Status of OpenChrome Project (June / July 2018)

I missed reporting the status of OpenChrome Project for June 2018, so I will make it up, and combine it with the July 2018 edition.

Ever since I started to work on developing ATI Technologies RAGE 128 graphics stack, inevitably, the time dedicated to developing OpenChrome graphics stack went down quite a bit.  I must admit that I am personally pretty bad at handling multiple tasks, so as a result, vast majority of the time went into working on RAGE 128 graphics stack, particularly xf86-video-r128 (RAGE 128 DDX).  You can verify this from RAGE 128 DDX commit log.

Anyway, I still managed to get a few things done with OpenChrome.  I helped figure out why standby resume was not working properly on Samsung NC20 netbook when Xubuntu 18.04 is used.  Also, I managed to fix VGA display issue of ECS VX900-I and VIA Embedded VE-900 mainboards when VGA and HDMI are both connected.  Furthermore, I found a remedy for VIA Embedded EPIA-M830 not restoring the display properly after standby resume.  OpenChrome DRM got the fix, but have not committed the fix for OpenChrome DDX yet due to personal laziness.  Please note that VIA Embedded EPIA-M830 still suffers from a different standby resume issue (complete freeze), and I do not think it is OpenChrome related (OpenChrome DDX and DRM are both affected).

I auctioned off VIA Embedded EPIA-M800 mainboard in June, but unfortunately, it got destroyed during transit (USPS Priority Mail).  There was a hole that penetrated even the VIA Embedded box inside a USPS Priority Mail flat rate box, and badly damaged the mainboard. If the seller packed the box better, that might have saved the mainboard.  It was insured, so I got the money back.  I was hoping to test its integrated DVI transmitter (some early revisions of VX800 chipset contained a TMDS transmitter to support DVI), but it may never happen.

The last item to report is that I restarted the development of GEM / TTM memory allocator code.  I hope to put more time into developing OpenChrome next month.

Rewriting OpenChrome DRM GEM / TTM modules (again . . . )

Back in February, I discussed about completely rewriting OpenChrome DRM’s GEM / TTM code here.  The original motivation for this activity was to see if the GEM / TTM code developed by the previous developer was the cause of runtime screen resolution X Server crash I was observing.  While I did not discuss the matter in detail, I did manage to rewrite GEM / TTM code back in February, but after testing the code, the runtime screen resolution crash was still occurring, so I pulled the plug developing this code further, concluding that it was not really necessary.  That being said, it allowed me to look into other possible causes, and solved the matter in early March.

As discussed here, the SSD I was using (ADATA Ultimate SU800 SSD) died a few months ago, and took away the code I spent a few weeks writing . . .  Since then, I decided to expand into developing a new r128 DRM with KMS support, and I now have more reasons to get this GEM / TTM code developed so that the new r128 DRM can reuse the code.  I do not want the new r128 DRM project to become another vaporware project (meaning, I am going to finish it unlike other Linux graphics developers who tried, but failed to finish several KMS supporting DRM projects in the early Year 2010s), and having some proven code reused helps with the r128 DRM development.

Actually, there are two other reasons why I will like to implement my own implementation of GEM / TTM code.  First is, DRM maintainer wants OpenChrome DRM to remove unfinished acceleration code before the code goes into the mainline tree.  The way I look at this request is that it is an opportunity to remove unmaintainable GEM / TTM code written by the previous developer, and replace it with my own implementation.  Second is, I have observed freezes on VX900 chipset.  This happened when changing the screen resolution during runtime.  Yes, this issue was remedied back in March 2018, but apparently, it is a different bug from the one fixed in March 2018.  This new bug was discovered only recently.  It involves a crash with TTM.  Hence, the impetus the replace the existing GEM / TTM code.

This time around, the new GEM / TTM code is safely inside the trustworthy SanDisk Extreme 64 GB USB 3.0 flash storage stick.  It will stay inside there until it is ready to replace the existing code.

The struggle to rewrite the aging xf86-video-r128 code

I have been spending quite a few hours rewriting the aging xf86-video-r128 (ATI Technologies RAGE 128 DDX) code this week.  Unfortunately, portions of the code are written in a way that makes it hard to separate EXA and XAA acceleration initialization.  Furthermore, it supports the virtually deprecated DRI1 as well.  The initialization of these items (EXA, XAA, and DRI1) were written in a way that is very confusing and fragile ways that it is taking a lot of time to carefully untangle them.

