Re: [PATCH] get_empty_inode() cleanup

2000-11-16 Thread Alexander Viro
On Thu, 16 Nov 2000, Tigran Aivazian wrote: > on the other hand, even 1 minute's thought reveals that making strict > logical separation between "consumers of inode with sb" and "consumers of > inode without sb" is probably worth the overhead of an extra function > call. So, I don't strongly

Re: [PATCH] get_empty_inode() cleanup

2000-11-16 Thread Tigran Aivazian
replying to myself before someone flames me :) I missed the word "inline" in Alexander's email (maybe because his code snippet did not have it) so the "issues" I raised are in fact "non-issues". (and if they were issues then there would be a serious bug in his patch -- namely new_inode would

Re: [PATCH] get_empty_inode() cleanup

2000-11-16 Thread Tigran Aivazian
On Thu, 16 Nov 2000, Tigran Aivazian wrote: > On Thu, 16 Nov 2000, Alexander Viro wrote: > > > Almost all (== all filesystem and then some) callers of > > get_empty_inode() follow it with > > inode->i_sb = some_sb; > > inode->i_dev = some_sb->s_dev; > > Some of them do it twice for

Re: [PATCH] get_empty_inode() cleanup

2000-11-16 Thread Tigran Aivazian
On Thu, 16 Nov 2000, Alexander Viro wrote: > Almost all (== all filesystem and then some) callers of > get_empty_inode() follow it with > inode->i_sb = some_sb; > inode->i_dev = some_sb->s_dev; > Some of them do it twice for no good reason (assign the same value, > even though

Re: [PATCH] get_empty_inode() cleanup

2000-11-16 Thread Tigran Aivazian
On Thu, 16 Nov 2000, Alexander Viro wrote: Almost all (== all filesystem and then some) callers of get_empty_inode() follow it with inode-i_sb = some_sb; inode-i_dev = some_sb-s_dev; Some of them do it twice for no good reason (assign the same value, even though neither

Re: [PATCH] get_empty_inode() cleanup

2000-11-16 Thread Tigran Aivazian
On Thu, 16 Nov 2000, Tigran Aivazian wrote: On Thu, 16 Nov 2000, Alexander Viro wrote: Almost all (== all filesystem and then some) callers of get_empty_inode() follow it with inode-i_sb = some_sb; inode-i_dev = some_sb-s_dev; Some of them do it twice for no good reason

Re: [PATCH] get_empty_inode() cleanup

2000-11-16 Thread Tigran Aivazian
replying to myself before someone flames me :) I missed the word "inline" in Alexander's email (maybe because his code snippet did not have it) so the "issues" I raised are in fact "non-issues". (and if they were issues then there would be a serious bug in his patch -- namely new_inode would

Re: [PATCH] get_empty_inode() cleanup

2000-11-16 Thread Alexander Viro
On Thu, 16 Nov 2000, Tigran Aivazian wrote: on the other hand, even 1 minute's thought reveals that making strict logical separation between "consumers of inode with sb" and "consumers of inode without sb" is probably worth the overhead of an extra function call. So, I don't strongly feel