Asus G73SW Notebook BSODs

[DeanP]

AVG had been causing issues with Windows 7 lately... I have no idea why the dump says unknown error.

 

[johnm123]

Ok, thanks Dean.

 

[cluberti]

The reason it says it's unknown is because your CPU is apparently in an invalid state:

3: kd> r
rax=0000000000000000 rbx=0000000000000000 rcx=0000000000000000
rdx=0000000000000000 rsi=0000000000000000 rdi=0000000000000000
rip=0000000000000000 rsp=0000000000000000 rbp=0000000000000000
 r8=0000000000000000  r9=0000000000000000 r10=0000000000000000
r11=0000000000000000 r12=0000000000000000 r13=0000000000000000
r14=0000000000000000 r15=0000000000000000
iopl=0         nv up di pl nz na pe nc
cs=0000  ss=0000  ds=0000  es=0000  fs=0000  gs=0000             efl=00000000
00000000`00000000 ??              ???

It can't tell you what's wrong, because there's just no information on the CPU for it to disassemble. We know from the .bugcheck output that it was an attempted write by something to address 0x00000040, which is going to fail (it's within the first 64K of a process, in this case the kernel, and all VA in the first 64K is marked PAGE_NOACCESS). Other than that, though, without a *complete* memory dump there's no way to tell what's happening (and even with one, I can't guarantee 100% that it'll be found, but usually it can be done).

 

[johnm123]

Thanks a lot cluberti, I'm not sure where to select "full" dumps, since the drop down menu only allows me to select between a minidump and a kernel dump (is that a full one?). I'll reproduce the BSOD with a kernel dump and post it here.

On that note, it is not a real BSOD, rather the screen turns black though all the fans are still running, lights still glowing and in that time it obviously creates a dump as well. Not sure if that helps or hints at something in particular. That black screen of death appears shortly after the windows symbol/animation, right before the login screen should appear.

Edit: This time the BSOD actually was blue again, not black. Not sure why the randomness...and I also couldn't sign in via "last known good configuration" anymore but had to get in via safemode to turn off verifier.exe. So it has 3 new dumps instead of one.

 

[cluberti]

Once you've selected Kernel Dump (hopefully that'll be enough - otherwise it's registry editing time) you will need to reboot for it to take effect. Once rebooted, additional crashes will create a file in \Windows called memory.dmp. That will be the file we will need.

 

[cluberti]

OK, I know why the context of the CPU is wrong - truecrypt. Not sure if that's the cause of the crash or not, but would it be possible for you to decrypt and remove this during troubleshooting? The reason I ask, is this is the last thing that was done before the box crashed. The originating IRP is pending in a truecrypt IRP, which seems to obfuscate everything (given what truecrypt does, I suppose this is expected):

