kernel/linux.git/drivers/video/fbdev/core, branch v4.19.16

fbcon: Do not takeover the console from atomic context

2018-08-10T15:23:02+00:00

Taking over the console involves allocating mem with GFP_KERNEL, talking to drm drivers, etc. So this should not be done from an atomic context. But the console-output trigger deferred console takeover may happen from an atomic context, which leads to "BUG: sleeping function called from invalid context" errors. This commit fixes these errors by doing the deferred takeover from a workqueue. Signed-off-by: Hans de Goede [b.zolnierkie: remove unused variable] Signed-off-by: Bartlomiej Zolnierkiewicz

fbcon: Only defer console takeover if the current console driver is the dummycon

2018-08-10T15:23:01+00:00

We rely on dummycon's output notifier mechanism to defer the takeover. If say vgacon is the current console driver then dummycon will never get used so its output notifier will also never get called and fbcon never takes over. This commit fixes this by only deferring the console takeover if the current console driver is the dummycon driver. This commit also moves the entirety of fbcon_start under the console_lock, since the conswitchp which fbcon_start now checks is protected by it. This commit also inlines fbcon_register_output_notifier, since we now need a #ifdef CONFIG_FRAMEBUFFER_CONSOLE_DEFERRED_TAKEOVER in fbcon_start anyways because of the write access to the deferred_takeover variable, this has the added advantage that it puts the dummycon_register_output_notifier() call directly after the "conswitchp != &dummy_con" comparison making it clear why that check is there. Note the arch setup code will set conswitchp to either dummy_con or vga_con, in the cases where it gets set to vga_con even though their is no vga_con present we rely on vga_con_startup() to set conswitchp to dummy_con. vga_con_startup() is guaranteed to happen before fb_console_init() as it gets called as a console_initcall where as fb_console_init() gets called as a subsys_initcall. Signed-off-by: Hans de Goede Signed-off-by: Bartlomiej Zolnierkiewicz

fbdev/core: Disable console-lock warnings when fb.lockless_register_fb is set

2018-07-31T11:06:58+00:00

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. v3: - manipulate ignore_console_lock_warning with atomic_{inc,dec} v2: - restore ignore_console_lock_warning if lock_fb_info() fails Signed-off-by: Thomas Zimmermann Acked-by: Hans de Goede Reviewed-by: Petr Mladek Reviewed-by: Sergey Senozhatsky Cc: Steven Rostedt (VMware) Cc: Andrew Morton Signed-off-by: Bartlomiej Zolnierkiewicz

fb: fix lost console when the user unplugs a USB adapter

2018-07-25T13:41:54+00:00

I have a USB display adapter using the udlfb driver and I use it on an ARM board that doesn't have any graphics card. When I plug the adapter in, the console is properly displayed, however when I unplug and re-plug the adapter, the console is not displayed and I can't access it until I reboot the board. The reason is this: When the adapter is unplugged, dlfb_usb_disconnect calls unlink_framebuffer, then it waits until the reference count drops to zero and then it deallocates the framebuffer. However, the console that is attached to the framebuffer device keeps the reference count non-zero, so the framebuffer device is never destroyed. When the USB adapter is plugged again, it creates a new device /dev/fb1 and the console is not attached to it. This patch fixes the bug by unbinding the console from unlink_framebuffer. The code to unbind the console is moved from do_unregister_framebuffer to a function unbind_console. When the console is unbound, the reference count drops to zero and the udlfb driver frees the framebuffer. When the adapter is plugged back, a new framebuffer is created and the console is attached to it. Signed-off-by: Mikulas Patocka Cc: Dave Airlie Cc: Bernie Thompson Cc: Ladislav Michl Cc: stable@vger.kernel.org [b.zolnierkie: preserve old behavior for do_unregister_framebuffer()] Signed-off-by: Bartlomiej Zolnierkiewicz

fbdev: fbmem: mark expected switch fall-through

2018-07-24T17:11:27+00:00

In preparation to enabling -Wimplicit-fallthrough, mark switch cases where we are expecting to fall through. Signed-off-by: Gustavo A. R. Silva Signed-off-by: Bartlomiej Zolnierkiewicz

