I asked a question in the Japanese vmware community, but since there seems to be nobody with similar case,
I will also post it to the English version.
I do not have much English is good at.
Because it is translating by google translation, it may be incomplete English, but please understand.
This sentence is very long. I'm sorry.
I am using a virtual machine (vmware player 12.5.7) with "MSI A88X-G43".
Immediately after starting the virtual machine, after several hours, etc. irregularly
"BSOD CLOCK_WATCHDOG_TIMEOUT (101)" occurs in the host OS.
When "vmware player" is executed with "MSI A88X-G43", in the same way
Is there anyone with an error "BSOD CLOCK_WATCHDOG_TIMEOUT (101)"?
Please tell me if there is a BSOD error occurring.
The CPU is AMD A10-7850K.
AMD CPU uses "vmware player" but BSOD CLOCK_WATCHDOG_TIMEOUT (101)
Is there anyone who doesn't happen?
Could you tell me the CPU, BIOS version information, and motherboard name of the machine that does not occur?
By the way, even with VMware 12.5.7 or earlier, the same BSOD is occurring, it is not a problem of VMware 12.5.7.
With various trial and error, with the virtual machine running,
If you do the following operation, you know that it always occurs in the environment where this failure occurs.
Press [Setting (T)] on "Startup and Recovery" in the [Control Panel] - [System and Security]
- [System] - [Advanced System Settings] - [Detail Setting] Tag.
(I think that the English version is somewhat different, because it is the display wording of the Japanese version)
At first, I thought of hardware and driver problems, and removed all connected devices such as USB and expansion cards.
And even with minimal configuration with CPU, memory, hard disk, power supply, mouse, keyboard
When running vmware player BSOD occurred.
I diagnosed memory and hard disk with the following tools.
Memory windows OS included tool, memtest86
HDD WD Data Lifeguard Diagnostic for Windows
By doing this, I confirmed that it is not abnormal in the hardware.
So I installed Windows 10 offline and clean install.
When running vmware player on this host OS, "BSOD CLOCK_WATCHDOG_TIMEOUT" similarly occurred.
In addition, we prepared another motherboard A88X-G43 of the same model number.
I have created another machine that has a separate power source, another CPU, another memory,
another mouse, another keyboard, another hard disk, and no identical parts at all. (See the end for details)
I cleanly installed Windows 10 offline on this machine.
If you run "vmware player" on this machine, the host OS similarly raises "BSOD CLOCK_WATCHDOG_TIMEOUT".
I tried using the guest OS which cleanly installed Windows 10, but similarly BSOD occurs.
Therefore, it is neither an individual defect of hard nor an OS problem.
Here I thought that there might be a problem with the BIOS.
Until then, I used the latest 3.C (3.12).
If this version is downgraded to 3.9, BSOD no longer occurs.
There are two A88X-G43 as above.
For both machines "BSOD CLOCK_WATCHDOG_TIMEOUT" will occur if "BIOS" is 3.10 or more.
If "BIOS" is 3.9, it does not occur at all.
Therefore, it is obvious that it is BUG of BIOS.
AMD provides microcode such as CPU, memory controller and HT bus controller in the form of AGESA code.
When examining the AMD AGESA version in the BIOS, it seems that it changed significantly between 3.9 and 3.10 (3.A).
The results of examining the BIOS and AGESA version with HWINFO 64 are as follows.
■BIOS where BSOD occurs
(1)BIOS Version: V3.12(3.C)
BIOS Date: 08/10/2016
AMD AGESA Version: CarrizoFM2r2PI V1.2.0.4X
(2)BIOS Version: V3.11(3.B)
BIOS Date: 04/08/2016
AMD AGESA Version: CarrizoFM2r2PI V1.2.0.3X
(3)BIOS Version: V3.10(3.A)
BIOS Date: 01/14/2016
AMD AGESA Version: CarrizoFM2r2PI V1.2.0.2X
■BIOS that BSOD does not occur
(1)BIOS Version: V3.9
AMD AGESA Version: GodavariPI V1.0.0.0
BIOS Date: 04/20/2015
This "BSOD" seems to occur when updating "BIOS" to V3.10 (2016/01/14) or later and using "vmware player" and using virtual instruction set of AMD CPU.
We analyzed complete memory dump at BSOD with windbg.
The CPU has four processors, but the 0, 1 and 2 processors were waiting for the completion of the 3rd processor.
Conversely, the third processor waited for the completion of 0, 1 and 2 processors, and it seemed to be deadlocked.
In this third processor, the program being executed was vmware.
It seemed like a problem of exclusive processing of memory and resources peculiar to multiprocessor.
When running vmware in this single processor state, BSOD no longer occurs.
However, it is slow with a single core, it can not be used at all times.
I thought to downgrade the BIOS to V3.9, which does not cause BSOD problems.
However, with this V3.9, connecting the Asmedia 1061 SATA Hostcortroler to the PCI-e causes another BSOD to occur frequently and the OS can not be started.
Therefore, it is necessary to use problematic V3.10 or later.
I am troubled because I can not avoid the BSOD problem by downgrading to V3.9.
The problematic BIOS, V3.10 seems to be released for compatibility with the new CPU Carrizo.
Unfortunately, it seems that the following fatal troubles are mixed from this version.
With old CPU (Trinity / Kaveri) BSOD occurs when virtual function is used.
For this motherboard, the carrizo core CPUs are "Athlon X4 845" and "Athlon X 4 835".
This product has no GPU and the clock is lower than the previous generations.
If there is no merit and the AGESA code is the cause, it is an annoying story.
However, the Microcode of the CPU in the AGESA code is as follows and there is no change.
AGESA code seems to be microcode other than CPU as a problem.
