* [PATCH] move IMMUTABLE|APPEND checks to notify_change()
@ 2006-08-08 11:44 Kirill Korotaev
2006-08-08 20:38 ` Al Viro
0 siblings, 1 reply; 6+ messages in thread
From: Kirill Korotaev @ 2006-08-08 11:44 UTC (permalink / raw)
To: Andrew Morton, viro, Linux Kernel Mailing List, Mishin Dmitry
[PATCH] move IMMUTABLE|APPEND checks to notify_change()
This patch moves lots of IMMUTABLE and APPEND flag checks
scattered all around to more logical place in notify_change().
Signed-Off-By: Dmitry Mishin <dim@openvz.org>
Signed-Off-By: Kirill Korotaev <dev@openvz.org>
--- ./fs/attr.c.immut 2006-06-18 05:49:35.000000000 +0400
+++ ./fs/attr.c 2006-08-08 15:15:59.000000000 +0400
@@ -109,6 +109,9 @@ int notify_change(struct dentry * dentry
struct timespec now;
unsigned int ia_valid = attr->ia_valid;
+ if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
+ return -EPERM;
+
mode = inode->i_mode;
now = current_fs_time(inode->i_sb);
--- ./fs/open.c.immut 2006-07-14 19:08:29.000000000 +0400
+++ ./fs/open.c 2006-08-08 15:19:58.000000000 +0400
@@ -252,10 +252,6 @@ static long do_sys_truncate(const char _
if (IS_RDONLY(inode))
goto dput_and_out;
- error = -EPERM;
- if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
- goto dput_and_out;
-
/*
* Make sure that there are no leases.
*/
@@ -316,10 +312,6 @@ static long do_sys_ftruncate(unsigned in
if (small && length > MAX_NON_LFS)
goto out_putf;
- error = -EPERM;
- if (IS_APPEND(inode))
- goto out_putf;
-
error = locks_verify_truncate(inode, file, length);
if (!error)
error = do_truncate(dentry, length, ATTR_MTIME|ATTR_CTIME, file);
@@ -385,10 +377,6 @@ asmlinkage long sys_utime(char __user *
/* Don't worry, the checks are done in inode_change_ok() */
newattrs.ia_valid = ATTR_CTIME | ATTR_MTIME | ATTR_ATIME;
if (times) {
- error = -EPERM;
- if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
- goto dput_and_out;
-
error = get_user(newattrs.ia_atime.tv_sec, ×->actime);
newattrs.ia_atime.tv_nsec = 0;
if (!error)
@@ -398,15 +386,9 @@ asmlinkage long sys_utime(char __user *
goto dput_and_out;
newattrs.ia_valid |= ATTR_ATIME_SET | ATTR_MTIME_SET;
- } else {
- error = -EACCES;
- if (IS_IMMUTABLE(inode))
- goto dput_and_out;
-
- if (current->fsuid != inode->i_uid &&
+ } else if (current->fsuid != inode->i_uid &&
(error = vfs_permission(&nd, MAY_WRITE)) != 0)
- goto dput_and_out;
- }
+ goto dput_and_out;
mutex_lock(&inode->i_mutex);
error = notify_change(nd.dentry, &newattrs);
mutex_unlock(&inode->i_mutex);
@@ -442,24 +424,14 @@ long do_utimes(int dfd, char __user *fil
/* Don't worry, the checks are done in inode_change_ok() */
newattrs.ia_valid = ATTR_CTIME | ATTR_MTIME | ATTR_ATIME;
if (times) {
- error = -EPERM;
- if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
- goto dput_and_out;
-
newattrs.ia_atime.tv_sec = times[0].tv_sec;
newattrs.ia_atime.tv_nsec = times[0].tv_usec * 1000;
newattrs.ia_mtime.tv_sec = times[1].tv_sec;
newattrs.ia_mtime.tv_nsec = times[1].tv_usec * 1000;
newattrs.ia_valid |= ATTR_ATIME_SET | ATTR_MTIME_SET;
- } else {
- error = -EACCES;
- if (IS_IMMUTABLE(inode))
- goto dput_and_out;
-
- if (current->fsuid != inode->i_uid &&