fbdev: fix modedb docs in fb_find_mode

2018-07-24T17:11:27+00:00

Fix up the indenting that confused sphinx. To make sure we don't have to make the examples unreadable with escaping just put them in as block quotes, that seems the simplest solution. Signed-off-by: Daniel Vetter [b.zolnierkie: ported over fbdev changes] Signed-off-by: Bartlomiej Zolnierkiewicz

fbcon: introduce for_each_registered_fb() helper

2018-07-24T17:11:26+00:00

Following pattern is often used: for (i = 0; i < FB_MAX; i++) { if (registered_fb[i]) { ... } } Therefore, as Andy's suggestion, for_each_registered_fb() helper can be introduced to make the code easier to read and write by reducing indentation level. It also saves few lines of code in each occurrence. This patch convert all part here at the same time. Suggested-by: Andy Shevchenko Signed-off-by: Yisheng Xie Acked-by: Hans de Goede Reviewed-by: Andy Shevchenko Cc: Kees Cook Cc: David Lechner Signed-off-by: Bartlomiej Zolnierkiewicz

fbdev: Distinguish between interlaced and progressive modes

2018-07-24T17:11:24+00:00

I discovered the problem when developing a frame buffer driver for the PlayStation 2 (not yet merged), using the following video modes for the PlayStation 3 in drivers/video/fbdev/ps3fb.c: }, { /* 1080if */ "1080if", 50, 1920, 1080, 13468, 148, 484, 36, 4, 88, 5, FB_SYNC_BROADCAST, FB_VMODE_INTERLACED }, { /* 1080pf */ "1080pf", 50, 1920, 1080, 6734, 148, 484, 36, 4, 88, 5, FB_SYNC_BROADCAST, FB_VMODE_NONINTERLACED }, In ps3fb_probe, the mode_option module parameter is used with fb_find_mode but it can only select the interlaced variant of 1920x1080 since the loop matching the modes does not take the difference between interlaced and progressive modes into account. In short, without the patch, progressive 1920x1080 cannot be chosen as a mode_option parameter since fb_find_mode (falsely) thinks interlace is a perfect match. Signed-off-by: Fredrik Noring Cc: "Maciej W. Rozycki" [b.zolnierkie: updated patch description] Signed-off-by: Bartlomiej Zolnierkiewicz

console/fbcon: Add support for deferred console takeover

2018-06-28T13:20:30+00:00

Currently fbcon claims fbdevs as soon as they are registered and takes over the console as soon as the first fbdev gets registered. This behavior is undesirable in cases where a smooth graphical bootup is desired, in such cases we typically want the contents of the framebuffer (typically a vendor logo) to stay in place as is. The current solution for this problem (on embedded systems) is to not enable fbcon. This commit adds a new FRAMEBUFFER_CONSOLE_DEFERRED_TAKEOVER config option, which when enabled defers fbcon taking over the console from the dummy console until the first text is displayed on the console. Together with the "quiet" kernel commandline option, this allows fbcon to still be used together with a smooth graphical bootup, having it take over the console as soon as e.g. an error message is logged. Note the choice to detect the first console output in the dummycon driver, rather then handling this entirely inside the fbcon code, was made after 2 failed attempts to handle this entirely inside the fbcon code. The fbcon code is woven quite tightly into the console code, making this to only feasible option. Reviewed-by: Daniel Vetter Signed-off-by: Hans de Goede Signed-off-by: Bartlomiej Zolnierkiewicz

fbcon: Call WARN_CONSOLE_UNLOCKED() where applicable

2018-06-28T13:20:28+00:00

Replace comments about places where the console lock should be held with calls to WARN_CONSOLE_UNLOCKED() to assert that it is actually held. Acked-by: Steven Rostedt (VMware) Reviewed-by: Daniel Vetter Reviewed-by: Sergey Senozhatsky Signed-off-by: Hans de Goede Signed-off-by: Bartlomiej Zolnierkiewicz