Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Topics - Dariusz Piatkowski

Pages: 1 2 [3] 4 5 ... 9
31
Internet / FF 45.x and CPU spikes
« on: October 05, 2020, 05:28:21 pm »
So if there is one complaint I have here about FF (ignoring the other blatant problems such as inability to view certain sites) it is that certain sites which do render correctly cause the CPU to spike and stay at the consistent 25-40% level across all the cores on my machine.

I have been able to track this down to specific sites, for example GMAIL is one. So long as the browser window (tab) that has GMAIL opened is focused on the spikes continue. Even after I switch to a different window (I run the XPager here, so a total of 9 virtual Desktops) the CPU spikes continue UNLESS I first open a new TAB in the same browser window that has the GMAIL tab and leave that as the last focused tab.

So for the most part I ignore it, let the spikes continue and just live with it.

However, a little over a month ago I was testing the 3.23.14 ACPI release on the AOS Testers list and found some weird stuff, specifically it was no longer possible to shut OFF a particular CPU core b/c the ACPI Power Manager would continually overwrite that change and effectively bring the core back on-line (that causes me a problem here, but that's a different story).

I logged an issue for this, the root cause has not been identified so far, but I priodically check the status of the ticket.

Well, I was out there last night and spotted another issue that was logged regarding the msg passing delays that are being seen, see https://mantis.arcanoae.com/view.php?id=3019. In that update a suggestion was made that the /NOHLT option produces a different result, a better performing one.

I was curious, so I tried it out. To my suprise here is what I found (see attached screenshot): the normal FF usage shows the spikes, but the instant FF is no longer the foreground application with focus the spikes go away. This is very different from the symptoms I described at the start of my post. The result actually is that the minute I am done looking at my GMAIL and switch to another window, another application, the spikes completely go away.

32
Applications / FreeRDP - we have options!!!
« on: September 30, 2020, 03:34:00 pm »
So I thought I would summarize my experience with this particular approach to working around our platform issues, one of them being a badly out-of-date web browser.

Some of you may be aware of the excellent FreeRDP (http://os2.snc.ru/product-freerdp.html) port that Andrey Vasilkin has made available. I have been using his previous version of the RDesktop (http://os2.snc.ru/product-rdesktop.html) for a while now. When he made the FreeRDP available I jumped at it because it offered something significant: the ability to run a remote APPLICATION as if it was a local app.

Now let me detail some of the pre-requisites because this isn't one of these solutions that everyone can use. But, if you have a Winx box running on your LAN (or even WAN) you certainly could deploy this approach.

In short, if you have such a Windows installation accessible to you, you can serve a particular APPLICATION on that HOST machine. FreeRDP is the client side of the RDP solution that will allow you to effectively just RDP to that single application the host is sharing. This is the key difference between something like RDesktop, which does a whole session sharing where the whole Win Desktop is available. As much as sometimes that's the preferred way to access, at other times this introduces a significant amount of overhead.

Subsequently, with FreeRDP you can certainly be a single-click away from having access to a Firefox 'whatever version' you need that you have running on that remote HOST, and it being served up locally on your OS/2 WPS literally as if it was just another PM window.

BUT

...there is of course a caveat that comes with this.

In order to be able to serve that application on the HOST side you need to run a Win Server OS, such as 2008, etc., basically something that has the Terminal Server (Remote Desktop Services) component up and running. So what is one to do if this is not an option?

Well, there is an alternative. Take a look at Thinkstuff (https://www.thinstuff.com/products), and in particular their XP/VS Terminal Server. There are several options available as far as the licensing configuration. The lightest option is the 'Lite' configuration, and at $39.99 USD that's a pretty low-entry to a viable way of "running" (double-quotes there b/c you're NOT really running it locally, this is not ODIN, rather only serving the display) most Winx applications locally on your OS/2 box.

So how did this pan out for me?

