Skip to contentSkip to conversation

xfsprogs: fix Out-of-bounds access in repair/dinode.c

PATCH In review Li Zhong 4 participants 11 messages Aug 12, 2013 – Oct 18, 2013

lore
Share & tools
lore

Li Zhong

author
#0 lore raw

PATCH 1/1 xfsprogs: fix Out-of-bounds access in repair/dinode.c

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +2 -2
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}
Signature · 6 lines
1.8.1.4

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Reviewed-by on v1 1/1
#1 lore raw

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
  repair/dinode.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
  	}

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 10 lines
repair/dinode.c +2 -2
 
  	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
  			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
  			break;
  		}
  	}

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#2 lore raw

Hey Rich and Li Zhong,

Quoted 34 lines · Rich Johnston

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 9 lines
repair/dinode.c +2 -2
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}

We also discussed this issue a bit in this thread:
http://oss.sgi.com/archives/xfs/2013-08/msg00228.html

Looks like the loop itself is incorrect and should be removed, and Eric has suggested that the conditional be changed to a memcmp in case the size of the pad changes in the future. Would either of you care to spin up another patch to clean it up?

Thanks,
Ben

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

author
#3 lore raw
Quoted 50 lines · Ben Myers

Hey Rich and Li Zhong,

Quoted 34 lines · Rich Johnston

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 9 lines
repair/dinode.c +2 -2
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}

We also discussed this issue a bit in this thread:
http://oss.sgi.com/archives/xfs/2013-08/msg00228.html

Quoted 4 lines

Looks like the loop itself is incorrect and should be removed, and Eric has suggested that the conditional be changed to a memcmp in case the size of the pad changes in the future. Would either of you care to spin up another patch to clean it up?

Hi Ben,

If I understand correctly, we need to change 16 to be a sizeof the di_pad2 array(like the fix attached below)?

It seems to me that the loop is needed to check all of the 16 entries in the array?

Thanks, Zhong

---

repair/dinode.c +2 -2
index b2b9a95..7469fc8 100644
@@ -182,10 +182,10 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 		platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
 	}
 
-	for (i = 0; i < 16; i++) {
+	for (i = 0; i < sizeof(dinoc->di_pad2)/sizeof(dinoc->di_pad2[0]); i++) {
 		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad2, 0, 16);
+			memset(dinoc->di_pad2, 0, sizeof(dinoc->di_pad2));
 			break;
 		}
 	}

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#4 lore raw

Hey Zhong,

Quoted 56 lines · Li Zhong
Quoted 50 lines · Ben Myers

Hey Rich and Li Zhong,

Quoted 34 lines · Rich Johnston

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 9 lines
repair/dinode.c +2 -2
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}

We also discussed this issue a bit in this thread:
http://oss.sgi.com/archives/xfs/2013-08/msg00228.html

Quoted 4 lines

Looks like the loop itself is incorrect and should be removed, and Eric has suggested that the conditional be changed to a memcmp in case the size of the pad changes in the future. Would either of you care to spin up another patch to clean it up?

Quoted 7 lines

Hi Ben,

If I understand correctly, we need to change 16 to be a sizeof the di_pad2 array(like the fix attached below)?

It seems to me that the loop is needed to check all of the 16 entries in the array?

You are correct. We need the loop...

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 185

Quoted 13 lines

Thanks, Zhong

---

repair/dinode.c +1 -1
index b2b9a95..7469fc8 100644
@@ -182,10 +182,10 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 		platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
 	}
 
-	for (i = 0; i < 16; i++) {
+	for (i = 0; i < sizeof(dinoc->di_pad2)/sizeof(dinoc->di_pad2[0]); i++) {

Just sizeof(dinoc->di_pad2) would be enough here, right? No need for the division I think.

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 188

Quoted 4 lines
repair/dinode.c +1 -1
 		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad2, 0, 16);
+			memset(dinoc->di_pad2, 0, sizeof(dinoc->di_pad2));
						  ^^^^^^^^^^^^^^^^^^^^^^

That guy will return 16, so we'll memset off the end of di_pad2. sizeof(dinoc->di_pad2[0]) would work.

Thanks!
-Ben

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

author
#5 lore raw
Quoted 80 lines · Ben Myers

Hey Zhong,

Quoted 56 lines · Li Zhong
Quoted 50 lines · Ben Myers

Hey Rich and Li Zhong,

Quoted 34 lines · Rich Johnston

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 9 lines
repair/dinode.c +2 -2
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}

We also discussed this issue a bit in this thread:
http://oss.sgi.com/archives/xfs/2013-08/msg00228.html

Quoted 4 lines

Looks like the loop itself is incorrect and should be removed, and Eric has suggested that the conditional be changed to a memcmp in case the size of the pad changes in the future. Would either of you care to spin up another patch to clean it up?

