From: Benny Halevy <bhalevy@primarydata.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>,
linux-fsdevel@vger.kernel.org,
NFS list <linux-nfs@vger.kernel.org>,
Christoph Hellwig <hch@infradead.org>,
Andreas Gruenbacher <agruen@suse.de>
Subject: Re: [PATCH] posix_acl: resolve compile dependency in posix_acl.h
Date: Thu, 10 Oct 2013 11:49:43 +0300 [thread overview]
Message-ID: <52566A27.3080204@primarydata.com> (raw)
In-Reply-To: <20131009154109.13c7248c53b97d96194ca8f9@linux-foundation.org>
On 2013-10-10 01:41, Andrew Morton wrote:
> On Wed, 02 Oct 2013 17:36:29 +0300 Benny Halevy <bhalevy@primarydata.com> wrote:
>
>> From: Benny Halevy <bhalevy@panasas.com>
>>
>> get_cached_acl is defined as inline in posix_acl.h
>> requiring the full definition of struct inode as it
>> dereferences its struct inode * parameter.
>
> That's very old code so you must have a peculiar config. Please
> describe the circumstances under which this occurs, because I'd like to
> avoid merging this patch.
>
Wow, sorry, you're right. It originated in 2.6.33 as far as I can see
and it is no longer needed.
>> --- a/include/linux/posix_acl.h
>> +++ b/include/linux/posix_acl.h
>> @@ -9,6 +9,7 @@
>> #define __LINUX_POSIX_ACL_H
>>
>> #include <linux/bug.h>
>> +#include <linux/fs.h>
>> #include <linux/slab.h>
>> #include <linux/rcupdate.h>
>
> A better fix is to undo all that crazy inlining in posix_acl.h.
>
>
> From: Andrew Morton <akpm@linux-foundation.org>
> Subject: posix_acl: uninlining
>
> Uninline vast tracts of nested inline functions in
> include/linux/posix_acl.h.
>
> This reduces the text+data+bss size of x86_64 allyesconfig vmlinux by 8026
> bytes.
>
> Also fixes an obscure build error reported by Benny: get_cached_acl()
> needs fs.h for struct inode internals.
Sorry for the stale report, I have no problem with 3.12-rc3.
>
> The patch also regularises the positioning of the EXPORT_SYMBOLs in
> posix_acl.c.
>
> Reported-by:: Benny Halevy <bhalevy@panasas.com>
ditto.
> Cc: Alexander Viro <viro@zeniv.linux.org.uk>
> Cc: J. Bruce Fields <bfields@fieldses.org>
> Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
> Cc: Benny Halevy <bhalevy@primarydata.com>
> Cc: Andreas Gruenbacher <agruen@suse.de>
> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
FWIW, I tested my linux-pnfs 3.12-rc3 tree with this patch
and it builds and causes no regression with the connectathon test suite over pnfs.
Benny
> ---
>
> fs/posix_acl.c | 84 +++++++++++++++++++++++++++++++++---
> include/linux/posix_acl.h | 78 ++-------------------------------
> 2 files changed, 85 insertions(+), 77 deletions(-)
>
> diff -puN fs/posix_acl.c~posix_acl-uninlining fs/posix_acl.c
> --- a/fs/posix_acl.c~posix_acl-uninlining
> +++ a/fs/posix_acl.c
> @@ -22,11 +22,80 @@
>
> #include <linux/errno.h>
>
> -EXPORT_SYMBOL(posix_acl_init);
> -EXPORT_SYMBOL(posix_acl_alloc);
> -EXPORT_SYMBOL(posix_acl_valid);
> -EXPORT_SYMBOL(posix_acl_equiv_mode);
> -EXPORT_SYMBOL(posix_acl_from_mode);
