Skip to contentSkip to conversation

f2fs: Fix gcc9 error -Werror=maybe-uninitialized

PATCH In review Michael Chang 4 participants 9 messages May 17, 2019 – May 29, 2019

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 f2fs: Fix gcc9 error -Werror=maybe-uninitialized

The function grub_get_node_path could return uninitialized offset with level == 0 if the block is greater than direct_index + 2*direct_blks + 2*indirect_blks + dindirect_blks. The uninitialized offset is then used by function grub_f2fs_get_block because level == 0 is valid and meaningful return to be processed.

The fix is to set level = -1 as return value by grub_get_node_path to signify an error that the input block cannot be handled. Any caller should therefore check level is negative or not before processing the output.

---
 grub-core/fs/f2fs.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
grub-core/fs/f2fs.c +6 -1
index 644653dbe..bb28b291b 100644
@@ -702,7 +702,7 @@ grub_get_node_path (struct grub_f2fs_inode *inode, grub_uint32_t block,
   grub_uint32_t dindirect_blks = indirect_blks * NIDS_PER_BLOCK;
   grub_uint32_t direct_index = DEF_ADDRS_PER_INODE;
   int n = 0;
-  int level = 0;
+  int level = -1;
 
   if (inode->i_inline & F2FS_INLINE_XATTR)
     direct_index -= F2FS_INLINE_XATTR_ADDRS;
@@ -712,6 +712,7 @@ grub_get_node_path (struct grub_f2fs_inode *inode, grub_uint32_t block,
   if (block < direct_index)
     {
       offset[n] = block;
+      level = 0;
       goto got;
     }
 
@@ -860,6 +861,10 @@ grub_f2fs_get_block (grub_fshelp_node_t node, grub_disk_addr_t block_ofs)
   int level, i;
 
   level = grub_get_node_path (inode, block_ofs, offset, noffset);
+
+  if (level < 0)
+    return -1;
+
   if (level == 0)
     return grub_le_to_cpu32 (inode->i_addr[offset[0]]);
 
Signature · 1 line
2.16.4
#1 lore raw

CC-ing [email protected]

Michael, thank you for posting the patch.

Neil, does it solve your problem?

Daniel

Quoted 51 lines · Michael Chang

The function grub_get_node_path could return uninitialized offset with level == 0 if the block is greater than direct_index + 2*direct_blks + 2*indirect_blks + dindirect_blks. The uninitialized offset is then used by function grub_f2fs_get_block because level == 0 is valid and meaningful return to be processed.

The fix is to set level = -1 as return value by grub_get_node_path to signify an error that the input block cannot be handled. Any caller should therefore check level is negative or not before processing the output.

---
 grub-core/fs/f2fs.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
grub-core/fs/f2fs.c +6 -2
index 644653dbe..bb28b291b 100644
@@ -702,7 +702,7 @@ grub_get_node_path (struct grub_f2fs_inode *inode, grub_uint32_t block,
   grub_uint32_t dindirect_blks = indirect_blks * NIDS_PER_BLOCK;
   grub_uint32_t direct_index = DEF_ADDRS_PER_INODE;
   int n = 0;
-  int level = 0;
+  int level = -1;
 
   if (inode->i_inline & F2FS_INLINE_XATTR)
     direct_index -= F2FS_INLINE_XATTR_ADDRS;
@@ -712,6 +712,7 @@ grub_get_node_path (struct grub_f2fs_inode *inode, grub_uint32_t block,
   if (block < direct_index)
     {
       offset[n] = block;
+      level = 0;
       goto got;
     }
 
@@ -860,6 +861,10 @@ grub_f2fs_get_block (grub_fshelp_node_t node, grub_disk_addr_t block_ofs)
   int level, i;
 
   level = grub_get_node_path (inode, block_ofs, offset, noffset);
+
+  if (level < 0)
+    return -1;
+
   if (level == 0)
     return grub_le_to_cpu32 (inode->i_addr[offset[0]]);
 
--

2.16.4

#2 lore raw

Hi Daniel & Michael

I've tested the new patch (in place of my attempt) and grub is building successfully so it looks good here!

Many thanks!
Neil

Quoted 61 lines · Daniel Kiper

CC-ing [email protected]

Michael, thank you for posting the patch.

Neil, does it solve your problem?

Daniel

Quoted 51 lines · Michael Chang

The function grub_get_node_path could return uninitialized offset with level == 0 if the block is greater than direct_index + 2*direct_blks + 2*indirect_blks + dindirect_blks. The uninitialized offset is then used by function grub_f2fs_get_block because level == 0 is valid and meaningful return to be processed.

The fix is to set level = -1 as return value by grub_get_node_path to signify an error that the input block cannot be handled. Any caller should therefore check level is negative or not before processing the output.

