On 01/13/2012 01:14 AM, Greg KH wrote:
>
> Again, care to send this to [email protected]?
>
> thanks,
>
> greg k-h
As reply to this mail I'm sending two versions of this patch:
One for the linux-3.0.y branch (annotated linux-3.0.y)
2nd for the linux-3.1.y branch (annotated linux-3.1.y)
Thanks
As mandated by the standard. In case of an IO error, a pNFS
objects layout driver must return it's layout. This is because
all device errors are reported to the server as part of the
layout return buffer.
This is implemented the same way PNFS_LAYOUTRET_ON_SETATTR
is done, through a bit flag on the pnfs_layoutdriver_type->flags
member. The flag is set by the layout driver that wants a
layout_return preformed at pnfs_ld_{write,read}_done in case
of an error.
(Though I have not defined a wrapper like pnfs_ld_layoutret_on_setattr
because this code is never called outside of pnfs.c and pnfs IO
paths)
Without this patch 3.[0-2] Kernels leak memory and have an annoying
WARN_ON after every IO error utilizing the pnfs-obj driver.
[This is the linux-3.0.y version of mainline [fe0fe83] patch]
CC: Stable Tree <[email protected]>
Signed-off-by: Trond Myklebust <[email protected]>
Signed-off-by: Boaz Harrosh <[email protected]>
---
fs/nfs/objlayout/objio_osd.c | 3 ++-
fs/nfs/pnfs.c | 12 ++++++++++++
fs/nfs/pnfs.h | 1 +
3 files changed, 15 insertions(+), 1 deletions(-)
diff --git a/fs/nfs/objlayout/objio_osd.c b/fs/nfs/objlayout/objio_osd.c
index 1d1dc1e..75fe694 100644
--- a/fs/nfs/objlayout/objio_osd.c
+++ b/fs/nfs/objlayout/objio_osd.c
@@ -1006,7 +1006,8 @@ static bool objio_pg_test(struct nfs_pageio_descriptor *pgio,
static struct pnfs_layoutdriver_type objlayout_type = {
.id = LAYOUT_OSD2_OBJECTS,
.name = "LAYOUT_OSD2_OBJECTS",
- .flags = PNFS_LAYOUTRET_ON_SETATTR,
+ .flags = PNFS_LAYOUTRET_ON_SETATTR |
+ PNFS_LAYOUTRET_ON_ERROR,
.alloc_layout_hdr = objlayout_alloc_layout_hdr,
.free_layout_hdr = objlayout_free_layout_hdr,
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 36d2a29..9951887 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -1119,6 +1119,14 @@ pnfs_ld_write_done(struct nfs_write_data *data)
data->mds_ops->rpc_release(data);
return 0;
}
+ if (NFS_SERVER(data->inode)->pnfs_curr_ld->flags &
+ PNFS_LAYOUTRET_ON_ERROR) {
+ /* Don't lo_commit on error, Server will needs to
+ * preform a file recovery.
+ */
+ clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(data->inode)->flags);
+ pnfs_return_layout(data->inode);
+ }
dprintk("%s: pnfs_error=%d, retry via MDS\n", __func__,
data->pnfs_error);
@@ -1167,6 +1175,10 @@ pnfs_ld_read_done(struct nfs_read_data *data)
return 0;
}
+ if (NFS_SERVER(data->inode)->pnfs_curr_ld->flags &
+ PNFS_LAYOUTRET_ON_ERROR)
+ pnfs_return_layout(data->inode);
+
dprintk("%s: pnfs_error=%d, retry via MDS\n", __func__,
data->pnfs_error);
status = nfs_initiate_read(data, NFS_CLIENT(data->inode),
diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index 9d147d9..bb8b324 100644
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -68,6 +68,7 @@ enum {
enum layoutdriver_policy_flags {
/* Should the pNFS client commit and return the layout upon a setattr */
PNFS_LAYOUTRET_ON_SETATTR = 1 << 0,
+ PNFS_LAYOUTRET_ON_ERROR = 1 << 1,
};
struct nfs4_deviceid_node;
--
1.7.2.3
As mandated by the standard. In case of an IO error, a pNFS
objects layout driver must return it's layout. This is because
all device errors are reported to the server as part of the
layout return buffer.