> +struct posix_acl **acl_by_type(struct inode *inode, int type)
> +{
> + switch (type) {
> + case ACL_TYPE_ACCESS:
> + return &inode->i_acl;
> + case ACL_TYPE_DEFAULT:
> + return &inode->i_default_acl;
> + default:
> + BUG();
> + }
> +}
> +EXPORT_SYMBOL(acl_by_type);
> +
> +struct posix_acl *get_cached_acl(struct inode *inode, int type)
> +{
> + struct posix_acl **p = acl_by_type(inode, type);
> + struct posix_acl *acl = ACCESS_ONCE(*p);
> + if (acl) {
> + spin_lock(&inode->i_lock);
> + acl = *p;
> + if (acl != ACL_NOT_CACHED)
> + acl = posix_acl_dup(acl);
> + spin_unlock(&inode->i_lock);
> + }
> + return acl;
> +}
> +EXPORT_SYMBOL(get_cached_acl);
> +
> +struct posix_acl *get_cached_acl_rcu(struct inode *inode, int type)
> +{
> + return rcu_dereference(*acl_by_type(inode, type));
> +}
> +EXPORT_SYMBOL(get_cached_acl_rcu);
> +
> +void set_cached_acl(struct inode *inode, int type, struct posix_acl *acl)
> +{
> + struct posix_acl **p = acl_by_type(inode, type);
> + struct posix_acl *old;
> + spin_lock(&inode->i_lock);
> + old = *p;
> + rcu_assign_pointer(*p, posix_acl_dup(acl));
> + spin_unlock(&inode->i_lock);
> + if (old != ACL_NOT_CACHED)
> + posix_acl_release(old);
> +}
> +EXPORT_SYMBOL(set_cached_acl);
> +
> +void forget_cached_acl(struct inode *inode, int type)
> +{
> + struct posix_acl **p = acl_by_type(inode, type);
> + struct posix_acl *old;
> + spin_lock(&inode->i_lock);
> + old = *p;
> + *p = ACL_NOT_CACHED;
> + spin_unlock(&inode->i_lock);
> + if (old != ACL_NOT_CACHED)
> + posix_acl_release(old);
> +}
> +EXPORT_SYMBOL(forget_cached_acl);
> +
> +void forget_all_cached_acls(struct inode *inode)
> +{
> + struct posix_acl *old_access, *old_default;
> + spin_lock(&inode->i_lock);
> + old_access = inode->i_acl;
> + old_default = inode->i_default_acl;
> + inode->i_acl = inode->i_default_acl = ACL_NOT_CACHED;
> + spin_unlock(&inode->i_lock);
> + if (old_access != ACL_NOT_CACHED)
> + posix_acl_release(old_access);
> + if (old_default != ACL_NOT_CACHED)
> + posix_acl_release(old_default);
> +}
> +EXPORT_SYMBOL(forget_all_cached_acls);
>
> /*
> * Init a fresh posix_acl
> @@ -37,6 +106,7 @@ posix_acl_init(struct posix_acl *acl, in
> atomic_set(&acl->a_refcount, 1);
> acl->a_count = count;
> }
> +EXPORT_SYMBOL(posix_acl_init);
>
> /*
> * Allocate a new ACL with the specified number of entries.
> @@ -51,6 +121,7 @@ posix_acl_alloc(int count, gfp_t flags)
> posix_acl_init(acl, count);
> return acl;
> }
> +EXPORT_SYMBOL(posix_acl_alloc);
>
> /*
> * Clone an ACL.
> @@ -146,6 +217,7 @@ posix_acl_valid(const struct posix_acl *
> return 0;
> return -EINVAL;
> }
> +EXPORT_SYMBOL(posix_acl_valid);
>
> /*
> * Returns 0 if the acl can be exactly represented in the traditional
> @@ -186,6 +258,7 @@ posix_acl_equiv_mode(const struct posix_
> *mode_p = (*mode_p & ~S_IRWXUGO) | mode;
> return not_equiv;
> }
> +EXPORT_SYMBOL(posix_acl_equiv_mode);
>
> /*
> * Create an ACL representing the file mode permission bits of an inode.
