Go Back   nV News Forums > Linux Support Forums > NVIDIA Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 09-28-07, 04:04 PM   #1
NV4TEHWIN
Registered User
 
Join Date: Sep 2007
Posts: 7
Default x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Argh.

From what I've researched on this, I'm leaning towards it being a BIOS problem re: 32-bit address registers, but since vesa driver works (albeit painfully slow) I'll ask...

I can't run X with nvidia. It either starts up with the black screen, or a wonderful psychedelic pattern. Sometimes after starting X it cold-boots the laptop, most of the time it just freezes up. I've tried all of the things in the Appendix: Known Issues section, including passing swiotlb=65536 and adjusting NVreg_RemapLimit accordingly. Since it's a core2 duo there's no IOMMU but software, I don't know if this is a good thing or a bad thing or even relevant at all.

All I want is a stable system with all 4G of my memory available.

Asus G2S-B1, BIOS v209 (most recent- 08/29/07), FC7 base, kernel 2.6.22.9 vanilla, compiled core2/xeon optimized, sparse memory, NVIDIA-Linux-x86_64-100.14.19-pkg2 compiles and loads without error...

Any help greatly appreciated.
Attached Files
File Type: bz2 nvidia-bug-report.log.bz2 (20.7 KB, 154 views)
File Type: bz2 Xorg.0.log.bz2 (5.5 KB, 138 views)
NV4TEHWIN is offline   Reply With Quote
Old 09-29-07, 06:13 AM   #2
chunkey
#!/?*
 
Join Date: Oct 2004
Posts: 662
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Wow,
Code:
ata1.00: exception Emask 0x2 SAct 0x3c SErr 0x0 action 0x2 frozen
ata1.00: (spurious completions during NCQ issue=0x0 SAct=0x3c FIS=005040a1:00000002)
ata1.00: cmd 60/50:10:6b:90:8d/00:00:0e:00:00/40 tag 2 cdb 0x0 data 40960 in
         res 50/00:70:bb:91:8d/00:00:0e:00:00/40 Emask 0x2 (HSM violation)
ata1.00: cmd 60/18:18:83:91:8d/00:00:0e:00:00/40 tag 3 cdb 0x0 data 12288 in
         res 50/00:70:bb:91:8d/00:00:0e:00:00/40 Emask 0x2 (HSM violation)
ata1.00: cmd 60/08:20:a3:91:8d/00:00:0e:00:00/40 tag 4 cdb 0x0 data 4096 in
         res 50/00:70:bb:91:8d/00:00:0e:00:00/40 Emask 0x2 (HSM violation)
ata1.00: cmd 60/70:28:bb:91:8d/00:00:0e:00:00/40 tag 5 cdb 0x0 data 57344 in
         res 50/00:70:bb:91:8d/00:00:0e:00:00/40 Emask 0x2 (HSM violation)
ata1: soft resetting port
Code:
NVRM: Xid (0001:00): 13, 0001 00000000 00005097 00000680 000206f2 00000080
NVRM: Xid (0001:00): 13, 0001 00000000 00005097 00000680 000206f2 00000080
NVRM: Xid (0001:00): 13, 0001 00000000 00005097 00000680 0002070b 00000080
NVRM: Xid (0001:00): 13, 0001 00000000 00005097 00000680 0002070b 00000080
NVRM: Xid (0001:00): 13, 0001 00000000 00005097 00000680 000206f7 00000080
NVRM: Xid (0001:00): 13, 0001 00000000 00005097 00000680 000206e2 00000080
NVRM: Xid (0001:00): 13, 0001 00000000 00005097 00000680 000206fd 00000080
NVRM: Xid (0001:00): 13, 0001 00000000 00005097 00000680 000206e1 00000080
Code:
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol 
nvidia: Unknown symbol
SATA errors, XIDs, unknown symbols?? sounds like your Laptop has some serious problems... Are you sure it isn't damaged? Anyway there is a kernel hack for the G1S with 4 GB...
chunkey is offline   Reply With Quote
Old 09-29-07, 05:16 PM   #3
NV4TEHWIN
Registered User
 
Join Date: Sep 2007
Posts: 7
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Nah, it's just brand spankin' new, the B1 revision of the G2S. (7700 Meron). The problem with buying just-released hardware, lack of support... The Hitachi SATA-II issue is known, reading kernel-dev I see they'll have a patch for it soon enough, the thing spits those errors but it works fine, right now it's just the video that's keeping me in Vista...

Uhm, isn't the diff you linked is for 32-bit not 64-bit? Oh wait, is this part of your original issue in this thread? Did you get it working? Dude speak! Details details... :-)

edit: Also, if you've already begun this process, what exactly are Xid errors, and the no decoder found errors, and the assigning a 64-bit address register but "no kernels support this yet" mean, etc.... It seems that nearly EVERYONE here seems to be complaining about roughly the same issue?? The threads here are quite difficult to glean information from, I guess I'll start sequentially reading them to find the ones with the gems...
NV4TEHWIN is offline   Reply With Quote
Old 09-29-07, 06:31 PM   #4
chunkey
#!/?*
 
Join Date: Oct 2004
Posts: 662
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Quote:
Originally Posted by NV4TEHWIN
Nah, it's just brand spankin' new, the B1 revision of the G2S. (7700 Meron). The problem with buying just-released hardware, lack of support... The Hitachi SATA-II issue is known, reading kernel-dev I see they'll have a patch for it soon enough, the thing spits those errors but it works fine, right now it's just the video that's keeping me in Vista...

Uhm, isn't the diff you linked is for 32-bit not 64-bit? Oh wait, is this part of your original issue in this thread? Did you get it working? Dude speak! Details details... :-)

edit: Also, if you've already begun this process, what exactly are Xid errors, and the no decoder found errors, and the assigning a 64-bit address register but "no kernels support this yet" mean, etc.... It seems that nearly EVERYONE here seems to be complaining about roughly the same issue?? The threads here are quite difficult to glean information from, I guess I'll start sequentially reading them to find the ones with the gems...
Well, think about it 4GB Ram and 32 Bit... That doesn't make sense
Of course it's made and tested on x86_64 arch (but it should work for IA32 too...).

And yes. Hack worked nicely for me. (However, I don't know if you have to modify it... Since the dmesg in your nvidia-bug-report.log is full of "Unknown Symbol" instead of "valuable" information.)

NVIDIA's FAQ
Quote:
Originally Posted by NVIDIA's FAQ
Q: My kernel log contains messages that are prefixed with "Xid"; what do these messages mean?

A: "Xid" messages indicate that a general GPU error occurred, most often due to the driver misprogramming the GPU or to corruption of the commands sent to the GPU. These messages provide diagnostic information that can be used by NVIDIA to aid in debugging reported problems.
So far... please try the patch and "maybe" all your XiDs will go away.
chunkey is offline   Reply With Quote
Old 10-01-07, 09:34 PM   #5
NV4TEHWIN
Registered User
 
Join Date: Sep 2007
Posts: 7
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Thanks. I tried your patch verbatim, but the same results. Psychedelic screen. (which you can't screen capture, it looks fine when I open the file in vista)

Thanks again. Got any other suggestions?
Attached Files
File Type: gz nvidia-bug-report.log.gz (28.8 KB, 180 views)
NV4TEHWIN is offline   Reply With Quote
Old 10-02-07, 06:32 AM   #6
chunkey
#!/?*
 
Join Date: Oct 2004
Posts: 662
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Quote:
Originally Posted by NV4TEHWIN
Thanks. I tried your patch verbatim, but the same results. Psychedelic screen. (which you can't screen capture, it looks fine when I open the file in vista)

Thanks again. Got any other suggestions?
well, according to your new bug-report.log the memory mappings are left untouched. So the addresses are maybe a bit different on the G2S.

add a
Code:
printk(KERN_INFO "PCI Device: mstart:%x - mend:%x\n", r->start, r->end);
right before the "if ((r->start == 0xbdf00000) && (r->end == 0xddefffff)) {" in arch/i386/pci/i386.c. This thing will fill your dmesg a bit so you should increase the "Kernel log buffer size" (CONFIG_LOG_BUF_SHIFT) a bit if you
see that the log is cut...

to find the right mapping just subtract the "mend" with "mstart"... and look for the one with the biggest delta. If you found it put the new "mstart" and "mend" values into the "if ((r->start == 0xbdf00000) && [...]".

you can also add:
Code:
printk(KERN_INFO "Changed GFX Bridge PREFETCH WINDOW\n");
between r->start = 0xc0000000; and r->end = 0xd0000000; in the same file for "debugging".
chunkey is offline   Reply With Quote
Old 10-02-07, 01:05 PM   #7
NV4TEHWIN
Registered User
 
Join Date: Sep 2007
Posts: 7
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Ack. While adding in the commenting lines I saw what I did wrong: booted into the wrong kernel and pasted that output - gee go figure it was the same!

OK my embarrassment aside, your patch defaults do match up to the largest delta and the corresponding main memory range advertised via lspci. My bad.

Code:
g2s kernel: PCI Device: mstart:9000 - mend:9fff
g2s kernel: PCI Device: mstart:f7d00000 - mend:fddfffff
g2s kernel: PCI Device: mstart:bdf00000 - mend:ddefffff
g2s kernel: PCI Device: mstart:fde00000 - mend:fdefffff
g2s kernel: PCI Device: mstart:fdf00000 - mend:fdffffff
g2s kernel: PCI Device: mstart:a000 - mend:afff
g2s kernel: PCI Device: mstart:fe000000 - mend:fe7fffff
g2s kernel: PCI Device: mstart:ddf00000 - mend:dfefffff
g2s kernel: PCI Device: mstart:b000 - mend:bfff
g2s kernel: PCI Device: mstart:fe800000 - mend:fe8fffff
g2s kernel: PCI Device: mstart:c000 - mend:cfff
g2s kernel: PCI Device: mstart:fe900000 - mend:fe9fffff
g2s kernel: PCI Device: mstart:fea00000 - mend:feafffff

01:00.0 VGA compatible controller: nVidia Corporation Unknown device 0407 (rev a1) (prog-if 00 [VGA])
	Subsystem: ASUSTeK Computer Inc. Unknown device 1515
	Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B-
	Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR-
	Latency: 0, Cache Line Size: 32 bytes
	Interrupt: pin A routed to IRQ 16
	Region 0: Memory at fc000000 (32-bit, non-prefetchable) [size16M]
	Region 1: Memory at c0000000 (64-bit, prefetchable) [size=256M]
	Region 3: Memory at fa000000 (64-bit, non-prefetchable) [size 32M]
	Region 5: I/O ports at 9c00 [size=128]
	Expansion ROM at fdde0000 [disabled] [size=128K]
However, it doesn't work. It's progress; I do get the desktop to draw. But if I do anything besides roll the mouse around, it deadlocks. Ctrl-Alt-Backspace will shut X down, but that console and every other console is now frozen and requires a hard boot. (impossible to ssh into, it's gone.)

Any other ideas?
NV4TEHWIN is offline   Reply With Quote
Old 10-02-07, 01:14 PM   #8
chunkey
#!/?*
 
Join Date: Oct 2004
Posts: 662
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Quote:
Originally Posted by NV4TEHWIN
Ack. While adding in the commenting lines I saw what I did wrong: booted into the wrong kernel and pasted that output - gee go figure it was the same!

OK my embarrassment aside, your patch defaults do match up to the largest delta and the corresponding main memory range advertised via lspci. My bad.

Code:
g2s kernel: PCI Device: mstart:9000 - mend:9fff
g2s kernel: PCI Device: mstart:f7d00000 - mend:fddfffff
g2s kernel: PCI Device: mstart:bdf00000 - mend:ddefffff
g2s kernel: PCI Device: mstart:fde00000 - mend:fdefffff
g2s kernel: PCI Device: mstart:fdf00000 - mend:fdffffff
g2s kernel: PCI Device: mstart:a000 - mend:afff
g2s kernel: PCI Device: mstart:fe000000 - mend:fe7fffff
g2s kernel: PCI Device: mstart:ddf00000 - mend:dfefffff
g2s kernel: PCI Device: mstart:b000 - mend:bfff
g2s kernel: PCI Device: mstart:fe800000 - mend:fe8fffff
g2s kernel: PCI Device: mstart:c000 - mend:cfff
g2s kernel: PCI Device: mstart:fe900000 - mend:fe9fffff
g2s kernel: PCI Device: mstart:fea00000 - mend:feafffff

01:00.0 VGA compatible controller: nVidia Corporation Unknown device 0407 (rev a1) (prog-if 00 [VGA])
	Subsystem: ASUSTeK Computer Inc. Unknown device 1515
	Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B-
	Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR-
	Latency: 0, Cache Line Size: 32 bytes
	Interrupt: pin A routed to IRQ 16
	Region 0: Memory at fc000000 (32-bit, non-prefetchable) [size16M]
	Region 1: Memory at c0000000 (64-bit, prefetchable) [size=256M]
	Region 3: Memory at fa000000 (64-bit, non-prefetchable) [size 32M]
	Region 5: I/O ports at 9c00 [size=128]
	Expansion ROM at fdde0000 [disabled] [size=128K]
However, it doesn't work. It's progress; I do get the desktop to draw. But if I do anything besides roll the mouse around, it deadlocks. Ctrl-Alt-Backspace will shut X down, but that console and every other console is now frozen and requires a hard boot. (impossible to ssh into, it's gone.)

Any other ideas?
Hmm, strange... Do you hear if the CPU fan goes nuts when it deadlocks? Or can you revive the laptop by killing X with ALT-SYSRQ-K (or -I, -E)?
(a updated nvidia-bug-report.log would be nice too.. just _boot_ into Runlevel 3.)
chunkey is offline   Reply With Quote

Old 10-02-07, 06:56 PM   #9
javich
Registered User
 
Join Date: Oct 2007
Posts: 4
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Quote:
Originally Posted by chunkey
Wow,

SATA errors, XIDs, unknown symbols?? sounds like your Laptop has some serious problems... Are you sure it isn't damaged? Anyway there is a kernel hack for the G1S with 4 GB...

Nice trick
This hack did the job.

I thought it would be ready in 2.6.22 but I had to patch the kernel.

Well at last my g1s is again up & running this time with 4GB

Tnx for the tip.
__________________
ASUS G1S-A1 4GB RAM, 8600M GT
javich is offline   Reply With Quote
Old 10-02-07, 07:16 PM   #10
chunkey
#!/?*
 
Join Date: Oct 2004
Posts: 662
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Quote:
Originally Posted by javich
Nice trick
This hack did the job.

I thought it would be ready in 2.6.22 but I had to patch the kernel.

Well at last my g1s is again up & running this time with 4GB

Tnx for the tip.
Nontheless, can you _please_ write a nice letter to ASUS and tell them to fix the BIOS?! Because this patch won't ever get merged into the kernel, for very good reasons!


Well, there's some more software for the G1S (and maybe G2S)... e.g.
- uvcvideo driver for the webcam
- asusoled a little program for the little 128x32 OLED Display
(you can use it to display any JPG/GIF/PNG (whatever libSDL-Image supports) on this little screen)
- a patch that adds some lmsensors stuff (you can read and SET the fan rpm!)
chunkey is offline   Reply With Quote
Old 10-02-07, 08:02 PM   #11
NV4TEHWIN
Registered User
 
Join Date: Sep 2007
Posts: 7
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

HEY! Ah! Damnit. It works fine. Zero Xid messages. It's libata-core needing that damn Hitachi TS722020K9SA00 to be blacklisted for spurious NCQ completion. I guess all the throughput from the startup of X is the first time the system needs enough disk I/O to queue up commands?? Well, doesn't matter why at this point. But it happened in the same place as I'm trying to get X video working, so I just assumed... So, now it's patched to be part of the ATA_HORKAGE_NONCQ collection and lo, I get into X. With nvidia. And everything behaves.

Thanks for all your assistance chunkey! You rock. Now the laptop does too.

And anyone else like javich who's finding this thread, if you have a Hitachi TS series drive in your Asus G Series, be on the lookout for those NCQ messages - add your drive to the blacklist in libata-core.c
Attached Files
File Type: gz nvidia-bug-report.log.gz (27.1 KB, 155 views)
NV4TEHWIN is offline   Reply With Quote
Old 10-02-07, 08:09 PM   #12
chunkey
#!/?*
 
Join Date: Oct 2004
Posts: 662
Default Re: x86_64/Core2Duo/8600M GT- Yet another "****INVALID MEM ALLOCATION**** person

Quote:
Originally Posted by NV4TEHWIN
HEY! Ah! Damnit. It works fine. Zero Xid messages. It's libata-core needing that damn Hitachi TS722020K9SA00 to be blacklisted for spurious NCQ completion. I guess all the throughput from the startup of X is the first time the system needs enough disk I/O to queue up commands?? Well, doesn't matter why at this point. But it happened in the same place as I'm trying to get X video working, so I just assumed... So, now it's patched to be part of the ATA_HORKAGE_NONCQ collection and lo, I get into X. With nvidia. And everything behaves.

Thanks for all your assistance chunkey! You rock. Now the laptop does too.

And anyone else like javich who's finding this thread, if you have a Hitachi TS series drive in your Asus G Series, be on the lookout for those NCQ messages - add your drive to the blacklist in libata-core.c


Don't forget the software... asusoled (new project is @ sourceforge.net) is very useful to _monitor_ the fan speed & cpu temp
chunkey is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
NV30 to be 256bit mem??? SurfMonkey Rumor Mill 16 10-04-02 05:29 AM

All times are GMT -5. The time now is 03:34 PM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.