[Qemu-devel] [PATCH v6 RESENT 0/2] sheepdog: add user-defined redundancy option

[Qemu-devel] [PATCH v6 RESENT 0/2] sheepdog: add user-defined redundancy option

[Liu Yuan]

v6:
 - update comment typo
 - remove is_number()

v5:
 - use pstrcpy instead of strncpy
 - fix a segfalt for 'null' string option string

v4:
 - fix do_sd_create that forgot to pass nr_copies
 - fix parse_redundancy dealing with replicated vdi

v3:
 - rework is_numeric

v2:
 - fix a typo in comment and commit log

 This patch set add one sheepdog specific option for qemu-img to control
 redundancy.

 This patch set is on top of Kevin's block tree.

Liu Yuan (2):
  sheepdog: refactor do_sd_create()
  sheepdog: support user-defined redundancy option

 block/sheepdog.c          |  127 +++++++++++++++++++++++++++++++++++++--------
 include/block/block_int.h |    1 +
 2 files changed, 105 insertions(+), 23 deletions(-)

-- 
1.7.9.5

[Qemu-devel] [PATCH v6 1/2] sheepdog: refactor do_sd_create()

[Liu Yuan]

We can actually use BDRVSheepdogState *s to pass most of the parameters.

Cc: Kevin Wolf <kwolf@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Liu Yuan <namei.unix@gmail.com>
---
 block/sheepdog.c |   37 +++++++++++++++----------------------
 1 file changed, 15 insertions(+), 22 deletions(-)

diff --git a/block/sheepdog.c b/block/sheepdog.c
index ef387de..66b3ea8 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -1464,9 +1464,7 @@ out:
     return ret;
 }
 
-static int do_sd_create(BDRVSheepdogState *s, char *filename, int64_t vdi_size,
-                        uint32_t base_vid, uint32_t *vdi_id, int snapshot,
-                        uint8_t copy_policy)
+static int do_sd_create(BDRVSheepdogState *s, uint32_t *vdi_id, int snapshot)
 {
     SheepdogVdiReq hdr;
     SheepdogVdiRsp *rsp = (SheepdogVdiRsp *)&hdr;
@@ -1483,11 +1481,11 @@ static int do_sd_create(BDRVSheepdogState *s, char *filename, int64_t vdi_size,
      * does not fit in buf?  For now, just truncate and avoid buffer overrun.
      */
     memset(buf, 0, sizeof(buf));
-    pstrcpy(buf, sizeof(buf), filename);
+    pstrcpy(buf, sizeof(buf), s->name);
 
     memset(&hdr, 0, sizeof(hdr));
     hdr.opcode = SD_OP_NEW_VDI;
-    hdr.vdi_id = base_vid;
+    hdr.vdi_id = s->inode.vdi_id;
 
     wlen = SD_MAX_VDI_LEN;
 
@@ -1495,8 +1493,8 @@ static int do_sd_create(BDRVSheepdogState *s, char *filename, int64_t vdi_size,
     hdr.snapid = snapshot;
 
     hdr.data_length = wlen;
-    hdr.vdi_size = vdi_size;
-    hdr.copy_policy = copy_policy;
+    hdr.vdi_size = s->inode.vdi_size;
+    hdr.copy_policy = s->inode.copy_policy;
 
     ret = do_req(fd, (SheepdogReq *)&hdr, buf, &wlen, &rlen);
 
@@ -1507,7 +1505,7 @@ static int do_sd_create(BDRVSheepdogState *s, char *filename, int64_t vdi_size,
     }
 
     if (rsp->result != SD_RES_SUCCESS) {
-        error_report("%s, %s", sd_strerror(rsp->result), filename);
+        error_report("%s, %s", sd_strerror(rsp->result), s->inode.name);
         return -EIO;
     }
 
@@ -1568,23 +1566,21 @@ static int sd_create(const char *filename, QEMUOptionParameter *options,
                      Error **errp)
 {
     int ret = 0;
-    uint32_t vid = 0, base_vid = 0;
-    int64_t vdi_size = 0;
+    uint32_t vid = 0;
     char *backing_file = NULL;
     BDRVSheepdogState *s;
-    char vdi[SD_MAX_VDI_LEN], tag[SD_MAX_VDI_TAG_LEN];
+    char tag[SD_MAX_VDI_TAG_LEN];
     uint32_t snapid;
     bool prealloc = false;
     Error *local_err = NULL;
 
     s = g_malloc0(sizeof(BDRVSheepdogState));
 
-    memset(vdi, 0, sizeof(vdi));
     memset(tag, 0, sizeof(tag));
     if (strstr(filename, "://")) {
-        ret = sd_parse_uri(s, filename, vdi, &snapid, tag);
+        ret = sd_parse_uri(s, filename, s->name, &snapid, tag);
     } else {
-        ret = parse_vdiname(s, filename, vdi, &snapid, tag);
+        ret = parse_vdiname(s, filename, s->name, &snapid, tag);
     }
     if (ret < 0) {
         goto out;
@@ -1592,7 +1588,7 @@ static int sd_create(const char *filename, QEMUOptionParameter *options,
 
     while (options && options->name) {
         if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
-            vdi_size = options->value.n;
+            s->inode.vdi_size = options->value.n;
         } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FILE)) {
             backing_file = options->value.s;
         } else if (!strcmp(options->name, BLOCK_OPT_PREALLOC)) {
@@ -1610,7 +1606,7 @@ static int sd_create(const char *filename, QEMUOptionParameter *options,
         options++;
     }
 
-    if (vdi_size > SD_MAX_VDI_SIZE) {
+    if (s->inode.vdi_size > SD_MAX_VDI_SIZE) {
         error_report("too big image size");
         ret = -EINVAL;
         goto out;
@@ -1645,12 +1641,11 @@ static int sd_create(const char *filename, QEMUOptionParameter *options,
             goto out;
         }
 
-        base_vid = s->inode.vdi_id;
         bdrv_unref(bs);
     }
 
     /* TODO: allow users to specify copy number */
-    ret = do_sd_create(s, vdi, vdi_size, base_vid, &vid, 0, 0);
+    ret = do_sd_create(s, &vid, 0);
     if (!prealloc || ret) {
         goto out;
     }
@@ -1833,8 +1828,7 @@ static int sd_create_branch(BDRVSheepdogState *s)
      * false bail out.
      */
     deleted = sd_delete(s);
-    ret = do_sd_create(s, s->name, s->inode.vdi_size, s->inode.vdi_id, &vid,
-                       !deleted, s->inode.copy_policy);
+    ret = do_sd_create(s, &vid, !deleted);
     if (ret) {
         goto out;
     }
@@ -2097,8 +2091,7 @@ static int sd_snapshot_create(BlockDriverState *bs, QEMUSnapshotInfo *sn_info)
         goto cleanup;
     }
 
-    ret = do_sd_create(s, s->name, s->inode.vdi_size, s->inode.vdi_id, &new_vid,
-                       1, s->inode.copy_policy);
+    ret = do_sd_create(s, &new_vid, 1);
     if (ret < 0) {
         error_report("failed to create inode for snapshot. %s",
                      strerror(errno));
-- 
1.7.9.5

[Qemu-devel] [PATCH v6 2/2] sheepdog: support user-defined redundancy option

[Liu Yuan]

Sheepdog support two kinds of redundancy, full replication and erasure coding.

# create a fully replicated vdi with x copies
 -o redundancy=x (1 <= x <= SD_MAX_COPIES)

# create a erasure coded vdi with x data strips and y parity strips
 -o redundancy=x:y (x must be one of {2,4,8,16} and 1 <= y < SD_EC_MAX_STRIP)

E.g, to convert a vdi into sheepdog vdi 'test' with 8:3 erasure coding scheme

$ qemu-img convert -o redundancy=8:3 linux-0.2.img sheepdog:test

Cc: Kevin Wolf <kwolf@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Liu Yuan <namei.unix@gmail.com>
---
 block/sheepdog.c          |   74 ++++++++++++++++++++++++++++++++++++++++++++-
 include/block/block_int.h |    1 +
 2 files changed, 74 insertions(+), 1 deletion(-)

diff --git a/block/sheepdog.c b/block/sheepdog.c
index 66b3ea8..144da4f 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -91,6 +91,14 @@
 #define SD_NR_VDIS   (1U << 24)
 #define SD_DATA_OBJ_SIZE (UINT64_C(1) << 22)
 #define SD_MAX_VDI_SIZE (SD_DATA_OBJ_SIZE * MAX_DATA_OBJS)
+/*
+ * For erasure coding, we use at most SD_EC_MAX_STRIP for data strips and
+ * (SD_EC_MAX_STRIP - 1) for parity strips
+ *
+ * SD_MAX_COPIES is sum of number of data strips and parity strips.
+ */
+#define SD_EC_MAX_STRIP 16
+#define SD_MAX_COPIES (SD_EC_MAX_STRIP * 2 - 1)
 
