]> git.proxmox.com Git - mirror_ubuntu-kernels.git/commit
fbcon: fix race condition between console lock and cursor timer
authorDave Airlie <airlied@redhat.com>
Tue, 21 Aug 2012 06:40:07 +0000 (16:40 +1000)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 21 Aug 2012 16:08:50 +0000 (09:08 -0700)
commitec5da7f8dc023fc4dcbdd42e87dad231d2c2812d
tree42762ef00fbfcb1c48580fa9baf65e90ca99bcb5
parent9160338de92c0305329be5163a76f849806e83de
fbcon: fix race condition between console lock and cursor timer

So we've had a fair few reports of fbcon handover breakage between
efi/vesafb and i915 surface recently, so I dedicated a couple of
days to finding the problem.

Essentially the last thing we saw was the conflicting framebuffer
message and that was all.

So after much tracing with direct netconsole writes (printks
under console_lock not so useful), I think I found the race.

  Thread A (driver load)    Thread B (timer thread)
    unbind_con_driver ->              |
    bind_con_driver ->                |
    vc->vc_sw->con_deinit ->          |
    fbcon_deinit ->                   |
    console_lock()                    |
        |                             |
        |                       fbcon_flashcursor timer fires
        |                       console_lock() <- blocked for A
        |
        |
  fbcon_del_cursor_timer ->
    del_timer_sync
    (BOOM)

Of course because all of this is under the console lock,
we never see anything, also since we also just unbound the active
console guess what we never see anything.

Hopefully this fixes the problem for anyone seeing vesafb->kms
driver handoff.

Signed-off-by: David Airlie <airlied@redhat.com>
Acked-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: stable@vger.kernel.org
Tested-by: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
drivers/video/console/fbcon.c