11 Aug, 2010
40 commits
-
Using:
gcc (GCC) 4.5.0 20100610 (prerelease)
with CONFIG_CONSOLE_TRANSLATIONS=n, the following warnings are seen:
drivers/char/vt_ioctl.c: In function ‘vt_ioctl’:
drivers/char/vt_ioctl.c:1309:4: warning: statement with no effect
drivers/char/vt.c: In function ‘vc_allocate’:
drivers/char/vt.c:774:3: warning: statement with no effect
drivers/video/console/vgacon.c: In function ‘vgacon_init’:
drivers/video/console/vgacon.c:587:3: warning: statement with no effect
drivers/video/console/vgacon.c: In function ‘vgacon_deinit’:
drivers/video/console/vgacon.c:606:2: warning: statement with no effect
drivers/video/console/fbcon.c: In function ‘fbcon_init’:
drivers/video/console/fbcon.c:1087:3: warning: statement with no effect
drivers/video/console/fbcon.c:1089:3: warning: statement with no effect
drivers/video/console/fbcon.c: In function ‘fbcon_set_disp’:
drivers/video/console/fbcon.c:1369:3: warning: statement with no effect
drivers/video/console/fbcon.c:1371:3: warning: statement with no effectThis is because several functions in include/linux/vt_kern.h are
defined to (0). Convert them to static inline functions to
silence the warnings and gain a bit of type safety.Signed-off-by: Kevin Winchester
Signed-off-by: Greg Kroah-Hartman -
Remove unnesessary casts from void*.
Signed-off-by: Kulikov Vasiliy
Signed-off-by: Greg Kroah-Hartman -
At the moment there is only one platform type supported and there is is
hard wired, but with these changes the infrastructure is now there for
anyone else to provide methods for their hardware.Signed-off-by: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
This device is used by some of the Intel MID platforms. It's not similar
enough to the MAX3100 to use the same driver.At this point the driver is specific to the platform and not generalised.
We will fix that later.Signed-off-by: jianwei.yang
Signed-off-by: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
When the console has been redirected, a hangup of the tty
will cause tty_release to be called under the big tty_mutex,
which leads to a deadlock because hangup is also called
under the BTM.This moves the BTM deeper into the tty_hangup function so
we can close the redirected tty without holding the BTM.
In case of pty, we now need to drop the BTM before
calling tty_vhangup.Signed-off-by: Arnd Bergmann
Acked-by: Alan Cox
Cc: Tony Luck
Cc: Thomas Gleixner
Cc: Andrew Morton
Cc: John Kacur
Cc: Al Viro
Cc: Ingo Molnar
Signed-off-by: Greg Kroah-Hartman -
The ldisc number now gets passed into ->set_ldisc.
Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
Most tty drivers may block while opening a device.
Since this possibly depends on another thread
closing it first and both threads may need the BTM,
we need to release it here.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
The tty locking now follows the rules for mutexes, so
we can replace the BKL usage with a new subsystem
wide mutex.Using a regular mutex here will change the behaviour
when blocked on the BTM from spinning to sleeping,
but that should not be visible to the user.Using the mutex also means that all the BTM is now
covered by lockdep.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
This changes all remaining users of tty_lock_nested
to be non-recursive, which lets us kill this function.
As a consequence, we won't need to keep the lock count
any more, which allows more simplifications later.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
Some wait_until_sent versions require the big
tty mutex, others don't and some callers of
wait_until_sent already hold it while other don't.
That leads to recursive use of the BTM in these
functions, which we're trying to get rid of.This turns all cleans up the locking there so
that the driver's wait_until_sent function
never takes the BTM itself if it is already
called with that lock held.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
We need to release the BTM in paste_selection() when
sleeping in tty_ldisc_ref_wait to avoid deadlocks
with tty_ldisc_enable.In tty_set_ldisc, we now always grab the BTM before
taking the ldisc_mutex in order to avoid AB-BA
deadlocks between the two.tty_ldisc_halt potentially blocks on a workqueue
function that takes the BTM, so we must release
the BTM before calling it.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
Calling wait_event_interruptible implicitly
releases the BKL when it sleeps, but we need
to do this explcitly when we have converted
it to a mutex.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
The wait_event_interruptible_timeout in acm_port_down is
never reached. Remove it to avoid possible deadlocks
with the big tty mutex if someone were to start using
the blocking version of acm_port_down.Signed-off-by: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
vgacon_do_font_op releases and reacquires the BTM while holding
console_sem. This violates the rule that BTM has to be the
outer lock whenever we hold both.There does not seem to be any reason to give up the BTM here,
so just stop doing that.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
tty_mutex is never taken with the BTM held, except for
two corner cases that are worked around here.
We give up the BTM before calling tty_release() in the
error path of tty_open().
Similarly, we reorder the locking in ptmx_open()
to get tty_mutex before the BTM.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
As a preparation for replacing the big kernel lock
in the TTY layer, wrap all the callers in new
macros tty_lock, tty_lock_nested and tty_unlock.Signed-off-by: Arnd Bergmann
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
Move termios initialization in open into uart_dtr_rts to make sure
it always gets called when necessary. Based on a suggestion from
Alan Cox.Alan writes:
Ok this sort of makes sense. Something isn't getting initialised and both
getty and minicom will do a termios set which is sorting it out.
