We already call remove_conflicting_framebuffers() on PCI BAR0 during pci-probe, no need to do that again during device loading.
This avoids calling into remove_conflicting_framebuffers() from within DRM ->load() callback, which might deadlock, once we make this call remove DRM-backed sysfb-devices, too.
Signed-off-by: David Herrmann dh.herrmann@gmail.com --- drivers/gpu/drm/mgag200/mgag200_main.c | 9 --------- 1 file changed, 9 deletions(-)
diff --git a/drivers/gpu/drm/mgag200/mgag200_main.c b/drivers/gpu/drm/mgag200/mgag200_main.c index 26868e5..0ee093c 100644 --- a/drivers/gpu/drm/mgag200/mgag200_main.c +++ b/drivers/gpu/drm/mgag200/mgag200_main.c @@ -117,20 +117,11 @@ static int mga_probe_vram(struct mga_device *mdev, void __iomem *mem) static int mga_vram_init(struct mga_device *mdev) { void __iomem *mem; - struct apertures_struct *aper = alloc_apertures(1); - if (!aper) - return -ENOMEM;
/* BAR 0 is VRAM */ mdev->mc.vram_base = pci_resource_start(mdev->dev->pdev, 0); mdev->mc.vram_window = pci_resource_len(mdev->dev->pdev, 0);
- aper->ranges[0].base = mdev->mc.vram_base; - aper->ranges[0].size = mdev->mc.vram_window; - - remove_conflicting_framebuffers(aper, "mgafb", true); - kfree(aper); - if (!devm_request_mem_region(mdev->dev->dev, mdev->mc.vram_base, mdev->mc.vram_window, "mgadrmfb_vram")) { DRM_ERROR("can't reserve VRAM\n");