+ } else if (current->fsuid != inode->i_uid &&
(error = vfs_permission(&nd, MAY_WRITE)) != 0)
- goto dput_and_out;
- }
+ goto dput_and_out;
mutex_lock(&inode->i_mutex);
error = notify_change(nd.dentry, &newattrs);
mutex_unlock(&inode->i_mutex);
@@ -638,9 +610,6 @@ asmlinkage long sys_fchmod(unsigned int
err = -EROFS;
if (IS_RDONLY(inode))
goto out_putf;
- err = -EPERM;
- if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
- goto out_putf;
mutex_lock(&inode->i_mutex);
if (mode == (mode_t) -1)
mode = inode->i_mode;
@@ -672,10 +641,6 @@ asmlinkage long sys_fchmodat(int dfd, co
if (IS_RDONLY(inode))
goto dput_and_out;
- error = -EPERM;
- if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
- goto dput_and_out;
-
mutex_lock(&inode->i_mutex);
if (mode == (mode_t) -1)
mode = inode->i_mode;
@@ -709,9 +674,6 @@ static int chown_common(struct dentry *
error = -EROFS;
if (IS_RDONLY(inode))
goto out;
- error = -EPERM;
- if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
- goto out;
newattrs.ia_valid = ATTR_CTIME;
if (user != (uid_t) -1) {
newattrs.ia_valid |= ATTR_UID;
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] move IMMUTABLE|APPEND checks to notify_change()
2006-08-08 11:44 [PATCH] move IMMUTABLE|APPEND checks to notify_change() Kirill Korotaev
@ 2006-08-08 20:38 ` Al Viro
2006-08-09 7:15 ` Dmitry Mishin
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Al Viro @ 2006-08-08 20:38 UTC (permalink / raw)
To: Kirill Korotaev
Cc: Andrew Morton, viro, Linux Kernel Mailing List, Mishin Dmitry
On Tue, Aug 08, 2006 at 03:44:07PM +0400, Kirill Korotaev wrote:
> [PATCH] move IMMUTABLE|APPEND checks to notify_change()
>
> This patch moves lots of IMMUTABLE and APPEND flag checks
> scattered all around to more logical place in notify_change().
NAK. For example, you are allowed to do unames(file, NULL) on
any file you own or can write to, whether it's append-only or
not. With your change that gets -EPERM.
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] move IMMUTABLE|APPEND checks to notify_change()
2006-08-08 20:38 ` Al Viro
@ 2006-08-09 7:15 ` Dmitry Mishin
2006-08-09 14:11 ` Al Viro
2006-08-09 9:07 ` Kirill Korotaev
2006-08-09 10:11 ` Kirill Korotaev
2 siblings, 1 reply; 6+ messages in thread
From: Dmitry Mishin @ 2006-08-09 7:15 UTC (permalink / raw)
To: Al Viro; +Cc: Kirill Korotaev, Andrew Morton, viro, Linux Kernel Mailing List
Do you meant utimes(file, NULL)?
But is it correct behaviour? Why then do you get -EPERM on utimes(file, smth)
if the file is append-only? And why do you get -EACCESS on utimes(file,
NULL), if this file is immutable?
Could you explain, why is it done so?
On Wednesday 09 August 2006 00:38, Al Viro wrote:
> On Tue, Aug 08, 2006 at 03:44:07PM +0400, Kirill Korotaev wrote:
> > [PATCH] move IMMUTABLE|APPEND checks to notify_change()
> >
> > This patch moves lots of IMMUTABLE and APPEND flag checks
> > scattered all around to more logical place in notify_change().
>
> NAK. For example, you are allowed to do unames(file, NULL) on
> any file you own or can write to, whether it's append-only or
> not. With your change that gets -EPERM.