My Windows server is a Win7 Pro 64-bit box. I installed the Lite version of the ThinStuff's 'XP/VS Terminal Server' and was able to configure a WPS object that invokes the Firefox browser on that Win7 machine and serves it out locally on my OS/2 box.

This is on a LAN (so pretty quick) and even watching YouTube videos that actually works. There are some audio issues I'm encountering, intermittend drop-outs, etc., but video and audio is already a heavy application, so the mere fact that I can get to numerous web-sites while still directly in front of my OS/2 screen is great!!!

Anyways, for what it's worth, there is an alternative some of us may be able to use.

33
Web applications / FF 45.x and resulting broken sites
« on: September 19, 2020, 03:50:13 pm »
I suppose this was inevitable, but I'm now starting to see consistent behaviour with multiple sites that no longer render in our FF 45.x browser.

Things were never smooth sailing in the past, but at least the previous symptoms were more of a nuisance as opposed to just a flat-out stop.

So here is the thing: before I completely decide that certain browser traffic must go on another device, can you guys concurr that you are seeing similar problem on one of these sites?

Take someting like https://www.linkedin.com. This was fine in the past, SLOOOWWW for sure, but between NoScript and uBlock I managed. Well, today that is no longer the case. When attempting to visit that site I simply get the traveling timer bar, but nothing else ever happens.

Make no mistake though, it is not like the browser isn't doing anything, the CPU cores spikes, I literally get 100% utilization showing, FF eventually becomes fairly non-responsive.

So I took a stab at this to try to at least figure out where the proble lies (heck, I thought maybe one of my addons is the culprit - it isn't as I tried this very same thing with a blank profile, same result). I relied on the FF built-in Performance monitor, did some capture and came away with some results, which is what I'd like to share.

Basically, it appears that our FF's Graphics engine/function goes into some serious effort and yet nothing is produced.

For those of you who may be more familiar with the FF inner-workings, does what you see in the attached screen capture make any sense?

Thanks!
-Dariusz

34
Programming / REXX and double quoted INPUT parameter strings
« on: September 14, 2020, 02:18:09 am »
So let me set the context of the question: I wrote a little REXX script to automate the execution of RSYNC. The goal is to have this scheduled in CRON2 to allow me to copy the contents of the LOCAL drives over to the NAS box, or if needed to a LOCAL to LOCAL copy (such as a backup run to another HDD).

The INPUT parameters to the REXX script are various LOCAL folders as well as the NAS box folder where I'm going to copy the stuff to, basically the SOURCE and TARGET folder destinations.

So while I have this working well enough to handle any folder names where no space exists in the name:

1) G:\DESKTOP\AUTOMOTIVE

...I have ran into some difficulties trying to get this working for the folders that do have a space:

2) G:\DESKTOP\MISC INFO

Clearly passing such a folder name as an un-double quoted value just causes the REXX script to interpret the first space as a argument separator, subsequently:

CLI  => rsync_backup NAS G:\DESKTOP\MISC INFO\ V:\PUBLIC\DOCUMENTS\MISC INFO

causes the following args to be identified:

Code: [Select]
SOURCE => G:\DESKTOP\MISC
TARGET => INFO\ V:\PUBLIC\DOCUMENTS\MISC INFO

Now if I do apply the double quotes, just like one would need to when attempting to work with such a folder through a CLI window:

CLI => rsync_backup NAS "G:\DESKTOP\MISC INFO\" "V:\PUBLIC\DOCUMENTS\MISC INFO"

causes the following args to be identified:

Code: [Select]
SOURCE => "G:\DESKTOP\MISC
TARGET => INFO\" "V:\PUBLIC\DOCUMENTS\MISC INFO"

The REXX code I use to pull in the arguments is pretty basic:

Code: [Select]
/* parse the command line to pick up the BACKUP_TYPE, SOURCE and TARGET locations */
arg backup_type source target

I looked through all my REXX PDFs and INFs but haven't found anything that spells out how to get around this.