16.2: kd> !thread fffffa80`0bbb5660
THREAD fffffa800bbb5660  Cid 023c.02d8  Teb: 000007fffffd4000 Win32Thread: fffff900c1b21c30 RUNNING on processor 5
IRP List:
    fffff98003ba6c10: (0006,03e8) Flags: 40000884  Mdl: 00000000
Not impersonating
DeviceMap                 fffff8a0014cd3f0
Owning Process            fffffa800bbadb30       Image:         svchost.exe
Attached Process          N/A            Image:         N/A
Wait Start TickCount      1497           Ticks: 1 (0:00:00:00.015)
Context Switch Count      639                 LargeStack
UserTime                  00:00:00.000
KernelTime                00:00:00.093
Win32 Start Address 0x000007fefd87a808
Stack Init fffff88003e8dc70 Current fffff88003e8c6e0
Base fffff88003e8e000 Limit fffff88003e85000 Call 0
Priority 9 BasePriority 8 UnusualBoost 0 ForegroundBoost 0 IoPriority 2 PagePriority 5
Child-SP          RetAddr           : Args to Child                                                           : Call Site
fffff880`03e8c950 fffff800`0429708a : 00000000`00000000 fffff800`00000000 13200000`0520a860 00000000`00000008 : nt!ExfAcquirePushLockExclusive+0x150
fffff880`03e8c9d0 fffff800`0428676e : 00000000`00000008 fffff880`014c7080 00000000`00000000 fffffa80`090ab180 : nt!MmAccessFault+0x1a6a
fffff880`03e8cb30 fffff880`014c7080 : fffff880`014c8ff1 00000000`00000010 fffff880`014c8f50 00000000`00000010 : nt!KiPageFault+0x16e (TrapFrame @ fffff880`03e8cb30)
fffff880`03e8ccc8 fffff880`014c8ff1 : 00000000`00000010 fffff880`014c8f50 00000000`00000010 00000000`00010282 : Ntfs!NtfsQueryDirectory
fffff880`03e8ccd0 fffff880`014cb8ef : fffffa80`0b5cee40 00000000`00000000 00000000`00000002 00000000`00000000 : Ntfs!NtfsCommonDirectoryControl+0xa1
fffff880`03e8cd10 fffff800`0472bc16 : fffff980`03aacc10 fffff980`03aacc10 fffffa80`0b5cee40 fffff880`03e8cd38 : Ntfs!NtfsFsdDirectoryControl+0x10f
fffff880`03e8cd80 fffff880`01218bcf : fffff980`03aacfb0 fffff880`03e8ce20 fffffa80`0bab2010 fffffa80`085e7e00 : nt!IovCallDriver+0x566
fffff880`03e8cde0 fffff880`0121baea : 00000000`00000000 00000000`00000000 00000000`00000000 fffffa80`0bab2010 : fltmgr!FltpLegacyProcessingAfterPreCallbacksCompleted+0x24f
fffff880`03e8ce70 fffff880`01239978 : fffffa80`0863cbc0 00000000`00000022 fffffa80`087e9010 fffffa80`0bab20c0 : fltmgr!FltPerformSynchronousIo+0x2ca
fffff880`03e8cf10 fffff880`0616fd71 : 00000000`00000000 00000000`00000020 fffff880`03e8cfd0 fffff880`03e8d060 : fltmgr!FltQueryDirectoryFile+0x98
fffff880`03e8cf50 fffff880`012398c2 : 00000000`00000034 fffffa80`0863cab0 fffffa80`0bbf5770 00000000`00000000 : luafv!LuafvNormalizeNameComponentEx+0x145
fffff880`03e8d050 fffff880`01239afe : 00000000`00000001 00000000`00000034 00000000`00000010 00000000`000007ff : fltmgr!FltpCallNormalizeNameComponentHandler+0x82
fffff880`03e8d0b0 fffff880`0123af81 : 00000000`00000050 00000000`00000050 00000000`00000035 00000000`00000000 : fltmgr!FltpExpandShortNames+0x14e
fffff880`03e8d110 fffff880`0123ae1e : fffffa80`0863cab0 fffff6fc`00020000 00000000`00000000 fffffa80`0bb42010 : fltmgr!FltpGetNormalizedFileNameWorker+0xc1
fffff880`03e8d150 fffff880`01226b9d : c00000bb`00000000 00000000`00000000 fffffa80`0bb42b00 fffff880`03e8e000 : fltmgr!FltpCreateFileNameInformation+0xee
fffff880`03e8d1b0 fffff880`01220bf6 : fffffa80`08421200 fffffa80`0bb42010 fffffa80`087e9108 fffff800`0428687d : fltmgr!HandleStreamListNotSupported+0x15d
fffff880`03e8d1f0 fffff880`01227b44 : fffffa80`0bbf6200 00000000`00000000 fffffa80`0bb42010 00000000`00000101 : fltmgr! ?? ::FNODOBFM::`string'+0x30f3
fffff880`03e8d270 fffff880`0619587b : fffffa80`0863cab0 00000000`00000000 fffffa80`0bb4dec0 fffffa80`0bbdaac0 : fltmgr!FltGetFileNameInformation+0x184
fffff880`03e8d300 fffff880`01218067 : fffffa80`0bbdab70 fffff880`03e8d428 00000000`00000000 00000000`00000000 : avgntflt+0x1487b
fffff880`03e8d3b0 fffff880`0121a9aa : fffffa80`0bbf6200 fffffa80`0bbf6200 fffffa80`0858ed00 fffffa80`087e9000 : fltmgr!FltpPerformPreCallbacks+0x2f7
fffff880`03e8d4b0 fffff880`012382a3 : fffff980`03ba6c10 fffff980`03ba6c10 fffff980`03ba6c10 fffffa80`0bbf6220 : fltmgr!FltpPassThroughInternal+0x4a
fffff880`03e8d4e0 fffff800`0472bc16 : fffff980`03ba6c10 00000000`00000002 00000000`00000040 00000000`00000000 : fltmgr!FltpCreate+0x293
fffff880`03e8d590 fffff800`04586495 : 00000000`00000005 fffffa80`0bc0fcc8 fffffa80`0bc0f010 fffffa80`090c5da0 : nt!IovCallDriver+0x566
fffff880`03e8d5f0 fffff800`04582d38 : fffffa80`08577940 fffff800`00000000 fffffa80`0bc0fb10 00000000`00000001 : nt!IopParseDevice+0x5a5
fffff880`03e8d780 fffff800`04583f56 : 00000000`00000000 fffffa80`0bc0fb10 00000000`00000000 fffffa80`067418a0 : nt!ObpLookupObjectName+0x588
fffff880`03e8d870 fffff800`0458585c : 00000000`0000029b 00000000`00000000 fffff8a0`015a8701 00000000`000007ff : nt!ObOpenObjectByName+0x306
fffff880`03e8d940 fffff800`04590478 : 00000000`00b7e948 00000000`c0100080 00000000`00b7e998 00000000`00b7e958 : nt!IopCreateFile+0x2bc
fffff880`03e8d9e0 fffff800`042878d3 : fffffa80`0bbb5660 fffff880`03e8db60 fffffa80`0bbb5660 fffff800`0457fa34 : nt!NtCreateFile+0x78
fffff880`03e8da70 00000000`76f6186a : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : nt!KiSystemServiceCopyEnd+0x13 (TrapFrame @ fffff880`03e8dae0)
00000000`00b7e8c8 00000000`00000000 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : 0x76f6186a

 

[johnm123]

Thanks for the breakdown.

Interesting, so truecrypt encrypts the whole...erm...kernel or what does it do? Because I only use it to encrypt a handful of personal documents and I disabled it to automatically start up with the system so I'm not sure why it even does this stuff. I also updated to the newest version but that didn't seem to help either.

I'll get rid of that program then and run verifier.exe again to see if this time the system can start up without a BSOD. Thanks again!

 

[cluberti]

Please do, and post back your results.

 

[cluberti]

Johnm123, I got the dump from your last PM, and frankly it doesn't give any further insight. Also, some of the context is screwed up, so it's not all there. Honestly, at this point, we need you to configure the system for a complete memory dump, and manually configure your paging file to be both min and max RAM+50MB. To configure for a complete memory dump:

1. Open sysdm.cpl
2. Click the "Advanced" tab, and click the "Settings" button in the Performance section
3. Click the "Advanced" tab, and click the "Change" button
4. Click "Custom size", and enter (in MB) the same value for both the "Initial size" and "Maximum size" boxes - your system's RAM+50MB (so if you have 8GB of RAM, you would enter 8242 - 8192 is RAM, +50 = 8242).
5. Open regedit
6. Browse to HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management
7. Change the value of the DWORD "CrashDumpEnabled" to 1
8. Reboot

Now, do NOT go back into the system control panel and touch anything - the moment you view the memory dump dropdown and then click OK (even if you change nothing), it'll change the CrashDumpEnabled value back to 2 (kernel dump only). If you do the above, in that order, and *only* the above, the next crash should be a *complete* memory dump, including all user and kernel-mode contexts (let's hope!).

 

[johnm123]

Thanks a lot for this step-by-step tutorial. I did everything except the regedit part because that "CrashDumpEnabled" option just wasn't there. Any ideas? (screenshot attached)

 

[cluberti]

Yes, you'll have to create it - it's a DWORD (32bit) value, and it must be called CrashDumpEnabled (no trailing spaces, written as-is).

 

[johnm123]

My bad, I assumed that this value/entry was already there, thanks.

So this is what I've done now:

1. Followed your instructions, set pagefile size to 8242mb and added the registry value, didn't touch system control panel afterwards.

2. Restarted computer so settings could be saved. Started verifier.exe again to select the soundcard and USB drivers again (using the settings posted in another thread on here) since those would cause the BSOD.

3.) Restarted, this time no BSOD occured as it used to at this point. So I deleted verifier.exe-settings and restarted. Then I activated verifier.exe again and selected ALL drivers (except microsoft) and restarted. No BSOD on startup this time either.

It's a bit hard to believe that setting that registry value or the pagefile size could have solved the problem (the pagefile size...could it?)

I'm not sure what to do from here on, just let the laptop run for a couple of hours with verifier.exe testing all those drivers and see if the BSOD shows up again? I also already ran memtest with 9 passes and 0 errors.

 

[cluberti]

Given how drivers start under verifier, it could be something that simple. I think at this point you should (if you can) at least leave it running for a short time - if it doesn't repro shortly, disable verifier, reboot, and hopefully it won't happen again. If it does, you'll at least have a crash dump that could be helpful in troubleshooting it this time .

 

[johnm123]

Thanks a lot again, cluberti...hopefully that was it now!

How short is "short" for verifier.exe to run? And I hope there is no kind of bad effect of having the pagefile in a fixed size like that?

Anyway, I'll keep you posted. verifier.exe ran for an hour already with all drivers selected as mentioned above, no BSOD THANK YOU!

 

[cluberti]

If verifier doesn't catch anything in about 24 hours, that's good enough considering the symptoms and this thread. As to fixed, for a system with 8GB of RAM what you're doing is something I would suggest.

 

[johnm123]

Sounds good, thank you! I'll let it run for 24hrs and report back on how it went.

 

[johnm123]

Hm, seems I just got lucky yesterday. After running it for several hours, I decided to test it again with restarting (activating verifier.exe, selecting all drivers except microsoft, restart) and the BSOD happened again. Weirdly enough, this time the file is actually smaller (~350mb) than the kernel dump (~380mb) - I hope this is the full one?

Anyway, I've pm'd you the download link. Hopefully this time the error can be tracked down.

Thanks,
John M.

 

[cluberti]

Well, this is a kernel dump (which is odd), but it most certainly has some good information - the BSOD is happening in the kernel, but autochk is also running:

16.4: kd> !thread fffffa8006735680
THREAD fffffa8006735680  Cid 0004.002c  Teb: 0000000000000000 Win32Thread: 0000000000000000 RUNNING on processor 4
IRP List:
    fffff98016092ee0: (0006,0118) Flags: 40060000  Mdl: 00000000
Not impersonating
DeviceMap                 fffff8a000007dc0
Owning Process            fffffa8006712ae0       Image:         System
Attached Process          N/A            Image:         N/A
Wait Start TickCount      482            Ticks: 1 (0:00:00:00.015)
Context Switch Count      775             
UserTime                  00:00:00.000
KernelTime                00:00:00.312
Win32 Start Address nt!ExpWorkerThread (0xfffff800042ee910)
Stack Init fffff8800355bc70 Current fffff8800355a710
Base fffff8800355c000 Limit fffff88003556000 Call 0
Priority 13 BasePriority 12 UnusualBoost 0 ForegroundBoost 0 IoPriority 2 PagePriority 5
Child-SP          RetAddr           : Args to Child                                                           : Call Site
00000000`00000000 00000000`00000000 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : 0x0
16.4: kd> !irp fffff98016092ee0
Irp is active with 1 stacks 3 is current (= 00000000)
 No Mdl: No System Buffer: Thread fffff800042f9530:  Irp is completed.  Pending has been returned
     cmd  flg cl Device   File     Completion-Context
 [  f, 0]   0 10 fffffa800a473b80 00000000 00000000-00000000    
        \Driver\DXGKrnl
   Args: 00000000 00000000 00000000 00000000
 
16.4: kd> !process 0 0
**** NT ACTIVE PROCESS DUMP ****
PROCESS fffffa8006712ae0
    SessionId: none  Cid: 0004    Peb: 00000000  ParentCid: 0000
    DirBase: 00187000  ObjectTable: fffff8a0000018c0  HandleCount:  97.
    Image: System
PROCESS fffffa800a3f8040
    SessionId: none  Cid: 01b4    Peb: 7fffffdc000  ParentCid: 0004
    DirBase: 1c428a000  ObjectTable: fffff8a0007f8700  HandleCount:  26.
    Image: smss.exe
PROCESS fffffa800a3ff630
    SessionId: none  Cid: 01c0    Peb: 7fffffdb000  ParentCid: 01b4
    DirBase: 1ccc72000  ObjectTable: fffff8a000a19f90  HandleCount:   2.
    Image: autochk.exe
 
16.4: kd> !thread fffffa800a279b60
THREAD fffffa800a279b60  Cid 01c0.01c4  Teb: 000007fffffde000 Win32Thread: 0000000000000000 WAIT: (Executive) KernelMode Alertable
    fffff880037b73a0  NotificationEvent
Not impersonating
DeviceMap                 fffff8a000007dc0
Owning Process            fffffa800a3ff630       Image:         autochk.exe
Attached Process          N/A            Image:         N/A
Wait Start TickCount      459            Ticks: 24 (0:00:00:00.374)
Context Switch Count      18             
UserTime                  00:00:00.000
KernelTime                00:00:00.000
Win32 Start Address 0x00000000ffa80660
Stack Init fffff880037b7c70 Current fffff880037b6fe0
Base fffff880037b8000 Limit fffff880037b2000 Call 0
Priority 9 BasePriority 8 UnusualBoost 0 ForegroundBoost 0 IoPriority 2 PagePriority 5
Child-SP          RetAddr           : Args to Child                                                           : Call Site
fffff880`037b7020 fffff800`042e9992 : fffffa80`0a279010 fffffa80`0a279b60 00000000`00000000 fffffa80`085ceb70 : nt!KiSwapContext+0x7a
fffff880`037b7160 fffff800`042ec1af : fffffa80`08ea4bc0 fffff880`0145bf6d fffff980`00000000 fffff980`16102ce0 : nt!KiCommitThreadWait+0x1d2
fffff880`037b71f0 fffff800`04582266 : 00000000`00000100 fffff800`00000000 fffffa80`0a279000 00000000`00000001 : nt!KeWaitForSingleObject+0x19f
fffff880`037b7290 fffff800`045822e7 : fffffa80`083f7840 00000000`00000002 fffffa80`085ceb70 00000000`00000000 : nt!FsRtlCancellableWaitForMultipleObjects+0x5e
fffff880`037b72f0 fffff880`01227605 : fffff880`037b73a0 00000000`00000002 fffffa80`085ceb70 fffffa80`08ea4bc0 : nt!FsRtlCancellableWaitForSingleObject+0x27
fffff880`037b7330 fffff880`01222971 : fffffa80`0840c480 fffffa80`085bc900 fffffa80`0a3e5610 fffffa80`085ceb70 : fltmgr!FltpFsControlMountVolume+0x385
fffff880`037b7400 fffff800`04787c16 : fffff980`16102c10 00000000`00000002 fffffa80`0840c480 fffffa80`0696a798 : fltmgr!FltpFsControl+0x101
fffff880`037b7460 fffff800`045502d7 : fffffa80`0840c480 fffff800`044e1250 fffffa80`0a279b60 fffffa80`085bc900 : nt!IovCallDriver+0x566
fffff880`037b74c0 fffff800`042d3a87 : fffff880`037b77e0 fffff880`037b7700 fffffa80`0855e900 fffff800`045e1501 : nt!IopMountVolume+0x28f
fffff880`037b7580 fffff800`045e2706 : 00000000`00000005 00000000`00000000 fffff880`037b77e0 fffff880`037b7640 : nt!IopCheckVpbMounted+0x1b7
fffff880`037b75f0 fffff800`045ded38 : fffffa80`0855e940 fffff800`00000000 fffffa80`0a3e5a50 fffff880`00000001 : nt!IopParseDevice+0x816
fffff880`037b7780 fffff800`045dff56 : 00000000`00000000 fffffa80`0a3e5a50 00000000`00100080 fffffa80`06741b40 : nt!ObpLookupObjectName+0x588
fffff880`037b7870 fffff800`045e185c : 00000000`00000000 00000000`00000000 00000000`00000001 00000000`00000000 : nt!ObOpenObjectByName+0x306
fffff880`037b7940 fffff800`045cd134 : 00000000`001cf8b8 00000000`00100080 00000000`001cf930 00000000`001cf960 : nt!IopCreateFile+0x2bc
fffff880`037b79e0 fffff800`042e38d3 : fffffa80`0a3ff630 0000007f`ffffffff fffffa80`00000400 00000980`00000000 : nt!NtOpenFile+0x58
fffff880`037b7a70 00000000`7790164a : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : nt!KiSystemServiceCopyEnd+0x13 (TrapFrame @ fffff880`037b7ae0)
00000000`001cf858 00000000`00000000 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : 0x7790164a

