Hi,
On 19-07-18 10:53, Petr Mladek wrote:
On Wed 2018-07-18 11:30:02, Thomas Zimmermann wrote:
If the console is unlocked during registration, the console subsystem generates significant amounts of warnings, which obfuscate actual debugging messages. Setting ignore_console_lock_warning while debugging console registration avoid the noise.
v2:
- restore ignore_console_lock_warning if lock_fb_info() fails
Signed-off-by: Thomas Zimmermann tzimmermann@suse.de
drivers/video/fbdev/core/fbmem.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-)
diff --git a/drivers/video/fbdev/core/fbmem.c b/drivers/video/fbdev/core/fbmem.c index 9e2f9d3c760e..432c26eeabfb 100644 --- a/drivers/video/fbdev/core/fbmem.c +++ b/drivers/video/fbdev/core/fbmem.c @@ -1627,6 +1627,7 @@ static int do_register_framebuffer(struct fb_info *fb_info) int i, ret; struct fb_event event; struct fb_videomode mode;
- bool saved_ignore_console_lock_warning = ignore_console_lock_warning;
Hmm, this approach is racy if there are other users saving/setting/restoring ignore_console_lock_warning in parallel. I mean that this works only when the entire safe/set/restore operation is nested or sequential.
We might need another approach if there are more users, e.g. use an atomic counter for ignore_console_lock_warning.
I noticed this would be racy too, but this only gets used when console-locking should be disabled when registering fbdev-s for debugging purposes at which point everything console related is racy already anyways, so I think this is fine as is.
On the other hand, I wonder if there ever will be other user. Also it is "just" for debugging. We could keep it simple for now. It might be enough to add a comment into include/linux/console.h, something like:
Ack, lets keep this simple / as is in v2 of the patch.
Regards,
Hans
/*
- Set ignore_console_lock_warning to true if you need to quiet
- WARN_CONSOLE_UNLOCKED() for debugging purposes. Might need
- another approach if manipulated by more users in parallel.
*/
Best Regards, Petr