--- a/libglusterfs/src/xlator.c
+++ b/libglusterfs/src/xlator.c
@@ -81,6 +81,7 @@ fill_defaults (xlator_t *xl)
SET_DEFAULT_FOP (fsetattr);
SET_DEFAULT_FOP (fallocate);
SET_DEFAULT_FOP (discard);
+ SET_DEFAULT_FOP (zerofill);
SET_DEFAULT_FOP (getspec);
diff --git a/libglusterfs/src/xlator.h b/libglusterfs/src/xlator.h
index d6296262a13..2f938c384fc 100644
--- a/libglusterfs/src/xlator.h
+++ b/libglusterfs/src/xlator.h
@@ -434,6 +434,14 @@ typedef int32_t (*fop_discard_cbk_t) (call_frame_t *frame,
struct iatt *preop_stbuf,
struct iatt *postop_stbuf, dict_t *xdata);
+typedef int32_t (*fop_zerofill_cbk_t) (call_frame_t *frame,
+ void *cookie,
+ xlator_t *this,
+ int32_t op_ret,
+ int32_t op_errno,
+ struct iatt *preop_stbuf,
+ struct iatt *postop_stbuf, dict_t *xdata);
+
typedef int32_t (*fop_lookup_t) (call_frame_t *frame,
xlator_t *this,
loc_t *loc,
@@ -665,6 +673,12 @@ typedef int32_t (*fop_discard_t) (call_frame_t *frame,
off_t offset,
size_t len,
dict_t *xdata);
+typedef int32_t (*fop_zerofill_t) (call_frame_t *frame,
+ xlator_t *this,
+ fd_t *fd,
+ off_t offset,
+ size_t len,
+ dict_t *xdata);
struct xlator_fops {
fop_lookup_t lookup;
@@ -711,6 +725,7 @@ struct xlator_fops {
fop_getspec_t getspec;
fop_fallocate_t fallocate;
fop_discard_t discard;
+ fop_zerofill_t zerofill;
/* these entries are used for a typechecking hack in STACK_WIND _only_ */
fop_lookup_cbk_t lookup_cbk;
@@ -757,6 +772,7 @@ struct xlator_fops {
fop_getspec_cbk_t getspec_cbk;
fop_fallocate_cbk_t fallocate_cbk;
fop_discard_cbk_t discard_cbk;
+ fop_zerofill_cbk_t zerofill_cbk;
};
typedef int32_t (*cbk_forget_t) (xlator_t *this,