hfsplus: Change finder_info to u32
Subsystems (3)
PATCH v2 1/2 hfsplus: Change finder_info to u32
The finder_info block in the hfsplus volume header is currently defined as an array of 8 bit values, but TN1150 defines it as being an array of 32 bit values. Fix for convenience.
--- fs/hfsplus/hfsplus_raw.h | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
fs/hfsplus/hfsplus_raw.h +1 -1
@@ -117,7 +117,7 @@ struct hfsplus_vh {__be32 write_count;__be64 encodings_bmp; -u8 finder_info[32 ]; +u32 finder_info[8 ];struct hfsplus_fork_raw alloc_file;struct hfsplus_fork_raw ext_file;
Signature · 1 line
1.7.7.1
PATCH v2 2/2 hfsplus: Add an ioctl to bless files
Making an hfsplus partition bootable requires the ability to "bless" a file by putting its inode number in the volume header. Doing this from userspace on a mounted filesystem is impractical since the kernel will write back the original values on unmount. Add an ioctl to allow userspace to update the volume header information based on the target file.
---
V2 - remember to fix the endianness.
Documentation/ioctl/ioctl-number.txt | 1 + fs/hfsplus/hfsplus_fs.h | 1 + fs/hfsplus/ioctl.c | 22 ++++++++++++++++++++++ 3 files changed, 24 insertions(+), 0 deletions(-)
Documentation/ioctl/ioctl-number.txt +1 -0
@@ -213,6 +213,7 @@ Code Seq#(hex) Include File Comments 'f' 00-0F fs/ext4/ext4.h conflict! 'f' 00-0F linux/fs.h conflict! 'f' 00-0F fs/ocfs2/ocfs2_fs.h conflict! +'f' 20-2F fs/hfsplus/hfsplus_fs.h 'g' 00-0F linux/usb/gadgetfs.h 'g' 20-2F linux/usb/g_printer.h 'h' 00-7F conflict! Charon filesystem
fs/hfsplus/hfsplus_fs.h +1 -0
@@ -315,6 +315,7 @@ static inline unsigned short hfsplus_min_io_size(struct super_block *sb)#define HFSPLUS_IOC_EXT2_GETFLAGS FS_IOC_GETFLAGS#define HFSPLUS_IOC_EXT2_SETFLAGS FS_IOC_SETFLAGS +#define HFSPLUS_IOC_BLESS _IO('f', 0x20)/* * Functions in any *.c used in other files
fs/hfsplus/ioctl.c +22 -0
@@ -20,6 +20,26 @@#include <asm/uaccess.h> #include "hfsplus_fs.h" +static int hfsplus_ioctl_bless (struct file *file,int __user *user_flags) +{ +struct inode *inode = file->f_dentry->d_inode; +struct hfsplus_sb_info *sbi =HFSPLUS_SB (inode->i_sb); +struct hfsplus_vh *vh = sbi->s_vhdr; +struct hfsplus_vh *bvh = sbi->s_backup_vhdr; + +if (!capable (CAP_SYS_ADMIN )) +return -EPERM ; + +mutex_lock (&sbi->vh_mutex); + vh->finder_info[0 ] = bvh->finder_info[0 ] = +cpu_to_be32 (parent_ino (file->f_dentry)); + vh->finder_info[1 ] = bvh->finder_info[1 ] =cpu_to_be32 (inode->i_ino); + vh->finder_info[5 ] = bvh->finder_info[5 ] = +cpu_to_be32 (parent_ino (file->f_dentry)); +mutex_unlock (&sbi->vh_mutex); +return 0 ; +} +static int hfsplus_ioctl_getflags (struct file *file,int __user *user_flags) {struct inode *inode = file->f_path.dentry->d_inode;
@@ -108,6 +128,8 @@ long hfsplus_ioctl(struct file *file, unsigned int cmd, unsigned long arg)return hfsplus_ioctl_getflags (file, argp);case HFSPLUS_IOC_EXT2_SETFLAGS :return hfsplus_ioctl_setflags (file, argp); +case HFSPLUS_IOC_BLESS : +return hfsplus_ioctl_bless (file, argp);default :return -ENOTTY ; }
Signature · 1 line
1.7.7.1
Re: [PATCH V2 2/2] hfsplus: Add an ioctl to bless files
+#define HFSPLUS_IOC_BLESS _IO('f', 0x20)
I'd probably move this to fs.h and follow the numbering there, otherwise we are bound to run into conflicts.
Quoted 1 line
+static int
hfsplus_ioctl_bless(struct file *file, int __user *user_flags)
Care to add a little comment on what this ioctl does, bless is a bit of a generic name.
fs/hfsplus/ioctl.c line 25
Quoted 2 lines
fs/hfsplus/ioctl.c +2
-0+{ +struct inode *inode = file->f_dentry->d_inode;
Please use file->f_path.dentry instead of the f_dentry define, please.
