Refactor file_systems to use the kernel's list
cover Refactor file_systems to use the kernel's list
struct file_system_type defines a next field which is used to link the file_system_type structs registered to the kernel. This patch set replaces the propietry linked list implementation with the kernel's list.h.
This change makes clearer interfaces and code (e.g. the change in register_filesystem and find_filesystem), and eliminates unnecessary usage of * and & operators.
Tested by comparing the lists in /proc/filesystems.
Carmeli Tamir (2):
Use list.h instead of file_system_type next
Changed unsigned param type to unsigned int
fs/filesystems.c | 69 ++++++++++++++++++++++++---------------------- include/linux/fs.h | 2 +- 2 files changed, 37 insertions(+), 34 deletions(-)
Signature · 1 line
2.19.1
PATCH 1/2 Use list.h instead of file_system_type next
From: Tamir <[email protected]>
Changed file_system_type next field to list_head and refactored the code to use list.h functions.
--- fs/filesystems.c | 68 ++++++++++++++++++++++++---------------------- include/linux/fs.h | 2 +- 2 files changed, 36 insertions(+), 34 deletions(-)
fs/filesystems.c +35 -33
@@ -31,7 +31,7 @@* Once the reference is obtained we can drop the spinlock. */ -static struct file_system_type *file_systems; +static LIST_HEAD (file_systems);static DEFINE_RWLOCK (file_systems_lock);/* WARNING: This can be used only if we _already_ own a reference */
@@ -46,14 +46,16 @@ void put_filesystem(struct file_system_type *fs)module_put (fs->owner); } -static struct file_system_type **find_filesystem (const char *name,unsigned len) +static struct file_system_type *find_filesystem (const char *name,unsigned len) { -struct file_system_type **p; -for (p = &file_systems; *p; p = &(*p)->next) -if (strncmp ((*p)->name, name, len) ==0 && - !(*p)->name[len]) -break ; -return p; +struct file_system_type *p; + +list_for_each_entry (p, &file_systems, fs_types) { +if (strncmp (p->name, name, len) ==0 && + !p->name[len]) +return p; + } +return NULL ; }/**
@@ -72,20 +74,21 @@ static struct file_system_type **find_filesystem(const char *name, unsigned len)int register_filesystem (struct file_system_type * fs) {int res =0 ; -struct file_system_type ** p; +struct file_system_type *p;if (fs->parameters && !fs_validate_description (fs->parameters))return -EINVAL ;BUG_ON (strchr (fs->name,'.' )); -if (fs->next) -return -EBUSY ; + +INIT_LIST_HEAD (&fs->fs_types); +write_lock (&file_systems_lock); p =find_filesystem (fs->name,strlen (fs->name)); -if (*p) +if (p) res = -EBUSY ;else - *p = fs; +list_add_tail (&fs->fs_types, &file_systems);write_unlock (&file_systems_lock);return res; }
@@ -106,19 +109,16 @@ EXPORT_SYMBOL(register_filesystem);int unregister_filesystem (struct file_system_type * fs) { -struct file_system_type ** tmp; +struct file_system_type *tmp;write_lock (&file_systems_lock); - tmp = &file_systems; -while (*tmp) { -if (fs == *tmp) { - *tmp = fs->next; - fs->next =NULL ; +list_for_each_entry (tmp, &file_systems, fs_types) { +if (fs == tmp) { +list_del (&tmp->fs_types);write_unlock (&file_systems_lock);synchronize_rcu ();return 0 ; } - tmp = &(*tmp)->next; }write_unlock (&file_systems_lock);
@@ -141,7 +141,8 @@ static int fs_index(const char __user * __name) err = -EINVAL ;read_lock (&file_systems_lock); -for (tmp=file_systems, index=0 ; tmp ; tmp=tmp->next, index++) { +list_for_each_entry (tmp, &file_systems, fs_types) { + index++;if (strcmp (tmp->name, name->name) ==0 ) { err = index;break ;
@@ -158,9 +159,11 @@ static int fs_name(unsigned int index, char __user * buf)int len, res;read_lock (&file_systems_lock); -for (tmp = file_systems; tmp; tmp = tmp->next, index--) +list_for_each_entry (tmp, &file_systems, fs_types) { + index--;if (index <=0 &&try_module_get (tmp->owner))break ; + }read_unlock (&file_systems_lock);if (!tmp)return -EINVAL ;
@@ -174,12 +177,13 @@ static int fs_name(unsigned int index, char __user * buf)static int fs_maxindex (void ) { -struct file_system_type * tmp; -int index; +struct list_head *pos; +int index =0 ;read_lock (&file_systems_lock); -for (tmp = file_systems, index =0 ; tmp ; tmp = tmp->next, index++) - ; +list_for_each (pos, &file_systems) { + index++; + }read_unlock (&file_systems_lock);return index; }
@@ -214,12 +218,12 @@ int __init get_filesystem_list(char *buf)struct file_system_type * tmp;read_lock (&file_systems_lock); - tmp = file_systems; -while (tmp && len <PAGE_SIZE -80 ) { +list_for_each_entry (tmp, &file_systems, fs_types) { +if (len >=PAGE_SIZE -80 ) +break ; len +=sprintf (buf+len,"%s\t%s\n" , (tmp->fs_flags &FS_REQUIRES_DEV ) ?"" :"nodev" , tmp->name); - tmp = tmp->next; }read_unlock (&file_systems_lock);return len;
