|
Issues Resolved in 4543
|
|
Reference No.
|
Description
|
Affected Component(s)
|
Affected OS(s)
|
Affected Project(s)
|
|
BugID:
1972786
|
System does not automatically switch into clone mode when
external display is attached.
|
MINIPORT
|
Windows* 2000, Windows* XP
|
Intel(R) 915GM Express Chipset, Intel(R) 945GM Express
Chipset
|
|
Resolution Description:
Root Cause: Two issues
fixed by this DCN. Issue (1): Root Cause: The problem is the detection of
first boot event. By looking at "FirstBootDone" in gfxinitialize,
if this key is NOT present, it becomes first boot and creates the key. On
subsequent boots, since this key is presented, it is NOT first boot.
Problem starting as the driver is
installed on a fresh OS. It is immidiately loaded, and a call to
GfxInitialize is made. Therefore we will read and set this key, and caused
the actual first boot after the driver install be considered as a
subsequent boot. Since we do not decicde the configuartion based on first
boot algorithm on a subsequent boot, issue happens.
Fix Description: Create the "FirstBootDone" Key only on a
display modeset. Since the driver load immediately, the installation does
not load the display driver. This key will not be created on an actual boot
after driver install but get created on the first display mode set.
Therefore we will be able to detect the first boot and subsequent boots
properly.
|
|
BugID:
1944478
|
Can't change/specify the resolution when using silent
install.
|
MINIPORT
|
Windows* 2000, Windows* XP
|
Intel(R) 915GM Express Chipset, Intel(R) 945GM Express
Chipset
|
|
Resolution Description:
Root Cause: Two issues
fixed by this DCN. Issue (2): Root Cause: When we detect the devices for
first boot algorithm, we are doing a passive detect, relying on the active
detect performed by GfxGetChildDescriptor which also does a Legacy CRT
detection. But before coming to first boot algorithm (GfxInitialize), we
also do a display detect without legacy CRT detection to get the encoder
list for building FSM in the event manager. The event manager removes the
CRT from the list if it is legacy, so the passive detect in boot algorithm
does not show CRT. Then we go to LFP SPSD thinking we have only one device.
Fix Description: Do an active detect for getting the connected devices
for first boot algorithm. To reduce the number of active detection calls,
do a passive detect in the same handler (Driver_Fresh_Install), which is OK
since this will always be preceeded by an active detect in the first boot
algorithm.
|
|
BugID:
1979612
|
DVI is not getting detected.
|
SOFTBIOS
|
Windows* XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root
Cause: The wrong VBT data for DDC pin resulted in the GPIO
port value that initialized to 0 (invalid) and hence EDID read fails for
DFP.
Fix Description: Adding a fail safe mechanism to initialize the value to
a default value even if the value in VBT is not valid.
|
|
BugID:
1981202
|
DCT “D3D Manage” test under the D3D tab of the secondary
Controller causes a system hang.
|
D3D
|
Windows* XP
|
Intel(R) 915GM Express Chipset, Intel(R) 945GM Express
Chipset, Intel(R) 915G Express Chipset, Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root
Cause: The Flush Buffer with Power Management is called
only when the primary surface in back with at least one back buffer to flip.
The previous condition was only check for the primary surface not in the
front. That is not enough to prevent other unrelated condition to get into
this call, causes the unrelated condition to hang.
Fix Description: Add more conditions to check for the primary surface,
primary surface not in the front and at least one back buffer count. Since
the need to fix the other bug with this condition of 800x600 resoultion,
this resolution condition is also added to allow only 800x600.
|
|
BugID:
1972852
|
Video driver is unstable on LCD default refresh rate.
|
MINIPORT
|
Windows* XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root
Cause: There is no mode with RR>56Hz in the EDID, so
the OS comes in 64 Hz.
Fix Description: We add the 60 Hz mode in established timings of EDID
of LFP panel if it does not support any mode with RR>=56Hz. When adding
this mode to Established timing, OS recognizes all the other modes with
lower Refresh rates. Similarly for Clone, we also add this modes to
Established timings in EDID if LFP which is part of CLONE has max
RR<56Hz.
|
|
BugID:
1972680
|
DCT “Check resolution for Dual View” test under GDI tab
repeatedly fails.
|
GDI
|
Windows* XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root
Cause: 64-bit Windows* XP sending smaller size buffer
appears to be the root cause.
Fix Description: As per DDK, in the event of an error, DrvGetModes DDI
should return 0. So the solution is display driver to return ZERO as buffer
size when buffer for mode retrieval is less than the modes available.
Subsequently, OS requests for mode table by passing correct buffer size.
|
|
Issues Resolved in 4532
|
|
Reference No.
|
Description
|
Affected Component(s)
|
Affected OS(s)
|
Affected Project(s)
|
|
BugID:
1976044
|
Display resolution turns to 800x600 after driver
installation.
|
SOFTBIOS
|
Windows* XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root
Cause: The VBT has multiple panel with same PnP ID value.
All the entries have same DTD timings. The VBTMANAGER_FillSimilarPnPId adds
4 of them into the LVDS timing structure. During the mode table building
when these modes are to be added, the preferred one gets overwritten with
the another one that has similar timings.
Fix Description: Not to add a panel’s detail in VBTMANAGER_FillSimilarPnPId
if the timings are same, since the PNP_ID method is to support timings
where the X and Y are same but different refresh rate.
|
|
BugID:
1666600
|
Error screen while running 3DLabs race game for short
time.
|
OGL
|
Windows* XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root
Cause: This patch covers up a memory corruption to the
pixel shader constant in the indirect state buffer when there are no commands
pending and the scene has not be initialized. In this case, the synctag in
gfxFlushSB and gfxFlushNB are left as NULL, so old tag for the pixel shader
constant is used which may lead to free this resource prematurely. Setting
the synctag's default value to the lastSyncTag prevents the resource free
too soon; however, this is only a patch and does NOT fix the root cause of
this problem, which is to be determined.
Fix Description: Avoided NULL syncTag.
|
|
Issues Resolved in 4512
|
|
Reference No.
|
Description
|
Affected Component(s)
|
Affected OS(s)
|
Affected Project(s)
|
|
BugID:
1968600
|
The LED of the monitor connected to ADD2 card did not turn
amber in power saving mode.
|
AIM
|
Windows* XP
|
Intel(R) 915G Express Chipset
|
|
Resolution Description:
Root Cause: For CRTs,
when we send the Opcode 05 (SetActiveOutputs) with value 0, it just turns off
the DAC, but H/V Sync signals are still ON. The LED on the monitor does not
turn amber in low power saving mode. For this to happen, the device driver
has to set Opcode 7D (SetDisplayPowerState) to 3 for turning off all
signals.
Fix Description: The reason is
for VGA bypass mode that data and syncs are separated output from CH7021.
Data outputted from DACs and H/V syncs are output from our sync pins. The Opcode 05 will only turn off the DACs
but not the syncs. To turn on/off sync, Opcode 7D needed to be programmed.
|
|
BugID:
1970425, 1972217
|
BSOD when running DirectDraw DCT test.
|
ROTATION
|
Windows* XP
|
Intel(R) 915GM Express Chipset, Intel(R) 945GM Express Chipset,
Intel(R) 915G Express Chipset, Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root
Cause: In line number 251 of file:
source\d3d\common\eWindows* XPorts.cpp, a typo occurred on this line,
return D3dDestroyFakeSurface( lpDestroySurface );.
Fix Description:
D3dDestroyFakeSurface should be D3dDestroyFakeSurfaceAlm.
|
|
BugID:
1969765
|
PnP monitor becomes "Default monitor" after
connected the analog monitor with DVI-I connecter.
|
SOFTBIOS
|
Windows* XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: Analog CRT
DDC lines with SDVO DVI lines were tied.
