black screen with kubuntu 8.04 x86_64 and fglrx 8.5

[hehe2]

If the extra memory isn't remapped, it will be hidden by the PCI memory mapping.

That said, I guess the OS could do remapping itself, but by the time it boots that may be too late.

I've made the test yesterday, disabling the memory remapping in the bios. That leads to 3.3Gb usable only, even with the 64Gb PAE kernel.

In fact, as soon as the PC starts, bios detects only 3.3Gb without memory remapping (startup bios display), and I presume the memory lost can't be recovered by the OS in a later time.

[kingtaurus]

Did you try using the mem=xxxxMB as a kernel option?

I've made the test yesterday, disabling the memory remapping in the bios. That leads to 3.3Gb usable only, even with the 64Gb PAE kernel.

In fact, as soon as the PC starts, bios detects only 3.3Gb without memory remapping (startup bios display), and I presume the memory lost can't be recovered by the OS in a later time.

[Kano]

only M not MB.

[kingtaurus]

Ooops ... stupid me .

You might have to do the following:

[X86-32] Use together with memmap= to avoid physical
address space collisions. Without memmap= PCI devices
could be placed at addresses belonging to unused RAM.

From (http://www.mjmwired.net/kernel/Docum...arameters.txt)

Kano ... is there an x86_64 kernel parameter like pci=nobios (x86_32)?

only M not MB.

[kingtaurus]

Ooops ... stupid me .

You might have to do the following:

[X86-32] Use together with memmap= to avoid physical
address space collisions. Without memmap= PCI devices
could be placed at addresses belonging to unused RAM.

From (http://www.mjmwired.net/kernel/Docum...arameters.txt)

Kano ... is there an x86_64 kernel parameter like pci=nobios (x86_32)?

MTRR is now setup properly using ubuntu-8.10 release candidate. My tables are now configured like:
gking@bender:~$ cat /proc/mtrr
reg01: base=0xe0000000 (3584MB), size= 512MB: uncachable, count=1
reg02: base=0x00000000 ( 0MB), size=4096MB: write-back, count=1
reg03: base=0x100000000 (4096MB), size= 512MB: write-back, count=1
reg04: base=0x120000000 (4608MB), size= 256MB: write-back, count=1

[alec]

this is good news!

[aslpavel]

2kingtaurus
could u provide exect kernel options that u use to achieve shift of pci device from 0xd000000 to 0xe0000000. I think porblems some how related to alignment of the "base" it should be aligned to the "size" but if we got 512Mb video ram and it mapped to 3328 by bios it cannot be alligned this way and when driver try to access higher ranges of memory it crashes system. Actualy catalyst work for me but randomly crashes when i use comositing.

Mb: Asus P5Q Pro
Mem: 2x2Gb
Cpu: Q9300
Video: HD4870

Thx

[kingtaurus]

You can set the mtrr by (hand). However I would caution against this. My mtrr tables are currently set to:

reg01: base=0x0e0000000 ( 3584MB), size=  512MB, count=1: uncachable
reg02: base=0x000000000 (    0MB), size= 4096MB, count=1: write-back
reg03: base=0x100000000 ( 4096MB), size=  512MB, count=1: write-back
reg04: base=0x120000000 ( 4608MB), size=  256MB, count=1: write-back

these are set by the kernel (I don't modify the tables by hand). Currently using Jaunty (9.04) with the corresponding

uname -a

being

Linux bender 2.6.28-11-generic #42-Ubuntu SMP Fri Apr 17 01:58:03 UTC 2009 x86_64 GNU/Linux

2kingtaurus
could u provide exect kernel options that u use to achieve shift of pci device from 0xd000000 to 0xe0000000. I think porblems some how related to alignment of the "base" it should be aligned to the "size" but if we got 512Mb video ram and it mapped to 3328 by bios it cannot be alligned this way and when driver try to access higher ranges of memory it crashes system. Actualy catalyst work for me but randomly crashes when i use comositing.

Mb: Asus P5Q Pro
Mem: 2x2Gb
Cpu: Q9300
Video: HD4870

Thx

[aslpavel]

actualy there is no use changin mtrr unless u device realy located on 0xe0000000 mine is located at 0xd0000000:

01:00.0 VGA compatible controller: ATI Technologies Inc RV770 [Radeon HD 4870] (prog-if 00 [VGA controller])
	Subsystem: ATI Technologies Inc Device 0502
	Flags: bus master, fast devsel, latency 0, IRQ 30
	Memory at d0000000 (64-bit, prefetchable) [size=256M]
	Memory at fe8e0000 (64-bit, non-prefetchable) [size=64K]
	I/O ports at c000 [size=256]
	Expansion ROM at fe8c0000 [disabled] [size=128K]
	Capabilities: [50] Power Management version 3
	Capabilities: [58] Express Legacy Endpoint, MSI 00
	Capabilities: [a0] MSI: Mask- 64bit+ Count=1/1 Enable+
	Capabilities: [100] Vendor Specific Information <?>
	Kernel driver in use: fglrx_pci
	Kernel modules: fglrx

i thought u find a way to shift device to 256Mb in address space
with compoziting enabled i got crashes like this:

Backtrace:
0: X(xorg_backtrace+0x26) [0x4eaff6]
1: X(xf86SigHandler+0x39) [0x487f89]
2: /lib/libc.so.6 [0x7f3ca7941270]
3: /usr/lib64/xorg/modules//libxaa.so [0x7f3ca5ce8dec]
4: /usr/lib64/xorg/modules//libxaa.so [0x7f3ca5ce95f0]
5: X [0x52c3d8]
6: X [0x51b63a]
7: X(Dispatch+0x364) [0x44a404]
8: X(main+0x44d) [0x430b2d]
9: /lib/libc.so.6(__libc_start_main+0xe6) [0x7f3ca792d5c6]
10: X [0x42ff19]

Fatal server error:
Caught signal 11.  Server aborting

and kernel log like this

May  6 23:17:46 maggot [fglrx:firegl_find_any_map] *ERROR* Invalid map handle!<3>[fglrx:drm_vm_close] *ERROR* map not found -> inconsistent kernel data!!! vma_start:0x7fbfc7228000,handle:0xd13f5000
May  6 23:17:46 maggot [fglrx:firegl_find_any_map] *ERROR* Invalid map handle!<3>[fglrx:drm_vm_close] *ERROR* map not found -> inconsistent kernel data!!! vma_start:0x7fbfc7229000,handle:0xd13f4000
May  6 23:17:46 maggot [fglrx:firegl_find_any_map] *ERROR* Invalid map handle!<3>[fglrx:drm_vm_close] *ERROR* map not found -> inconsistent kernel data!!! vma_start:0x7fbfc722a000,handle:0xd13f0000
May  6 23:17:46 maggot [fglrx:firegl_takedown] *ERROR* map 0xffff8800cbcb7a80 - type UNKNOWN 16
May  6 23:17:46 maggot [fglrx:firegl_takedown] *ERROR* map 0xffff8800cd1cfe40 - type UNKNOWN 16
May  6 23:17:46 maggot [fglrx:firegl_takedown] *ERROR* map 0xffff8800cdd9d120 - type UNKNOWN 16
May  6 23:17:46 maggot [fglrx:firegl_takedown] *ERROR* map 0xffff8800cdd9de40 - type UNKNOWN 16
May  6 23:17:46 maggot [fglrx:firegl_takedown] *ERROR* map 0xffff8800cc116c60 - type UNKNOWN 16

After this the only way to make it work again is reboot, if i try
to start X again it hangs whole system even ssh connetion fails
Any suggestions?