Hello, please test:
1. run ndcucfg on any console (e.g. telnet) and enter "core clock" assume the result is about 792MHz.
2. Copy the DLL from the attachment to ffsdisk
3. Do registering according the txt file from attachment (e.g. by console command "ndcucfg -B\<path_and_name_of_txt>)
4. Reboot and check "core clock" again, the result should be 400MHz and heat should be drastically reduced
Can you confirm?
Hello,
no sorry. I stuck a bit while implementation, it's no as easy as I thought.
I try to provide you a version with 400MHz fixed clock today - do you think it is useful? Nevertheless I will proceed and spend additional time while you test.
Hello,
i wil try it until KW32/2022.
Hello,
this information is sufficient. I will add the dvfc feature to a kernel based on "XIPiMX6_C8E_V240_Perdum_180316". This is your custom kernel. I will let you know when you can download it.
Hello,
we can try to deploy dvfc driver which sets cpu-/perclock according the current demands.
Which kernel-/HW-/OS versions you are using (see HLKM\platform)?
Hello,
by default ActiveSync should be enabled on device side:
[HKEY_LOCAL_MACHINE\Drivers\USB\FunctionDrivers]
"DefaultClientDriver"="Serial_Class"
Does it work on W7 or XP? So the problem is not located on device side.
W10 support for MobileDeviceCenter is tricky. Please check under "Services" if it is active. Or try to google.
Hello,
very good that you found the cause of the issue! And thanks for share the information.
Hello,
can you please test with V1.40 from 210723. Here all WEC updates up to March 2021 included.
"CreateFile" was only meant as a sample, you should check all DLL imports for possible memory leaks!
Assume to disable dvfc should not cause a memory leak.
>> I have implemented GlobalMemoryStatus, so it is easier to check the memory status.
<< So maybe you can relate memory increase to specail code section?
Hello,
.NET code should not cause memory leaks because of the garbage collector. Probably is "SerialParser.ByteArrayToString" only the trigger when it try to allocate some 100kB? Which kernel you are using so I can also check for Windows Embedded Compact updates may we have a bug inside .NET.
Do you have any DLL imports included using native code? For example if you call several times "CreateFile( <deviceX>)" without any "CloseFile(<forhandledevicex)" you will loose some bytes each time! Embedded Compact allocates RAM in junks of 4kB. So if you loose each time 8 bytes for example you will recognize the effect after 512 times.
In a first step I would include sections with DLL imports with "GlobalMemoryStatus" and print out the results on the debug line or screen.
Hope this helps.
Hello,
i put a SSH patch into the download under fsimxsx WEC7 BSP. Please try.
Extract the file into the folder SSH under ..\common\src\netdcu (maybe some pathes have to be adapt, please check the bib).
Enablke SSH by:
[HKLM\comm\sshd]
IsEnabled=dword:1
UserList = <your name>
You can add the files via bib-file to image, e.g project.bib:
sshd.dll $(_FLATRELEASEDIR)\sshd.dll NK SH
...
But it is better to wait for the patch (i will send you bib and reg file too, so we do not need the catalog at all, it is more easy for me:).
Hello Vladimir,
you are right, there are some Registry *) setting required to enable the feature. But I am not sure if the both DLLs above are sufficiet for deployment but you may try it.
*) Regitry in neducfg notation:
reg open \comm\sshd
reg set value IsEnabled dword 1
reg set value UserList string <your name>
reg save
REM ensure enter in last line
Hello Vladimir,
assume the DLLs are not added to the BSP. I will try to provide you a patch within the next two weeks,
Dear Vladimir,
>> Can you please tell me when do you expect to have EFUSA9Xr2-V2-W13 available for buyers
<< Best would be to contact our sales office for order samples and talk about delivery date. SW should not be a big issue we have the same PHY already in use on efusA9r2.
Dear mbazdar,
you need at least new binary of Ethernet driver (which you can integrate into older standard kernel or BSP).
But i recommend to use future EBoot, kernel, BSP which are designed for automatically detect new board type efusA9Xr2. We start implemetation as soon we have the new efusA9r2 on our desk.
Hello,
command "R" should reset complete Registry to kernels default, hence also HKLM\comm\... .
Hello,
>> But now i have the next problem, we have an application in which we use the F&S boot screen. My attempt to load the file 'PSD140708.bin' together with the current bootloader (V1.12) and NBoot (430 LB) did not work (spin forever)
<< We have to create a new PSD (because here the partition size is stored in blocks not in MB).
- Use the card with the already installed old PSD (display is ajusted here)
- Step into the EBoot
- Set new partition settings, using command "P": e.g. 24MB for OS rest for ffsdisk, wait for success ...
- Download kernel again
- Now you can create new PSD using NDCUCfg command: "PSD read <filename>.bin".
Hope this helps.
Hello,
I put you a kernel into PicoCOM4 download (folder "test").
Please chekc if it works with your app (it is V1.11 based only Flashdriver for new Flash was added).
Hope this helps.
Hello,
I will (try) to provide you a "XIPNKPC4_CORE_CF35_V117_XXXXXX" with the previous version of .NET DLLs. Also I will check if there are a newer vesion available or if the problem is known. But WCE6 is EOL so I think we have to switch to older libraries.
This may take about fives days.
Hello,
>> We found from the hardware documentation that PIN 77 (LOUT) and PIN 78 (ROUT) have multiple assignments ...
<< If audio works we can be sure that codec is assembled, so this should not be the problem.
>> We also noticed that the audio is much less clear than the PicoCOM4
<< Is there a noise or a delay witfhin the sound?
In the firts case you may contact our HW department.
In second case try:
[HKEY_LOCAL_MASCHINE\Drivers\BuiltIn\PicoCOMA5\WaveDevI2C]
Priority256=dword:60 ;still below wavedev but above all other periphericals
*or*
Priority256=dword:5e ;above wavedev
Await first vale will improve.
Hope this helps.