Quoted 7 lines

Hi Ben,

If I understand correctly, we need to change 16 to be a sizeof the di_pad2 array(like the fix attached below)?

It seems to me that the loop is needed to check all of the 16 entries in the array?

You are correct. We need the loop...

Quoted 11 lines

Thanks, Zhong

---

repair/dinode.c +0 -0
index b2b9a95..7469fc8 100644
@@ -182,10 +182,10 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 		platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
 	}
 
Quoted 5 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 185

Quoted 2 lines
repair/dinode.c +1 -1
-	for (i = 0; i < 16; i++) {
+	for (i = 0; i < sizeof(dinoc->di_pad2)/sizeof(dinoc->di_pad2[0]); i++) {

Just sizeof(dinoc->di_pad2) would be enough here, right? No need for the division I think.

I think the division could be used to calculate the number of elements in the array(16). But yes, in this case, as the element in the array is 1 byte in size, so it's the same whether we add the division or not.

Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 188

Quoted 5 lines
repair/dinode.c +1 -1
 		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad2, 0, 16);
+			memset(dinoc->di_pad2, 0, sizeof(dinoc->di_pad2));
						  ^^^^^^^^^^^^^^^^^^^^^^

That guy will return 16, so we'll memset off the end of di_pad2. sizeof(dinoc->di_pad2[0]) would work.

I think we need to memset the whole array, if we find some non-zero value there.

Thanks, Zhong

Quoted 4 lines

Thanks!
-Ben

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#6 lore raw
In reply to Ben Myers
Quoted 50 lines · Ben Myers

Hey Rich and Li Zhong,

Quoted 34 lines · Rich Johnston

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 9 lines
repair/dinode.c +2 -2
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}

We also discussed this issue a bit in this thread:
http://oss.sgi.com/archives/xfs/2013-08/msg00228.html

Quoted 4 lines

Looks like the loop itself is incorrect and should be removed, and Eric has suggested that the conditional be changed to a memcmp in case the size of the pad changes in the future. Would either of you care to spin up another patch to clean it up?

I think I was confused; it seems fine as it is in git, not sure what I was thinking.

memcmp can't use a bare "0" as an arg, so it's not ideal to use either.

Not a huge fan of the hard-coded 16, but I think the code is correct now; we can probably move on to real problems. ;)

-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#7 lore raw

Hey,

Quoted 56 lines · Eric Sandeen
Quoted 50 lines · Ben Myers

Hey Rich and Li Zhong,

Quoted 34 lines · Rich Johnston

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 9 lines
repair/dinode.c +2 -2
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}

We also discussed this issue a bit in this thread:
http://oss.sgi.com/archives/xfs/2013-08/msg00228.html

Quoted 4 lines

Looks like the loop itself is incorrect and should be removed, and Eric has suggested that the conditional be changed to a memcmp in case the size of the pad changes in the future. Would either of you care to spin up another patch to clean it up?

Quoted 7 lines

I think I was confused; it seems fine as it is in git, not sure what I was thinking.

memcmp can't use a bare "0" as an arg, so it's not ideal to use either.

Not a huge fan of the hard-coded 16, but I think the code is correct now; we can probably move on to real problems. ;)

185 for (i = 0; i < 16; i++) {
186 if (dinoc->di_pad2[i] != 0) {

187                         __dirty_no_modify_ret(dirty);
188                         memset(dinoc->di_pad2, 0, 16);
189                         break;

190 }
191 }

D'oh! I was mistaken too!

I was thinking that line 188 read 'memset(&dinoc->di_pad2[i], 0, 16);' and that we were going off the end of the array as i increased... Teach me to look a little closer.

I agree that the current code is fine.

Apologies to Rich and Zhong.

-Ben

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

author
#8 lore raw
Quoted 72 lines · Ben Myers

Hey,

Quoted 56 lines · Eric Sandeen
Quoted 50 lines · Ben Myers

Hey Rich and Li Zhong,

Quoted 34 lines · Rich Johnston

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 9 lines
repair/dinode.c +2 -2
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}

We also discussed this issue a bit in this thread:
http://oss.sgi.com/archives/xfs/2013-08/msg00228.html

Quoted 4 lines

Looks like the loop itself is incorrect and should be removed, and Eric has suggested that the conditional be changed to a memcmp in case the size of the pad changes in the future. Would either of you care to spin up another patch to clean it up?