 #define SD_INODE_SIZE (sizeof(SheepdogInode))
 #define CURRENT_VDI_ID 0
@@ -1495,6 +1503,7 @@ static int do_sd_create(BDRVSheepdogState *s, uint32_t *vdi_id, int snapshot)
     hdr.data_length = wlen;
     hdr.vdi_size = s->inode.vdi_size;
     hdr.copy_policy = s->inode.copy_policy;
+    hdr.copies = s->inode.nr_copies;
 
     ret = do_req(fd, (SheepdogReq *)&hdr, buf, &wlen, &rlen);
 
@@ -1562,6 +1571,60 @@ out:
     return ret;
 }
 
+/*
+ * Sheepdog support two kinds of redundancy, full replication and erasure
+ * coding.
+ *
+ * # create a fully replicated vdi with x copies
+ * -o redundancy=x (1 <= x <= SD_MAX_COPIES)
+ *
+ * # create a erasure coded vdi with x data strips and y parity strips
+ * -o redundancy=x:y (x must be one of {2,4,8,16} and 1 <= y < SD_EC_MAX_STRIP)
+ */
+static int parse_redundancy(BDRVSheepdogState *s, const char *opt)
+{
+    struct SheepdogInode *inode = &s->inode;
+    const char *n1, *n2;
+    long copy, parity;
+    char p[10];
+
+    pstrcpy(p, sizeof(p), opt);
+    n1 = strtok(p, ":");
+    n2 = strtok(NULL, ":");
+
+    if (!n1) {
+        return -EINVAL;
+    }
+
+    copy = strtol(n1, NULL, 10);
+    if (copy > SD_MAX_COPIES || copy < 1) {
+        return -EINVAL;
+    }
+    if (!n2) {
+        inode->copy_policy = 0;
+        inode->nr_copies = copy;
+        return 0;
+    }
+
+    if (copy != 2 && copy != 4 && copy != 8 && copy != 16) {
+        return -EINVAL;
+    }
+
+    parity = strtol(n2, NULL, 10);
+    if (parity >= SD_EC_MAX_STRIP || parity < 1) {
+        return -EINVAL;
+    }
+
+    /*
+     * 4 bits for parity and 4 bits for data.
+     * We have to compress upper data bits because it can't represent 16
+     */
+    inode->copy_policy = ((copy / 2) << 4) + parity;
+    inode->nr_copies = copy + parity;
+
+    return 0;
+}
+
 static int sd_create(const char *filename, QEMUOptionParameter *options,
                      Error **errp)
 {
@@ -1602,6 +1665,11 @@ static int sd_create(const char *filename, QEMUOptionParameter *options,
                 ret = -EINVAL;
                 goto out;
             }
+        } else if (!strcmp(options->name, BLOCK_OPT_REDUNDANCY)) {
+            ret = parse_redundancy(s, options->value.s);
+            if (ret < 0) {
+                goto out;
+            }
         }
         options++;
     }
@@ -1644,7 +1712,6 @@ static int sd_create(const char *filename, QEMUOptionParameter *options,
         bdrv_unref(bs);
     }
 
-    /* TODO: allow users to specify copy number */
     ret = do_sd_create(s, &vid, 0);
     if (!prealloc || ret) {
         goto out;
@@ -2416,6 +2483,11 @@ static QEMUOptionParameter sd_create_options[] = {
         .type = OPT_STRING,
         .help = "Preallocation mode (allowed values: off, full)"
     },
+    {
+        .name = BLOCK_OPT_REDUNDANCY,
+        .type = OPT_STRING,
+        .help = "Redundancy of the image"
+    },
     { NULL }
 };
 
diff --git a/include/block/block_int.h b/include/block/block_int.h
index a48731d..b90862f 100644
--- a/include/block/block_int.h
+++ b/include/block/block_int.h
@@ -53,6 +53,7 @@
 #define BLOCK_OPT_COMPAT_LEVEL      "compat"
 #define BLOCK_OPT_LAZY_REFCOUNTS    "lazy_refcounts"
 #define BLOCK_OPT_ADAPTER_TYPE      "adapter_type"
+#define BLOCK_OPT_REDUNDANCY        "redundancy"
 
 typedef struct BdrvTrackedRequest {
     BlockDriverState *bs;
-- 
1.7.9.5

Re: [Qemu-devel] [PATCH v6 RESENT 0/2] sheepdog: add user-defined redundancy option

[Liu Yuan]

On Thu, Nov 07, 2013 at 10:56:36PM +0800, Liu Yuan wrote:
> v6:
>  - update comment typo
>  - remove is_number()

Anyone pick this set up?