What I am trying to do here is to untangle and isolate the initialization of these standards so that I can disable them without causing issues.  Furthermore, by implementing the initialization process this way, it will later make it easier to add DRI2 and KMS support down the road.

For now, I am trying to untangle the code without breaking the existing code.  I expect this to continue for several weeks.

xf86-video-r128 Version 6.11 released

I released xf86-video-r128 DDX for ATI Technologies RAGE 128 sometime yesterday.  It is a bug fix release where some RAGE 128 Pro cards get their VGA mistakenly recognized as DVI.  Worse, EDID (Extended Display Identification) based automatic detection of display was not working properly, and this was leading to the display complaining about “out of range” display signal being sent from the graphics chip.  This situation required the user to specify the monitor refresh rate inside xorg.conf.  With this release, there is no longer the need to do this.

At least the fix I committed fixes one bug, and there has been no new release for a while, so I just went ahead and released a new one.  Starting with Version 6.11, I will be changing the version naming convention, and patch level (i.e., the third number) will be going up more frequently as I rewrite / add new code.  For the next few months, most of my development effort for RAGE 128 DDX will be to stabilize the existing DDX UMS (User Mode Setting) code.

As of Version 6.11 release, the code contains these two major issues.

  • Standby resume freezes the computer upon resume
  • Occasional display of junk on the screen (i.e., when scrolling within a window or dragging a window)

I am actively working on both issues, but have not found the solution to the issues yet.

Anyway, if you support my work fixing and enhancing abandoned old graphics device drivers, please consider a small donation through my donation link here.

OpenChrome DRM drm-next-4.18 branch is now EOL

OpenChrome DRM drm-next-4.18 branch is now EOL.  The bleeding edge development has now transitioned to drm-next-4.19 branch.  Due to DRM maintainer changing the Linux DRM subsystem upstream repository location, I lost sync with the repository for a while.  Besides that, Linux 4.17-rc5 had file system (BTRFS) related boot issues, so I had to wait until Linux 4.18-rc1 or rc2 to update the drm-next-4.18 branch and then transition to drm-next-4.19 branch.  Now that I caught up with the bleeding edge of DRM subsystem, I plan to pour more time into OpenChrome DRM development along with continuing the development of xf86-video-r128 DDX.

My first bug fix patch for RAGE 128 DDX posted

Now that I have commit privilege for x.org X Server components, I was able to push some commits into the xf86-video-r128 DDX (RAGE 128 DDX) repository.  However, these commits did not alter the behavior of the code too much (they were almost all code refactoring commits), and did not fix any known problems of the code.

A few days ago when I was setting up my 64-bit x86 (x86-64 or AMD64) instruction set capable computer with an AGP slot for the purpose of kicking RAGE 128 DDX development into high gear, I plugged in RAGE 128 Pro Ultra 16 MB AGP graphics card into the AGP slot.  Unfortunately, the monitor connected to the graphics card complained about “out of range” signal coming out of the VGA connector.  I remembered that this particular card had this same issue back in Year 2015, and I reported the VGA display detection issue of RAGE 128 Pro Ultra back then.  Somehow, the bug report was closed with “RESOLVED FIXED” status even though the bug is still there.  Note that this bug affects certain models of RAGE 128 Pro, and not all models of RAGE 128 Pro are affected. The particular graphics card I used here is a pull from a Dell desktop PC.

I remembered that there is a workaround for this bug, and it is to limit the display frequency of the attached monitor via xorg.conf file.  With this workaround, display appears to work okay, but now the display is misidentified as DVI rather than VGA.  Since this workaround is merely a temporary band-aid, I decided to analyze the RAGE 128 DDX code to see what is going on, and try to fix the issue.  After analyzing the code, I came to the conclusion that the current RAGE 128 DDX code does not really detect the available display resources very well, and it will require a lot of work to try to fix the code.