Fix Description: In
IntCRTEncoder.c the VBT settings, ie. GPIO PinPair was not updated
properly. Also the SDVO slave address should be updated.
|
|
BugID:
1969174
|
LCD is not correctly displayed after booting with
graphics driver installed.
|
SOFTBIOS
|
Windows* XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root
Cause: In INTLVDSENCODER_IsDualChannelLVDS, variable
m_bDualChannel is set to true only if the bits 4 and 5 of 61180 are set,
and bFound is set to false. In later part of the code for 14x10, although
the user has a single channel LVDS; we set it to dual channel with m_bDualChannel to true. As per the ssf,
the values for 14x10 are 61180 => 40000300h ( Bit 5:4 are resetted), but
we still set m_bDualChannel to true in ' if (bFound == FALSE) '.
Fix Description: Put the bFound out side the if check for bit 5;4.
|
|
BugID:
1969297
|
VGA and TV is not recognized when VGA and TV is connected
to port replicator.
|
SOFTBIOS
|
Windows* XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root
Cause: In VBTMANAGER_GetDispConfRemovalList, a vbt config
removal entry is valid if both pipe A and pipe B entries are valid. If pipe
A entry is valid but pipe B is not, then we still add the entry in to
config removal table. E.g. LFP, CRT encoders are present and DFP encoder is
not present for a given system. Per current logic,
ENCODERMANAGER_GetDisplayIDListFromBitmaskPattern on pipe A returns a list
of display configuration to be removed as CRT. On pipe B, EFP + LFP
configuration fails since EFP encoder is not present. But the config
removal table still gets updated. The final display configuration to be
removed is now Pipe A => CRT and Pipe B => NULL, which is incorrect.
Fix Description: The entry is valid only if the display configuration
removal entry is valid for pipe A and pipe B rather than being valid on
pipe A or pipe B only.
|
|
BugID:
1951241
|
COPP Session with CGMS-A enabled causes unsmooth
playback.
|
DVD
|
Windows* XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root
Cause: The cause is a long delay to write CGMSA registers.
If three packets are written consecutively (i.e. when reasserting the
protection level), the amount of time spent doing this will cause the
unsmooth playback.
Fix Description: A better resolution timer was used to send a CGMSA
packet every third of a second instead of sending all three every second.
Additionally, a bug was fixed in GfxTimer() that prevented it from calling
timers other that the first timer in the list.
|
|
Issues Resolved in 4504
|
|
Reference No.
|
Description
|
Affected Component(s)
|
Affected OS(s)
|
Affected Project(s)
|
|
BugID:
1969098
|
Changing DVD Movie title while in Full DOS screen will
cause BSOD.
|
GDI
|
Windows Media* Center
|
Intel(R) 915GM Express Chipset, Intel(R) 945GM Express
Chipset, Intel(R) 915G Express Chipset, Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root
Cause: When we go to FSDOS, in the AssertMode call, we
will make pSharedInfo NULL. When we eject the current DVD which is playing
and switch to a different DVD, GHALInsertTag function will get a call to
insert tag from a D3D function. The D3D function will be in a while loop
until the GHALInsertTag returns True. But since sharedinfo is NULL,
GHALInsertTag will keep on returning false and this resulting in BSOD.
Fix Description: Do not return false even when sharedinfo is NULL.
Sufficient care is taken in the function when sharedinfo is NULL.
|
|
BugID:
1704120
|
No Device Radio Button is seen on CUI.
|
MINIPORT
|
Windows* XP
|
Intel(R) 915GM Express Chipset
|
|
Resolution Description:
Root
Cause: From 14.19 onwards, CUI queries Event Manager for
all valid configurations. All the
configurations in the machine array (created during boot time) are
validated using following criteria: Test 1) Configuration should be valid
according to SB, 2) Connectivity Check: A state is valid if all devices in
the configuration are connected otherwise state is invalid, and 3) Overlay
check: If overlay is running and display switching is blocked while overlay
is running, any state other than the current state is considered as
invalid.