#1"AMD A10-7850K", the BIOS version 3.9 - 3.12 was all the same as "6003106".
#2"AMD A4-5300", BIOS version 3.9 - 3.10 was the same as "6001119".
However, as mentioned above, although the version of the AGESA code has certainly changed,
it is a situation where it can not be concluded that this is the cause.
Because recent UEFI BIOS is considerably huge, complicated, compressed and stored in 8 MB.
Besides the AGESA code, it is made up of more than 200 modules.
It contains various programs (initialization code and firmware driver) as below.
1)chipet
2) Storage controller
3) LAN controller
4) Clock generator
5) Power supply IC type etc.,
In addition, many modules such as MSI company and BIOS maker AMI company are built in.
I took the difference between the modules of BIOS 3.9 and 3.10.
Looking at it seems that not only the AGESA code but also many modules have been changed. (See attached file)
The yellow module shows that there is a difference.
However, there are many things that only change a few bytes of the BIOS version.
Therefore, I think there are fewer modules that have been modified or changed.
Some of these modules have been greatly changed.
I think this module is likely to have bugs.
Like me, if I run vmware on AMD platform, is there anyone with "BSOD CLOCK_WATCHDOG_TIMEOUT (101)"
occurring in the host OS?
Likewise, if there are people with BSOD errors, please let me know.
AMD CPU uses "vmware player" but BSOD CLOCK_WATCHDOG_TIMEOUT (101)
Is there anyone who doesn't happen?
Could you tell me the CPU, BIOS version information, and motherboard name of the machine that does not occur?
The specifications of the A88X-G43 are as follows.
[Current Computer]
Computer Brand Name: MSI MS-7793
UEFI Boot: Present
[Motherboard]
Motherboard Model: MSI A88X-G43 (MS-7793)
Motherboard Chipset: AMD A88X (Bolton-D4)
Motherboard Slots: 1xPCI, 5xPCI Express x1, 1xPCI Express x2, 2xPCI Express x4, 1xPCI Express x16
PCI Express Version Supported: v3.0
USB Version Supported: v3.0
[BIOS]
BIOS Manufacturer: American Megatrends
BIOS Date: 08/10/2016
BIOS Version: V3.12
AMD AGESA Version: CarrizoFM2r2PI V1.2.0.4X
UEFI BIOS: Capable
Super-IO/LPC Chip: Fintek F71879A/F71889A
BIOS ----------------------------------------------------------------------
BIOS Vendor: American Megatrends Inc.
BIOS Version: V3.12
BIOS Release Date: 08/10/2016
BIOS Start Segment: F000
BIOS Size: 8192 KBytes
System BIOS Version: 4.6
ISA Support: Not Present
MCA Support: Not Present
EISA Support: Not Present
PCI Support: Present
PC Card (PCMCIA) Support: Not Present
Plug-and-Play Support: Not Present
APM Support: Not Present
Flash BIOS: Present
BIOS Shadow: Present
VL-VESA Support: Not Present
ESCD Support: Not Present
Boot from CD: Present
Selectable Boot: Present
BIOS ROM Socketed: Present
Boot from PC Card: Not Present
EDD Support: Present
NEC PC-98 Support: Not Present
ACPI Support: Present
USB Legacy Support: Present
AGP Support: Not Present
I2O Boot Support: Not Present
LS-120 Boot Support: Not Present
ATAPI ZIP Drive Boot Support: Not Present
IEE1394 Boot Support: Not Present
Smart Battery Support: Not Present
BIOS Boot Specification Support: Present
Function key-initiated Network Service Boot Support: Not Present
Targeted Content Distribution Support: Present
UEFI Specification Support: Present
The specifications of the CPU are as follows.
# 1st machine
CPU information
[General Information]
Processor Name: AMD A10-7850K
CPU ID: 00630F01
Extended CPU ID: 00630F01
CPU Brand Name: AMD A10-7850K Radeon R7, 12 Compute Cores 4C+8G
CPU Stepping: KV-A1
CPU Code Name: Steamroller/Kaveri
Microcode Update Revision: 6003106
# 2st machine
CPU information
[General Information]
Processor Name: AMD A4-5300
CPU ID: 00610F01
Extended CPU ID: 00610F01
CPU Brand Name: AMD A4-5300 APU with Radeon(tm) HD Graphics
CPU Stepping: TN1-A1 (Trinity)
CPU Code Name: Piledriver/Trinity
Microcode Update Revision: 6001119
Thank you for reading very long sentences.
I understood the new reproduction condition.
I found that it happens when the BIOS / UEFI mode of the host OS is in UEFI mode.
It does not occur in BIOS mode.
Procedure summarizing the conditions that we know until now will be as follows.
(1) After the generation of the AGESA code Carrizo FM 2 r 2 PI
It is not before, it will be caused by the BIOS upteme.
(2) The number of CPUs of the host OS is multi-core (2 cores or more)
In the host OS msconfig.exe boot tag, boot advanced option,
When the number of processors is limited to one, no BSOD occurs.
In order to limit the number of processors of the host OS to 1, at the same time, the number of processors of the guest OS is also set to 1.
(3) When the BIOS / UEFI mode of the host OS is UEFI mode
When asking MSI, although AMD is still on sale, it is said that AMD's microcode AGESA code is not corrected, it is taking a bad response.
AMD AGESA Version before BIOS 3.9: If you return to GodavariPI V 1.0.0.0, this problem will not occur, so I think that it can be modified technically.
Reproduction procedure
Press [Setting (T)] on "Startup and Recovery"
in the [Control Panel] - [System and Security]
- [System] - [Advanced System Settings] - [Detail Setting] Tag.
(I think that the English version is somewhat different,
because it is the display wording of the Japanese version)