Quoted 7 lines

I think I was confused; it seems fine as it is in git, not sure what I was thinking.

memcmp can't use a bare "0" as an arg, so it's not ideal to use either.

Not a huge fan of the hard-coded 16, but I think the code is correct now; we can probably move on to real problems. ;)

185 for (i = 0; i < 16; i++) {
186 if (dinoc->di_pad2[i] != 0) {

187                         __dirty_no_modify_ret(dirty);
188                         memset(dinoc->di_pad2, 0, 16);
189                         break;
Quoted 8 lines

190 }
191 }

D'oh! I was mistaken too!

I was thinking that line 188 read 'memset(&dinoc->di_pad2[i], 0, 16);' and that we were going off the end of the array as i increased... Teach me to look a little closer.

I see, hehe

Quoted 4 lines

I agree that the current code is fine.

Apologies to Rich and Zhong.

No problem, and after the discussion, maybe we could improve the code by removing the hard-coded 16 :)

Thanks, Zhong

Quoted 3 lines

-Ben

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

author
#9 lore raw
In reply to Eric Sandeen
Quoted 56 lines · Eric Sandeen
Quoted 50 lines · Ben Myers

Hey Rich and Li Zhong,

Quoted 34 lines · Rich Johnston

Looks good, thanks for the patch Li Zhong. it has been committed.

--Rich

commit e7c05095f5baa9cd2e35a6de03d7dd9f51dd3910

Author: Li Zhong <[email protected]>
Date:   Mon Aug 12 06:11:01 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
 repair/dinode.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 	}
 
Quoted 9 lines

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 9 lines
repair/dinode.c +2 -2
 	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
 			break;
 		}
 	}

We also discussed this issue a bit in this thread:
http://oss.sgi.com/archives/xfs/2013-08/msg00228.html

Quoted 4 lines

Looks like the loop itself is incorrect and should be removed, and Eric has suggested that the conditional be changed to a memcmp in case the size of the pad changes in the future. Would either of you care to spin up another patch to clean it up?

Quoted 7 lines

I think I was confused; it seems fine as it is in git, not sure what I was thinking.

memcmp can't use a bare "0" as an arg, so it's not ideal to use either.

Not a huge fan of the hard-coded 16, but I think the code is correct now; we can probably move on to real problems. ;)

OK :) Or maybe we could improve it with the calculation using sizeof as below(which I posted in another thread)?

Thanks, Zhong

---

repair/dinode.c +2 -2
index b2b9a95..7469fc8 100644
@@ -182,10 +182,10 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
 		platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
 	}
 
-	for (i = 0; i < 16; i++) {
+	for (i = 0; i < sizeof(dinoc->di_pad2)/sizeof(dinoc->di_pad2[0]); i++) {
 		if (dinoc->di_pad2[i] != 0) {
 			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad2, 0, 16);
+			memset(dinoc->di_pad2, 0, sizeof(dinoc->di_pad2));
 			break;
 		}
 	}

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#10 lore raw
In reply to Li Zhong

This has been committed.

Thanks
--Rich

commit e6efb967e61a366dbe877f34e220e32866e7db42

Author: Li Zhong <[email protected]>
Date:   Tue Aug 27 01:58:34 2013 +0000

xfsprogs: fix Out-of-bounds access in repair/dinode.c

Quoted 21 lines · Li Zhong

Following is reported by coverity in bug 1061528:

187 __dirty_no_modify_ret(dirty);

CID 1061528 (#1 of 1): Out-of-bounds access (OVERRUN)53. overrun-buffer-arg: Overrunning array "dinoc->di_pad" of 6 bytes by passing it to a function which accesses it at byte offset 15 using argument "16UL".
188 memset(dinoc->di_pad, 0, 16);

It seems that di_pad here should be di_pad2, as sekharan pointed out.

---
  repair/dinode.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
repair/dinode.c +0 -0
index e607f0b..94bf2f8 100644
@@ -183,9 +183,9 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
  	}

repair/dinode.c clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) · line 191

Quoted 10 lines
repair/dinode.c +2 -2
 
  	for (i = 0; i < 16; i++) {
-		if (dinoc->di_pad[i] != 0) {
+		if (dinoc->di_pad2[i] != 0) {
  			__dirty_no_modify_ret(dirty);
-			memset(dinoc->di_pad, 0, 16);
+			memset(dinoc->di_pad2, 0, 16);
  			break;
  		}
  	}

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs