Issues
Resolved in Build 1246
|
Reference
No.
|
Description
|
Affected
Component(s)
|
Affected
OS(s)
|
Affected
Project(s)
|
BugID:
2292583, 2329581
|
Mouse Pointer Arrow sometimes has a bigger shadow.
|
Miniport
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset
|
Resolution Description:
Root Cause: The issue
happens because the cursor buffer gets update before it came into effect by
changing the cursor registers (will take effect in next VBLANK). So, there
is mismatch between cursor buffer data and cursor registers type. This
results into corruption for a brief moment. The solution is to write new
cursor data into extra cursor memory and then use the extra memory address
to enable the different type of cursor. Since, in Windows Vista* there are
two cursor buffers are initialized for achieving the independent rotation.
So, we use the second buffer when there is change in the type of cursor and
after cursor data is written into new buffer, we update the cursor address
register to jump to the new buffer. Changed the total cursor memory buffer
allocation to get memory for 4 cursor buffers in GMM. 2 extra buffers for
Independent rotation case and extended desktop.
|
BugID:
2221895, 2332626, 2335186, 2335423
|
Functional Failure / SoftBIOS - Test ID 8650 - displayed
mode is switched to CRT only after changing the display mode from CRT+TV EXTD
to CRT+TV DDC while setting CRT to 2048x1536x32bit@60Hz. [64-bit REG]
|
Miniport, SoftBIOS
|
Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset
|
Resolution Description:
Root Cause: This issue is
happening because we are failing RFVN call while applying clone
configuration CRT+TV for 20 x 15 mode. Note that in this case TV supports
only up to 10X7 and the CRT has no EDID. While trying toapplying 20X15 on
it we fail to find native mode index for TV (which supports only up to
10x7). Here we are getting the entire mode list because of existing clone
mode pruning logic in SB which is as follows: 1) If both the devices have EDID,Prune the mode list
based on the max resolution of least capable device (i.e. remove all the
resolutions beyond this). 2) Else return the whole mode list (upto20X15).
This logic was OK for Windows* XP which support panning, but it is creating
problem for Windows Vista* which doesn’t support panning. As a solution I
have changed this behavior by using a flag which will be set by Windows
Vista* MP. This flag will indicate to SB that it has to return the pruned
clone mode list even in the case of EDID less device (i.e. enabling the
step 1 for EDID less device as well for Windows Vista*). This will result
in getting the mode list only up to 10X7 in this case and user will not be
in a position to apply a mode which is not supported by TV.
|
BugID:
2077788, 2329497, 2439006
|
iStdLib Hashtable Optimization.
|
Direct3D*, GHAL3D, OpenGL*
|
Windows* XP, Windows* XP-64, Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset
|
Resolution Description:
Root Cause: Algorithm
changes to the iStdLib CHashTable class.
Changing the first level binary tree to an array-based look-up
table. Local testing includes 3DMark03, 05, 06 andQuake3 and Doom3 on
Windows* XP32.
|
BugID:
2438187
|
Enable LP Writes (for BLK).
|
Miniport
|
Windows* XP, Windows* XP-64, Windows Vista*, Windows Vista*-64
|
Intel® 3 Series Chipset Family-B
|
Resolution Description:
Root Cause: Setting bit
11 of 0x20e4.
|
BugID:
2436409
|
Black screen on Windows Vista* restart with integrated
graphics.
|
Miniport
|
Windows Vista*
|
Intel® 945GM Express Chipset
|
Resolution Description:
Root Cause: Removing display
turn on call during ResetDevice resolves the issue. As per MSFT, we can
ignore ResetDevice call for shutdown/reboot case.
Fix Description: Ignore ResetDevice call for shutdown/reboot case.
|
BugID:
2333108
|
MIT-HOU- System unable to display Dual View when toggling
/w "Fn+F4".
|
SoftBIOS
|
Windows* XP, Windows Vista*
|
Intel® GM965 Express Chipset
|
Resolution Description:
Root Cause: The bModeCurrentlyNotSupported
in MODETABLE_Entry was not initialized to FALSE which was resulted in
garbage value and ISVALIDMODEENTRY to valuate to TRUE always. This fix also
ensures all variables are initialized in specified files.
|
BugID:
2331879
|
When main memory increased, the "Max Graphics
Memory" is not allocated according to the DVMT 4.0 spec.
|
CUI2
|
Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset, Intel® GM965 Express Chipset
|
Resolution Description:
Root Cause: If Windows
Vista* OS was managing the gfx memory, and the max gfx memory reported by
driver was not matching with that. Used a DX10 OS API to get the max gfx
memory from OS.
|
BugID:
2331469
|
Workaround: Incompatibilities with >32-bit Physical
Addresses.
|
GMM
|
Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset, Intel® G965 Express Chipset,
Intel® GM965 Express Chipset
|
Resolution Description:
Root Cause:
|
BugID:
2294686
|
[Windows Vista*32/Intel®
G965 Express Chipset/15.0]Hotplug detection does not work - Need fix on
latest Chip Rev.
|
AIM
|
Windows* XP, Windows Vista*
|
Intel® G965 Express Chipset
|
Resolution Description:
Root Cause: The hotplug
detection workaround, which was implemented for previous Chrontel cards,
needs to be extended for latest Chrontel 7315 Rev 2.1 card, too. WA: To
extend the card checks for the new revision, we made some code changes to
fix additional issues cause by above solution. Also address peer review
comments, the initialization of bDisplayEEDIDFilter is moved from
UAIM_InitializeDevice to UAIM_ResetDeviceCaps.
|
BugID:
2200609
|
HW Cursor shows corrupted shadow on dragging desktop
icon.
|
Miniport
|
Windows* XP
|
Intel® 945GM Express Chipset
|
Resolution Description:
Root Cause: While
dragging the desktop icon, the cursor image get corrupted for a brief moment.This
is because the cursor buffer gets update before it came into effect by
changing the cursor registers (will take effect in next VBLANK). So, there
is mismatch between cursor buffer data and cursor registers type. This
results into corruption for a brief moment. The solution is to write new
cursor data into extra cursor memory and then use the extra memory address
to enable the different type of cursor.
|
BugID:
2115096, 2171383, 2211026, 2290798, 2325928
|
MMIO flip corruption fix.
|
DVD
|
Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset, Intel® 945GM Express
Chipset, Intel® GM965 Express Chipset, Intel® 945G Express Chipset
|
Resolution Description:
Root Cause:
|
Issues
Resolved in Build 1229
|
Reference
No.
|
Description
|
Affected
Component(s)
|
Affected
OS(s)
|
Affected
Project(s)
|
BugID:
2292924
|
Intel® 3 Series Chipset Family: Enabling TV as secondary
blanks other displays.
|
Resource Manager
|
Windows Vista*
|
Intel® 3 Series Chipset Family-B
|
Resolution Description:
Root Cause: 1) Windows Vista* for Extended Desktop the
first call can be either primary/secondary. Swapping of plane/pipe is not
happening due to current checks. Current swapping logic of DPDD is same as
DPSM. Swapping logic needs to be changed to be independent of call sequence
and should depend on the previous plane-pipe configuration and new
plane-pipe configuration. Swap should be done properly if any change in
pipe/plane assignments to the display device. 2) Bandwidth Issues where seen in case of
pipe swap. And also Due to pipe-swap GRM is getting notified and release
device is called in between commit path due to swap handler. Thus releasing
the wrong plane. And CRT blanks out. 3) Reworked on DPDD code flow. Made
code changes and added proper checks where ever required. Done changes in
Miniport/Grm.
|
BugID:
2327163
|
W - DTM - Hazing on 24-bit panels.
|
SoftBIOS
|
Windows Vista*
|
Intel® GM965 Express Chipset
|
Resolution Description:
Root
Cause: 24-bit panel is hazing. SB uses LVDS port reg. value
in VBT to program dithering bit and clock control bits for the panel. But
this value is not updated accordingly, if the color depth is chosen as
24-bit. So, the issue happens.
Fix Description: Now, we look into the color depth and channel
selection variables for the panel and use the appropriate values to program
port reg.
|
BugID:
2333226, 2333359, 2333806, 2334473
|
GMM Node and Heap Synchronization.
|
GMM
|
Windows Vista*, Windows Vista*-64
|
Intel® 3 Series Chipset Family-B, Intel® G965 Express
Chipset, Intel® 945GM Express Chipset, Intel® GM965 Express Chipset
|
Resolution Description:
Root Cause: GMM has multiple
heaps and a single node manager. Each heap pushes and pops nodes from the
node manager for its own use. Threads can be accessing various heaps (which
intern acquires/frees nodes) simultaneiously, resulting in race conditions
w/in the node manager. Gaurds must be used to prevent corruption of nodes
and the node manager. Multiple threads could be accessing a single heap
simultaneously if the APIENTRY is reentrant. Gaurds must be put in for
those heaps. DCN also fixes independent rotation on BLB.
|
BugID:
2331469
|
Workaround: Incompatibilities with >32-bit Physical
Addresses.
|
GMM
|
Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset, Intel® G965 Express Chipset,
Cantiga, Intel® GM965 Express Chipset
|
Resolution Description:
Root Cause: Removed "pop-up" page subtraction from
SavedStolenSize computation. (Since the subtraction is not matched in GMM's
usage of stolen memory, it seems this would be causing less stolen memory to
be saved/restored than we actually could be using.)
|
BugID:
2327998
|
Turn off Adaptive Deinterlacing for Intel® 945GM Express
Chipset / Windows Vista*.
|
DVD
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset
|
Resolution Description:
Root Cause: Adaptive
Deinterlacing is being turned off on Intel® 945GM Express Chipset / Windows
Vista* due to performance issues.
|
BugID:
2327916
|
Intel® 3 Series Chipset Family: No hardware detection
sound for Dollar DVI monitor.
|
AIM, SoftBIOS
|
Windows Vista*
|
Intel® 3 Series Chipset Family-B
|
Resolution Description:
Root
Cause: DVI-I hotplug is not handled properly in Softbios.
If there is an interrupt at SDVO port, Softbios selects the first encoder
object from the encoder list which matches with that port. But in DVI-I
case, both TMDS and RGB can be at the same port. This results in wrong
encoder object selection. Consequently, interrupt is not cleared properly.
Fix Description: Clear interrupts for all encoder objects correspoding
to the specified port.
|
BugID:
2332452
|
LFP Polarity of H-Sync and V-Sync are incorrect.
|
SoftBIOS
|
Windows* XP, Windows Vista*
|
Intel® GM965 Express Chipset
|
Resolution Description:
Root
Cause: In GetLVDSTimingInfo(), we were not getting the flags
from the DTD timing and we were saving the restoring the values as were
before. Because of which even if we specify the Negative polarity in the
DTD timing we were not actually populating the same in theflags so the
register 61180 was not getting updated with proper polarity. It is always
set default values. Fix Done: Done changes, along with the timing also
populate the Flags taken from DTD timing. Removed the code of save and
restore.
|
BugID:
2329497
|
Stack underflow detection fix.
|
Direct3D*, GHAL3D, OpenGL*
|
Windows* XP, Windows Vista*
|
Intel® G965 Express Chipset
|
Resolution Description:
Root Cause: Fix stack underflow
detection where underflow was incorrectly detected one level too soon when
the max stack depth of the shader was equal to the max stack depth allowed
before system thread is required to be invoked.
|
BugID:
2200307
|
In DDC Refresh rates not enumerated for Digital TV.
(HDMI)
|
GDI
|
Windows* XP, Windows* XP-64
|
Intel® G965 Express Chipset
|
Resolution Description:
Root
Cause: Modes 8x6, 10x7, 11x8, 12x6 @60Hz removed for secondary
while doing intersection of secondary and clone mode list.
Fix Description: The logic in the Filtermodelist function was fine
tuned tobe in sync with the OS property page. This issue comes up whenever
the current RR of a mode not in EDID is greater than the maxRR for last
EDID mode. The logic is incomplete for this particular case and the RR was
not considered. Function was modified to check the minRR of current non
EDID mode to be less than maxRR for last EDID mode and then RR from minRR
to MaxRR is considered.
|
BugID:
2329629
|
Functional Failure / Power Cons - ForceTDR app causes CRT
to exit Power Savings mode but no image is displayed.
|
SoftBIOS
|
Windows Vista*, Windows Vista*-64
|
Intel® GM965 Express Chipset
|
Resolution Description:
Root Cause: This issue
happens when we have LFP and CRT connected with the system with
configuration set as LFP only and we force the TDR (with the help of an App),
Monitor sync turns on and it sows some horizontal lines intermittently.
This is happening because TDR causes Port registers values to get all 0s
which has following effects for analog port. As the LFP is also associated
with Pipe 0 (Gen-4 system), as soon as we enable pipe 0 after TDR, sync
starts coming on CRT causing undesirable behavior. As a solution for this I
have added a TDRHandler function in port controller which will disable all
the ports. This function will get called by SB TDR handler after Graphics
Device gets reset.
|
BugID:
2256824
|
FBC/S2DDT is enabled on PwrConsTool application of the
Client system while running OpenGL*_Tunnel application with modified BIOS.
|
Power Conservation
|
Windows* XP, Windows* XP-64, Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset, Intel® GM965 Express
Chipset
|
Resolution Description:
Root
Cause: For Intel® 945GM Express Chipset and beyond, the
FBC feature no longer needs to be notified about OpenGL applications
running. The flipping restrictions no longer apply after Intel® 915GM
Express Chipset. So, the testing needs to be modified for Intel® 945GM
Express Chipset and beyond. We will need to fix the code for Intel® 915GM
Express Chipset, though, so that OpenGL will properly get disabled in this
scenario.
Fix Description: 1) Platforms before Intel® 945GM Express Chipset do
not support FBC running with ANY kind of flipping. Therefore, checking is done on pre-Intel®
945GM Express Chipset platforms for Direct3D* and OpenGL being active, and,
if they are, disable FBC. 2) Platforms before Intel® GM965 Express Chipset
support synchronous flipping, but not asynchronous flipping. So, on
platforms before Intel® GM965 Express Chipset, we check for asynchronous
flipping from both Direct3D* and OpenGL. If it is occurring, then turn off
FBC. 3) Cleaned up the genearl Display Driver and GHAL areas where event
notifications to Power Conservation are generated.
|
Issues
Resolved in Build 1219
|
Description
|
Affected
Component(s)
|
Affected
OS(s)
|
Affected
Project(s)
|
|
First Event for Windows Vista*.
|
CUI2
|
Windows Vista*, Windows Vista*-64
|
Intel® GM965 Express Chipset
|
|
Resolution Description:
Root Cause: To implement
the first event behavior for Windows Vista*.
|
GHAL3D CURBE and Surface State
Streaming Optimization.
|
Direct3D*, GHAL3D, OpenGL*
|
Windows* XP, Windows* XP-64, Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset
|
|
Resolution Description:
Root Cause: Adding new indirect
state model for handling CURBE data and surface states. This model changes default model from a
caching model to a streaming model, however both models still exist in the
code. Tests ran on 3Dmark03, 3DMark05, 3DMark06, FEAR, Sims2, Quake3, Doom3
on Windows* XP, and 3DMark03 on Windows Vista*.
|
Task Sceduler Engine Has Stopped Working is displayed on
reboot with graphics driver installed.
|
CUI2
|
Windows Vista*, Windows Vista*-64
|
Intel® GM965 Express Chipset
|
|
Resolution Description:
Root Cause: If the call
is coming from TMM will not execute GetDevicemodes.
|
Abnormal screen happened in external monitor after resume
S3(Stand By) state in Dual Display Clone mode.
|
SoftBIOS
|
Windows Vista*
|
Intel® 945GM Express Chipset
|
|
Resolution Description:
Root
Cause: This issue has two parts. First is when 10x6 mode is
applied on LFP + SDVO_VGA DDC config (with FULL_SCREEN), the display on CRT
comes cornered to upper left. This happens as the ModesManager_CenterImage
is not called for FULL_SCREEN. Second part is after S3-resume in this
configuration, display on CRT comes stretched abnormally. This is because
of the CRT display being cornered to upper left part and stretching gets
removed if the VGA centering (BIT25|24 in 0x71400) is enabled. This
stretching also happens if there is upper-left cornered display on Int_LVDS
or DFP in LFP+DFP DDC config on CRL, and here also stretching vanishes as
soon as VGA centering is set. So, this upper-left cornered display in
FULL_SCREEN causes the issue.
Fix Description: For FULL SCREEN as well on SDVO VGA,
MODESMANAGER_CenterImage will get called.
|
Corruption on windowed OpenGL* apps in Windows Vista*
with Aero On.
|
OpenGL*
|
Windows Vista*, Windows Vista*-64
|
Intel® 3 Series Chipset Family-B, Intel® G965 Express Chipset,
Intel® G965 Express Chipset, Intel® 945GM Express Chipset, Cantiga, Intel®
GM965 Express Chipset, Intel® 945G Express Chipset
|
|
Resolution Description:
Root Cause: The problem is
caused by a corner case in the scissor to drawing rectangle optimization.
|
Sometimes rotation hot keys do not work on Windows
Vista*.
|
CUI2
|
Windows Vista*
|
Intel® 3 Series Chipset Family-B
|
|
Resolution Description:
Root Cause: Since
rotation caps is always returned as 0x07 for Windows Vista*, we added code
to check only if rotation is or is not enabled on Windows Vista*.
|
Doom3 Gen3 Windows Vista*
Hang Fix.
|
OpenGL*
|
Windows* XP, Windows* XP-64, Windows Vista*, Windows Vista*-64
|
Intel® 3 Series Chipset Family-B, Intel® G965 Express
Chipset, Intel® 945GM Express Chipset, Intel® 945G Express Chipset
|
|
Resolution Description:
Root Cause: Fixes Doom3
on Windows Vista* Intel® 945G Express
Chipset, Intel® 3 Series Chipset Family, and Intel® 945GM Express Chipset.
This issue was caused by OpenGL programming GHAL3D with an improper state
setting.
|
Fix for proper default values for unused "X"
channels in vertex element formats.
|
Direct3D*, GHAL3D, OpenGL*
|
Windows* XP, Windows Vista*
|
Intel® G965 Express Chipset, Intel® GM965 Express Chipset
|
|
Resolution Description:
Root Cause: fix unused
"X" channels (such as in R10G10B10X2_USCALED) so that they have
the proper defaults values. When the driver is run in HWVP mode, this fix
will allow the WHQL test VShader.exe to fully pass.
|
Issues
Resolved in Build 1195
|
Reference
No.
|
Description
|
Affected
Component(s)
|
Affected
OS(s)
|
Affected
Project(s)
|
BugID:
2077576, 2326924
|
GHAL3D VS - constant buffer optimization for indirectly
addressed constants.
|
Direct3D*, GHAL3D, OpenGL*
|
Windows* XP, Windows* XP-64, Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset, Intel® GM965 Express Chipset
|
Resolution Description:
Root Cause: Optimization for
HW vertex processing. This optimization gets applied for workloads
containing constants with indirect registers. This optimization will
attempt to batch up 'send' instructions to minimize the number of sends
that we make.
|
BugID:
2327772
|
Cannot toggle display via FN+F4 in Extended Desktop.
|
Miniport
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset, Intel® GM965 Express
Chipset
|
Resolution Description:
Root Cause: The issue is that
with Hotkey persistence set to ON in VBT, we are unable to come out of
extended desktop as we could in Graphics driver on Windows* XP. As part of
fixing the issue, we have tried to keep the behavior Windows* XP parity
w.r.t two VBT bits. One is Hotkey persistence and another is enabling
MDS/Clone on Hotkey event. With these changes, to exit out of EDS on an
ACPI Hotkey, Hotkey persistence has to be ON while to enter into EDS with
ACPI Hotkey both Hotkey persistence and enabling MDS/Clone on Hotkey event
should be set. This is basically the behavior we see in Windows* XP.
|
BugID:
2327409
|
Visual anomaly / DVD - Test ID 37278 - Media Player 11,
InterVideo WinDVD 8 - Blank playback displayed while playing flower_wmv_2mbps_cbr_29.97fps_na_8s.wmv
movie clip.
|
DVD
|
Windows Vista*, Windows Vista*-64
|
Intel® 3 Series Chipset Family-B, Intel® 945GM Express
Chipset, Intel® 945G Express Chipset
|
Resolution Description:
Root Cause: This DCN
fixes three issues: 1) When allocating a resource (surface), DXVA was
calculating the height of Y+U+V and then calling CreateResource(). This is
incorrect, because GMM already calculates the height, so DXVA must not do
that. 2) The condition to enable BT.709 Color Space Conversion was
incorrect, because it was not checking the Source FourCC. 3) The height for
the V-plane was being set incorrectly in HybridPlanartoPlanar conversion.
|
BugID:
2326213
|
CUI cannot select 1080i correctly on 1080p monitor.
|
CUI2
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset
|
Resolution Description:
Root Cause: User will be able
to see and apply interlaced modes when the display is inactive. Extended
Desktop position settings also works fine for interlaced modes.
|
BugID:
2325298
|
The Primary and secondary Displays blank out on resume from
hibernation for EDS CRT+HDTV.
|
Miniport
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset
|
Resolution Description:
Root Cause: The Primary
and Secondary Displays blank out on resume from hibernation for EDS CRT+HDTV.
It is actually not a blackout on both Primary and Secondary, but a blackout
on Primary only. Secondary blacking out (or empty screen ) with password
enabled during resume is an expected behavior. The issue is happening
because of a bug in the code, by which the source address is not
reprogrammed to proper values after a pipe/plane swap. Due to multiple
SetSourceAddress calls that OS sends in normal mode sets, the issue was
getting hidden in most of the cases. On resume from hibernation, with password
screen enabled, since we do not get subsequent SetSourceAddress calls the
issue got exposed.
Fix Description: If GRM recommended a Pipe/Plane swap, program the
source address properly. This would be done only if there is a change in the
planes being used, and will be ignored if GRM recommended a pipe/plane
swap, but the planes have not been swapped. ULT Details: Since these
changes would only affect Pipe/Plane swap, tested MDS with CRT+HDTV where
GRM would recommend a swap. Tested mode sets, S3/S4 resume under the same
config.
|
BugID:
2290995
|
OS Name is Not Appearing In CUI Information page.
|
CUI2
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset, Intel® GM965 Express
Chipset
|
Resolution Description:
Root Cause: OS name for
Windows Vista* is read from registry and will be shown on information page.
|
Issues
Resolved in Build 1179
|
Reference
No.
|
Description
|
Affected
Component(s)
|
Affected
OS(s)
|
Affected
Project(s)
|
ID: 629845
|
Gfx Reset WA for Bearlake-B.
|
SoftBIOS
|
Windows Vista*
|
Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution
Description:
Root Cause: BL-B silicon validation identified a bug
in the gfx reset entry logic of CI unit. This bug results in a hang that
prevents gfx from being reset. The bug shows up only in certain boundary
conditions where new requests are presented to CI unit on the display
interface and reset request is asserted to CI simultaneously
Fix Description: To work around
the bug, it is required that overlay, cursor and display engines are turned
off before the driver initiates a gfx soft reset.
|
ID: 630038
|
Bearlake enabling on Vista OGL.
|
OGL
|
Windows Vista*-64
|
Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution Description:
Root Cause: Very low risk fix for enabling OGL on Vista
Bearlake. Adding support for the Bearlake product ID.
|
Issues
Resolved in Build 1166
|
Reference
No.
|
Description
|
Affected
Component(s)
|
Affected
OS(s)
|
Affected
Project(s)
|
BugID:
2221928
|
Resume from S4 blanks out on HDTV.
|
CUI COM
|
Windows Vista*, Windows Vista*-64
|
Intel® 945G Express Chipset, Intel® Q33/Q35/G33 Express
Chipsets-B
|
Resolution Description:
Root Cause: Removing current
config check for CUICOM.
Fix Description: Fixed it
appropiately.
|
BugID:
2198158, 2255316
|
On test 606 with DDEX 4 and 5 on Windows Vista* 32 the
donut shapes have a black square box around them.
|
Direct3D*, Miniport
|
Windows Vista*
|
Intel® G965 Express Chipset, Intel® Q33/Q35/G33 Express
Chipsets-B
|
Resolution Description:
Root Cause: This appears
to be caused by an improper loading of Gamma Ramp values to the palette of an
8 bit mode. The typical gamma ramp table values make the screen data appear
as grayscale, which is incorrect and does not match the last call to
SetPalette(). There is also another related problem involving coming back
from power saving to grayscale when running a 8-bit application like
Foxbear. The file that generates this improper call is in PathManager.c and
is in the PathUpdateActiveGammaRamp() function.
Fix Description: Fixed it
appropiately.
|
BugID:
2099757
|
Corruption is displayed while running the 3DMark05
application at 180 degree rotation mode.
|
Power Conservation, Rotation
|
Windows* 2000, Windows* XP, Windows* 20003, Windows
Media* Center, Windows* XP-64, Windows* 20003-64, Windows Vista*, Windows Vista*-64
|
Intel® GM965 Express Chipset, Intel® Q33/Q35/G33 Express
Chipsets-B
|
Resolution Description:
Root Cause: Asynchronous flip is broken in HW in HW 180 degree case.
Fix Description: CheckForAsyncFlipSupport()
is the new function introduced which will update
pSharedData->bAsyncFlipsupport once at the time of DirectDraw* driver
loading. After which, all componnents should depend on pSharedData member
to know the avaialability of the feature.
|
BugID:
2294395
|
DTM-EDID 1.3 and DDC – Failing Log.
|
Miniport
|
Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset, Intel® 945GM Express
Chipset, Intel® 945G Express Chipset, Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution Description:
Root Cause: Back merge problem with IDDC and Downscalar check-in.
Fix Description: Fixed it
appropiately.
|
BugID:
2211334
|
Fn+F3 does not work after [A driver sttopped responding
and has recoverd] message is displayed.
|
Miniport
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset, Intel® Q33/Q35/G33 Express
Chipsets-B
|
Resolution Description:
Root
Cause: Once TDR happens, the driver does a Graphics Device
Reset with which all the MMIO and I/O register gets initialized to default
values. Due to which ActiveToggle list table index and GR18 I/O
registerwere also getting zeroed out as a result of which Display switch
was blocked.
Fix Description: So now once a TDR happens we save the state of scratch
registers before a graphics reset and restore the relevant fields after the
reset is done. Changes involved in this solution: Changes are in SoftBios
and Miniport. It involves adding TDR handler in GMCH and VBIOSHandler
class.
|
BugID:
2292243
|
The resolution and color bit change to the lowest setting
after switching output destination.
|
CUI2
|
Windows Vista*
|
Intel® 945GM Express Chipset
|
Resolution Description:
Root Cause: With Windows*
XP driver on Windows Vista*, MinDispMode call was returning 6X4@8bpp
because the check for Windows Vista* OS was missing.
Fix Description: Added this
check.
|
BugID:
2172939
|
Dragging content off left of screen causes player to
hang.
|
KMD
|
Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset, Intel® Q33/Q35/G33 Express
Chipsets-B
|
Resolution Description:
Root Cause: Negative
values in the 3DSTATE_DRAWING_RECTANGLE state command in the Present
DDI. The X and Y MinClip values are
unsigned words as described in the BSpec.
Negative values there are interpreted as large positive values,
which cause the entire Present blit to be clipped out, which causes the
screen not to be updated.
Fix Description: Clamp
MinClip X and Y values to zero in 3DSTATE_DRAWING_RECTANGLE in Gen4 Present
code. Do not allow negative values.
|
BugID:
2223147
|
When system boots, noise appears after progress bar of
Windows Vista*.
|
SoftBIOS
|
Windows* XP, Windows Vista*
|
Intel® 945GM Express Chipset, Intel® GM965 Express
Chipset, Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution Description:
Root Cause: During Post,
VBIOS uses High Res Plane to display OEM Logo’s. This data in the frame
buffer is not cleared and hence the corruption. SB do not program Start Address
as part of setmode. There are cases that Start Address is not available
during setmode in Windows Vista*. Windows Vista* OS will call Driver to
BlankVideo before setmode and enables back the video after setmode through
SetSourceVisibilty DDI. But SB internally disables and enables back the
video as part of setmode. This causes the corruption. This was required for
Windows* XP to synchronize with render. The same is not required in Windows
Vista*.
Fix Description: Added a new flag bVideoBlankingReqd in setmode
interface. Clients have to set/reset this flag based on their need.
Windows* XP setmode path has to set this flag and Windows Vista* setmode
path has to reset this flag. We need to set this flag in all cases within
SB where we use SetMode Interface.
|
BugID:
2290237
|
CLONE no longer functional and loss of monitor after lid
event (5 second wait between lid close and lid open).
|
Miniport
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset, Intel® Q33/Q35/G33 Express
Chipsets-B
|
Resolution Description:
Root
Cause: On lid close/open, we get two notifications from
system BIOS. 1) Lid switch notification(mandatory) „³ Drivershould notify OS
about addition/removal of LFP depending on lid open/close case
respectively. Driver does not
indicate a need of display change since we rely on TMM for handling lid
events. 2) Display switch notification(optional) -- Driver needs to ignore
this notification in Windows Vista* (this was added for Windows* XP since
lid notification does not get propagated to driver via. VALIDATE_CHILD
IOCTL.)
Fix Description: Handle events without any dependency on Event Manager
priority scheme. 1) During lid switch notification - Do not update lid
status. Just notify OS about addition/removal of LFP. 2) During display
switch notification (for Non Opregion case) -- Check dock status. If there is a change
in dock status: Dock event. Update
dock status and ignore this notification. Check lid status. If there is a change in lid status „³ lid
event. Update lid status and ignore this notification. No change in
lid/dock status: Service ACPI hotkey event.
|
BugID:
2197021
|
Flicker is displayed on HDTV after switching the desktop
resolution to 1766x1000 with HDTV_1080i59 video standard.
|
TVout
|
Windows* XP, Windows Media* Center, Windows Vista*,
Windows Vista*-64
|
Intel® GM965 Express Chipset, Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution Description:
Root
Cause: 3 Tap Filter was not enabled, hence the text
readability issue.
Fix Description: Following things needs to be followed to enable 3 Tap Filters:
1) If the vertical filter is being used the pipe interlace bit should be
off and the TV interlace bit should be on. 2) If the vertical filter is not
being used the pipe interlace bit should be on and the TV interlace bit
should be off. 3) If Vertical filters are enabled then we have to program
the Field_swap mode (BIT15 of 68000) to Normal
mode. 4) Load 3 Tap Filter Coefficients as given by Todd. Currently only
one set of coefficients is given and its does not have the granularity as
found in 5 Tap filters where we have atleast 11 sets for each of the
sharpness value. 5) 3 Tap Filter needs to be enabled only when the
display’s resolution is greater than 1024.
|
BugID:
2142742, 2221906, 2287905
|
Gamma issues in Windows Vista*.
