Hi, I'm testing Jerome Glisse's drm-next-radeon branch which provides Radeon Kernel-modesetting support on a Samsung UP1500 with a PCI Radeon X1550. See http://cgit.freedesktop.org/~glisse/drm-next/log/?h=drm-next-radeon While doing so, I've hit a few snags, and placed DRM_INFOs and printks throughout code in question. On Alpha, it hangs after calling fb_notifier_call_chain in register_framebuffer. Jerome and I think that the hang may be caused by being unable to read the PCI ROM. The radeon_read_bios function makes use of pci_map_rom, which calls pci_get_rom_size. In pci_get_rom_size, readb(image) doesn't return 0x55. It falls back to read_disabled functions which appear to succeed. Why would the ROM not be able to be read by readb? Is this a general problem, or Alpha-specific? Attached are relevant dmesg logs and printk patches (against commit 7582b4332602dcf03bb0c79f17e0cf89eeec6623) to give context. Thanks, Matt Turner
Attachment:
radeon-kms.log
Description: Binary data
Attachment:
radeon-kms-debugging.patch
Description: Binary data
