Skip to contentSkip to conversation

sysv: use BUILD_BUG_ON instead of runtime check

PATCH Applied Pavel Skripkin 1 participant 3 messages Aug 13, 2021 – Sep 27, 2021

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 sysv: use BUILD_BUG_ON instead of runtime check

There was runtime checks about sizes of struct v7_super_block and struct sysv_inode. If one of these checks fail kernel will panic. Since these values are known on complite time let's use BUILD_BUG_ON(), because it's standard mechanism for validation checking at build time

Fixes 1da177e4c3f4 ("Linux-2.6.12-rc2")
---
 fs/sysv/super.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
fs/sysv/super.c +2 -4
index cc8e2ed155c8..d1def0771a40 100644
@@ -474,10 +474,8 @@ static int v7_fill_super(struct super_block *sb, void *data, int silent)
 	struct sysv_sb_info *sbi;
 	struct buffer_head *bh;
 
-	if (440 != sizeof (struct v7_super_block))
-		panic("V7 FS: bad super-block size");
-	if (64 != sizeof (struct sysv_inode))
-		panic("sysv fs: bad i-node size");
+	BUILD_BUG_ON(sizeof(struct v7_super_block) != 440);
+	BUILD_BUG_ON(sizeof(struct sysv_inode) != 64);
 
 	sbi = kzalloc(sizeof(struct sysv_sb_info), GFP_KERNEL);
 	if (!sbi)
Signature · 1 line
2.32.0
#1 lore raw
Quoted 19 lines · Pavel Skripkin

There was runtime checks about sizes of struct v7_super_block and struct sysv_inode. If one of these checks fail kernel will panic. Since these values are known on complite time let's use BUILD_BUG_ON(), because it's standard mechanism for validation checking at build time

Fixes 1da177e4c3f4 ("Linux-2.6.12-rc2")
---
  fs/sysv/super.c | 6 ++----
  1 file changed, 2 insertions(+), 4 deletions(-)
fs/sysv/super.c +0 -0
index cc8e2ed155c8..d1def0771a40 100644
@@ -474,10 +474,8 @@ static int v7_fill_super(struct super_block *sb, void *data, int silent)
  	struct sysv_sb_info *sbi;
  	struct buffer_head *bh;

fs/sysv/super.c static int v7_fill_super(struct super_block *sb, void *data, int silent) · line 481

Quoted 10 lines
fs/sysv/super.c +2 -4
 
-	if (440 != sizeof (struct v7_super_block))
-		panic("V7 FS: bad super-block size");
-	if (64 != sizeof (struct sysv_inode))
-		panic("sysv fs: bad i-node size");
+	BUILD_BUG_ON(sizeof(struct v7_super_block) != 440);
+	BUILD_BUG_ON(sizeof(struct sysv_inode) != 64);
 
  	sbi = kzalloc(sizeof(struct sysv_sb_info), GFP_KERNEL);
  	if (!sbi)

Hi, David and Alexander!

Maybe, this one can go through one of your trees? I didn't find any sysv related trees, so..

Lore link to the patch: https://lore.kernel.org/lkml/[email protected]/

Thank you!

With regards,
Pavel Skripkin

#2 lore raw
Quoted 37 lines · Pavel Skripkin
Quoted 19 lines · Pavel Skripkin

There was runtime checks about sizes of struct v7_super_block and struct sysv_inode. If one of these checks fail kernel will panic. Since these values are known on complite time let's use BUILD_BUG_ON(), because it's standard mechanism for validation checking at build time

Fixes 1da177e4c3f4 ("Linux-2.6.12-rc2")
---
  fs/sysv/super.c | 6 ++----
  1 file changed, 2 insertions(+), 4 deletions(-)
fs/sysv/super.c +0 -0
index cc8e2ed155c8..d1def0771a40 100644
@@ -474,10 +474,8 @@ static int v7_fill_super(struct super_block *sb, void *data, int silent)
  	struct sysv_sb_info *sbi;
  	struct buffer_head *bh;

fs/sysv/super.c static int v7_fill_super(struct super_block *sb, void *data, int silent) · line 481

Quoted 10 lines
fs/sysv/super.c +2 -4
 
-	if (440 != sizeof (struct v7_super_block))
-		panic("V7 FS: bad super-block size");
-	if (64 != sizeof (struct sysv_inode))
-		panic("sysv fs: bad i-node size");
+	BUILD_BUG_ON(sizeof(struct v7_super_block) != 440);
+	BUILD_BUG_ON(sizeof(struct sysv_inode) != 64);
 
  	sbi = kzalloc(sizeof(struct sysv_sb_info), GFP_KERNEL);
  	if (!sbi)

Hi, David and Alexander!

Maybe, this one can go through one of your trees? I didn't find any sysv related trees, so..

Quoted 6 lines

Lore link to the patch:
https://lore.kernel.org/lkml/[email protected]/

Thank you!

Hi, Andrew,

Maybe this one can go through your tree?

With regards,
Pavel Skripkin