For now, I will settle for a quick fix to the code that only benefits RAGE 128 on x86 platform.  This is because the fix relies on video BIOS supplied available display resources table, and it is unclear if there was something comparable to this on Apple PowerMac platform.  If you have any information about it, please contact me since I do not wish to break the code for PowerMac users.  Furthermore, if you have RAGE Mobility 128, M3, or M4, please test the patch as well to see if FP on your laptop is still working.  With this fix, I no longer have to limit the display frequency of the attached monitor, and X Server recognizes the attached monitor as VGA.  The patch is here and my supplemental post is here.

If I were to repost the instructions on how to compile the RAGE 128 DDX with the patch applied, this is what it looks like. Please note that ‘(“. . .”)’ portion should be replaced with actual storage location, and I assume that you are using Ubuntu based OS.

sudo cp /usr/lib/xorg/modules/drivers/r128_drv.so (“Anywhere appropriate to store a backup”)
sudo apt-get install git
sudo apt-get build-dep xserver-xorg-video-r128
(“sudo apt-get build-dep xserver-xorg-video-r128-hwe-16.04” for Ubuntu 16.04 with HWE)
mkdir (“Whatever appropriate name for a directory”)
git clone git://anongit.freedesktop.org/xorg/driver/xf86-video-r128
cd xf86-video-r128
patch -p 1 < (“File name of the patch”)
./autogen.sh –prefix=/usr CFLAGS=”-Wall”
(‘./autogen.sh –prefix=/usr –enable-dri CFLAGS=”-Wall”‘ if you want DRI1 enabled)
make
sudo make install

If it did not work, you should be able to restore the r128 DDX this way.

sudo cp (“Location where the backup of r128_drv.so was stored”)/r128_drv.so /usr/lib/xorg/modules/drivers/

If the patch does not cause a regression with PowerMac or RAGE Mobility based laptops, I will go ahead and release a new version of xf86-video-r128 DDX with this fix in the next few weeks.  Please note that this fix is just the first step, and I plan to fix several more issues of the code over the next few months.

If you appreciate my work on RAGE 128 graphics stack, please financially support me through a small donation via PayPal.  Part of the donation will pay for the increased electricity cost of running fairly power hungry AMD Opteron 165 on ASRock 939Dual-SATA2 mainboard.

Currently working on getting GPIO based I2C bus working for OpenChrome DDX

I am currently working on getting MSI CN700T mainboard’s DVI to properly handle the screen resolutions the monitor actually supports. At the present time, the DVI coming out of this mainboard can display certain screen resolutions X Server provides by default if EDID was not received by the DDX. I do not find this behavior to be very desirable. The issue I am observing here is that VIA Technologies VT1632A DVI transmitter is connected to GPIO2 / 3 pins to emulate I2C bus, but it is not working for the purpose of receiving EDID. That being said, it works for the purpose of detecting and controlling VT1632A. Please note that the code I am talking about is the existing OpenChrome DDX UMS (User Mode Setting) code for the upcoming Version 0.7 release.

Strange twist (?) to this saga is that if I let OpenChrome DRM to perform KMS (Kernel Mode Setting), its GPIO based I2C code works well, and I am able to obtain EDID from the monitor.

Finally was able to commit into xf86-video-r128 upstream repository

After weeks of delays, I was notified today that I now have the commit privilege to commit into xorg related components. Prior to this, I published 4 patches related to xf86-video-r128 DDX over at xorg-driver-ati mailing list. I used git-am to apply the patches to my local xf86-video-r128 repository.

git am (“Name of the patch”)

Since I had 4 patches, I repeated the above operation 4 times. Then, I pushed the changes to the upstream repository.

git push

Here is the first of the 4 patches I committed into the xf86-video-r128 DDX upstream repository.

Taking advantage of the software infrastructure Radeon graphics stack developers set up for RAGE 128 graphics stack development

Since my initial focus of my foray into developing r128 graphics stack is to first fix its broken standby resume behavior (Again, I really hate having to leave a room with an idling desktop computer.), I wanted to analyze the issue using the same development strategy I used successfully with OpenChrome Project. What I mean by “successfully” here is that over the past 2 years, I have been able to substantially improve the standby resume reliability of OpenChrome graphics stack. Unlike in the past, I can now put most of my VIA Chrome based computers, including laptops, to sleep and wake it up fairly reliably. In fact when one starts doing this, you happen to stumble across network device drivers that do not handle standby resume very well (it either freezes the system or cannot renegotiate with the network after standby resume). What was crucial in analyzing various standby resume issues was the existence of a backdoor register manipulation tool. This tool allows the developer to play around with hardware registers to figure out what is going on with the display and do some simple experiments before writing the software code to fix the issue.

