fbdev: efifb: Cleanup fb_info in .fb_destroy rather than .remove
authorJavier Martinez Canillas <javierm@redhat.com>
Thu, 5 May 2022 22:05:40 +0000 (00:05 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 18 May 2022 08:26:48 +0000 (10:26 +0200)
commitcd3c8abbdc4ea1c67db992da45e6fd39595d278d
treef8cb232b471d2537e029a8bd0b55d17925d0e5f7
parent02eef429bf4a46396e4408d8e4ca2ebed20650d1
fbdev: efifb: Cleanup fb_info in .fb_destroy rather than .remove

[ Upstream commit d258d00fb9c7c0cdf9d10c1ded84f10339d2d349 ]

The driver is calling framebuffer_release() in its .remove callback, but
this will cause the struct fb_info to be freed too early. Since it could
be that a reference is still hold to it if user-space opened the fbdev.

This would lead to a use-after-free error if the framebuffer device was
unregistered but later a user-space process tries to close the fbdev fd.

To prevent this, move the framebuffer_release() call to fb_ops.fb_destroy
instead of doing it in the driver's .remove callback.

Strictly speaking, the code flow in the driver is still wrong because all
the hardware cleanupd (i.e: iounmap) should be done in .remove while the
software cleanup (i.e: releasing the framebuffer) should be done in the
.fb_destroy handler. But this at least makes to match the behavior before
commit 27599aacbaef ("fbdev: Hot-unplug firmware fb devices on forced removal").

Fixes: 27599aacbaef ("fbdev: Hot-unplug firmware fb devices on forced removal")
Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20220505220540.366218-1-javierm@redhat.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
drivers/video/fbdev/efifb.c