Skip to contentSkip to conversation

cmirrord:fix endian issue on s390

PATCH New Liuhua Wang 2 participants 4 messages Sep 5, 2014 – Sep 15, 2014

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 cmirrord:fix endian issue on s390

Cmirrord has endian bugs, which cause failure to lvcreate a mirrored lv on s390.
- data_size is uint32, should not use xlate64 to convert, which will cause data_size 0 after xlate.
- request_type and data_size still used by local(v5_data_switch), should convert later. if request_type xlate too early, it will cause request_type judge error; if data_size xlate too early, it will cause coredump in case DM_ULOG_CLEAR_REGION.
- when receiving package in clog_request_from_network. vp[0] will always be little endian. we could use xlate64(vp[0]) == vp[0] to decide if the local node is little endian or not.

---
 daemons/cmirrord/compat.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
daemons/cmirrord/compat.c +5 -4
index 3f7a9b5..4955f50 100644
@@ -126,13 +126,14 @@ static int v5_endian_to_network(struct clog_request *rq)
 
 	u_rq->error = xlate32(u_rq->error);
 	u_rq->seq = xlate32(u_rq->seq);
-	u_rq->request_type = xlate32(u_rq->request_type);
-	u_rq->data_size = xlate64(u_rq->data_size);
 
 	rq->originator = xlate32(rq->originator);
 
 	v5_data_endian_switch(rq, 1);
 
+	u_rq->request_type = xlate32(u_rq->request_type);
+	u_rq->data_size = xlate32(u_rq->data_size);
+
 	return size;
 }
 
@@ -167,7 +168,7 @@ static int v5_endian_from_network(struct clog_request *rq)
 	u_rq->error = xlate32(u_rq->error);
 	u_rq->seq = xlate32(u_rq->seq);
 	u_rq->request_type = xlate32(u_rq->request_type);
-	u_rq->data_size = xlate64(u_rq->data_size);
+	u_rq->data_size = xlate32(u_rq->data_size);
 
 	rq->originator = xlate32(rq->originator);
 
@@ -187,7 +188,7 @@ int clog_request_from_network(void *data, size_t data_len)
 
 	switch (version) {
 	case 5: /* Upstream */
-		if (version == unconverted_version)
+		if (version == vp[0])
 			return 0;
 		break;
 	case 4: /* RHEL 5.[45] */
Signature · 1 line
1.9.0
#1 lore raw

PATCH 1/1 cmirrord:fix endian issue on s390

Yes, this is the sort of patch I'm looking for. Can you confirm it works for s390 and x86[_64]? That patch looks good to me otherwise.

brassow

Quoted 61 lines · Liuhua Wang

Cmirrord has endian bugs, which cause failure to lvcreate a mirrored lv on s390.
- data_size is uint32, should not use xlate64 to convert, which will cause data_size 0 after xlate.
- request_type and data_size still used by local(v5_data_switch), should convert later. if request_type xlate too early, it will cause request_type judge error; if data_size xlate too early, it will cause coredump in case DM_ULOG_CLEAR_REGION.
- when receiving package in clog_request_from_network. vp[0] will always be little endian. we could use xlate64(vp[0]) == vp[0] to decide if the local node is little endian or not.

---
daemons/cmirrord/compat.c | 9 +++++----

1 file changed, 5 insertions(+), 4 deletions(-)
daemons/cmirrord/compat.c +5 -4
index 3f7a9b5..4955f50 100644
@@ -126,13 +126,14 @@ static int v5_endian_to_network(struct clog_request *rq)
 
 	u_rq->error = xlate32(u_rq->error);
 	u_rq->seq = xlate32(u_rq->seq);
-	u_rq->request_type = xlate32(u_rq->request_type);
-	u_rq->data_size = xlate64(u_rq->data_size);
 
 	rq->originator = xlate32(rq->originator);
 
 	v5_data_endian_switch(rq, 1);
 
+	u_rq->request_type = xlate32(u_rq->request_type);
+	u_rq->data_size = xlate32(u_rq->data_size);
+
 	return size;
 }
 