> @@ -207,6 +280,7 @@ posix_acl_from_mode(umode_t mode, gfp_t
> acl->a_entries[2].e_perm = (mode & S_IRWXO);
> return acl;
> }
> +EXPORT_SYMBOL(posix_acl_from_mode);
>
> /*
> * Return 0 if current is granted want access to the inode
> diff -puN include/linux/posix_acl.h~posix_acl-uninlining include/linux/posix_acl.h
> --- a/include/linux/posix_acl.h~posix_acl-uninlining
> +++ a/include/linux/posix_acl.h
> @@ -94,78 +94,12 @@ extern int posix_acl_chmod(struct posix_
> extern struct posix_acl *get_posix_acl(struct inode *, int);
> extern int set_posix_acl(struct inode *, int, struct posix_acl *);
>
> -#ifdef CONFIG_FS_POSIX_ACL
> -static inline struct posix_acl **acl_by_type(struct inode *inode, int type)
> -{
> - switch (type) {
> - case ACL_TYPE_ACCESS:
> - return &inode->i_acl;
> - case ACL_TYPE_DEFAULT:
> - return &inode->i_default_acl;
> - default:
> - BUG();
> - }
> -}
> -
> -static inline struct posix_acl *get_cached_acl(struct inode *inode, int type)
> -{
> - struct posix_acl **p = acl_by_type(inode, type);
> - struct posix_acl *acl = ACCESS_ONCE(*p);
> - if (acl) {
> - spin_lock(&inode->i_lock);
> - acl = *p;
> - if (acl != ACL_NOT_CACHED)
> - acl = posix_acl_dup(acl);
> - spin_unlock(&inode->i_lock);
> - }
> - return acl;
> -}
> -
> -static inline struct posix_acl *get_cached_acl_rcu(struct inode *inode, int type)
> -{
> - return rcu_dereference(*acl_by_type(inode, type));
> -}
> -
> -static inline void set_cached_acl(struct inode *inode,
> - int type,
> - struct posix_acl *acl)
> -{
> - struct posix_acl **p = acl_by_type(inode, type);
> - struct posix_acl *old;
> - spin_lock(&inode->i_lock);
> - old = *p;
> - rcu_assign_pointer(*p, posix_acl_dup(acl));
> - spin_unlock(&inode->i_lock);
> - if (old != ACL_NOT_CACHED)
> - posix_acl_release(old);
> -}
> -
> -static inline void forget_cached_acl(struct inode *inode, int type)
> -{
> - struct posix_acl **p = acl_by_type(inode, type);
> - struct posix_acl *old;
> - spin_lock(&inode->i_lock);
> - old = *p;
> - *p = ACL_NOT_CACHED;
> - spin_unlock(&inode->i_lock);
> - if (old != ACL_NOT_CACHED)
> - posix_acl_release(old);
> -}
> -
> -static inline void forget_all_cached_acls(struct inode *inode)
> -{
> - struct posix_acl *old_access, *old_default;
> - spin_lock(&inode->i_lock);
> - old_access = inode->i_acl;
> - old_default = inode->i_default_acl;
> - inode->i_acl = inode->i_default_acl = ACL_NOT_CACHED;
> - spin_unlock(&inode->i_lock);
> - if (old_access != ACL_NOT_CACHED)
> - posix_acl_release(old_access);
> - if (old_default != ACL_NOT_CACHED)
> - posix_acl_release(old_default);
> -}
> -#endif
> +struct posix_acl **acl_by_type(struct inode *inode, int type);
> +struct posix_acl *get_cached_acl(struct inode *inode, int type);
> +struct posix_acl *get_cached_acl_rcu(struct inode *inode, int type);
> +void set_cached_acl(struct inode *inode, int type, struct posix_acl *acl);
> +void forget_cached_acl(struct inode *inode, int type);
> +void forget_all_cached_acls(struct inode *inode);
>
> static inline void cache_no_acl(struct inode *inode)
> {
> _
>
next prev parent reply other threads:[~2013-10-10 8:49 UTC|newest]
Thread overview: 139+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-09-26 18:36 [PATCH RFC v0 0/49] pnfsd-dlm Benny Halevy
2013-09-26 18:39 ` [PATCH RFC v0 01/49] pnfsd: Define CONFIG_PNFSD Benny Halevy
2013-09-26 18:39 ` [PATCH RFC v0 02/49] pnfsd: define NFSDDBG_PNFS Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 03/49] pnfsd: return pnfs flags on exchange_id Benny Halevy
2013-09-26 21:55 ` J. Bruce Fields
2013-09-27 1:09 ` Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 04/49] pnfsd: don't set up back channel on create_session for ds Benny Halevy
2013-09-26 22:01 ` J. Bruce Fields
2013-09-27 1:20 ` Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 05/49] pnfsd: introduce pnfsd header files Benny Halevy