---
 grub-core/fs/f2fs.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
grub-core/fs/f2fs.c +6 -2
index 644653dbe..bb28b291b 100644
@@ -702,7 +702,7 @@ grub_get_node_path (struct grub_f2fs_inode *inode, grub_uint32_t block,
   grub_uint32_t dindirect_blks = indirect_blks * NIDS_PER_BLOCK;
   grub_uint32_t direct_index = DEF_ADDRS_PER_INODE;
   int n = 0;
-  int level = 0;
+  int level = -1;
 
   if (inode->i_inline & F2FS_INLINE_XATTR)
     direct_index -= F2FS_INLINE_XATTR_ADDRS;
@@ -712,6 +712,7 @@ grub_get_node_path (struct grub_f2fs_inode *inode, grub_uint32_t block,
   if (block < direct_index)
     {
       offset[n] = block;
+      level = 0;
       goto got;
     }
 
@@ -860,6 +861,10 @@ grub_f2fs_get_block (grub_fshelp_node_t node, grub_disk_addr_t block_ofs)
   int level, i;
 
   level = grub_get_node_path (inode, block_ofs, offset, noffset);
+
+  if (level < 0)
+    return -1;
+
   if (level == 0)
     return grub_le_to_cpu32 (inode->i_addr[offset[0]]);
 
--

2.16.4

#3 lore raw
Quoted 4 lines · Neil MacLeod

Hi Daniel & Michael

I've tested the new patch (in place of my attempt) and grub is building successfully so it looks good here!

Thanks! I allowed myself to add your Tested-by. I hope this is not a problem for you.

Michael, thank you for posting the patch.

This and other fixes posted earlier are now in the tree.

Daniel

#4 lore raw
Quoted 1 line

Thanks! I allowed myself to add your Tested-by. I hope this is not a problem for you

Not at all!

Quoted 1 line

This and other fixes posted earlier are now in the tree.

Latest grub HEAD (53e70d30cf0d18e6c28bab0ab8d223a90d3e1b46) continues to fail when building f2fs.c with gcc-9.1 due to a packed member issue
- Adrian (John? - sorry!) proposed a fix[1] and Michael confirmed it is correct, but that fix doesn't yet appear to be in the master tree, any ideas?

Thanks
Neil

1. http://lists.gnu.org/archive/html/grub-devel/2019-05/msg00096.html

Quoted 15 lines · Daniel Kiper
Quoted 4 lines · Neil MacLeod

Hi Daniel & Michael

I've tested the new patch (in place of my attempt) and grub is building successfully so it looks good here!

Thanks! I allowed myself to add your Tested-by. I hope this is not a problem for you.

Michael, thank you for posting the patch.

This and other fixes posted earlier are now in the tree.

Daniel

#5 lore raw
Quoted 12 lines · Neil MacLeod
Quoted 1 line

Thanks! I allowed myself to add your Tested-by. I hope this is not a problem for you

Not at all!

Quoted 1 line

This and other fixes posted earlier are now in the tree.

Latest grub HEAD (53e70d30cf0d18e6c28bab0ab8d223a90d3e1b46) continues to fail when building f2fs.c with gcc-9.1 due to a packed member issue
- Adrian (John? - sorry!) proposed a fix[1] and Michael confirmed it is correct, but that fix doesn't yet appear to be in the master tree, any ideas?

Quoted 4 lines

Thanks
Neil

1. http://lists.gnu.org/archive/html/grub-devel/2019-05/msg00096.html

Yes, we need the patch from John to solve reported gcc-9 build issue entirely. From the latest update he will do soon. :)

http://lists.gnu.org/archive/html/grub-devel/2019-05/msg00099.html

Thanks,
Michael

Quoted 17 lines · Daniel Kiper
Quoted 15 lines · Neil MacLeod
Quoted 4 lines

Hi Daniel & Michael

I've tested the new patch (in place of my attempt) and grub is building successfully so it looks good here!

Thanks! I allowed myself to add your Tested-by. I hope this is not a problem for you.

Michael, thank you for posting the patch.

This and other fixes posted earlier are now in the tree.

Daniel

#6 lore raw
Quoted 16 lines · Michael Chang
Quoted 15 lines · Neil MacLeod
Quoted 1 line

Thanks! I allowed myself to add your Tested-by. I hope this is not a problem for you

Not at all!

Quoted 1 line

This and other fixes posted earlier are now in the tree.

Latest grub HEAD (53e70d30cf0d18e6c28bab0ab8d223a90d3e1b46) continues to fail when building f2fs.c with gcc-9.1 due to a packed member issue
- Adrian (John? - sorry!) proposed a fix[1] and Michael confirmed it is correct, but that fix doesn't yet appear to be in the master tree, any ideas?

Thanks
Neil

Quoted 6 lines
Quoted 1 line

1. http://lists.gnu.org/archive/html/grub-devel/2019-05/msg00096.html

Yes, we need the patch from John to solve reported gcc-9 build issue entirely. From the latest update he will do soon. :)

http://lists.gnu.org/archive/html/grub-devel/2019-05/msg00099.html

John, ping? Could you send us the patch soon? IMO this is the last thing which blocks the release.

Daniel

#7 lore raw
Quoted 4 lines · Daniel Kiper
Quoted 1 line

http://lists.gnu.org/archive/html/grub-devel/2019-05/msg00099.html

John, ping? Could you send us the patch soon? IMO this is the last thing which blocks the release.

Just a second, I will send the patch today. Sorry for the delay.

Adrian

Signature · 4 lines
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - [email protected]
`. `'   Freie Universitaet Berlin - [email protected]
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913
#8 lore raw
In reply to Daniel Kiper
Quoted 7 lines · Daniel Kiper
Quoted 4 lines

Yes, we need the patch from John to solve reported gcc-9 build issue entirely. From the latest update he will do soon. :)

http://lists.gnu.org/archive/html/grub-devel/2019-05/msg00099.html

John, ping? Could you send us the patch soon? IMO this is the last thing which blocks the release.

Working on this now. Just need to perform a test build with gcc-9 now.

Adrian

Signature · 4 lines
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - [email protected]
`. `'   Freie Universitaet Berlin - [email protected]
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913