Also given that we use the dentry almost as often I'd pull it into a local variable as well.
fs/hfsplus/ioctl.c line 38
Quoted 6 lines
fs/hfsplus/ioctl.c +6
-0+mutex_lock (&sbi->vh_mutex); + vh->finder_info[0 ] = bvh->finder_info[0 ] = +cpu_to_be32 (parent_ino (file->f_dentry)); + vh->finder_info[1 ] = bvh->finder_info[1 ] =cpu_to_be32 (inode->i_ino); + vh->finder_info[5 ] = bvh->finder_info[5 ] = +cpu_to_be32 (parent_ino (file->f_dentry));
Any idea why we write the parent twice?
Not directly relevant, but where do you plan to put the userspace to call this ioctl?
Re: [PATCH V2 2/2] hfsplus: Add an ioctl to bless files
Quoted 4 lines · Christoph Hellwig
+#define
HFSPLUS_IOC_BLESS_IO('f', 0x20)I'd probably move this to
fs.hand follow the numbering there, otherwise we are bound to run into conflicts.
Ok. Any problem with leaving something filesystem specific in there?
Quoted 4 lines
Quoted 1 line
+static int
hfsplus_ioctl_bless(struct file *file, int __user *user_flags)Care to add a little comment on what this ioctl does, bless is a bit of a generic name.
Will do.
Quoted 4 lines
fs/hfsplus/ioctl.c line 25
Quoted 2 lines
fs/hfsplus/ioctl.c +2
-0+{ +struct inode *inode = file->f_dentry->d_inode;Please use
file->f_path.dentryinstead of thef_dentrydefine, please.
Ok.
Quoted 10 lines
fs/hfsplus/ioctl.c line 38
Quoted 6 lines
fs/hfsplus/ioctl.c +6
-0+mutex_lock (&sbi->vh_mutex); + vh->finder_info[0 ] = bvh->finder_info[0 ] = +cpu_to_be32 (parent_ino (file->f_dentry)); + vh->finder_info[1 ] = bvh->finder_info[1 ] =cpu_to_be32 (inode->i_ino); + vh->finder_info[5 ] = bvh->finder_info[5 ] = +cpu_to_be32 (parent_ino (file->f_dentry));Any idea why we write the parent twice?
From the spec:
"finderInfo[0] contains the directory ID of the directory containing the bootable system (for example, the System Folder in Mac OS 8 or 9, or /System/Library/CoreServices in Mac OS X). It is zero if there is no bootable system on the volume. This value is typically equal to either finderInfo[3] or finderInfo[5]."
finderInfo[3] is the OS8/9 system folder, finderInfo[5] is the OS X one. I'd guess it's just to indicate whether the boot media should default to OS8/9 or OS X, but it's not entirely clear.
Quoted 2 lines
Not directly relevant, but where do you plan to put the userspace to call this ioctl?
There's a git repo at git://cavan.codon.org.uk/mactel-boot
Signature · 1 line
Matthew Garrett | [email protected]
Re: [PATCH V2 2/2] hfsplus: Add an ioctl to bless files
Quoted 7 lines · Matthew Garrett
Quoted 4 lines · Christoph Hellwig
+#define
HFSPLUS_IOC_BLESS_IO('f', 0x20)I'd probably move this to
fs.hand follow the numbering there, otherwise we are bound to run into conflicts.Ok. Any problem with leaving something filesystem specific in there?
Leaving it in hfsplus sounds fine, but I'd avoid using 'f' then just to reduce the chance for overlap.
Quoted 11 lines
Quoted 1 line
From the spec:
"finderInfo[0] contains the directory ID of the directory containing the bootable system (for example, the System Folder in Mac OS 8 or 9, or /System/Library/CoreServices in Mac OS X). It is zero if there is no bootable system on the volume. This value is typically equal to either finderInfo[3] or finderInfo[5]."
finderInfo[3] is the OS8/9 system folder, finderInfo[5] is the OS X one. I'd guess it's just to indicate whether the boot media should default to OS8/9 or OS X, but it's not entirely clear.
Care to add this explanation as a comment?
Re: [PATCH V2 2/2] hfsplus: Add an ioctl to bless files
Quoted 11 lines · Christoph Hellwig
Quoted 7 lines · Matthew Garrett
Quoted 4 lines · Christoph Hellwig
+#define
HFSPLUS_IOC_BLESS_IO('f', 0x20)I'd probably move this to
fs.hand follow the numbering there, otherwise we are bound to run into conflicts.Ok. Any problem with leaving something filesystem specific in there?
Leaving it in hfsplus sounds fine, but I'd avoid using 'f' then just to reduce the chance for overlap.
Using the same range but putting it in a different file seems like a good way to have someone miss it. I added the range to ioctl-list.txt so in theory nobody should step on it - do you think that's likely to be inadequate?