@@ -231,12 +235,10 @@ static int filesystems_proc_show(struct seq_file *m, void *v)struct file_system_type * tmp;read_lock (&file_systems_lock); - tmp = file_systems; -while (tmp) { +list_for_each_entry (tmp, &file_systems, fs_types) {seq_printf (m,"%s\t%s\n" , (tmp->fs_flags &FS_REQUIRES_DEV ) ?"" :"nodev" , tmp->name); - tmp = tmp->next; }read_unlock (&file_systems_lock);return 0 ;
@@ -255,7 +257,7 @@ static struct file_system_type *__get_fs_type(const char *name, int len)struct file_system_type *fs;read_lock (&file_systems_lock); - fs = *(find_filesystem (name, len)); + fs =find_filesystem (name, len);if (fs && !try_module_get (fs->owner)) fs =NULL ;read_unlock (&file_systems_lock);
include/linux/fs.h +1 -1
@@ -2182,7 +2182,7 @@ struct file_system_type { const char *, void *);void (*kill_sb) (struct super_block *);struct module *owner; -struct file_system_type * next; +struct list_head fs_types;/* All registered file_system_type structs */ struct hlist_head fs_supers;struct lock_class_key s_lock_key;
Signature · 1 line
2.19.1
Al Viro
maintainerRe: [PATCH 1/2] Use list.h instead of file_system_type next
Quoted 4 lines · Carmeli Tamir
From: Tamir <[email protected]>
Changed
file_system_typenext field tolist_headand refactored the code to uselist.hfunctions.
... except that list_head is not a good match here. For one thing, we never walk that thing backwards. For another, filesystem can be used without ever going through register_filesystem(), making your data structure quite a mess - e.g. use of list_empty() (a perfectly normal list.h primitive) on it might oops on some instances.
IOW, what you are making is not quite list_head and pretending it to be list_head is asking for serious headache down the road.
Frankly, what's the point? Reusing an existing data type, to avoid DIY is generally a good advice, but then you'd better make sure that existing type *does* fit your needs and that your creation is playing by that type's rules.
This patch does neither...
Re: [PATCH 1/2] Use list.h instead of file_system_type next
In reply to Carmeli TamirQuoted 2 lines · Carmeli Tamir
Changed
file_system_typenext field tolist_headand refactored the code to uselist.hfunctions.
What might be interesting is getting rid of this list and using an XArray instead. This would be a more in-depth change; getting rid of the rwlock in favour of using RCU accesses for the read-side and the xa_lock for write accesses to the filesystem list.
Re: [PATCH 1/2] Use list.h instead of file_system_type next
I just found it weird that there is a proprietary implementation of a linked list while surely the kernel already offers well established data structures.
IMO, the current code is a bit hard to understand, especially the addition of a new struct to the list in the line "*p = fs" after find_filesystem returned the last member.
Correct, I'm not familiar with all the use cases of the code.
I'm not sure that XArray is a good choice since there is no notion of an index attached to the pointer, it's really just a linked list of pointers.
Tell me if you think there is any way to improve my suggestion. Thanks for you attention.
Quoted 9 lines · Matthew Wilcox
Quoted 2 lines · Carmeli Tamir
Changed
file_system_typenext field tolist_headand refactored the code to uselist.hfunctions.What might be interesting is getting rid of this list and using an XArray instead. This would be a more in-depth change; getting rid of the rwlock in favour of using RCU accesses for the read-side and the
xa_lockfor write accesses to the filesystem list.
Re: [PATCH 1/2] Use list.h instead of file_system_type next
Quoted 3 lines · Tamir Carmeli
I just found it weird that there is a proprietary implementation of a linked list while surely the kernel already offers well established data structures.
It's a singly linked list rather than a doubly linked list.
Quoted 4 lines
IMO, the current code is a bit hard to understand, especially the addition of a new struct to the list in the line "*p = fs" after
find_filesystemreturned the last member.
Correct, I'm not familiar with all the use cases of the code.
It looks like a fairly standard implementation of a singly-linked list in C to me.
Quoted 3 lines
I'm not sure that XArray is a good choice since there is no notion of an index attached to the pointer, it's really just a linked list of pointers.