2013-09-29 11:43 ` Christoph Hellwig
2013-09-29 12:12 ` Benny Halevy
2013-09-29 12:13 ` Christoph Hellwig
2013-09-29 12:20 ` Benny Halevy
2013-09-29 12:21 ` Christoph Hellwig
2013-09-29 12:35 ` Christoph Hellwig
2013-09-30 15:23 ` Benny Halevy
2013-10-01 13:19 ` Christoph Hellwig
2013-10-01 1:05 ` Boaz Harrosh
2013-10-01 13:33 ` Christoph Hellwig
2013-10-02 11:35 ` Benny Halevy
2013-10-02 16:06 ` Christoph Hellwig
2013-10-01 20:30 ` J. Bruce Fields
2013-10-02 11:36 ` Benny Halevy
2013-10-02 16:07 ` Christoph Hellwig
2013-10-03 6:02 ` Benny Halevy
2013-10-03 9:55 ` Christoph Hellwig
2013-10-03 12:29 ` Benny Halevy
2013-10-03 12:37 ` Christoph Hellwig
2013-10-03 13:12 ` Ric Wheeler
2013-10-03 13:17 ` Christoph Hellwig
2013-10-03 13:18 ` Ric Wheeler
2013-10-03 14:19 ` Benny Halevy
2013-10-03 14:21 ` Christoph Hellwig
2013-10-03 14:24 ` Ric Wheeler
2013-10-03 14:38 ` Benny Halevy
2013-10-01 1:41 ` Boaz Harrosh
2013-10-01 19:43 ` J. Bruce Fields
2013-09-26 18:40 ` [PATCH RFC v0 06/49] pnfsd: define pnfs_export_operations Benny Halevy
2013-09-27 14:39 ` J. Bruce Fields
2013-09-29 10:53 ` Benny Halevy
2013-09-29 12:14 ` Christoph Hellwig
2013-09-26 18:40 ` [PATCH RFC v0 07/49] pnfsd: add pnfs export option Benny Halevy
2013-09-27 14:36 ` J. Bruce Fields
2013-09-29 10:51 ` Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 08/49] pnfsd: layout verify Benny Halevy
2013-09-27 14:44 ` J. Bruce Fields
2013-09-29 11:16 ` Benny Halevy
2013-10-01 20:38 ` J. Bruce Fields
2013-10-02 11:42 ` Benny Halevy
2013-10-01 22:12 ` J. Bruce Fields
2013-09-26 18:40 ` [PATCH RFC v0 09/49] pnfsd: initial stub Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 10/49] pnfsd: use sbid hash table to map super_blocks to devid major identifiers Benny Halevy
2013-10-01 22:14 ` J. Bruce Fields
2013-10-02 14:32 ` Benny Halevy
2013-10-02 15:24 ` J. Bruce Fields
2013-10-11 19:56 ` Christoph Hellwig
2013-10-13 6:11 ` Benny Halevy
2013-10-13 11:08 ` Christoph Hellwig
2013-10-13 12:44 ` Benny Halevy
2013-10-14 14:15 ` Christoph Hellwig
2013-09-26 18:40 ` [PATCH RFC v0 11/49] NFSD: introduce exp_xdr.h Benny Halevy
2013-09-29 12:15 ` Christoph Hellwig
2013-09-30 15:25 ` Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 12/49] pnfsd: get device list/info Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 13/49] pnfsd: filelayout: " Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 14/49] pnfsd: provide helper for xdr encoding of deviceid Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 15/49] pnfsd: add helper functions for identifying DS filehandles Benny Halevy
2013-09-26 18:40 ` [PATCH RFC v0 16/49] pnfsd: accept all ds stateids Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 17/49] DEBUG: nfsd: more client_lock asserts Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 18/49] pnfsd: nfs4_assert_state_locked Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 19/49] pnfsd: layout get Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 20/49] pnfsd: filelayout: layout encoding Benny Halevy
2013-09-29 12:16 ` Christoph Hellwig
2013-10-01 1:15 ` Boaz Harrosh
2013-10-01 13:34 ` Christoph Hellwig
2013-10-01 6:04 ` Benny Halevy
2013-10-02 14:27 ` Benny Halevy
2013-10-02 16:09 ` Christoph Hellwig
2013-09-26 18:41 ` [PATCH RFC v0 21/49] nfsd: no need to unhash_stid before free Benny Halevy
2013-10-11 19:37 ` Christoph Hellwig
2013-10-13 6:23 ` Benny Halevy
2013-10-13 19:28 ` J. Bruce Fields
