Ok, firstly, thank Dan Carpenter for pointing out my spelling error and I upload a PoC and compile it to a binary, hope that helpful for you to test it. regards, butt3rflyh4ck On Tue, Jul 14, 2020 at 12:12 AM Daniel Vetter <daniel@xxxxxxxx> wrote: > > Adding Thomas, who's the main author for vram helpers. > -Daniel > > On Fri, Jul 10, 2020 at 1:53 PM Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote: > > > > On Fri, Jul 10, 2020 at 04:24:03PM +0800, butt3rflyh4ck wrote: > > > I report a bug (in linux-5.8.0-rc4) found by syzkaller. > > > > > > kernel config: https://github.com/butterflyhack/syzkaller-fuzz/blob/master/v5.8.0-rc4.config > > > > > > I test the reproducer and crash too. > > > > > > In the drm_em_vram_t() function, ttm_bo_init() function call > > ^^^^^^^^^^^^^ > > This a typo. The function name is drm_gem_vram_init(). > > > > > ttm_bo_init_reserved(), > > > the ttm_bo_init_reserved() function call ttm_bo_put(), it will free > > > gbo->bo that is struct ttm_buffer_object. > > > > > > then, goto the err_drm_gem_object_release lable, > > > drm_gem_object_release() function will free gbo->bo.base, so cause use > > > after free. > > > > > > > There is a third free in drm_gem_vram_create(). This is a triple free > > bug. The correct place to free this is in drm_gem_vram_create() because > > that's where it was allocated. > > > > This code is quite subtle so I'm not going to attempt to fix it because > > I can't test it. > > > > regards, > > dan carpenter > > > > > -- > Daniel Vetter > Software Engineer, Intel Corporation > http://blog.ffwll.ch
Attachment:
uaf-drm_gem_object_release
Description: Binary data
Attachment:
uaf-drm_gem_object_release.c
Description: Binary data
_______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel
