BUG: AdobeARM v1.824.460.1067 getting stuck in memory after error "NtReadVirtualMemory unicode_path failed, error: -2147483635"
BUG: AdobeARM v1.824.460.1067 (from Adobe Acrobat Reader MUI v24.001.20629) getting stuck in memory after error "NtReadVirtualMemory unicodepath failed, error: -2147483635". Can download files (AcroRdrDCUpd2400120643MUIincr.msp and ReaderDCManifest3.msi on 04/01) and (AcroRdrDCUpd2400220687MUI.msp and ReaderDCManifest3.msi yesterday, so on 04/12) but is unable to continue installing updates because mostly looping on error "NtReadVirtualMemory unicodepath failed, error: -2147483635" and also sporadically also meet error "NtReadVirtualMemory readpeb failed, error: -1073741790".
-
Rob Mer commented
After some more investigation I also did for another bug I've commented, I've also been able to verify that current Adobe technical article I've initially considered as a knowledgeable reference indeed contains partially invalid details. On my client Adobe Acrobat Reader was initially installed from its OEM (Dell) and so I guess they used some Adobe own approved setup method. During years I constantly continued to upgrade it until Reader XI MUI, that I later upgraded to Acrobat Reader DC v15.023.20070, then immediately to v17.009.20044 and then, sometime later, to v18.011.20055 and since its version continued to upgrade until current v24.002.20687 I believe that Adobe article "Acrobat Reader migration to 64-bit" needs more corrections because since registry key HKEY_LOCAL_MACHINE\SOFTWARE\Adobe\Adobe Acrobat\DC\Installer\ do not exist, it doesn't even consider that Acrobat Reader DC MUI (32bit) when installed on an x64 bit OS will instead just use a very different registry key path (HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Adobe\Acrobat Reader\DC\Installer\) that I correctly found and that is populated with lot of registry values but not with 'SCAPackageLevel' that is described in above mentioned technical article and does not even exists.
This also confirms that error described into this bug should also not be related with missing registry keys (a doubt I had since I verified that my OS registry was missing completely missing HKEY_LOCAL_MACHINE\SOFTWARE\Adobe\ registry key path, so above mentioned errors may not be caused from an unexplicable deletion that indeed never happened. 0:-) -
Rob Mer commented
Update: After manually restarting BITS service and 'Adobe Acrobat Update Service' (=AdobeARMservice, so armsvc.exe) once more I awaited for next scheduled run to see if anything changed. But, unfortunately same issue continued again (see effects in AdobeARM_207ter.log as always from %SystemRoot%\Temp), so I decided to see if activating a manual update from within Adobe Reader DC could work.
And this time update started correctly and then completed fine (see contents of AdobeARM_208bis.log at the end where it's possible to see that this time '... Command Line: ...' line is indeed not empty and with all needed arguments so I hope all this really helped to avoid past errors I described in previous posts). -
Rob Mer commented
After some more investigations I did on my client (also from same AdobeARM_196.log I already uploaded) I found that both same "NtReadVirtualMemory unicode_path failed, error: -2147483635" (corresponding to 0xFFFFFFFF8000000d) and also "NtReadVirtualMemory read_peb failed, error: -1073741790" (corresponding to 0xFFFFFFFFc0000022) were also temporarily only experienced on 3/28 right during last upgrade to Acrobat Reader v24.001.20629 that I still have and then also on 04/03 that I thought was 1st earlier evidence.
Now according to Bing Copilot additional help I decided to ask for, both error messages correspond to constant STATUS_ACCESS_DENIED in Windows OS but for current AdobeARM.exe v1.824.460.1067 main problem is that when currently finding 1st error (more frequent) or 2nd one (less frequent) is now not really adding into own log more additional details that might better help an end user really figure out or understand why they're being logged.
So, I would really hope that a next ARM release in future might better be able to log more conditions about own status when both situations may happen again so to possibly help an end user like me, or even Adobe yourself (because I also saw that you do seem to often self-get back a copy of my logs when "Command Line: /ArmElevate /CollectFiles /Svc /USER:<myAlias> /SESSIONID:1").
P.S. May all this be anyhow related to AdobeARM.exe only starting with an empty "Command Line: " but anyway "Using registered preference AUTO_ALL", or even with some other unclear situation related to BITS service own status that is used by AdobeARM when downloading own files ?
I'm saying this because in the past, right before upgrade to Acrobat Reader v24.001.20629 really succeeded I noticed some other BITS related errors (also from AdobeARM_196.log I included before) which I tried to possibly solve by manually forcing BITS to restart before next scheduled AdobeARM occurrence was going to take place:
"...
[2024-03-28 19:36:40:0400] ARMDownloader:: BITS_Job GetError() error: -2145386481
[2024-03-28 19:36:40:0400] ARMDownloader:: BG_JOB_STATE_CONNECTING: errorCount1
[2024-03-28 19:36:40:0450] all BITs attempts failed
[2024-03-28 19:36:40:0450] Error Code: -2145386481
[2024-03-28 19:36:40:0450] ** Setting Error Condition: 140000
[2024-03-28 19:36:40:0913] BITs failed, download manifest Job is not in Transferred state
[2024-03-28 19:36:48:0591] valid after BB download: C:\Users\<myAlias>\AppData\Local\Adobe\ARM\S\ArmManifest3.msi
[2024-03-28 19:36:48:0697] Newer version ARM update is not available
..." -
Rob Mer commented
Because after some more investigation I was also able to find all downloaded files already reported initially into C:\ProgramData\Adobe\ARM\Reader_24.001.20629 folder, my next step now will be to soon try to manually execute AcroRdrDCUpd2400120643_MUI_incr.msp and then AcroRdrDCUpd2400220687_MUI.msp only to see if manual installation complete correctly (but because I'm unsure what ReaderDCManifest3.msi may do and yesterday I found no reference details on it).
-
Rob Mer commented
As you can see from initially enclosed AdobeARM.log issue seems to have started on 04/05 and I casually noticed it because I saw that AdobeARM.exe was remaining in memory more time that expected. But in reality, this issue indeed started on 04/03 as you can see from a previous copy of same file I had saved on same day and before AdobeARM.log on 04/05 even reset his size after it had grown too much (reaching 3.57MiB).