Skip to contentSkip to conversation

staging: erofs: using switch-case while checking the inode type.

PATCH In review Pratik Shinde 4 participants 8 messages Aug 29, 2019 Staging

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 staging: erofs: using switch-case while checking the inode type.

while filling the linux inode, using switch-case statement to check the type of inode.
switch-case statement looks more clean.

---
 drivers/staging/erofs/inode.c | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)
drivers/staging/erofs/inode.c +12 -6
index 4c3d8bf..2d2d545 100644
@@ -190,22 +190,28 @@ static int fill_inode(struct inode *inode, int isdir)
 	err = read_inode(inode, data + ofs);
 	if (!err) {
 		/* setup the new inode */
-		if (S_ISREG(inode->i_mode)) {
+		switch (inode->i_mode & S_IFMT) {
+		case S_IFREG:
 			inode->i_op = &erofs_generic_iops;
 			inode->i_fop = &generic_ro_fops;
-		} else if (S_ISDIR(inode->i_mode)) {
+			break;
+		case S_IFDIR:
 			inode->i_op = &erofs_dir_iops;
 			inode->i_fop = &erofs_dir_fops;
-		} else if (S_ISLNK(inode->i_mode)) {
+			break;
+		case S_IFLNK:
 			/* by default, page_get_link is used for symlink */
 			inode->i_op = &erofs_symlink_iops;
 			inode_nohighmem(inode);
-		} else if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode) ||
-			S_ISFIFO(inode->i_mode) || S_ISSOCK(inode->i_mode)) {
+			break;
+		case S_IFCHR:
+		case S_IFBLK:
+		case S_IFIFO:
+		case S_IFSOCK:
 			inode->i_op = &erofs_generic_iops;
 			init_special_inode(inode, inode->i_mode, inode->i_rdev);
 			goto out_unlock;
-		} else {
+		default:
 			err = -EIO;
 			goto out_unlock;
 		}
Signature · 1 line
2.9.3
#1 lore raw

Hi Pratik,

Quoted 5 lines · Pratik Shinde

while filling the linux inode, using switch-case statement to check the type of inode.
switch-case statement looks more clean.

No, that is not the case, see __ext4_iget() in fs/ext4/inode.c. and could you write patches based on latest staging tree? erofs is now in "fs/" rather than "drivers/staging". and I will review it then.

p.s. if someone argues here or there, there will be endless issues since there is no standard at all.

Thanks,
Gao Xiang

Quoted 46 lines
---
 drivers/staging/erofs/inode.c | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)
drivers/staging/erofs/inode.c +12 -6
index 4c3d8bf..2d2d545 100644
@@ -190,22 +190,28 @@ static int fill_inode(struct inode *inode, int isdir)
 	err = read_inode(inode, data + ofs);
 	if (!err) {
 		/* setup the new inode */
-		if (S_ISREG(inode->i_mode)) {
+		switch (inode->i_mode & S_IFMT) {
+		case S_IFREG:
 			inode->i_op = &erofs_generic_iops;
 			inode->i_fop = &generic_ro_fops;
-		} else if (S_ISDIR(inode->i_mode)) {
+			break;
+		case S_IFDIR:
 			inode->i_op = &erofs_dir_iops;
 			inode->i_fop = &erofs_dir_fops;
-		} else if (S_ISLNK(inode->i_mode)) {
+			break;
+		case S_IFLNK:
 			/* by default, page_get_link is used for symlink */
 			inode->i_op = &erofs_symlink_iops;
 			inode_nohighmem(inode);
-		} else if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode) ||
-			S_ISFIFO(inode->i_mode) || S_ISSOCK(inode->i_mode)) {
+			break;
+		case S_IFCHR:
+		case S_IFBLK:
+		case S_IFIFO:
+		case S_IFSOCK:
 			inode->i_op = &erofs_generic_iops;
 			init_special_inode(inode, inode->i_mode, inode->i_rdev);
 			goto out_unlock;
-		} else {
+		default:
 			err = -EIO;
 			goto out_unlock;
 		}
Signature · 1 line
2.9.3
#2 lore raw

Hi Gao,

Sorry I didn't pull the latest tree. I will do the necessary. Anyways, don't you think it will be cleaner to have a switch case statement rather than if-else statement.

--Pratik

Quoted 72 lines · Gao Xiang,

Hi Pratik,

Quoted 5 lines · Pratik Shinde

while filling the linux inode, using switch-case statement to check the type of inode.
switch-case statement looks more clean.

No, that is not the case, see __ext4_iget() in fs/ext4/inode.c. and could you write patches based on latest staging tree? erofs is now in "fs/" rather than "drivers/staging". and I will review it then.

p.s. if someone argues here or there, there will be endless issues since there is no standard at all.

