Almico's Bug Tracker

View Issue Details Jump to Notes ] Issue History ] Print ]
IDProjectCategoryView StatusDate SubmittedLast Update
0001341SpeedFanOtherpublic2009-01-26 11:212011-03-19 12:07
Reporterdiscostu 
Assigned Toalfredo 
PrioritynormalSeveritycrashReproducibilitysometimes
StatusresolvedResolutionfixed 
PlatformAbit NF7-S v2.0OSWindows XP SP3OS Version4.35
Product Version4.35 
Target VersionFixed in Version4.43 
Summary0001341: Speedfan may freeze in automatic mode causing 100% CPU usage
DescriptionThis issue is similar to 0001093:
http://www.bugtrack.almico.com/bug_view_advanced_page.php?bug_id=1093 [^]

If speedfan is configured for automatic fan speed and I run a 3D app or a media player, sometimes (but regularly) the PC stalls after a while because speedfan is using 100% CPU time. In this situation, the speedfan user interface becomes unresponsive (cannot open the system tray menu) and all fan speeds seem to be frozen at current values - very dangerous, especially because of permanent full CPU load! Speedfan never recovers and needs to be terminated using the Windows task manager. When Speedfan is restarted, all fan speeds are reset to 100% and normal operation resumes.
This situation only occurs if speedfan is configured to automatic mode. As a workaround, I can switch speedfan to manual mode and set all fans to 100% before starting a 3D/media application.
Steps To Reproduce1. Configure Speedfan for Automatic fan speed
2. Run intensive 3D/media app

Result: Speedfan freezes causing permanent 100% CPU load.
Additional InformationI am currently using Speedfan v4.35, because of issue 1295 introduced in v4.36:
http://www.bugtrack.almico.com/view.php?id=1295 [^]
The freezing behaviour seems to be introduced recently, i.e. in version 4.35 or shortly before that, as I've been using Speedfan for years now on this PC and cannot recall this from happening before, with older Speedfan versions.
FYI, I attached a Speedfan debug information dump file.
TagsFreeze, V9999
Motherboard Model
Video Card ModelAsus V9999 (nVidia 6800)
Attached Filestxt file icon Speedfan 4.35 NF7-S.txt [^] (50,375 bytes) 2009-01-26 11:21 [Show Content]
? file icon debug.nfo [^] (43,477 bytes) 2009-02-08 17:21
? file icon speedfansens.cfg [^] (4,963 bytes) 2009-02-08 17:22
? file icon F75373S.cfg [^] (8,822 bytes) 2009-02-08 17:22

- Relationships
related to 0001295acknowledgedalfredo Fan control for Asus V9999 no longer works in version 4.36 

-  Notes
(0004366)
discostu (reporter)
2009-01-26 14:51

The problem seems to be related to the Asus V9999 videocard. When Speedfan is causing the 100% CPU load, the main window continuously displays messages like the following:

  SMBus msg : 14:40:07.937 Unsuccessful SELB $18 to $2E

Note that port 2Eh is connected to the Fintek F75375S on the V9999, as Speedfan shows on startup:

  F75375S found on SMBus at $2E

I'm currently using the nVidia GeForce driver version 185.20, but the problem first occurred when I was still using a driver from the 170 series.
(0004373)
discostu (reporter)
2009-01-28 19:04
edited on: 2009-01-28 19:05

I have some more information regarding this issue.

I have rolled back to Speedfan v4.34, but the 100% CPU issue still occurs. Lately I've been doing some web development using Visual Studio 2008, IE and Firefox and the problem occurs several times per day. This means that the issue is not restricted to graphics or multimedia applications, as I originally presumed. Also I just noticed the 100% CPU load while returning from the standard (2D) windows screensaver.
Is it possible that the cause of this problem is related to changes in the recent nVidia drivers, i.e. 180 and possibly 170 series? Besides from updating the nVidia graphics driver, I haven't made any significant hardware/driver configuration changes to my system for quite some time.

Note: when the spin lock occurs, Speedfan is stuck forever at 100% CPU and never returns to normal again. But after a manual restart, it always resumes normal operation including V9999 measurements. This seems to contradict any corruption in the internal nVidia driver state or hardware state. Could the problem be caused by a threading synchronization error in Speedfan?

(0004380)
discostu (reporter)
2009-02-04 21:49
edited on: 2009-02-04 21:54

Hello Alfredo,

using SysInternals Process Explorer, I have made a couple of stack traces when Speedfan was using 100% CPU, before closing and restarting the application. I noticed from the CPU graph that roughly half of the load occurred in kernel space.

Speedfan stack traces while using 100% CPU / nVidia SMBUS error:

==================================================================
Occurrence 1:

ntoskrnl.exe+0x47f3
ntoskrnl.exe!PsGetContextThread+0x329
ntdll.dll!KiFastSystemCallRet
speedfan.exe+0xd47b7
speedfan.exe+0xd54ea
speedfan.exe+0xd4834
speedfan.exe+0xd4de6
speedfan.exe+0xdba49
speedfan.exe+0xdbcc1
speedfan.exe+0xd7504

ntoskrnl.exe+0x47f3
ntoskrnl.exe!PsGetContextThread+0x329
ntoskrnl.exe!FsRtlInitializeFileLock+0x83f
hal.dll+0x2c35
hal.dll!HalMakeBeep+0x3f9
ntoskrnl.exe!ZwYieldExecution+0xb78
ntdll.dll!KiFastSystemCallRet
speedfan.exe+0xd47b7
speedfan.exe+0xd54ea
speedfan.exe+0xd4834
speedfan.exe+0xd4de6
speedfan.exe+0xdba49
speedfan.exe+0xdbcc1
speedfan.exe+0xd7504

ntoskrnl.exe!ExReleaseResourceLite+0x1a3
ntoskrnl.exe!PsGetContextThread+0x329
ntoskrnl.exe!FsRtlInitializeFileLock+0x83f
hal.dll!HalClearSoftwareInterrupt+0x341
hal.dll!HalMakeBeep+0x3f9
ntoskrnl.exe!ZwYieldExecution+0xb78
ntdll.dll!KiFastSystemCallRet
speedfan.exe+0xd47b7
speedfan.exe+0xd54ea
speedfan.exe+0xd4834
speedfan.exe+0xd4ee6
speedfan.exe+0xdba8f
speedfan.exe+0xdbcc1
speedfan.exe+0xd72d2
speedfan.exe+0x11c22f
speedfan.exe+0xfb193
speedfan.exe+0x2297b5
speedfan.exe+0x98906
speedfan.exe+0x82c9b
speedfan.exe+0x82b01
speedfan.exe+0x2abb1
kernel32.dll!RegisterWaitForInputIdle+0x49

==================================================================
Occurrence 2:

ntoskrnl.exe+0x47f3
ntoskrnl.exe!PsGetContextThread+0x329
ntoskrnl.exe!FsRtlInitializeFileLock+0x83f
hal.dll+0x2c35
hal.dll!HalMakeBeep+0x3f9
ntoskrnl.exe!ZwYieldExecution+0xb78
ntdll.dll!KiFastSystemCallRet
speedfan.exe+0xd47b7
speedfan.exe+0xd54ea
speedfan.exe+0xd4834
speedfan.exe+0xd4de6

ntoskrnl.exe!ExReleaseResourceLite+0x1a3
ntoskrnl.exe!PsGetContextThread+0x329
ntoskrnl.exe!FsRtlInitializeFileLock+0x83f
hal.dll!HalClearSoftwareInterrupt+0x341
hal.dll!HalMakeBeep+0x3f9
ntoskrnl.exe!ZwYieldExecution+0xb78
ntdll.dll!KiFastSystemCallRet
speedfan.exe+0xd47b7
speedfan.exe+0xd54ea
speedfan.exe+0xd4834
speedfan.exe+0xd4de6
speedfan.exe+0xdba49
speedfan.exe+0xdbcc1
speedfan.exe+0xd7504

ntoskrnl.exe!ExReleaseResourceLite+0x1a3
ntoskrnl.exe!PsGetContextThread+0x329
ntoskrnl.exe!FsRtlInitializeFileLock+0x83f
hal.dll!HalClearSoftwareInterrupt+0x341
speedfan.exe+0xd3a6b
speedfan.exe+0xd509e
speedfan.exe+0xd5260
speedfan.exe+0xd54ea
speedfan.exe+0xd4834
speedfan.exe+0xd4de6
speedfan.exe+0xdba49
speedfan.exe+0xdbcc1
speedfan.exe+0xd7504

ntoskrnl.exe!ExReleaseResourceLite+0x1a3
ntoskrnl.exe!PsGetContextThread+0x329
ntoskrnl.exe!FsRtlInitializeFileLock+0x83f
hal.dll!HalClearSoftwareInterrupt+0x341
hal.dll!HalMakeBeep+0x3f9
ntoskrnl.exe!ZwYieldExecution+0xb78
ntdll.dll!KiFastSystemCallRet
speedfan.exe+0xd47b7
speedfan.exe+0xd54ea
speedfan.exe+0xd4834
speedfan.exe+0xd4de6
speedfan.exe+0xdba49
speedfan.exe+0xdbcc1
speedfan.exe+0xd7504

Hope this helps,

Zef

(0004383)
alfredo (manager)
2009-02-06 11:49

Did you upgrade to SpeedFan 4.37?
(0004397)
discostu (reporter)
2009-02-08 17:21
edited on: 2009-02-08 17:24

Hello Alfredo,

thank you very much for looking into this.

I have recently upgraded to SpeedFan 4.37. However this version does not display the Asus V9999 temperatures or fan speeds, nor will it let me control the V9999 fan speed. It seems like the F75375 chip is no longer detected. This also happens with SpeedFan 4.36, as described here: http://www.bugtrack.almico.com/view.php?id=1295. [^]
Because SpeedFan 4.36+ doesn't detect the F75375 chip, it won't try to read the nVidia SMBUS at all and the 100% CPU problem doesn't occur - but at the expense of loosing control of the video card temps and fans.

SpeedFan 4.37 displays the following messages at startup:

Win9x:NO 64Bit:NO GiveIO:YES SpeedFan:YES
I/O properly initialized
Linked ISA BUS at $0290
Linked nVidia nForce2 SMBUS at $5000
Linked nVidia nForce2 SMBUS at $5100
Found nVidia GeForce 6800
Linked nVidiaI2C0 SMBUS at $3D403E3F
Linked nVidiaI2C1 SMBUS at $3D403637
Linked nVidiaI2C2 SMBUS at $3D405051
Scanning ISA BUS at $0290...
F71882F check found $FF $FF $11 $01 $FF
Winbond W83627HF (ID=$21) found on ISA at $290
SuperIO Chip=Winbond W83627HF
Scanning nForce2 SMBus at $5000...
Scanning nForce2 SMBus at $5100...
Scanning I2CNVidia SMBus at $3D403E3F...
Address $2D appears to be WRITE ONLY...
Address $2E appears to be WRITE ONLY...
Address $48 appears to be WRITE ONLY...
Address $49 appears to be WRITE ONLY...
Address $4C appears to be WRITE ONLY...
Address $50 appears to be WRITE ONLY...
Scanning I2CNVidia SMBus at $3D403637...
Address $2D appears to be WRITE ONLY...
Address $2E appears to be WRITE ONLY...
Address $48 appears to be WRITE ONLY...
Address $49 appears to be WRITE ONLY...
Address $4C appears to be WRITE ONLY...
Address $50 appears to be WRITE ONLY...
Scanning I2CNVidia SMBus at $3D405051...
Address $2D appears to be WRITE ONLY...
Address $2E appears to be WRITE ONLY...
Address $48 appears to be WRITE ONLY...
Address $49 appears to be WRITE ONLY...
Address $4C appears to be WRITE ONLY...
Found ACPI temperature (49.0C)
End of detection

In SpeedFan 4.35 and earlier, I think the F75375S chip was found at address $2E, as this address is mentioned a couple of times in the configuration file "speedfansens.cfg" for v4.35.

Also note that RivaTuner (up to current version 2.22) allows me to monitor the V9999 temps and fans. This utility has a special plugin with an associated configuration file for the F75375 chip that is installed in a sub folder of the application.

For your information, I will upload the file "debug.nfo" that Speedfan generates at startup, and an older version of "speedfansens.cfg" containing V9999 information. I will also attach the RivaTuner F75375 plugin configuration file.

Greetings,

Zef

(0004472)
alfredo (manager)
2009-02-16 10:04

The latest SpeedFan uses NVAPI to access the I2C bus. I don't know why, but on some video cards it seems that some devices on that bus can't be reached.
I'm trying to contact nVidia, but I never got any answer. If you can help me to reach somebody at nVidia it would be great and I would certainly be able to fix this issue in a few hours.
(0004474)
alfredo (manager)
2009-02-16 10:12

Comparing this issue to 0001295, it seems that 4.37 can access the Fintek chip on one computer and can't access it on another one. It would be great if you could tell me the exact nVidia driver revision that you are using.
(0004478)
discostu (reporter)
2009-02-16 20:20
edited on: 2009-02-16 20:28

Hello Alfredo,

as mentioned in another note that I posted with issue 1295:
  http://www.bugtrack.almico.com/view.php?id=1295 [^]
Speedfan does NOT detect the Fintek chip. The behavior mentioned in the original 1295 issue description was probably caused by older (v4.35) configuration files. Using a new configuration, Speedfan v4.37 does not (or falsely) detect the Fintek chip.

I am currently using the nVidia Geforce 185.20 graphics driver. Unfortunately I do not (and probably never will have) any inside contacts in the nVidia corporation. But I am willing to send a very humble begging message if you can give me an address ;p

Greetings,

Zef


- Issue History
Date Modified Username Field Change
2009-01-26 11:21 discostu New Issue
2009-01-26 11:21 discostu Status new => assigned
2009-01-26 11:21 discostu Assigned To => alfredo
2009-01-26 11:21 discostu File Added: Speedfan 4.35 NF7-S.txt
2009-01-26 11:21 discostu Video Card Model => Asus V9999 (nVidia 6800)
2009-01-26 11:23 discostu Tag Attached: Freeze
2009-01-26 14:51 discostu Note Added: 0004366
2009-01-28 19:04 discostu Note Added: 0004373
2009-01-28 19:05 discostu Note Edited: 0004373
2009-02-04 21:49 discostu Note Added: 0004380
2009-02-04 21:51 discostu Note Edited: 0004380
2009-02-04 21:54 discostu Note Edited: 0004380
2009-02-06 11:49 alfredo Note Added: 0004383
2009-02-06 11:49 alfredo Status assigned => acknowledged
2009-02-08 17:21 discostu Note Added: 0004397
2009-02-08 17:21 discostu File Added: debug.nfo
2009-02-08 17:22 discostu File Added: speedfansens.cfg
2009-02-08 17:22 discostu File Added: F75373S.cfg
2009-02-08 17:24 discostu Note Edited: 0004397
2009-02-08 17:26 discostu Tag Attached: V9999
2009-02-16 10:04 alfredo Note Added: 0004472
2009-02-16 10:05 alfredo Relationship added related to 0001295
2009-02-16 10:12 alfredo Note Added: 0004474
2009-02-16 10:14 alfredo Relationship added related to 0001337
2009-02-16 10:15 alfredo Relationship deleted related to 0001337
2009-02-16 20:20 discostu Note Added: 0004478
2009-02-16 20:27 discostu Note Edited: 0004478
2009-02-16 20:28 discostu Note Edited: 0004478
2011-03-19 12:07 alfredo Status acknowledged => resolved
2011-03-19 12:07 alfredo Resolution open => fixed
2011-03-19 12:07 alfredo Fixed in Version => 4.43


Copyright © 2000 - 2018 MantisBT Team
Time: 0.4119 seconds.
memory usage: 8,237 KB
Powered by Mantis Bugtracker