[PATCH 1/5] FIX: Unfreeze array if reshape_array wasn"t succeded

If reshape_array does not success we should not leave
array freezed.

Signed-off-by: Krzysztof Wojcik
---
Grow.c | 5 ++++-
1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/Grow.c b/Grow.c
index c5f83a8..7c5edae 100644
--- a/Grow.c
+++ b/Grow.c
@@ -1567,7 +1567,10 @@ int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
sync_metadata(st);
rv = reshape_array(container, cfd, fd, devname,
st, &info, force, backup_file, quiet, 0);
- frozen = 0;
+ if (rv)
+ frozen = 1;
+ else
+ frozen = 0;
}
release:
if (frozen > 0)

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 2/5] Add raid1->raid0 takeover support

Add support for raid1 to raid0 takeover operation in user space.
This patch includes support for native and imsm metadata.

Signed-off-by: Krzysztof Wojcik
---
Grow.c | 39 ++++++++++++++++++++++++++++-----------
super-intel.c | 6 ++++++
2 files changed, 34 insertions(+), 11 deletions(-)

diff --git a/Grow.c b/Grow.c
index 7c5edae..a8da20a 100644
--- a/Grow.c
+++ b/Grow.c
@@ -650,15 +650,20 @@ void abort_reshape(struct mdinfo *sra)
sysfs_set_str(sra, NULL, "sync_max", "max");
}

-int remove_disks_on_raid10_to_raid0_takeover(struct supertype *st,
- struct mdinfo *sra,
- int layout)
+int remove_disks_for_takeover(struct supertype *st,
+ struct mdinfo *sra,
+ int layout)
{
int nr_of_copies;
struct mdinfo *remaining;
int slot;

- nr_of_copies = layout & 0xff;
+ if (sra->array.level == 10)
+ nr_of_copies = layout & 0xff;
+ else if (sra->array/level == 1)
+ nr_of_copies = sra->array.raid_disks;
+ else
+ return 1;

remaining = sra->devs;
sra->devs = NULL;
@@ -911,8 +916,18 @@ char *analyse_change(struct mdinfo *info, struct reshape *re)
switch (info->array.level) {
case 1:
/* RAID1 can convert to RAID1 with different disks, or
- * raid5 with 2 disks
+ * raid5 with 2 disks, or
+ * raid0 with 1 disk
*/
+ if (info->new_level == 0) {
+ re->level = 0;
+ re->before.data_disks = info->array.raid_disks / 2;
+ re->after.data_disks = re->before.data_disks;
+ re->before.layout = 0;
+ re->backup_blocks = 0;
+ re->parity = 0;
+ return NULL;
+ }
if (info->new_level == 1) {
if (info->delta_disks == UnSet)
/* Don't know what to do */
@@ -1449,15 +1464,17 @@ int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
size = array.size;
}

- /* ========= check for Raid10 -> Raid0 conversion ===============
+ /* ========= check for Raid10/Raid1 -> Raid0 conversion ===============
* current implementation assumes that following conditions must be met:
- * - far_copies == 1
- * - near_copies == 2
+ * - RAID10:
+ * - far_copies == 1
+ * - near_copies == 2
*/
- if (level == 0 && array.level == 10 && sra &&
- array.layout == ((1 << 8) + 2) && !(array.raid_disks & 1)) {
+ if ((level == 0 && array.level == 10 && sra &&
+ array.layout == ((1 << 8) + 2) && !(array.raid_disks & 1)) ||
+ (level == 0 && array.level == 1 && sra)) {
int err;
- err = remove_disks_on_raid10_to_raid0_takeover(st, sra, array.layout);
+ err = remove_disks_for_takeover(st, sra, array.layout);
if (err) {
dprintf(Name": Array cannot be reshaped\n");
if (cfd > -1)
diff --git a/super-intel.c b/super-intel.c
index b25d4fb..461fb0c 100644
--- a/super-intel.c
+++ b/super-intel.c
@@ -6785,6 +6785,12 @@ enum imsm_reshape_type imsm_analyze_change(struct supertype *st,
check_devs = 1;
}
break;
+ case 1:
+ if (geo->level == 0) {
+ change = CH_TAKEOVER;
+ check_devs = 1;
+ }
+ break;
case 5:
if (geo->level != 0)
change = CH_LEVEL_MIGRATION;

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 3/5] Mistake in raid1->raid5 migration

1. Mistake in target level comparison.
2. Initialize reshape->after.data_disks field
to proper spares_needed calculation
---
Grow.c | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/Grow.c b/Grow.c
index a8da20a..e1e509e 100644
--- a/Grow.c
+++ b/Grow.c
@@ -941,11 +941,12 @@ char *analyse_change(struct mdinfo *info, struct reshape *re)
return NULL;
}
if (info->array.raid_disks == 2 &&
- info->array.raid_disks == 5) {
+ info->new_level == 5) {
/* simple in-place conversion */
re->level = 5;
re->parity = 1;
re->before.data_disks = 1;
+ re->after.data_disks = 1;
re->before.layout = ALGORITHM_LEFT_SYMMETRIC;
re->backup_blocks = 0;
return NULL;

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 4/5] FIX: Validate input in ping_monitor function

For native case we do not have monitor running so we have to
return without pinging.
Moreover we have NULL as input parameter. We should to avoid
segmentation fault.