Thanks,
Gao Xiang

Review

Quoted 5 lines
---
 drivers/staging/erofs/inode.c | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)
(quoted diff) +0 -0

b/drivers/staging/erofs/inode.c

drivers/staging/erofs/inode.c static int fill_inode(struct inode *inode, int

Quoted 4 lines

index 4c3d8bf..2d2d545 100644

drivers/staging/erofs/inode.c +0 -0
@@ -190,22 +190,28 @@ static int fill_inode(struct inode *inode, int

isdir)

Quoted 10 lines
drivers/staging/erofs/inode.c +3 -2
      err = read_inode(inode, data + ofs);
      if (!err) {
              /* setup the new inode */
-             if (S_ISREG(inode->i_mode)) {
+             switch (inode->i_mode & S_IFMT) {
+             case S_IFREG:
                      inode->i_op = &erofs_generic_iops;
                      inode->i_fop = &generic_ro_fops;
-             } else if (S_ISDIR(inode->i_mode)) {
+                     break;

drivers/staging/erofs/inode.c static int fill_inode(struct inode *inode, int isdir)

Quoted 10 lines
drivers/staging/erofs/inode.c +3 -2
+             case S_IFDIR:
                      inode->i_op = &erofs_dir_iops;
                      inode->i_fop = &erofs_dir_fops;
-             } else if (S_ISLNK(inode->i_mode)) {
+                     break;
+             case S_IFLNK:
                      /* by default, page_get_link is used for symlink */
                      inode->i_op = &erofs_symlink_iops;
                      inode_nohighmem(inode);
-             } else if (S_ISCHR(inode->i_mode) ||

S_ISBLK(inode->i_mode) ||

drivers/staging/erofs/inode.c static int fill_inode(struct inode *inode, int · line 204

Quoted 1 line
drivers/staging/erofs/inode.c +0 -1
-                     S_ISFIFO(inode->i_mode) ||

S_ISSOCK(inode->i_mode)) {

drivers/staging/erofs/inode.c static int fill_inode(struct inode *inode, int isdir)

Quoted 7 lines
drivers/staging/erofs/inode.c +5 -0
+                     break;
+             case S_IFCHR:
+             case S_IFBLK:
+             case S_IFIFO:
+             case S_IFSOCK:
                      inode->i_op = &erofs_generic_iops;
                      init_special_inode(inode, inode->i_mode,

inode->i_rdev);

Quoted 9 lines
drivers/staging/erofs/inode.c +1 -2
                      goto out_unlock;
-             } else {
+             default:
                      err = -EIO;
                      goto out_unlock;
              }
--

2.9.3

#3 lore raw
Quoted 5 lines · Pratik Shinde

Hi Gao,

Sorry I didn't pull the latest tree. I will do the necessary. Anyways, don't you think it will be cleaner to have a switch case statement rather than if-else statement.

I think so, but that's another personal choise and no urgent as well (It is just a cleanup to some extent).

I am very happy that you send a patch about this, but we have to take care of handling "fall through" properly at least, and I don't want to introduce some extra compile warnings instead at this time.

EROFS is sensitive for now and I have no idea what the "real" point is.

Thanks,
Gao Xiang

Quoted 79 lines

--Pratik

Quoted 72 lines · Gao Xiang,

Hi Pratik,

Quoted 5 lines · Pratik Shinde

while filling the linux inode, using switch-case statement to check the type of inode.
switch-case statement looks more clean.

No, that is not the case, see __ext4_iget() in fs/ext4/inode.c. and could you write patches based on latest staging tree? erofs is now in "fs/" rather than "drivers/staging". and I will review it then.

p.s. if someone argues here or there, there will be endless issues since there is no standard at all.

Thanks,
Gao Xiang

Review

Quoted 5 lines
---
 drivers/staging/erofs/inode.c | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)
(quoted diff) +0 -0

b/drivers/staging/erofs/inode.c

drivers/staging/erofs/inode.c static int fill_inode(struct inode *inode, int

Quoted 4 lines

index 4c3d8bf..2d2d545 100644

drivers/staging/erofs/inode.c +0 -0
@@ -190,22 +190,28 @@ static int fill_inode(struct inode *inode, int

isdir)

Quoted 10 lines
drivers/staging/erofs/inode.c +3 -2
      err = read_inode(inode, data + ofs);
      if (!err) {
              /* setup the new inode */
-             if (S_ISREG(inode->i_mode)) {
+             switch (inode->i_mode & S_IFMT) {
+             case S_IFREG:
                      inode->i_op = &erofs_generic_iops;
                      inode->i_fop = &generic_ro_fops;
-             } else if (S_ISDIR(inode->i_mode)) {
+                     break;

drivers/staging/erofs/inode.c static int fill_inode(struct inode *inode, int isdir)

Quoted 10 lines
drivers/staging/erofs/inode.c +3 -2
+             case S_IFDIR:
                      inode->i_op = &erofs_dir_iops;
                      inode->i_fop = &erofs_dir_fops;
-             } else if (S_ISLNK(inode->i_mode)) {
+                     break;
+             case S_IFLNK:
                      /* by default, page_get_link is used for symlink */
                      inode->i_op = &erofs_symlink_iops;
                      inode_nohighmem(inode);
-             } else if (S_ISCHR(inode->i_mode) ||