Fix Description: Overlay running and display switch blocked with
overlay running is treated as a separate case. So in normal scenarios, only
Test#1 and Test #2 are performed to validate the configuration. Special
Case (Overlay running and display switching blocked with overlay running):
In this case, we do not need to iterate through the machine array (contains
all the valid states - build during boot time).
|
|
Issues Resolved in 4474
|
|
Reference No.
|
Description
|
Affected Component(s)
|
Affected OS(s)
|
Affected Project(s)
|
|
BugID: 1704170
|
Divide
by zero error.
|
AIM
|
TABLET
PC, Windows Media* Center, Windows*XP
|
Intel(R) 855GM CHIPSET
|
|
Resolution Description:
Root Cause: GetOemScalingData
is using a zero value for vga_pixels as a
divisor.
Fix Description: Added a condition to check vga_pixels for zero value and ensured that 'divide by
zero' error does not occur.
|
|
BugID: 1737111
|
A blank
screen is displayed on LFP after switching the Command Prompt to full
screen mode.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 915G Express Chipset, Intel(R) 945G
Express Chipset
|
|
Resolution Description:
Root Cause: 1) Blanks out when toggling
between FULL_SCREEN and CENTERED_IMAGE. 2)Wrong information in scratch
register 71410
Details:
The Scratch register (71410) is falsely updated as integrated LVDS on pipeA instead of SDVO LVDS. 3) LFP blanks out when
toggling between FULL_SCREEN CENTERED_IMAGE.
Fix Description: 1) Replace timing info dwHActive
and dwVActive with Gdi
values. No need to change all the timing strutctures
as UAIM uses only hActive, vActive
and dotclock in UAIM code. 2) Corrected wrong
values. 3) Stretching is disabled if the requested resolution is same as
native panel resolution, irrespective of whether the current compensation
is Full Screen or Centered.
|
|
BugID: 1704970
|
There
are two Digital Displays under Display Devices.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: EDID Read of SDVO-CRT was
failing because of extra STOP Cycle in switchtoDDC2 Call. Hence, even if a
CRT was not connected, it showed it as a legacy CRT. Also removed the check
for VBT bit for DVI-I.
Fix Description: Look into the EDID and decide
if its CRT or DVI.
|
|
BugID: 1765428
|
If you
set display on LFP only with FSDOS in full screen and then close the lid,
FSDOS does not switch to DFP at all.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause:
Corrected High resolution multiplers
properly.
|
|
BugID: 1946001
|
SMI
calls log are all wrong.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 915GM Express Chipset
|
|
Resolution Description:
Root Cause: New SoftBios
is not printing the final value of SWSMI register. New SB sets bot0 to 1
where printing of the value happens before that.
Fix
Description: Move the print
statement and print the value which will be written instead of any input
parameters. Also included format within simple %x to print out full 4/8
digits.
|
|
BugID: 1765396
|
Cannot switch
display to TV when connected via AverMedia
CX25905 ADD2 card.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 915GM Express Chipset
|
|
Resolution Description:
Root Cause: With latest driver display switches
to TV. But display corruption is seen because of improper programming of
the following registers. Bit 9:12 (clock phase) should be programmed to 9
in Broadwater. Bit 3:4 in sdvo port (hsync polarity and vsync
polarity). These bits are newly introduced in Broadwater. These bits should
be set based on timing structure (DTD) given by the encoder. Since this is
newly introduced, some encoders are not giving proper values in the DTD. (DTD’s are generated through CreatePreferredTiming
opcodes) Hotplug Support:
KCH enables Hot-plug support only for Lakeport, modified the conditions to
include Broadwater also. Updated dot clock ranges for setting proper MNP
values.
Fix Description: Corrected the mentioned issues based on
Broadwater B-spec.
|
|
BugID: 1798293
|
Restore
Defaults button is always active in TV settings page.
|
CUI2
|
Windows*XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: After clicking on restore defaults
and if the advanced TV page is open, the value in DACMODERGB, DACMODEYC do
not match with the default values.
Fix Description: When checking for default values to grey
out the restore defaults button, there is no need to check for DACMODERGB,
DACMODEYC which are connector types and not used in CUI.
|
|
BugID: 1947782
|
Help Tip
is not seen when you right click on "Close" button in information
page in the CUI.
|
CUI2
|
Windows*XP
|
Intel(R) 915GM Express Chipset
|
|
Resolution Description:
Root Cause:
Help was not called for information page close button.