You don't need to attach an index to the pointer; you can just use xa_alloc() to store it at the first available index.
kbuild test robot
build botRe: [PATCH 1/2] Use list.h instead of file_system_type next
In reply to Carmeli TamirThank you for the patch! Yet something to improve:
Hi Carmeli,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on linus/master]
[also build test ERROR on v5.1 next-20190506]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Carmeli-Tamir/Use-list-h-instead-of-file_system_type-next/20190506-214032
config: riscv-allyesconfig (attached as .config)
compiler: riscv64-linux-gcc (GCC) 8.1.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=8.1.0 make.cross ARCH=riscv
If you fix the issue, kindly add following tag
All errors (new ones prefixed by >>):
>> fs/ocfs2/super.c:1250:3: error: 'struct file_system_type' has no member named 'next'
.next = NULL
^~~~
vim +1250 fs/ocfs2/super.c
ccd979bdb Mark Fasheh 2005-12-15 1243
ccd979bdb Mark Fasheh 2005-12-15 1244 static struct file_system_type ocfs2_fs_type = {
ccd979bdb Mark Fasheh 2005-12-15 1245 .owner = THIS_MODULE,
ccd979bdb Mark Fasheh 2005-12-15 1246 .name = "ocfs2",
152a08366 Al Viro 2010-07-25 1247 .mount = ocfs2_mount,
8ed6b2370 Goldwyn Rodrigues 2014-04-03 1248 .kill_sb = kill_block_super,
1ba9da2ff Mark Fasheh 2006-09-08 1249 .fs_flags = FS_REQUIRES_DEV|FS_RENAME_DOES_D_MOVE,
ccd979bdb Mark Fasheh 2005-12-15 @1250 .next = NULL
ccd979bdb Mark Fasheh 2005-12-15 1251 };
914177054 Eric W. Biederman 2013-03-07 1252 MODULE_ALIAS_FS("ocfs2");ccd979bdb Mark Fasheh 2005-12-15 1253
:::::: The code at line 1250 was first introduced by commit
:::::: ccd979bdbce9fba8412beb3f1de68a9d0171b12c [PATCH] OCFS2: The Second Oracle Cluster Filesystem
:::::: TO: Mark Fasheh <[email protected]> :::::: CC: Joel Becker <[email protected]>
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
kbuild test robot
build botRe: [PATCH 1/2] Use list.h instead of file_system_type next
In reply to Carmeli TamirThank you for the patch! Yet something to improve:
Hi Carmeli,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on linus/master]
[also build test ERROR on v5.1 next-20190506]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Carmeli-Tamir/Use-list-h-instead-of-file_system_type-next/20190506-214032
config: i386-randconfig-i2-201918 (attached as .config)
compiler: gcc-5 (Debian 5.5.0-3) 5.4.1 20171010 reproduce:
# save the attached .config to linux build tree
make ARCH=i386 If you fix the issue, kindly add following tag
All errors (new ones prefixed by >>):
>> fs/ocfs2/super.c:1250:2: error: unknown field 'next' specified in initializer
.next = NULL
^
vim +/next +1250 fs/ocfs2/super.c
ccd979bdb Mark Fasheh 2005-12-15 1243
ccd979bdb Mark Fasheh 2005-12-15 1244 static struct file_system_type ocfs2_fs_type = {
ccd979bdb Mark Fasheh 2005-12-15 1245 .owner = THIS_MODULE,
ccd979bdb Mark Fasheh 2005-12-15 1246 .name = "ocfs2",
152a08366 Al Viro 2010-07-25 1247 .mount = ocfs2_mount,
8ed6b2370 Goldwyn Rodrigues 2014-04-03 1248 .kill_sb = kill_block_super,
1ba9da2ff Mark Fasheh 2006-09-08 1249 .fs_flags = FS_REQUIRES_DEV|FS_RENAME_DOES_D_MOVE,
ccd979bdb Mark Fasheh 2005-12-15 @1250 .next = NULL
ccd979bdb Mark Fasheh 2005-12-15 1251 };
914177054 Eric W. Biederman 2013-03-07 1252 MODULE_ALIAS_FS("ocfs2");ccd979bdb Mark Fasheh 2005-12-15 1253
:::::: The code at line 1250 was first introduced by commit
:::::: ccd979bdbce9fba8412beb3f1de68a9d0171b12c [PATCH] OCFS2: The Second Oracle Cluster Filesystem
:::::: TO: Mark Fasheh <[email protected]> :::::: CC: Joel Becker <[email protected]>
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
PATCH 2/2 Changed unsigned param type to unsigned int
In reply to Carmeli TamirFrom: Tamir <[email protected]>
Fixed a checkpatch warning for usage of unsigned type. Submitted as different patch in the series since it's not related to the change, just wanted to fix checkpatch warnings from it.
--- fs/filesystems.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
fs/filesystems.c +2 -1
@@ -46,7 +46,8 @@ void put_filesystem(struct file_system_type *fs)module_put (fs->owner); } -static struct file_system_type *find_filesystem (const char *name,unsigned len) +static struct file_system_type *find_filesystem (const char *name, +unsigned int len) {struct file_system_type *p;
Signature · 1 line
2.19.1