Honestly, I'm not sure how to read this. However, at this point, it does appear likely to be a driver - however, I have my suspicions that neither the USB or soundcard driver is at fault, and more likely your video card/driver or something responsible for power. What happens if you run verifier only on these?

16.4: kd:x86> lmivm nvlddmkm
start             end                 module name
fffff880`0f258000 fffff880`0feb3300   nvlddmkm   (deferred)             
    Symbol file: nvlddmkm.sys
    Image path: \SystemRoot\system32\DRIVERS\nvlddmkm.sys
    Image name: nvlddmkm.sys
    Timestamp:        Thu Mar 17 04:08:59 2011 (4D81C19B)
    CheckSum:         00C6915C
    ImageSize:        00C5B300
    Translations:     0000.04b0 0000.04e4 0409.04b0 0409.04e4

16.4: kd:x86> lmivm atkwmiacpi64
start             end                 module name
fffff880`02ff7000 fffff880`03000000   atkwmiacpi64   (deferred)             
    Symbol file: atkwmiacpi64.sys
    Image path: \??\C:\Program Files (x86)\ASUS\ATK Package\ATK WMIACPI\atkwmiacpi64.sys
    Image name: atkwmiacpi64.sys
    Timestamp:        Mon Jul 26 01:56:49 2010 (4C4D23A1)
    CheckSum:         00013A7D
    ImageSize:        00009000
    Translations:     0000.04b0 0000.04e4 0409.04b0 0409.04e4

There are other drivers here that could be at fault, so it might make sense to uninstall anything you don't absolutely need, and if the system crashes again, hopefully the next dump is useful. Otherwise, the only other real (viable) option is a rebuild - attaching a live kernel debugger over a null-modem cable from another machine would work for debugging, but that would require a second machine, a specialized null-modem cable, and someone there locally (or remoted it) to debug it when it crashed.

 

[johnm123]

Weird, I followed what you wrote a few posts back to the letter so I'm not sure why it still gives a kernel dump. A full dump would also be MEMORY.DMP in c:\windows\, correct? I looked into regedit again and CrashDumpEnabled is still set to 1 so I'm not sure why it didn't make a full one

I ran verifier on both drivers you gave me above but with no reproducable BSOD after restarting several times with verifier. It would be bad if it's something important as this because the nvidia one is obviously my display/graphics driver and the ATK package is from Asus itself to make the FN buttons work (though I have to admit right now they don't work at all).

What you suggest sounds quite complicated because neither have I heard of a null-modem (though a G search could fix that), nor do I know someone around that could do the debugging. So at this point I'm almost considering to just delete the whole thing with DBAN and completely reinstall, but this sounds like a shot-gun approach and I'm not sure if that would actually fix anything.

Can we safely exclude a hardware error at this point? As I previously said, 9 passes of memtest didn't cause any errors though I haven't run a hard disk checking tool yet except the one from windows.

 

[Dave76]

You can run memtest86+ again, it is possible to not give errors a few times then show the errors on the next test, just a fact of electronics.
I've seen this happen many times, and the occasional RAM error can cause havoc with drivers.


If you don't know the HDD brand, D/L SIW - System Information for Windows, it will tell you.
Go to the manufacturers website and get their HDD diagnostics program.

Here are two,
Seagate SeaTools
Western Digital Lifeguard


For the CPU

CPU - Stress Test with Prime95
Prime95 Torture test
If memtest86+ passes then run Prime95 torture test.
First open Real Temp to monitor CPU temperatures, don't go over 77 degC.
Open Prime95 and stop the test, in Advanced tab select 'Round off checking', in the Options tab select 'Torture Test...', in the window that opens select 'Blend', after 'Number of torture test threads to run' enter 8, then click OK to start the test.