|
Miniport
|
Windows Vista*, Windows Vista*-64
|
Intel® G965 Express Chipset, Intel® 945GM Express
Chipset, Intel® 945G Express Chipset, Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution Description:
Root Cause: WA for Gamma implementation
in Windows Vista* - maintain a status bit which gets set whenever there is
a CommitVidPN with at least 1 path OR SetSourceVisibility, with visibility
as TRUE. In UpdateActiveVidPnPresentPath, if this bit is set then get gamma
applied by CUI from registry, apply that and reset the bit. If not set,
apply gamma sent by OS or default as requested by OS. This bit gets reset
when CommitVidPn comes with no path or when SetSourceVisibility gets called
to turn off visibility. On Windows Vista*64 the issue continues to happen
for DC.
Fix Description: Fixed it
appropiately.
|
BugID:
2255316
|
Adjust UMD code to make it easier to debug performance
issues.
|
Direct3D*
|
Windows Vista*
|
Intel® G965 Express Chipset, Intel® 945G Express Chipset,
Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution Description:
Root Cause: Need to make easier to debug performance issues.
Fix Description: Modified the
driver code to make debugging command buffer usage percentage easier. Allows UMD to printout command buffer
usage and breaks apart reasons for flushes.
|
BugID:
2255620, 2257115
|
Display mode cannot be switched on CRT only mode via
pressing the ACPI hot key.
|
Miniport
|
Windows* 2000, Windows* XP, Windows* 20003, Windows
Media* Center, Windows* XP-64, Windows* 20003-64, Tablet Pc
|
Intel® GM965 Express Chipset, Intel® Q33/Q35/G33 Express
Chipsets-B
|
Resolution Description:
Root Cause: One issue was caused because OpRegion
memory was updated of connected devices wrongly with LFP as connected even
when lid was closed. The other issue was happening because OpRegion
implementation of driver was not taking care of Display configuration
removal Table.
Fix Description: The fix was to tweak the
function GfxGetAttachedDisplayList function in
such a way that if OpRegion update flag is set, then the device list gets
updated. Adding devices for CUI Static, Hotkey Static and LFP detection are
handled have a new function which will again make passive detection call
but without ENABLE_REPORT_LFP flag set. The above will be done during
enumeration itself. While discussing this issue, we thought of another
issue where in with CUI static bit set, during detection in boot, we will
not report disconnected devices and the same will be updated in OpRegion
memory. However, when the system comes up, we would want to have OpRegion
memory updated with CUI static devices of CRT and TV updated. So, we update
OpRegion memory with Static device and Hotkey device flag set across any
events.
The
fix was also to check during the determination of next configuration to be
set from the toggle list, whether it is valid or not by using Event manager
interfaces
|
BugID:
2211334
|
Fn+F3 does not work after "A driver stopped
responding and has recoverd" message is displayed.
|
Miniport
|
Windows Vista*, Windows Vista*-64
|
Intel® 945GM Express Chipset, Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution Description:
Root
Cause: Once TDR happens, driver does a Graphics Device
Reset with which all the MMIO and I/O register gets initialized to default values.
Due to which ActiveToggle list table index and GR18 I/O registerwere also
getting zeroed out as a result of which Display switch was blocked.
Fix Description: So now once
a TDR happens we save the state of scratch registers before a graphics
reset and restore the relevant fields after the reset is done.
|
BugID:
2222232, 2222651
|
DVD playback content flashing when HWA is enabled.
|
DVD
|
Windows Vista*
|
Intel® G965 Express Chipset, Intel® G965 Express Chipset,
Intel® GM965 Express Chipset, Intel® Q33/Q35/G33 Express Chipsets-B
|
Resolution Description:
Root Cause: Add floating
point save/restore in DXVA.
|
BugID:
2292571
|
Hotplug detection does not work.
|
AIM
|
Windows* XP, Windows Vista*
|
Intel® G965 Express Chipset, Intel® Q33/Q35/G33 Express
Chipsets-B
|
Resolution Description:
Root
Cause: When hot plug is enabled, it generates an interrupt
but driver does not receive this interrupt as interrupt is not enabled in
the GMCH by this time. As a result interrupt state is not cleared in the
encoder and consequently it blocks all future hot plug interrupts.
Fix Description: Fixed the
issue by putting a workaround in UAIMEH. Added a new function
UAIM_HotPlugInterruptClearWA() in UAIMEHDFP.c, which gets called after
enabling hotplug. In this function, it is verified to see if there is any
hotplug interrupt source at the output corresponding to the pAIMObj. If
there is an interrupt event at that output, interrupt state is cleared.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|