For OpenChrome, one developer who was once VIA Technologies’s open source community liaison developed such a tool about 9 years ago, and I have to admit it was very helpful in analyzing various OpenChrome DDX / DRM bugs. Since such a tool worked out fairly well for me, I was thinking I will like to have something similar for the r128 graphics stack.

If I recall what I wrote previously, ATI Technologies RAGE 128 and early Radeon like R100 and RV100 actually have similar CRTC (display engine) registers. Hence, I figured maybe I can try to run a tool called radeontool with a RAGE 128, and maybe it will work.

So, I cloned the radeontool Git repository and built the binary.

git clone git://people.freedesktop.org/~airlied/radeontool
cd radeontool
./autogen.sh

Here is how to dump all registers.

sudo ./radeontool regmatch ‘*’

Here is how to disable CRTC1 display.

sudo ./radeontool regset CRTC_EXT_CNTL 0x00000448

Here is how to enable CRTC1 display.

sudo ./radeontool regset CRTC_EXT_CNTL 0x00000048

The good news is that I was successfully able to toggle a register bit to manipulate (disrupt) the display. Now, I need to figure out how to prevent the complete freeze after resuming from standby . . .

More on r128 DDX standby resume support status

I did mention in an earlier post that I will discuss in more detail about r128 DDX’s present standby resume functionality. In general, when it comes to dealing with standby resume, it puts the developer in a tough spot if the OS completely locks up after standby resume. In my case, I am sort of in this situation presently with the following computer.

  • AMD Sempron 3200+ (Socket 939)
  • ASRock 939DUAL-SATA2
  • 1 GB DDR400 DDR SDRAM (256 MB x 4)
  • ATI Technologies RAGE 128 Pro 32 MB (I think it is Xpert 2000 Pro.)

If you are into PC mainboards, perhaps ASRock 939DUAL-SATA2 mainboard was probably one of the most exotic mainboard to ever exist. It comes with native Gen 1 PCIe x16 and AGP 8X slots (Note: A real AGP slot, not those “fake” AGP slots hooking up AGP cards to the South Bridge’s PCI bus.) It even comes with a AM2CPU upgrade card slot. What happens is that if I put the computer into standby and then resume it, the keyboard connected to it will respond for a few seconds (i.e., Num Lock or Caps Lock toggles), but it will soon freeze completely.

I also have another computer to mainly to work with RAGE 128.

  • Intel Celeron 2.0A GHz
  • SOYO Computer P4S Dragon Ultra
  • 1 GB DDR266 DDR SDRAM (512 MB x 2)
  • ATI Technologies RAGE 128 Pro 32 MB (I think it is Xpert 2000.)

For this board, it comes with a Universal AGP Pro slot, hence, I can use it with RAGE 128 (RAGE 128 does not support AGP 1.5 V signaling). Intel did not make chipsets that supported AGP 3.3 V signaling cards for their Pentium 4 chipsets (i.e., Intel 845 and 850), but SiS and VIA Technologies did. Furthermore, this mainboard supports ACPI S3 State, and it is actually pretty rare to find a mainboard that has a Universal AGP slot that happens to support ACPI S3 State. Anyway, with this mainboard, when I resume the computer from standby, the computer beeps and keyboard does not respond at all. Nothing is displayed on the screen.

Going back to the ASRock 939DUAL-SATA2 mainboard based computer, when it comes out of standby, if I frantically try to transition to VT (Virtual Terminal) and back to XFCE desktop, it is able to display “something.” The “something” here is a totally distorted display. Based on my experience of fixing most OpenChrome standby resume issues, I will assume that r128 DDX is not reinitializing all necessary registers for a proper operation.

Typically in this type of a situation, comparing the register values for before and after standby allows me to figure out which registers are causing display issue. But before I can do that, I will need to figure out how to regain control of the computer after I resume from standby.