S_ISBLK(inode->i_mode) ||

drivers/staging/erofs/inode.c static int fill_inode(struct inode *inode, int · line 204

Quoted 1 line
drivers/staging/erofs/inode.c +0 -1
-                     S_ISFIFO(inode->i_mode) ||

S_ISSOCK(inode->i_mode)) {

drivers/staging/erofs/inode.c static int fill_inode(struct inode *inode, int isdir)

Quoted 7 lines
drivers/staging/erofs/inode.c +5 -0
+                     break;
+             case S_IFCHR:
+             case S_IFBLK:
+             case S_IFIFO:
+             case S_IFSOCK:
                      inode->i_op = &erofs_generic_iops;
                      init_special_inode(inode, inode->i_mode,

inode->i_rdev);

Quoted 9 lines
drivers/staging/erofs/inode.c +1 -2
                      goto out_unlock;
-             } else {
+             default:
                      err = -EIO;
                      goto out_unlock;
              }
--

2.9.3

#4 lore raw
Quoted 4 lines · Gao Xiang

I am very happy that you send a patch about this, but we have to take care of handling "fall through" properly at least, and I don't want to introduce some extra compile warnings instead at this time.

I can't apply the patch so I maybe missed something. I don't see a fall through issue. We have the code so you could use ^^^^^^^^ to indicate which lines have a fall through problem.

Quoted 3 lines

EROFS is sensitive for now and I have no idea what the "real" point is.

What does "sensitive" mean here? Now that it's out of staging we aren't applying clean up patches?

regards,
dan carpenter

#5 lore raw

Hi Dan,

Quoted 11 lines · Dan Carpenter
Quoted 4 lines · Gao Xiang

I am very happy that you send a patch about this, but we have to take care of handling "fall through" properly at least, and I don't want to introduce some extra compile warnings instead at this time.

I can't apply the patch so I maybe missed something. I don't see a fall through issue. We have the code so you could use ^^^^^^^^ to indicate which lines have a fall through problem.

Quoted 1 line
Quoted 5 lines
Quoted 2 lines

EROFS is sensitive for now and I have no idea what the "real" point is.

What does "sensitive" mean here? Now that it's out of staging we aren't applying clean up patches?

Of course not, I mean we should avoid "fall through" problem but I have no time to verify this patch since I am fixing what hch said as well.

Thanks,
Gao Xiang

Quoted 4 lines

regards,
dan carpenter

#6 lore raw

Hi Dan,

Quoted 19 lines · Gao Xiang

Hi Dan,

Quoted 16 lines · Dan Carpenter
Quoted 4 lines · Gao Xiang

I am very happy that you send a patch about this, but we have to take care of handling "fall through" properly at least, and I don't want to introduce some extra compile warnings instead at this time.

I can't apply the patch so I maybe missed something. I don't see a fall through issue. We have the code so you could use ^^^^^^^^ to indicate which lines have a fall through problem.

Quoted 3 lines

EROFS is sensitive for now and I have no idea what the "real" point is.

What does "sensitive" mean here? Now that it's out of staging we aren't applying clean up patches?

Again, due to language obstacle, I have to give a detailed explanation of what "sensitive" meant here.

I meant it seems all topic had no relationship with EROFS at all, but someone mentioned erofs in different topics, I have no idea what is wrong with EROFS, and I have no idea where it is like POS.

As your unlikely/likely concern, I think I discussed with you earlier, in fact, most of EROFS unlikely is due to error handling paths, which are meaningful in my thought.

If you argue that it has little performance difference, I think unlikely in IS_ERR can also be killed as well since for most use cases in Linux it is true that they are little performance difference at all. But I think totally it will have some impact.

Anyway, I didn't find some formal standard about this, and I remove all of these as you like.

Thanks,
Gao Xiang

Quoted 12 lines

Of course not, I mean we should avoid "fall through" problem but I have no time to verify this patch since I am fixing what hch said as well.

Thanks,
Gao Xiang

Quoted 4 lines

regards,
dan carpenter

#7 lore raw
In reply to Gao Xiang
Quoted 10 lines · Gao Xiang

Hi Pratik,

Quoted 5 lines · Pratik Shinde

while filling the linux inode, using switch-case statement to check the type of inode.
switch-case statement looks more clean.

No, that is not the case, see __ext4_iget() in fs/ext4/inode.c.

Unnecessary complaining.

Quoted 6 lines

and could you write patches based on latest staging tree? erofs is now in "fs/" rather than "drivers/staging". and I will review it then.

p.s. if someone argues here or there, there will be endless issues since there is no standard at all.

More complaining... It doesn't matter if you can find ext4 that looks like dog poop, that's irrelevant.

regards,
dan carpenter