@@ -167,7 +168,7 @@ static int v5_endian_from_network(struct clog_request *rq)
 	u_rq->error = xlate32(u_rq->error);
 	u_rq->seq = xlate32(u_rq->seq);
 	u_rq->request_type = xlate32(u_rq->request_type);
-	u_rq->data_size = xlate64(u_rq->data_size);
+	u_rq->data_size = xlate32(u_rq->data_size);
 
 	rq->originator = xlate32(rq->originator);
 
@@ -187,7 +188,7 @@ int clog_request_from_network(void *data, size_t data_len)
 
 	switch (version) {
 	case 5: /* Upstream */
-		if (version == unconverted_version)
+		if (version == vp[0])
 			return 0;
 		break;
 	case 4: /* RHEL 5.[45] */
Signature · 6 lines
1.9.0

--
lvm-devel mailing list
lvm-devel at redhat.com
https://www.redhat.com/mailman/listinfo/lvm-devel
#2 lore raw

PATCH 1/1 cmirrord:fix endian issue on s390

Hi Brassow,

Thanks for you quick response.

Quoted 1 line
Quoted 1 line
Quoted 1 line

On 9/9/2014 at 01:54 AM, in message

<[email protected]>, Brassow Jonathan

Quoted 4 lines · <[email protected]> wrote:

Yes, this is the sort of patch I'm looking for. Can you confirm it works for s390 and x86[_64]? That patch looks good to me otherwise.

brassow

I did some tests(lvcreate linear/-m1,mkfs,read/write)on both s390x and x86_64. It works well.

Best regards
Liuhua(lwang at suse.com)

Quoted 82 lines · Liuhua Wang
Quoted 2 lines

Cmirrord has endian bugs, which cause failure to lvcreate a mirrored

lv on s390.

Quoted 1 line

- data_size is uint32, should not use xlate64 to convert, which will

cause data_size 0

Quoted 2 lines

after xlate.
- request_type and data_size still used by local(v5_data_switch),

should convert later.

Quoted 1 line

if request_type xlate too early, it will cause request_type judge

error;

Quoted 1 line

if data_size xlate too early, it will cause coredump in case

DM_ULOG_CLEAR_REGION.

Quoted 1 line

- when receiving package in clog_request_from_network. vp[0] will

always be little endian.

Quoted 1 line
         we could use xlate64(vp[0]) == vp[0] to decide if the local node 

is little endian or not.

daemons/cmirrord/compat.c static int v5_endian_to_network(struct clog_request

Quoted 12 lines

---
daemons/cmirrord/compat.c | 9 +++++----

1 file changed, 5 insertions(+), 4 deletions(-)
daemons/cmirrord/compat.c +0 -0
index 3f7a9b5..4955f50 100644
@@ -126,13 +126,14 @@ static int v5_endian_to_network(struct clog_request 

*rq)

daemons/cmirrord/compat.c static int v5_endian_from_network(struct clog_request

Quoted 17 lines
daemons/cmirrord/compat.c +3 -2
 	u_rq->error = xlate32(u_rq->error);
 	u_rq->seq = xlate32(u_rq->seq);
-	u_rq->request_type = xlate32(u_rq->request_type);
-	u_rq->data_size = xlate64(u_rq->data_size);
 
 	rq->originator = xlate32(rq->originator);
 
 	v5_data_endian_switch(rq, 1);
 
+	u_rq->request_type = xlate32(u_rq->request_type);
+	u_rq->data_size = xlate32(u_rq->data_size);
+
 	return size;
 }
 
@@ -167,7 +168,7 @@ static int v5_endian_from_network(struct clog_request 

*rq)

daemons/cmirrord/compat.c int clog_request_from_network(void *data, size_t

Quoted 9 lines
daemons/cmirrord/compat.c +1 -1
 	u_rq->error = xlate32(u_rq->error);
 	u_rq->seq = xlate32(u_rq->seq);
 	u_rq->request_type = xlate32(u_rq->request_type);
-	u_rq->data_size = xlate64(u_rq->data_size);
+	u_rq->data_size = xlate32(u_rq->data_size);
 
 	rq->originator = xlate32(rq->originator);
 