So...any ideas how to handle this?

Thanks!

FYI: I've attached the actual rsync_backup.cmd file, but with a TXT extension.

35
Setup & Installation / how to tell DIR which date to show?
« on: September 12, 2020, 02:22:57 am »
Hmm...seems like this should be easy to do, and yet no hits so far as to what the answer is.

I have a JFS IFS here, DIR at CLI shows me the 'last write' date instead of 'creation date'. There is also the 'last access' date that's available.

Now, I spent years using HPFS386 and for the life of me can't ever recall having this issue, so I suspect something about JFS is different.

I have 'SET DIRCMD=/v' defined in CONFIG.SYS, but I am thinking this should not have an impact here as that only puts the byte counts into human-friendly format.

Is there a DIR option (or some system-wide config perhaps) that would allow me to force a choice of the date which should be reported?

To be honest when I'm physically at the CLI window it is because I want to see some of the details and from that perspective I care much more about the 'create date' than anything else.

Thanks!
-Dariusz

36
Internet / Replacing the OS/2 URL object - DIY version...
« on: June 16, 2020, 02:39:11 pm »
Hi everyone,

I recently moved all of my DOCUMENTS (you know, the personal stuff that accumulates over the years, PDFs, text files, notes, etc...LOL) from my OS/2 box to the NAS. The NAS does not support EAs, and as expected all of my URL objects (the WPS object that you to double-click thus openning up your web browser) have now become simple text files on the NAS.

Alright, so I knew this was going to happen and now am trying to figure out a way to deal with it.

The NAS files that used to be the OS/2 URL objects are now a simply text file with the one-liner content being the URL itself. I figured that if I can read the contents of that file and pass that onto Firefox as a command line parameter that should in fact give me a "poor man's" replacement for the WPS UL object.

I need a little help in making that into a one-liner though. The alternative is of course to do up a small REXX script, but if I can parse the stuff as a one-lines, well, that's better since I can simply create the program object for it and convert that into my own DIY "OS/2 URL Object" solution.

So here is a scenario I'm looking at:

1) URL text file => 'CBC - URL'

2) content of the above text file
http://www.cbc.ca/news

3) current DIY URL object CLI
g:\apps\tcipip\firefox\firefox.exe &head "cbc - url"

4) sample execution

The CLI produces the following, but the firefox window pops up on it's own and the URL is not passed to it

Code: [Select]
[V:\PUBLIC\DOCUMENTS\Dariusz]g:\apps\tcpip\firefox\firefox.exe &head "cbc - url"

http://www.cbc.ca/news

The idea here being that I can use the HEAD utility to simply pull that single line from the text file (which is the actual URL) and pass that onto Firefox as a CLI argument.

My problem is that testing this out I clearly do not have the right CLI definition, simply put the "&head" must feed INTO firefox.exe and I don't know how to do this in our OS/2 world.

I spent a little time going through various OS/2 help files but didn't spot anything that worked for me.

So can anyone give me the proper syntax for the CLI structure to pass an output of the 2nd command as input to the first command?

Thanks!

37
Networking / Samba client - correct username / pass setups...
« on: June 11, 2020, 07:43:56 pm »
So I've got just the Samba client on my OS/2 box here, it connects to multiple shares on the LAN.

Ever since I had installed AOO way back when I have had 'SET USER=...' entry in my CONFIG.SYS. Didn't really pay much attention to it, but now that I'm spending a lot more time with the Samba stuff I'm realizing that this little environment variable has had an impact on the Samba functionality.

Specificially, the client will look for the USER variable and if it finds it (which it does here) it uses that as the default authentication username. Since my current value in the CONFIG.SYS does not match the ID I have defined on my multiple LAN devices, I now have a bit of a problem. Now of course I can spec a different name, and when my shares are configured through NetDrive (which is how most of them are) this USER variable has very little impact. However, when doing stuff at CLI the situation is quite different.

Further on, looking at the 'kLIBC - User Manager' I now of course have that current CONFIG.SYS value showing, but the correct username does not exist.

So, can anyone explain the impact (potential, maybe none?) of actually creating a 'NEW kLIBC - User Manager' account with the LAN username I want to use?

The bigger goal here is to configure the Samba client side of the house on my OS/2 box so that the true LAN ID/PASS combinations are stored locally on my OS/2 box (in the Samba pass file presumably) and by default that combination is used do that I do not have to enter it at the CLI level.

Thanks!

38
Programming / Git : SVN plugin
« on: May 10, 2020, 08:27:26 pm »
Has anyone here tried (or succeeded in) installing the SVN plug-in for the Git system?

I have Git 2.11.0 installed and wanted to pull some source from several SVNs repos. Instead of installing SVN as a separate version control system I figured I would just try to utilizes Git instead.

However, to do so you need to install git-svn and that's where I'm running into a problem, a pre-requisite is missing and unfortunately trying to install it causes another error which I can not figure out.

Specifically, git-svn has the following dependency: perl-Git-SVN.noarch 2.11.0-4.oc00, which in turn shows this final error:

Code: [Select]
...
dependency: perl(SVN::Core)
 Unsatisfied dependency
dependency: perl(SVN::Delta)
 Unsatisfied dependency
dependency: perl(SVN::Ra)
 Unsatisfied dependency
...

What does "unsatisfied dependency" mean? How does one deal with this? Or is this a case of the RPM package itself having the wrong configuration?

I have no idea what is provided by the problem packages, but is this a case where perhaps I could try the RPM '-nodeps' parm?

Thanks!

39
Programming / Memory 'footprint' API?
« on: April 29, 2020, 07:12:26 pm »
In short, how do I take a snapshot of what's currently loaded?

I'd like to cover both the low and high areas.

What I am not looking for is a summary level stuff. Instead I want to see DLL/object level stuff, where it is sitting and so forth.

I suppose the closest I can describe it to is what the Theseus => Linear Memory Map => Show Process Contexts provides, except that I want to struture a memory map which shows each process currently running and which DLLs it links to and where these DLLs reside in memory.

Is there a starting point in terms of OS/2 APIs I should look at?

40
Programming / OS2AHCI interface?
« on: April 13, 2020, 03:20:08 pm »
Having recently made the move from DANIS506 to AHCI I've lost some of the IDE reporting functionality that was provided through little utilities like DiskMon, or even SmartMon.

So I thought I'd take a stab at building the AHCI versions of these, first one up being DiskMon since that provides a real-time snapshot of drive's throughput and temperature metrics, and I thought it would be a good excercise to try out my current GCC environment.

Alright, so here is the thing, I know I need to 're-point' the interface to the OS2AHCI, namely something like this:

Code: [Select]
  rc = DosOpen ("\\DEV\\IBMS506$", &hDevice, &ActionTaken, 0,  FILE_SYSTEM,
OPEN_ACTION_OPEN_IF_EXISTS, OPEN_SHARE_DENYNONE |
OPEN_FLAGS_NOINHERIT | OPEN_ACCESS_READONLY, NULL);
  if (rc) exit (rc);

needs to read more along the lines of this:

Code: [Select]
  rc = DosOpen ("\\DEV\\OS2AHCI$", &hDevice, &ActionTaken, 0,  FILE_SYSTEM,
OPEN_ACTION_OPEN_IF_EXISTS, OPEN_SHARE_DENYNONE |
OPEN_FLAGS_NOINHERIT | OPEN_ACCESS_READONLY, NULL);
  if (rc) exit (rc);

But i'm guessing it's nowhere near as simple as that. I mean I did make the change, the program succeesfully built and attempting to run it locked up the machine...LOL, I was expecting that. Strangely where it locked up was on a temperature fetch, and that may be more due to the fact that DiskMon is using the SMART reporting functionality of SmartMon (or so say the comments). I'm sure it's more complicated, but that is at least my starting point.

OK, so back to the post subject line: where do I obtain the definition of OS2AHCI interface?

I guess the only thing I can compare it to are the code samples that are included with the DANIS506 stuff, and those are very clear. I found nothing on the AOS pages on this, but before I log a ticket or post to the Testers' List I thought I'd ask here first.

Thanks!

41
Storage / How to force a folder 'refresh' from REXX?
« on: April 11, 2020, 04:30:43 pm »
I want to be able to run an on-demand (scheduled most likely) forced refresh of a particular folder, is there a call in REXX that would allow me to do so?

Basically I'm thinking the equivalent of 'RMB=>Refresh Now'.

Reason being: the built-in logic of XWP refresh causes frequent traps here (not hard trap, but soft trap) where XWP folders stop populating, this always requires either a Desktop re-start, or a complete re-boot. Since shutting this feature OFF I have been able to maintain pretty good WPS stability and rarely have the need today to force a re-boot.

However, there are certain folders, like my NAS LOG directory (where I'm sending all the various system LOGs) which of course are being populated very frequently and which my machine has a hard time detecting on a timely basis. Result being that from CLI I can see LOG udpates having been done, but through WPS these updated entries do not show up.

My thinking therefore is to schedule a forced refresh, run every 5 mins (I will play around with this) so that such particular folders are always up-to-date.

The NAS box access is through NetDrive and is a Samba share, I am currently using RL Walsh's OO.EXE to set the default folder view on several NAS folders to XVIEW, but as best as I can tell there is no option there for me to do a folder refresh.

Thanks!

42
Web applications / Google Photos - pics no longer showing?
« on: April 03, 2020, 06:20:50 am »
I was attempting to download some pics off of my mobile device and noticed that Google Photos (https://photos.google.com/) no longer actually shows the images. Instead of a photo I see a grey square/frame, but that's it.

Is anyone else seeing this?

I see no Google 'service outage' updates out there but having recently ran some RPM package updates I'm wondering if one of them messed things up. So before I go digging through them all I'd like to know if this is just my issue?

Thanks!

43
Hardware / PCIe - SATA3 card choices?
« on: March 29, 2020, 03:53:58 am »
The other thread about building an OS/2 hardware system got me thinking about where the current hardware I have could use a little "help".

All in all I have a working system, but as I toss the occasional upgrade at it here and there, the original foundational pieces are starting to show their age.

Case in point, the MSI motherboard (880G-E45) only has a SATA2 controller. A couple of years ago I installed a Samsung 850 Evo SSD and completely moved away from a HPFS386 setup to JFS. Working fine, no issues. But since 850 Evo is SATA3 capable I haven't really had the chance to benefit from that capability.

All the numerous arguments around whether a typical home user would benefit or not aside, the point is clear: SATA3 is faster than SATA2.

So I'm curious, has anyone attempted to use a PCIe SATA3 add-on controller card? If so, what chipset did you find supported on our platform, and was that in standard SATA or AHCI mode?

Here is the current output of SysBench for my 850 Evo SSD:

Code: [Select]
Disk I/O disk 1-2: 238473 MB - Samsung SSD 850 EVO 250GB
   Avg. data access time :       --.---    milliseconds
   Cache/Bus xfer rate   :      125.379    Megabytes/second
   Track 0 xfer rate fwd :      228.815    Megabytes/second
   Middle trk rate fwds. :      231.438    Megabytes/second
   Last track rate bwds. :      217.636    Megabytes/second
   Average Transfer rate :      225.963    Megabytes/second
   Disk use CPU load     :        2.830    percent
   -----------------------------------------------------------------------
   Total                 :       --.---    Disk I/O-marks

...and here is the output of DiskIO for the same drive:

Code: [Select]
DISKIO - Fixed Disk Benchmark, Version 1.18z
(C) 1994-1998 Kai Uwe Rommel
(C) 2004 madded2

Number of fixed disks: 6
Number of CD-ROM drives: 1

Dhrystone 2.1 C benchmark routines (C) 1988 Reinhold P. Weicker
Dhrystone benchmark for this CPU: 4327544 runs/sec

Hard disk 2: 255 sides, 30401 cylinders, 63 sectors per track = 238472 MB
Drive cache/bus transfer rate: 114199 k/sec
Data transfer rate on cylinder 0   : 205139 k/sec
Data transfer rate on cylinder 30399: 192094 k/sec
CPU usage by full speed disk transfers: 23%
Average data access time: Disk read error.
Multithreaded disk I/O (4 threads): 118139 k/sec, 14% CPU usage


Thanks!

44
Setup & Installation / command output redirection as input
« on: February 16, 2020, 07:56:57 pm »
This is an attempt on my part to convert a Unix RPM package dependencies command to OS/2 CLI command.

I have the following Unix command: yum deplist $(rpm -q libaio)

...tried the following:
Code: [Select]
[G:\](rpm -q libaio) 2>&1 |(yum deplist)
Loaded plugins: changelog, downloadonly, ps, replace, verify
Error: Need to pass a list of pkgs to deplist
 Mini usage:

deplist PACKAGE...

List a package's dependencies

...clearly this does not work, and I'm literally clue-less on the subjet of OS/2 command output/input re-direct.

Found the following two pages which seemed to make sense:
1) https://www.tavi.co.uk/os2pages/cmd.html
2) https://www.robvanderwoude.com/redirection.php

But none of these have yielded any results...so where am I going wrong with this?

45
Applications / FreeRDP and shared memory consumption...
« on: February 16, 2020, 05:37:05 pm »
For those of you using Digi's FreeRDP port, I'm curious what type of shared memory consumption you're seeing.

A quick summary if you are not familiar with the app: this is a RDP client, but a fairly sophisticated one. Previously I was using rdesktop, which was great, but it did not allow you to share just a single application. FreeRDP can do exactly that. So if you have a WIN box on your LAN you can certainly run any apps you want on that machine and simply bring them to your OS/2 desktop for ease of use. Heck, taken a step further I suppose you could even run a headless setup, or even a permanent VM on OS/2 which just serves up the specific application/stuff you want...but I'm getting off topic here.

Alright, so I'm seeing a massive shared memory consumption issue on my box. When FreeRDP starts up I literally have about 100M go away, this is despite the DLLs being marked as highmem for both code & data segment (Digi confirmed that's the intended setup).

He has reviewed this over on his box and is not seeing this problem. I was able to actually peg this down to any post 20191111 release. As it turns out that version loads almost everything entirely into highmem. Theseus shows just a small amount of about 1M in the shared area, everything else (about 99M) is loaded in highmem.

This is nearly the opposite of what I was seeing with the newer releases, latest one being 20200211. I tested in both OS2 and OS4 kernels, results are the same.

My setup is using a 3840x1200 resolution display, both OS/2 client and remote host. On the OS/2 side this is made up of two-head display (2 LCD panels of 1920x1200 native resolution in a side-by-side config) driven by SNAP video drivers. The post 20191111 FreeRDP releases can be somewhat made to handle this by putting them into custom resolution as opposed to FullScreen (which is how I run them normally). So I am guessing maybe this has some bearing on the shared memory consumption symptom I'm seeing...but that's just a speculation. In my testing forcing the display resolution down to something like 1920x1200 did not seem to improve the situation drastically.

So the question I have is: if you are running FreeRDP, what version is it, and what are you seeing in terms of shared memory consumption?

EDIT
====
Attached my DLLTree for the working and non-working releases. Some small differences, but I see nothing that blatantly pops out as causing the massive memory use.

Pages: 1 2 [3] 4 5 ... 9