2013-09-26 18:41 ` [PATCH RFC v0 22/49] nfsd: cleanup free_stid Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 23/49] pnfsd: layout state allocation Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 24/49] pnfsd: process the layout stateid Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 25/49] pnfsd: layout state per client tracking Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 26/49] pnfsd: layout state per file tracking Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 27/49] pnfsd: hash layouts on layout state Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 28/49] pnfsd: support layout segment merging Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 29/49] pnfsd: support layout_type attribute Benny Halevy
2013-09-29 12:17 ` Christoph Hellwig
2013-10-01 1:21 ` Boaz Harrosh
2013-10-01 8:32 ` Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 30/49] pnfsd: make pnfs server return layout_blksize when the client asks for it Benny Halevy
2013-09-26 18:41 ` [PATCH RFC v0 31/49] pnfsd: add support for per-file layout_types attribute Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 32/49] pnfsd: per block device dlm data server list cache Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 33/49] pnfsd: Add IP address validation to nfsd4_set_pnfs_dlm_device() Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 34/49] pnfsd: new nfsd filesystem file: pnfs_dlm_device Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 35/49] pnfsd: nfsd4_pnfs_dlm_getdeviter Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 36/49] pnfsd: nfsd4_pnfs_dlm_getdevinfo Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 37/49] pnfsd: make /proc/fs/nfsd/pnfs_dlm_device report dlm device list Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 38/49] pnfsd: nfsd4_pnfs_dlm_layoutget Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 39/49] pnfsd: DLM file layout only support read iomode layouts Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 40/49] pnfsd: add dlm file layout layout-type Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 41/49] pnfsd: dlm pnfs_export_operations Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 42/49] pnfsd: gfs2: use generic file layout pnfs operations vector Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 43/49] pnfsd: release state lock around iput in put_nfs4_file Benny Halevy
2013-09-29 12:19 ` Christoph Hellwig
2013-10-01 13:31 ` Benny Halevy
2013-10-01 13:37 ` Christoph Hellwig
2013-10-02 14:17 ` Benny Halevy
2013-10-02 15:26 ` J. Bruce Fields
2013-10-11 19:47 ` Christoph Hellwig
2013-10-13 6:26 ` Benny Halevy
2013-09-26 18:42 ` [PATCH RFC v0 44/49] posix_acl: resolve compile dependency in posix_acl.h Benny Halevy
2013-09-29 12:19 ` Christoph Hellwig
2013-10-02 14:17 ` Benny Halevy
2013-10-02 14:36 ` [PATCH] " Benny Halevy
2013-10-09 22:41 ` Andrew Morton
2013-10-10 8:49 ` Benny Halevy [this message]
2013-09-26 18:42 ` [PATCH RFC v0 45/49] nfs: resolve compile dependency in nfs_xdr.h Benny Halevy
2013-09-29 12:19 ` Christoph Hellwig
2013-10-02 14:19 ` Benny Halevy
2013-09-26 18:43 ` [PATCH RFC v0 46/49] pnfsd: layout return generic implementation Benny Halevy
2013-09-26 18:43 ` [PATCH RFC v0 47/49] pnfsd: pnfs_expire_client Benny Halevy
2013-09-26 18:43 ` [PATCH RFC v0 48/49] pnfsd: return on close Benny Halevy
2013-09-26 18:43 ` [PATCH RFC v0 49/49] pnfsd: dlm set return_on_close to true Benny Halevy
2013-09-26 19:44 ` [PATCH RFC v0 0/49] pnfsd-dlm J. Bruce Fields
2013-09-26 20:06 ` Benny Halevy
2013-09-27 13:31 ` Boaz Harrosh
2013-09-27 13:34 ` Benny Halevy
2013-09-27 16:37 ` Boaz Harrosh
2013-09-27 20:19 ` Benny Halevy
2013-10-01 0:23 ` Boaz Harrosh
2013-10-01 0:29 ` Boaz Harrosh
2013-10-02 6:02 ` Benny Halevy
2013-09-29 11:42 ` Christoph Hellwig
2013-09-29 11:54 ` Benny Halevy
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=52566A27.3080204@primarydata.com \
--to=bhalevy@primarydata.com \
--cc=agruen@suse.de \
--cc=akpm@linux-foundation.org \
--cc=hch@infradead.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-nfs@vger.kernel.org \
--cc=viro@zeniv.linux.org.uk \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).