Summary: | [C51PV] Garbage on initial Console Screen | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Product: | xorg | Reporter: | baldur | ||||||||
Component: | Driver/nouveau | Assignee: | Nouveau Project <nouveau> | ||||||||
Status: | RESOLVED DUPLICATE | QA Contact: | Xorg Project Team <xorg-team> | ||||||||
Severity: | critical | ||||||||||
Priority: | medium | ||||||||||
Version: | unspecified | ||||||||||
Hardware: | x86-64 (AMD64) | ||||||||||
OS: | Linux (All) | ||||||||||
Whiteboard: | |||||||||||
i915 platform: | i915 features: | ||||||||||
Attachments: |
|
Created attachment 67248 [details]
lspci -vv output
Created attachment 67250 [details]
/var/log/dmesg from fedora
as there are some initial lines missing from my dmesg output, this is what fedora had stored...
does it work if you boot with mem=2G in kernel command line? yes it works! (with mem=2G) But memory is already low on my system, so limiting it to 2G is not an option for me. This is an 64 bit system. So i am wondering why i have a 32bit (2G?) problem? Your card has problems addressing memory above 2G. As a temporary workaround you can apply patch from bug 46557. *** This bug has been marked as a duplicate of bug 46557 *** Marcin: As I mentioned in 46557, I'm not convinced that the problem is hardware. It could easily be some widening bug in C code. It smells like sign-extension of a 32 bit quantity that should be zero-extended. But it could be the truncation of a wider-than-32-bit quantity to 32 bits. And, of course, it could be hardware. Why would the 32-bit Linux have no trouble with the identical hardware? I can add another data point, which indicates it is not the hardware. This card works "flawless" with the nvidia driver since 4.5 years. It is only my curiosity, why i try nouveau from time to time. As recompiling the kernel on every new update from fedora is not an option i am going back to the nvidia driver. But at least i know now how to install a "fresh" fedora (mem=2G), so your help is greatly appreciated! D. Hugh Redelmeier: If you can prove it's "widening bug", please do (by fixing it). Memory used by the kernel in 32-bit mode is allocated in first GB, so you can't hit this bug there. Dietmar: Nvidia driver may: a) init the hardware in a way we don't, fixing this bug b) workaround it internally by allocating memory in first 2G |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.
Created attachment 67247 [details] dmesg output Dear all, after having bad experiences with nouveau (using fedora 15,16,17) , which resulted in an immediate lockup of my old server machine, i have given noveau another chance to run on a fedora kernel: Linux version 3.5.3-1.fc17.x86_64 (mockbuild@) The good is news is - it doesn't lead to immediate lockup of the machine anymore. The bad news. It still leads to a scrambled console screen, which does not contain any useful information. It seems that there is an "INVALID_CMD" issued by the noveau driver. Can anybody help me to resolve this problem?