Thanks
Yuan

Re: [Qemu-devel] [PATCH v6 RESENT 0/2] sheepdog: add user-defined redundancy option

[Liu Yuan]

On Mon, Nov 18, 2013 at 01:58:50PM +0800, Liu Yuan wrote:
> On Thu, Nov 07, 2013 at 10:56:36PM +0800, Liu Yuan wrote:
> > v6:
> >  - update comment typo
> >  - remove is_number()
> 
> Anyone pick this set up?

Ping again. Without this two patches, qemu-img can't create|convert erasure coded
images for sheepdog. I hope this two can go with release of v1.7 if possible.

Thanks
Yuan

Re: [Qemu-devel] [PATCH v6 RESENT 0/2] sheepdog: add user-defined redundancy option

[Stefan Hajnoczi]

On Thu, Nov 21, 2013 at 01:22:53AM +0800, Liu Yuan wrote:
> On Mon, Nov 18, 2013 at 01:58:50PM +0800, Liu Yuan wrote:
> > On Thu, Nov 07, 2013 at 10:56:36PM +0800, Liu Yuan wrote:
> > > v6:
> > >  - update comment typo
> > >  - remove is_number()
> > 
> > Anyone pick this set up?
> 
> Ping again. Without this two patches, qemu-img can't create|convert erasure coded
> images for sheepdog. I hope this two can go with release of v1.7 if possible.

QEMU entered "hard freeze" on 6th of November:
http://qemu-project.org/Planning/1.7

This means only bug fixes should be merged.  Right now we're up to the
final -rc and the QEMU 1.7 release is scheduled for 22nd of November
(tomorrow).

The best I can do is review and apply it to the block-next branch for
qemu.git merge once QEMU 1.8 development begins next week.

Stefan

Re: [Qemu-devel] [PATCH v6 RESENT 0/2] sheepdog: add user-defined redundancy option

[Stefan Hajnoczi]

On Thu, Nov 07, 2013 at 10:56:36PM +0800, Liu Yuan wrote:
> v6:
>  - update comment typo
>  - remove is_number()
> 
> v5:
>  - use pstrcpy instead of strncpy
>  - fix a segfalt for 'null' string option string
> 
> v4:
>  - fix do_sd_create that forgot to pass nr_copies
>  - fix parse_redundancy dealing with replicated vdi
> 
> v3:
>  - rework is_numeric
> 
> v2:
>  - fix a typo in comment and commit log
> 
>  This patch set add one sheepdog specific option for qemu-img to control
>  redundancy.
> 
>  This patch set is on top of Kevin's block tree.
> 
> Liu Yuan (2):
>   sheepdog: refactor do_sd_create()
>   sheepdog: support user-defined redundancy option
> 
>  block/sheepdog.c          |  127 +++++++++++++++++++++++++++++++++++++--------
>  include/block/block_int.h |    1 +
>  2 files changed, 105 insertions(+), 23 deletions(-)

Thanks, applied to my block-next tree:
https://github.com/stefanha/qemu/commits/block-next

Stefan

Re: [Qemu-devel] [PATCH v6 RESENT 0/2] sheepdog: add user-defined redundancy option

[Liu Yuan]

On Thu, Nov 21, 2013 at 09:23:59AM +0100, Stefan Hajnoczi wrote:
> On Thu, Nov 21, 2013 at 01:22:53AM +0800, Liu Yuan wrote:
> > On Mon, Nov 18, 2013 at 01:58:50PM +0800, Liu Yuan wrote:
> > > On Thu, Nov 07, 2013 at 10:56:36PM +0800, Liu Yuan wrote:
> > > > v6:
> > > >  - update comment typo
> > > >  - remove is_number()
> > > 
> > > Anyone pick this set up?
> > 
> > Ping again. Without this two patches, qemu-img can't create|convert erasure coded
> > images for sheepdog. I hope this two can go with release of v1.7 if possible.
> 
> QEMU entered "hard freeze" on 6th of November:
> http://qemu-project.org/Planning/1.7
> 
> This means only bug fixes should be merged.  Right now we're up to the
> final -rc and the QEMU 1.7 release is scheduled for 22nd of November
> (tomorrow).
> 
> The best I can do is review and apply it to the block-next branch for
> qemu.git merge once QEMU 1.8 development begins next week.
> 

Thanks for your explanation. It is helpful :)

Yuan