Signed-off-by: Krzysztof Wojcik
---
msg.c | 9 +++++++--
1 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/msg.c b/msg.c
index 76e74e7..b97ebec 100644
--- a/msg.c
+++ b/msg.c
@@ -206,8 +206,13 @@ int fping_monitor(int sfd)
/* give the monitor a chance to update the metadata */
int ping_monitor(char *devname)
{
- int sfd = connect_monitor(devname);
- int err = fping_monitor(sfd);
+ int sfd, err;
+
+ if (!devname)
+ return -1;
+
+ sfd = connect_monitor(devname);
+ err = fping_monitor(sfd);

close(sfd);
return err;

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 5/5] FIX: Validate input in ping_manager function

For native case we do not have manager running so we have to
return without pinging.
Moreover we have NULL as input parameter. We should to avoid
segmentation fault.

Signed-off-by: Krzysztof Wojcik
---
msg.c | 8 +++++++-
1 files changed, 7 insertions(+), 1 deletions(-)

diff --git a/msg.c b/msg.c
index b97ebec..2c770ab 100644
--- a/msg.c
+++ b/msg.c
@@ -432,7 +432,13 @@ void unblock_monitor(char *container, const int unfreeze)
*/
int ping_manager(char *devname)
{
- int sfd = connect_monitor(devname);
+ int sfd;
+
+ if (!devname)
+ return -1;
+
+ sfd = connect_monitor(devname);
+
struct metadata_update msg = { .len = -1 };
int err = 0;


--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/5] FIX: Unfreeze array if reshape_array wasn"tsucceded

On Mon, 24 Jan 2011 15:17:04 +0100
Krzysztof Wojcik wrote:

> If reshape_array does not success we should not leave
> array freezed.
>
> Signed-off-by: Krzysztof Wojcik
> ---
> Grow.c | 5 ++++-
> 1 files changed, 4 insertions(+), 1 deletions(-)
>
> diff --git a/Grow.c b/Grow.c
> index c5f83a8..7c5edae 100644
> --- a/Grow.c
> +++ b/Grow.c
> @@ -1567,7 +1567,10 @@ int Grow_reshape(char *devname, int fd, int
> quiet, char *backup_file, sync_metadata(st);
> rv = reshape_array(container, cfd, fd, devname,
> st, &info, force, backup_file,
> quiet, 0);
> - frozen = 0;
> + if (rv)
> + frozen = 1;
> + else
> + frozen = 0;

reshape_array is responsible for unfreezing the array in this case, and
I'm fairly sure that it already does!

NeilBrown

> }
> release:
> if (frozen > 0)

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/5] Add raid1->raid0 takeover support

On Mon, 24 Jan 2011 15:17:12 +0100
Krzysztof Wojcik wrote:

> Add support for raid1 to raid0 takeover operation in user space.
> This patch includes support for native and imsm metadata.
>
> Signed-off-by: Krzysztof Wojcik
> ---
> Grow.c | 39 ++++++++++++++++++++++++++++-----------
> super-intel.c | 6 ++++++
> 2 files changed, 34 insertions(+), 11 deletions(-)
>
> diff --git a/Grow.c b/Grow.c
> index 7c5edae..a8da20a 100644
> --- a/Grow.c
> +++ b/Grow.c
> @@ -650,15 +650,20 @@ void abort_reshape(struct mdinfo *sra)
> sysfs_set_str(sra, NULL, "sync_max", "max");
> }
>
> -int remove_disks_on_raid10_to_raid0_takeover(struct supertype *st,
> - struct mdinfo *sra,
> - int layout)
> +int remove_disks_for_takeover(struct supertype *st,
> + struct mdinfo *sra,
> + int layout)
> {
> int nr_of_copies;
> struct mdinfo *remaining;
> int slot;
>
> - nr_of_copies = layout & 0xff;
> + if (sra->array.level == 10)
> + nr_of_copies = layout & 0xff;
> + else if (sra->array/level == 1)
> + nr_of_copies = sra->array.raid_disks;
> + else
> + return 1;
>
> remaining = sra->devs;
> sra->devs = NULL;
> @@ -911,8 +916,18 @@ char *analyse_change(struct mdinfo *info, struct
> reshape *re) switch (info->array.level) {
> case 1:
> /* RAID1 can convert to RAID1 with different disks,
> or
> - * raid5 with 2 disks
> + * raid5 with 2 disks, or
> + * raid0 with 1 disk
> */
> + if (info->new_level == 0) {
> + re->level = 0;
> + re->before.data_disks =
> info->array.raid_disks / 2;
> + re->after.data_disks = re->before.data_disks;

dividing raid_disks by 2 doesn't make any sense at all.
It have fixed it so that it does make sense, and applied.