This is implemented the same way PNFS_LAYOUTRET_ON_SETATTR
is done, through a bit flag on the pnfs_layoutdriver_type->flags
member. The flag is set by the layout driver that wants a
layout_return preformed at pnfs_ld_{write,read}_done in case
of an error.
(Though I have not defined a wrapper like pnfs_ld_layoutret_on_setattr
because this code is never called outside of pnfs.c and pnfs IO
paths)
Without this patch 3.[0-2] Kernels leak memory and have an annoying
WARN_ON after every IO error utilizing the pnfs-obj driver.
[This is the linux-3.1.y version of mainline [fe0fe83] patch]
CC: Stable Tree <[email protected]>
Signed-off-by: Trond Myklebust <[email protected]>
Signed-off-by: Boaz Harrosh <[email protected]>
---
fs/nfs/objlayout/objio_osd.c | 3 ++-
fs/nfs/pnfs.c | 12 ++++++++++++
fs/nfs/pnfs.h | 1 +
3 files changed, 15 insertions(+), 1 deletions(-)
diff --git a/fs/nfs/objlayout/objio_osd.c b/fs/nfs/objlayout/objio_osd.c
index d0cda12..bd80718 100644
--- a/fs/nfs/objlayout/objio_osd.c
+++ b/fs/nfs/objlayout/objio_osd.c
@@ -1015,7 +1015,8 @@ static const struct nfs_pageio_ops objio_pg_write_ops = {
static struct pnfs_layoutdriver_type objlayout_type = {
.id = LAYOUT_OSD2_OBJECTS,
.name = "LAYOUT_OSD2_OBJECTS",
- .flags = PNFS_LAYOUTRET_ON_SETATTR,
+ .flags = PNFS_LAYOUTRET_ON_SETATTR |
+ PNFS_LAYOUTRET_ON_ERROR,
.alloc_layout_hdr = objlayout_alloc_layout_hdr,
.free_layout_hdr = objlayout_free_layout_hdr,
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index ee73d9a..6f6d43a 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -1177,6 +1177,15 @@ void pnfs_ld_write_done(struct nfs_write_data *data)
put_lseg(data->lseg);
data->lseg = NULL;
dprintk("pnfs write error = %d\n", data->pnfs_error);
+ if (NFS_SERVER(data->inode)->pnfs_curr_ld->flags &
+ PNFS_LAYOUTRET_ON_ERROR) {
+ /* Don't lo_commit on error, Server will needs to
+ * preform a file recovery.
+ */
+ clear_bit(NFS_INO_LAYOUTCOMMIT,
+ &NFS_I(data->inode)->flags);
+ pnfs_return_layout(data->inode);
+ }
}
data->mds_ops->rpc_release(data);
}
@@ -1271,6 +1280,9 @@ void pnfs_ld_read_done(struct nfs_read_data *data)
put_lseg(data->lseg);
data->lseg = NULL;
dprintk("pnfs write error = %d\n", data->pnfs_error);
+ if (NFS_SERVER(data->inode)->pnfs_curr_ld->flags &
+ PNFS_LAYOUTRET_ON_ERROR)
+ pnfs_return_layout(data->inode);
}
data->mds_ops->rpc_release(data);
}
diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index 1509530..53d593a 100644
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -68,6 +68,7 @@ enum {
enum layoutdriver_policy_flags {
/* Should the pNFS client commit and return the layout upon a setattr */
PNFS_LAYOUTRET_ON_SETATTR = 1 << 0,
+ PNFS_LAYOUTRET_ON_ERROR = 1 << 1,
};
struct nfs4_deviceid_node;
--
1.7.2.3
On Wed, Jan 18, 2012 at 11:56:21PM +0200, Boaz Harrosh wrote:
> On 01/13/2012 01:14 AM, Greg KH wrote:
> >
> > Again, care to send this to [email protected]?
> >
> > thanks,
> >
> > greg k-h
>
> As reply to this mail I'm sending two versions of this patch:
>
> One for the linux-3.0.y branch (annotated linux-3.0.y)
I've applied this one.
> 2nd for the linux-3.1.y branch (annotated linux-3.1.y)
As 3.1.y is end-of-life, I didn't apply this one, but thanks anyway.
greg k-h