@@ -187,7 +188,7 @@ int clog_request_from_network(void *data, size_t 

data_len)

Quoted 15 lines
daemons/cmirrord/compat.c +1 -1
 	switch (version) {
 	case 5: /* Upstream */
-		if (version == unconverted_version)
+		if (version == vp[0])
 			return 0;
 		break;
 	case 4: /* RHEL 5.[45] */
Signature · 6 lines
1.9.0

--
lvm-devel mailing list
lvm-devel at redhat.com 
https://www.redhat.com/mailman/listinfo/lvm-devel 

--
lvm-devel mailing list
lvm-devel at redhat.com
https://www.redhat.com/mailman/listinfo/lvm-devel

#3 lore raw

PATCH 1/1 cmirrord:fix endian issue on s390

In reply to Liuhua Wang

committed. I chose the author to be the person who posted the patch.

brassow

Gitweb:        http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=829e5a40373a2ef1721dc9210350c86c4b9d70f7
Commit:        829e5a40373a2ef1721dc9210350c86c4b9d70f7
Parent:        e9216eedfe8189947c506041d8fc9e879fac6d58
Author:        Liuhua Wang <[email protected]>
AuthorDate:    Sat Sep 13 02:08:28 2014 -0500
Committer:     Jonathan Brassow <[email protected]>
CommitterDate: Mon Sep 15 16:08:35 2014 -0500
Quoted 61 lines · Liuhua Wang

Cmirrord has endian bugs, which cause failure to lvcreate a mirrored lv on s390.
- data_size is uint32, should not use xlate64 to convert, which will cause data_size 0 after xlate.
- request_type and data_size still used by local(v5_data_switch), should convert later. if request_type xlate too early, it will cause request_type judge error; if data_size xlate too early, it will cause coredump in case DM_ULOG_CLEAR_REGION.
- when receiving package in clog_request_from_network. vp[0] will always be little endian. we could use xlate64(vp[0]) == vp[0] to decide if the local node is little endian or not.

---
daemons/cmirrord/compat.c | 9 +++++----

1 file changed, 5 insertions(+), 4 deletions(-)
daemons/cmirrord/compat.c +5 -4
index 3f7a9b5..4955f50 100644
@@ -126,13 +126,14 @@ static int v5_endian_to_network(struct clog_request *rq)
 
 	u_rq->error = xlate32(u_rq->error);
 	u_rq->seq = xlate32(u_rq->seq);
-	u_rq->request_type = xlate32(u_rq->request_type);
-	u_rq->data_size = xlate64(u_rq->data_size);
 
 	rq->originator = xlate32(rq->originator);
 
 	v5_data_endian_switch(rq, 1);
 
+	u_rq->request_type = xlate32(u_rq->request_type);
+	u_rq->data_size = xlate32(u_rq->data_size);
+
 	return size;
 }
 
@@ -167,7 +168,7 @@ static int v5_endian_from_network(struct clog_request *rq)
 	u_rq->error = xlate32(u_rq->error);
 	u_rq->seq = xlate32(u_rq->seq);
 	u_rq->request_type = xlate32(u_rq->request_type);
-	u_rq->data_size = xlate64(u_rq->data_size);
+	u_rq->data_size = xlate32(u_rq->data_size);
 
 	rq->originator = xlate32(rq->originator);
 
@@ -187,7 +188,7 @@ int clog_request_from_network(void *data, size_t data_len)
 
 	switch (version) {
 	case 5: /* Upstream */
-		if (version == unconverted_version)
+		if (version == vp[0])
 			return 0;
 		break;
 	case 4: /* RHEL 5.[45] */
Signature · 6 lines
1.9.0

--
lvm-devel mailing list
lvm-devel at redhat.com
https://www.redhat.com/mailman/listinfo/lvm-devel