Bug 16497
Summary: | Address space collision: host bridge window conflicts with Video ROM | ||
---|---|---|---|
Product: | Drivers | Reporter: | MM (c3904a4a) |
Component: | Other | Assignee: | Bjorn Helgaas (bjorn) |
Status: | CLOSED CODE_FIX | ||
Severity: | high | CC: | akpm, alan, bjorn, fedotov.i.f, florian, jbarnes, nao.nakashima |
Priority: | P1 | ||
Hardware: | All | ||
OS: | Linux | ||
Kernel Version: | 2.6.35 | Subsystem: | |
Regression: | No | Bisected commit-id: | |
Attachments: |
cat /proc/cpuinfo
cat /proc/version cat /proc/modules cat /proc/ioports cat /proc/iomem lspci -vvv cat /proc/scsi/scsi git bisect log (linux-staging-next-2.6) |
Description
MM
2010-08-03 05:38:25 UTC
We don't appear to have a category for PCI. I cc'ed Jesse. Might be an acpi issue though. Ok great, so any word back on this from Jesse? I hope we can get this fixed soon. THANKS Same bug. Tested with versions: 2.6.35, 2.6.37-rc2 1. pci_root PNP0A08:00: host bridge window [mem 0xfed45000-0xffffffff] ... pci_root PNP0A08:00: address space collision: host bridge window [mem ...] conflicts with video ROM [mem ...] Then nothing is displayed and num lock blinks (kernel panic?) 2. Ubuntu 10.04, Gentoo Created attachment 37462 [details]
cat /proc/cpuinfo
Created attachment 37472 [details]
cat /proc/version
Created attachment 37482 [details]
cat /proc/modules
Created attachment 37492 [details]
cat /proc/ioports
Created attachment 37502 [details]
cat /proc/iomem
Created attachment 37512 [details]
lspci -vvv
Created attachment 37522 [details]
cat /proc/scsi/scsi
Above is for 2.6.32-25 on Ubuntu 10.10 because 2.6.35 doesn`t work. Also I tried to git bisect this bug on linux-staging-next-2.6 tree some time ago, but fail because same versions was marked good and bad. I think that this bug can be enabled/disabled in .config Created attachment 37552 [details]
git bisect log (linux-staging-next-2.6)
Notice good/bad commit 494a43bb188c10b9b21375bbfe2691e41e3cb145
This is because different .config used.
I'm still getting this bug with 2.6.36.1 PCI: Using host bridge windows from ACPI; if necessary, use "pci=nocrs" and report a bug pci_root PNP0A08:00: address space collision: host bridge window [mem 0x000cc000-0x000cffff] conflicts with Video ROM [mem 0x000c0000-0x000ce9ff] I forgot to mention if I use lilo like this I don't see the message; append="pci=nocrs vt.default_utf8=0" By my understanding is I shouldn't need to use pci=nocrs? THANKS I apologize for missing this bug report. I happened to trip over it while researching bug #38522. Das: Are you seeing a functional problem in addition to the host bridge window conflict warning? The message itself isn't something that should cause any problems, and I plan to reduce it from KERN_WARN to KERN_INFO, so you won't see it any more. If there's something that doesn't work, in addition to the message, please attach a complete dmesg log from a recent kernel, 3.0 if possible (with "pci=nocrs" if necessary), and details about what is broken. If you have to use "pci=nocrs" to boot, that is a bug that I want to fix. Nao: It sounds like you see the message and, in addition, you have a hang. If the hang still occurs when you boot with "pci=nocrs", it's likely unrelated. Can you collect any more information about the hang, e.g., by booting with "ignore_loglevel" and using a serial console or taking a photo of the screen? I don't see any problems other then the log message, that made me wonder and I do use pci=nocrs... THANKS Actually let me back up here, I do see one problem, if I go from kernel 2.6.36.4 to anything higher, like .37, .38, or .39 with the Radeon KMS I'm using in the kernel, when the Radeon KMS kicks in the first few lines of messages appear in the middle of my screen then after that they appear as they should going down the far left side of the screen. I'll paste below my kernel config maybe someone can tell me why this configuration works fine in 36.4 but doesn't in newer kernels? Here's my kernel config for 2.6.36.4 http://pastebin.com/Cxhadb0K THANKS You should stop using "pci=nocrs" if you can. It won't cause you any problems, but it can cover up bugs that I'd like to fix. The Radeon KMS problem is completely unrelated, and it would be great if you could open a separate bug report for that. Maybe even attach a couple photos to make the problem concrete. Ok I'll remove the pci=nocrs, what would you like to see regarding this in dmesg I presume? THANKS Here's the patch I intend to use to fix the alarming error message: https://bugzilla.kernel.org/show_bug.cgi?id=38522#c7 That should be enough to resolve this bug. I'm going to treat this bug report as merely "message should be KERN_INFO, not KERN_ERR". If you discover an issue when you remove "pci=nocrs", e.g., you machine stops booting or something, we should open a new bug report for that. And if Nao can still reproduce the hang, we should open a new bug report for that, too. *** This bug has been marked as a duplicate of bug 38522 *** Das, I don't need a dmesg log or anything unless you see a problem when you remove "pci=nocrs". If you do see a problem, just open a new defect (or email the dmesg log to me), so we don't clutter this bug report. Well I'd feel better if you looked at the dmesg to make sure, better safe then sorry. :) http://pastebin.com/D88h1DES THANKS I think you're safe :) I don't see any devices that are having problems. The only thing I see is what looks like a minor BIOS bug: 215: bus: [00, 0f] on node 0 link 0 220: PCI: MMCONFIG for domain 0000 [bus 00-0f] at [mem 0xf7000000-0xf7ffffff] (base 0xf7000000) 236: ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff]) At 215, amd_bus.c reads chipset config and finds a host bridge leading to PCI buses 00-0f. At 220, the ACPI MCFG table tells us the MMCONFIG area for buses 00-0f. But at 236, the ACPI PNP0A08 device tells us the host bridge leads to buses 00-ff. This probably should be 00-0f, to match the ranges at 215 and 220. But this isn't a problem for you because the highest device you have is 08:00.0, so you don't need to use any buses above 0f. Ok great thanks... Also if you know anyone that deals with this bug; https://bugzilla.kernel.org/show_bug.cgi?id=38592 I'd appreciate you letting them and someone getting on this. I submitted this at the beginning of the year and I lost it and submitted this over, I've waited now almost 6 months on this with no word... THANKS TYPO; I meant I'd appreciate you letting them know... THANKS Hello, Let's please not marked this as resolved, the patch in the other bug report does not work and this is now in the 3.0 kernel as well, which I was hoping was going to be fixed in 3.0 and it was not... :( This is the output below from 3.0; pci_root PNP0A08:00: address space collision: host bridge window [mem 0x000cc000-0x000cffff] conflicts with Video ROM [mem 0x000c0000-0x000ce9ff] THANKS A patch referencing this bug report has been merged in Linux v3.1-rc1: commit 43d786ed4df4c54cb8802a523748a7d78130a2cb Author: Bjorn Helgaas <bhelgaas@google.com> Date: Sat Jul 2 10:47:12 2011 -0600 x86/PCI: reduce severity of host bridge window conflict warnings |