tty,vcs removing con_buf/conf_buf_mtx
authorJiri Olsa <jolsa@redhat.com>
Mon, 7 Feb 2011 18:31:25 +0000 (19:31 +0100)
committerGreg Kroah-Hartman <gregkh@suse.de>
Thu, 17 Feb 2011 19:13:19 +0000 (11:13 -0800)
commitfcdba07ee390d9d9c15de8b2a17baef689284fcc
tree607e8d1ca60006fd97c11e5c07740273d5b91895
parentdc1892c4bc6960121ca4c8023a07c815cfd689be
tty,vcs removing con_buf/conf_buf_mtx

seems there's no longer need for using con_buf/conf_buf_mtx
as vcs_read/vcs_write buffer for user's data.

The do_con_write function, that was the other user of this,
is currently using its own kmalloc-ed buffer.

Not sure when this got changed, as I was able to find this code
in 2.6.9, but it's already gone as far as current git history
goes - 2.6.12-rc2.

AFAICS there's a behaviour change with the current change.
The lseek is not completely mutually exclusive with the
vcs_read/vcs_write - the file->f_pos might get updated
via lseek callback during the vcs_read/vcs_write processing.

I tried to find out if the prefered behaviour is to keep
this in sync within read/write/lseek functions, but I did
not find any pattern on different places.

I guess if user end up calling write/lseek from different
threads she should know what she's doing. If needed we
could use dedicated fd mutex/buffer.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
drivers/tty/vt/vc_screen.c
drivers/tty/vt/vt.c
include/linux/vt_kern.h