Thanks,
NeilBrown


> + re->before.layout = 0;
> + re->backup_blocks = 0;
> + re->parity = 0;
> + return NULL;
> + }
> if (info->new_level == 1) {
> if (info->delta_disks == UnSet)
> /* Don't know what to do */
> @@ -1449,15 +1464,17 @@ int Grow_reshape(char *devname, int fd, int
> quiet, char *backup_file, size = array.size;
> }
>
> - /* ========= check for Raid10 -> Raid0 conversion
> ===============
> + /* ========= check for Raid10/Raid1 -> Raid0 conversion
> ===============
> * current implementation assumes that following conditions
> must be met:
> - * - far_copies == 1
> - * - near_copies == 2
> + * - RAID10:
> + * - far_copies == 1
> + * - near_copies == 2
> */
> - if (level == 0 && array.level == 10 && sra &&
> - array.layout == ((1 << 8) + 2) && !(array.raid_disks &
> 1)) {
> + if ((level == 0 && array.level == 10 && sra &&
> + array.layout == ((1 << 8) + 2) && !(array.raid_disks &
> 1)) ||
> + (level == 0 && array.level == 1 && sra)) {
> int err;
> - err = remove_disks_on_raid10_to_raid0_takeover(st,
> sra, array.layout);
> + err = remove_disks_for_takeover(st, sra,
> array.layout); if (err) {
> dprintf(Name": Array cannot be reshaped\n");
> if (cfd > -1)
> diff --git a/super-intel.c b/super-intel.c
> index b25d4fb..461fb0c 100644
> --- a/super-intel.c
> +++ b/super-intel.c
> @@ -6785,6 +6785,12 @@ enum imsm_reshape_type
> imsm_analyze_change(struct supertype *st, check_devs = 1;
> }
> break;
> + case 1:
> + if (geo->level == 0) {
> + change = CH_TAKEOVER;
> + check_devs = 1;
> + }
> + break;
> case 5:
> if (geo->level != 0)
> change = CH_LEVEL_MIGRATION;
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-raid"
> in the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH 3/5] Mistake in raid1->raid5 migration

On Mon, 24 Jan 2011 15:17:20 +0100
Krzysztof Wojcik wrote:

> 1. Mistake in target level comparison.

I had already fixed that one thanks...

> 2. Initialize reshape->after.data_disks field

But not that one. fixed now.

Thanks.
NeilBrown

> to proper spares_needed calculation
> ---
> Grow.c | 3 ++-
> 1 files changed, 2 insertions(+), 1 deletions(-)
>
> diff --git a/Grow.c b/Grow.c
> index a8da20a..e1e509e 100644
> --- a/Grow.c
> +++ b/Grow.c
> @@ -941,11 +941,12 @@ char *analyse_change(struct mdinfo *info,
> struct reshape *re) return NULL;
> }
> if (info->array.raid_disks == 2 &&
> - info->array.raid_disks == 5) {
> + info->new_level == 5) {
> /* simple in-place conversion */
> re->level = 5;
> re->parity = 1;
> re->before.data_disks = 1;
> + re->after.data_disks = 1;
> re->before.layout = ALGORITHM_LEFT_SYMMETRIC;
> re->backup_blocks = 0;
> return NULL;

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH 4/5] FIX: Validate input in ping_monitor function

On Mon, 24 Jan 2011 15:17:29 +0100
Krzysztof Wojcik wrote:

> For native case we do not have monitor running so we have to
> return without pinging.
> Moreover we have NULL as input parameter. We should to avoid
> segmentation fault.

I have not applied this patch or the next one.
I would rather not call ping_XXX when not needed, rather than have it
check if it was needed or not.

NeilBrown

>
> Signed-off-by: Krzysztof Wojcik
> ---
> msg.c | 9 +++++++--
> 1 files changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/msg.c b/msg.c
> index 76e74e7..b97ebec 100644
> --- a/msg.c
> +++ b/msg.c
> @@ -206,8 +206,13 @@ int fping_monitor(int sfd)
> /* give the monitor a chance to update the metadata */
> int ping_monitor(char *devname)
> {
> - int sfd = connect_monitor(devname);
> - int err = fping_monitor(sfd);
> + int sfd, err;
> +
> + if (!devname)
> + return -1;
> +
> + sfd = connect_monitor(devname);
> + err = fping_monitor(sfd);
>
> close(sfd);
> return err;
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-raid"
> in the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html