--
Thanks,
Dmitry.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] move IMMUTABLE|APPEND checks to notify_change()
2006-08-09 7:15 ` Dmitry Mishin
@ 2006-08-09 14:11 ` Al Viro
0 siblings, 0 replies; 6+ messages in thread
From: Al Viro @ 2006-08-09 14:11 UTC (permalink / raw)
To: Dmitry Mishin
Cc: Kirill Korotaev, Andrew Morton, viro, Linux Kernel Mailing List
On Wed, Aug 09, 2006 at 11:15:12AM +0400, Dmitry Mishin wrote:
> Do you meant utimes(file, NULL)?
> But is it correct behaviour? Why then do you get -EPERM on utimes(file, smth)
> if the file is append-only? And why do you get -EACCESS on utimes(file,
> NULL), if this file is immutable?
>
> Could you explain, why is it done so?
RTFPOSIX...
Short version:
* immutable files are immutable, including metadata
* append-only files may be touched (when you write to the end), which
means that you can touch them. Which is what utimes(file, NULL) does.
* you can not truncate append-only file, overwrite already written
data or set timestamps to arbitrary values.
That's where the difference between utimes(file, NULL) and utimes(file, p)
is - the former basically is a write-without-write ("touch foo") and the
latter directly assigns to timestamps. Permissions needed for these are
obviously different.
Please, read POSIX/SuS when modifying behaviour of syscalls. Really.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] move IMMUTABLE|APPEND checks to notify_change()
2006-08-08 20:38 ` Al Viro
2006-08-09 7:15 ` Dmitry Mishin
@ 2006-08-09 9:07 ` Kirill Korotaev
2006-08-09 10:11 ` Kirill Korotaev
2 siblings, 0 replies; 6+ messages in thread
From: Kirill Korotaev @ 2006-08-09 9:07 UTC (permalink / raw)
To: Al Viro; +Cc: Andrew Morton, viro, Linux Kernel Mailing List, Mishin Dmitry
>>[PATCH] move IMMUTABLE|APPEND checks to notify_change()
>>
>>This patch moves lots of IMMUTABLE and APPEND flag checks
>>scattered all around to more logical place in notify_change().
>
>
> NAK. For example, you are allowed to do unames(file, NULL) on
> any file you own or can write to, whether it's append-only or
> not. With your change that gets -EPERM.
Al, will you ACK the patch allowing to set current times in notify_change() for
APPEND files?
However, I'd like to see an explanation on dim@'s question why semantics
are different and current time is allowed to be set, while arbitrary time is not.
Kirill
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] move IMMUTABLE|APPEND checks to notify_change()
2006-08-08 20:38 ` Al Viro
2006-08-09 7:15 ` Dmitry Mishin
2006-08-09 9:07 ` Kirill Korotaev
@ 2006-08-09 10:11 ` Kirill Korotaev
2 siblings, 0 replies; 6+ messages in thread
From: Kirill Korotaev @ 2006-08-09 10:11 UTC (permalink / raw)
To: Al Viro; +Cc: Andrew Morton, viro, Linux Kernel Mailing List, Mishin Dmitry
Al Viro wrote:
> On Tue, Aug 08, 2006 at 03:44:07PM +0400, Kirill Korotaev wrote:
>
>>[PATCH] move IMMUTABLE|APPEND checks to notify_change()
>>
>>This patch moves lots of IMMUTABLE and APPEND flag checks
>>scattered all around to more logical place in notify_change().
>
>
> NAK. For example, you are allowed to do unames(file, NULL) on
> any file you own or can write to, whether it's append-only or
> not. With your change that gets -EPERM.
>
Does such check in notify_change() looks better for you?
notify_change():
if (IS_IMMUTABLE(inode))
return -EPERM;
if (IS_APPEND(inode) &&
(ia_valid & ~(ATTR_CTIME | ATTR_MTIME | ATTR_ATIME)))
return -EPERM;
Thanks,
Kirill
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2006-08-09 14:11 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-08-08 11:44 [PATCH] move IMMUTABLE|APPEND checks to notify_change() Kirill Korotaev
2006-08-08 20:38 ` Al Viro
2006-08-09 7:15 ` Dmitry Mishin
2006-08-09 14:11 ` Al Viro
2006-08-09 9:07 ` Kirill Korotaev
2006-08-09 10:11 ` Kirill Korotaev
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox