lots of "bridge window ... (disabled)" messages
Hi Bjorn,
I just booted Linus latest git tree on my Tukwila system. I'm seeing a lot of bridge windows marked as disabled. Here's the start of a diff of dmesg output comparing a kernel from last week to today's kernel (git 01e7770 vs. b72c409):
< pci 0000:00:01.0: bridge window [mem 0xfff00000 - 000fffff pref] reg reading
---
Quoted 1 line
pci 0000:00:01.0: bridge window [mem 0xfff00000-0x000fffff pref] (disabled)
222,224c222,224
< pci 0000:00:03.0: bridge window [io f000 - 0000] reg reading < pci 0000:00:03.0: bridge window [mem 0xfff00000 - 0x000fffff] reg reading < pci 0000:00:03.0: bridge window [mem 0xfff00000 - 000fffff pref] reg reading
---
Quoted 3 lines
pci 0000:00:03.0: bridge window [io 0xf000-0x0000] (disabled) pci 0000:00:03.0: bridge window [mem 0xfff00000-0x000fffff] (disabled) pci 0000:00:03.0: bridge window [mem 0xfff00000-0x000fffff pref] (disabled)
233c233
< pci 0000:00:05.0: bridge window [mem 0xfff00000 - 000fffff pref] reg reading
---
Quoted 1 line
pci 0000:00:05.0: bridge window [mem 0xfff00000-0x000fffff pref] (disabled)
... many more such diffs.
This is the same system that you helped out with a couple of months ago that had some issues in _CRS tables with ranges marked as "consumer". The BIOS team fixed the one range that was needed to allow access to my EHCI - but they didn't touch anything else.
Are these new checks related to the same sort of issues in _CRS? Or is this some other stuff? I don't appear to have lost access to any devices in today's kernel.
-Tony
Quoted 10 lines · Tony
Hi Bjorn,
I just booted Linus latest git tree on my Tukwila system. I'm seeing a lot of bridge windows marked as disabled. Here's the start of a diff of dmesg output comparing a kernel from last week to today's kernel (git 01e7770 vs. b72c409):
< pci 0000:00:01.0: bridge window [mem
0xfff00000-000fffffpref] reg reading
---Quoted 1 line
pci 0000:00:01.0: bridge window [mem 0xfff00000-0x000fffff pref] (disabled)
Nothing changed here except the message. The bridge windows are disabled in both cases (same values in base & limit, with base > limit); I just changed the message from "reg reading" to "(disabled)" because I thought that would be easier to interpret.
Bjorn