This is occurring because the generic block_til_ready sets
ASYNCB_NORMAL_ACTIVE so the termios updating gets skipped.This patch should cure it and then we can think about doing it more
elegantly by getting the serial layer to use tty_port_open, kfifo and
the like and removing the tons of repeated crap in all the drivers.Signed-off-by: Arnd Bergmann
Reported-by: Tony Luck
Cc: Alan Cox
Cc: Tony Luck
Signed-off-by: Greg Kroah-Hartman -
Our code now rather closely resembles the helper, so switch to it.
Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
The port mutex protects port->tty, but these paths never need to walk from
port->tty. They do need the low level lock as the API expects that but they
already also take it.Thus we can drop the extra mutex lock calls here.
Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
We can make this the same as the ones that will be needed by the tty_port
helper logic that we want to move to but still call them from the existing
code base.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
We want to push the lock/unlock into the helper functions so that we
can prepare to move to using the tty_port helper. The expansion initially
comes out a bit ugly but its worth the temporary expansion IMHO just so
we can produce a nice testable series of changes.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
This takes all the tty references through the expected interface points so
we can refcount them.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
The vt layer isn't safely handling reference counts to tty object on the input
side. Add a tty port structure to the vt layer in order to implement this using
the standard helpers.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
The virtual console layer uses the BKL for various things that don't really
need it. Clean them out.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
Pass down the ldisc number so that the drivers don't have to peek into the
tty object themselves. This lets us get rid of another case of back referencing
port to tty which we don't want (because of races versus hangup/close).Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
One or two drivers go poking back into the tty from the termios setting
routine in unsafe ways. We don't need to pass the tty down because the
[ab]users are just using it to get at things they can get at anyway.This leaves low_latency setting to sort out along with set_ldisc use.
Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
Make it robust against hang up events. In most cases we can do this simply
by passing the right things in the first place.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
Use the port mutex and port lock to fix the various races. The locking
still isn't totally consistent but its better than before. Wants switching
to the port helpers.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
This now refcounts but doesn't actually check the reference was obtained in
all the places it should.Signed-off-by: Alan Cox
Signed-off-by: Greg Kroah-Hartman -
Use the port mutex instead
Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
The lock is no longer needed for wait until sent paths so this can go
Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
Use the port mutext for config setting, the rest is locked sufficiently
anyway that the BKL makes no odds.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
We don't need it while waiting and we can lock the ioctls using the port
mutex. While at it eliminate use of the hangup mutex and switch to the port
mutex.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
We can use the port mutex for this and also for the hangup path so removing
the problematic use of the hangup mutex in this driver. Fix up the locking
on the various port flags while we are at it.Ultimately this driver needs to be using tty_port_ helpers which would sort
this out far better.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
As with the others we can use the port mutex to get the needed locking
properties and fix the race with open.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
The riscom8 board uses lock_kernel to protect bits of the port setting
ioctl logic. We can use the port mutex for this as the logic is internal
and will also lock set versus open (a locking property that has been lost
somewhere along the way)Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
This lets us avoid problems with races on the flag changes
Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
Remove unneeded tty layer lock kernel bits. Relock the needed bits using the
port mutex. The istallion still has brd state races but those are not new
or introduced by the removal of the lock_kernel logic.Signed-off-by: Alan Cox
Cc: Arnd Bergmann
Signed-off-by: Greg Kroah-Hartman -
Jesse's initial patch commit said:
"At panic time (i.e. when oops_in_progress is set) we should try a bit
harder to update the screen and make sure output gets to the VT, since
some drivers are capable of flipping back to it.So make sure we try to unblank and update the display if called from a
panic context."I've enhanced this to add a flag to the vc that console layer can set to
indicate they want this behaviour to occur. This also adds support to
fbcon for that flag and adds an fb flag for drivers to indicate they want
to use the support. It enables this for KMS drivers.Signed-off-by: Dave Airlie
Signed-off-by: Jesse Barnes
Acked-by: James Simmons
Cc: Alan Cox
Signed-off-by: Andrew Morton
Signed-off-by: Greg Kroah-Hartman -
This patch is against the 2.6.34 source.
Paraphrased from the 1989 BSD patch by David Borman @ cray.com:
These are the changes needed for the kernel to support
LINEMODE in the server.There is a new bit in the termios local flag word, EXTPROC.
When this bit is set, several aspects of the terminal driver
are disabled. Input line editing, character echo, and mapping
of signals are all disabled. This allows the telnetd to turn
off these functions when in linemode, but still keep track of
what state the user wants the terminal to be in.New ioctl:
TIOCSIG Generate a signal to processes in the
current process group of the pty.There is a new mode for packet driver, the TIOCPKT_IOCTL bit.
When packet mode is turned on in the pty, and the EXTPROC bit
is set, then whenever the state of the pty is changed, the
next read on the master side of the pty will have the TIOCPKT_IOCTL
bit set. This allows the process on the server side of the pty
to know when the state of the terminal has changed; it can then
issue the appropriate ioctl to retrieve the new state.Since the original BSD patches accompanied the source code for telnet
I've left that reference here, but obviously the feature is useful for
any remote terminal protocol, including ssh.The corresponding feature has existed in the BSD tty driver since 1989.
For historical reference, a good copy of the relevant files can be found
here:http://anonsvn.mit.edu/viewvc/krb5/trunk/src/appl/telnet/?pathrev=17741
Signed-off-by: Howard Chu
Cc: Alan Cox
Signed-off-by: Greg Kroah-Hartman