Signature · 1 line
Matthew Garrett | [email protected]
PATCH v2 1/1 hfsplus: Add an ioctl to bless files
In reply to Christoph HellwigMaking an hfsplus partition bootable requires the ability to "bless" a file by putting its inode number in the volume header. Doing this from userspace on a mounted filesystem is impractical since the kernel will write back the original values on unmount. Add an ioctl to allow userspace to update the volume header information based on the target file.
---
Kept the ioctl in the hfs code, but moved it to a different range to reduce reduce the chances of someone stepping on it with another filesystem.
Documentation/ioctl/ioctl-number.txt | 1 + fs/hfsplus/hfsplus_fs.h | 5 +++++ fs/hfsplus/ioctl.c | 34 ++++++++++++++++++++++++++++++++++ 3 files changed, 40 insertions(+), 0 deletions(-)
Documentation/ioctl/ioctl-number.txt +1 -0
@@ -218,6 +218,7 @@ Code Seq#(hex) Include File Comments 'h' 00-7F conflict! Charon filesystem <mailto:[email protected]> 'h' 00-1F linux/hpet.h conflict! +'h' 80-8F fs/hfsplus/ioctl.c 'i' 00-3F linux/i2o-dev.h conflict! 'i' 0B-1F linux/ipmi.h conflict! 'i' 80-8F linux/i8k.h
fs/hfsplus/hfsplus_fs.h +5 -0
@@ -317,6 +317,11 @@ static inline unsigned short hfsplus_min_io_size(struct super_block *sb)/* +* hfs+-specific ioctl for making the filesystem bootable +*/ +#define HFSPLUS_IOC_BLESS _IO('h', 0x80) + +/* * Functions in any *.c used in other files */
fs/hfsplus/ioctl.c +34 -0
@@ -20,6 +20,38 @@#include <asm/uaccess.h> #include "hfsplus_fs.h" +/* +* "Blessing" an HFS+ filesystem writes metadata to the superblock informing +* the platform firmware which file to boot from +*/ +static int hfsplus_ioctl_bless (struct file *file,int __user *user_flags) +{ +struct dentry *dentry = file->f_path.dentry; +struct inode *inode = dentry->d_inode; +struct hfsplus_sb_info *sbi =HFSPLUS_SB (inode->i_sb); +struct hfsplus_vh *vh = sbi->s_vhdr; +struct hfsplus_vh *bvh = sbi->s_backup_vhdr; + +if (!capable (CAP_SYS_ADMIN )) +return -EPERM ; + +mutex_lock (&sbi->vh_mutex); + +/* Directory containing the bootable system */ + vh->finder_info[0 ] = bvh->finder_info[0 ] = +cpu_to_be32 (parent_ino (dentry)); + +/* Bootloader */ + vh->finder_info[1 ] = bvh->finder_info[1 ] =cpu_to_be32 (inode->i_ino); + +/* Per spec, the OS X system folder - same as finder_info[0] here */ + vh->finder_info[5 ] = bvh->finder_info[5 ] = +cpu_to_be32 (parent_ino (dentry)); + +mutex_unlock (&sbi->vh_mutex); +return 0 ; +} +static int hfsplus_ioctl_getflags (struct file *file,int __user *user_flags) {struct inode *inode = file->f_path.dentry->d_inode;
@@ -108,6 +140,8 @@ long hfsplus_ioctl(struct file *file, unsigned int cmd, unsigned long arg)return hfsplus_ioctl_getflags (file, argp);case HFSPLUS_IOC_EXT2_SETFLAGS :return hfsplus_ioctl_setflags (file, argp); +case HFSPLUS_IOC_BLESS : +return hfsplus_ioctl_bless (file, argp);default :return -ENOTTY ; }
Signature · 1 line
1.7.7.6
Re: [PATCH V2] hfsplus: Add an ioctl to bless files
Quoted 7 lines · Matthew Garrett
Making an hfsplus partition bootable requires the ability to "bless" a file by putting its inode number in the volume header. Doing this from userspace on a mounted filesystem is impractical since the kernel will write back the original values on unmount. Add an ioctl to allow userspace to update the volume header information based on the target file.
Signed-off-by Matthew Garrett <[email protected]>
Looks good, thanks.
Re: [PATCH V2] hfsplus: Add an ioctl to bless files
In reply to Matthew GarrettQuoted 8 lines · Matthew Garrett
Making an hfsplus partition bootable requires the ability to "bless" a file by putting its inode number in the volume header. Doing this from userspace on a mounted filesystem is impractical since the kernel will write back the original values on unmount. Add an ioctl to allow userspace to update the volume header information based on the target file.
Signed-off-by Matthew Garrett <[email protected]>---
Works nicely, great stuff.
Thanks,
Henrik
Quoted 5 lines · Matthew Garrett
The
finder_infoblock in the hfsplus volume header is currently defined as an array of 8 bit values, but TN1150 defines it as being an array of 32 bit values. Fix for convenience.Signed-off-by Matthew Garrett <[email protected]>
Looks good, I'll queue it up.