Summary: | x11-drivers/nvidia-drivers-313.18 with kernel 3.7.4 - NVRM: This PCI I/O region assigned to your NVIDIA device is invalid: BAR1 is 0M @ 0x0 (PCI:0000:03:00.0) | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Roman Gruber <roman.gruber> |
Component: | [OLD] Library | Assignee: | Jeroen Roovers (RETIRED) <jer> |
Status: | RESOLVED DUPLICATE | ||
Severity: | normal | CC: | jer, kernel, tomwij, xarthisius |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
URL: | https://bugzilla.kernel.org/show_bug.cgi?id=53181 | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
emerge --info
kernel 3.3.8 working /var/log/messages kernel 3.7.4-gentoo and x11-drivers/nvidia-drivers-313.18 not working gentoo 3.7.4 kernel config 3.3.8 kernel config (the working kernel) lspci kernel 3.3.8 working |
Description
Roman Gruber
2013-01-29 17:43:29 UTC
Created attachment 337252 [details]
emerge --info
Created attachment 337254 [details]
kernel 3.3.8 working /var/log/messages
Created attachment 337256 [details]
kernel 3.7.4-gentoo and x11-drivers/nvidia-drivers-313.18 not working
nvidia-driver 313.18 and 310.19(which works with 3.3.8) doesnt works ...
Created attachment 337258 [details]
gentoo 3.7.4 kernel config
Created attachment 337260 [details]
3.3.8 kernel config (the working kernel)
Created attachment 337262 [details]
lspci kernel 3.3.8 working
Importance is to be set by the maintainers. > 3.6.11 I run =sys-kernel/gentoo-sources-3.6.11 with =x11-drivers/nvidia-drivers-313.18 and I can't reproduce this; so, this is specific to your hardware configuration. > Jan 29 18:14:08 localhost kernel: [ 110.181043] NVRM: The system BIOS may have misconfigured your GPU. This is the first occurence of something that tells us what may have happened, this one is particularly interesting as this would indicate your BIOS does something incorrectly; suggestion: Can you try upgrading your BIOS? > REading on google there may be a way to set this area with setpci. Do you have a link to this, it can put some light on the details this involves. > as several kernel worked in teh past, and even the working nvidia drivers dont work anymore with this newer kernel, its a kernel bug. I can write "as several nvidia drivers worked in the past, and even the working kernels dont work anymore with this newer nvidia driver, its a nvidia bug" as well, this doesn't really mean anything. All we know is that there is a high chance that this is a bug in your BIOS, this would mean the BIOS developers have the main responsibility to fix this if that's the case. If we put BIOS out of the view (because it's not always easy to get them to do that), then you can still wonder whether the bug lies with Nvidia or the kernel. And since Nvidia is proprietary we can't really tell this soon. Though, setpci works so it seems a quirk fix from the kernel is possible. > Please if I should submit a bug report somewhere else pls post me the relevant links and I do it. 1) Can you try upgrading your BIOS? 2) Can you try to contact your BIOS developers about this? 3) Could you link to that setpci trick you have found? 4) Could you report this at http://bugzilla.kernel.org and comment here with a link to that bug? 5) If they won't do it, can you try to contact Nvidia about this? 1) Can you try upgrading your BIOS? 2) Can you try to contact your BIOS developers about this? No, its an ASUSG70SG Laptop, 3 Years old. Asus dont support GNU LINUX. 3) Could you link to that setpci trick you have found? e.g like that is the problem.. http://www.nvnews.net/vbulletin/showthread.php?t=42906 Region 1: Memory at <ignored> (64-bit, prefetchable) well what i have written down and treid from google, but is not working. and what was always suggested something like that. and i changed it so it fits my chard, my card is at 03.00.0 instead of 01.00.0 setpci -s 03.00.0 BASE_ADDRESS_0=0x0xf2000000 honestly i have a knowledge about hex numbers. but if its now BASE_ADDRESS_1 or 0 or something else ... i only know ihave to set the range to 0xf2000000 as you can see on the attachmetns i made, becasue Region1 is ignored. 4) Could you report this at http://bugzilla.kernel.org and comment here with a link to that bug? OKAY, sure will do. 5) If they won't do it, can you try to contact Nvidia about this? well thats the issue. i read their forum and they hardly fix anything. ------------ Personal opinion: In my understandings the Bios should set the memory regions and the IRQ and such stuff. And the kernel overwrites this areas or changes these settings. There are userland tools or kernel settings to get over the "borked" bios. But thats a matter of defintion whats "borked" and whats not. Thank you for your response. I have read that linux has diferent memory mappings and thats the problem. as some memoery mappings overwrite block(which seems the case here) "used" areas. Thank you for reporting upstream. > setpci -s 03.00.0 BASE_ADDRESS_0=0x0xf2000000 03.00.0 should be 03:00.0 according to `setpci --help`'s listed formats. 0x0xf2000000 should be 0xf2000000, you had 0x twice. Also, the _0 behind BASE_ADDRESS might have to be _1 or something else. You can figure out which region is set incorrectly by executing `lspci -vv` and looking for your graphics card there. So, it should be something along the lines of > setpci -s 03:00.0 BASE_ADDRESS_1=0xf2000000 *** This bug has been marked as a duplicate of bug 447566 *** |