Fix Description: Ported the code to do so.
|
|
BugID: 1947007
|
Tab is getting
disabled after applying the changes to Overlay Settings in CUI.
|
CUI2
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: When the focus to the dialog is lost,
all messages to the dialog will not go to a proper control in the dialog.
Fix Description: When the focus is lost set the focus back
again.
|
|
BugID: 1944241
|
LFP blanks
or panning occurs after removing CRT while in S3.
|
CUI2
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: In ConfigPersist.cpp::GetCurrentConfig, EDS is called with ENUM_CURRENT_SETTINGS
and ENUM_REGISTRY_SETTINGS. If the modes returned are different,
ENUM_REGISTRY_SETTINGS data is taken as the current config.
In these scenarios, the panning mode is returned in ENUM_CURRENT_SETTINGS
and the actual mode (which is also stored in the persistence registry) is
returned in ENUM_REGISTRY_SETTINGS. So the persistence functions think the
current mode is the same as the mode that need to be applied from registry.
That’s why it does not run ApplyConfig. So SyncOSAndDriverThread is not started. So when login
happens, the event is not processed by persistence.
Fix
Description: While comparing the
configuration, current config is taken from CUI
Service method and config to apply is taken from
persistence method GetCurrentConfig. A new
function is introduced to verify if these two configs
are the same or not. Also the logic used in GetCurrentConfig
is modified to rectify the issues that are faced due to usage of EDS call
for Display2 in non-MDS case.
|
|
BugID: 1698365
|
NULL SharedInfo pointer in GetRingBufferSpace
crashes system.
|
GDI
|
Windows*XP
|
Intel(R) 915G Express Chipset
|
|
Resolution Description:
Root Cause: In GetRingBufferSpace
function, we update the PGhalFctTable structure
using the information from PDeviceSharedInfo. If PDeviceSharedInfo[1] (first device info) is Null then we use PDeviceSharedInfo[2] (second device info). But here we
are not checking whether PDeviceSharedInfo[2] is Null or and the issue is happening when both
device info are Null. The issue does not arise if sufficient care was taken
by checking the status of Pdev before calling the
lower functions like GetRingBufferSpace function.
This case does not arise in GDI as we will check for Pdev
status.
Fix Description: A Null check was introduced for the
secondary device info before updating the PGhalFctTable.
If this is also null then in the else case, we will return Null. This is to
ensure that the crash doesn’t happen in Ghal.
However corruption can happen as the calling function will get the Null
pointer. In all GDI functions, Pdev status will
be checked before passing down. This is not the case with DD or D3D
functions. Hence similar Null checks were introduced in all possible Ghal functions which can be called from DD or D3D
functions and where a Null SharedInfo pointer can
be accessed
|
|
BugID: 1947009
|
In
Extended mode, when changing primary display device Color depth, second display
device screen cannot extend.
|
GIG,
SOFT BIOS
|
Windows*XP
|
Intel(R) 945GM Express Chipset, Intel(R) 945G
Express Chipset, Intel(R) 915GM Express Chipset, Intel(R) 915G Express
Chipset, Intel(R) 955GM Express Chipset
|
|
Resolution Description:
Root Cause: SoftBios
internally uses the GMCH_CONFIG data which is supposed to be up-to-date. In
this case it receives a call to enable VGA on PipeB
due to which it internally disabled plane A & B. Now GMCH config will say PipeA has
NULL_PLANE and PipeB has VGA plane, but later it
receives a call to set new mode in pipeB. It does
not receive any call in PipeA to set last mode.
So the internal gmch config
data will be out of sync.
Fix Description: Softbios GAL
layer disables secondary plane as part of enabling VGA plane. This resulted
in a GDI WA where they used to forcefully enable secondary plane during
hires mode set on primary. A better fix would be to do this within Softbios GAL layer where it exactly knows that it
disabled secondary plane & hence should re-enable it when coming out of
VGA mode. Even with this, there will be a slight flicker on secondary since
plane transitions are occuring display plane->vga plane(blank
screen)->display plane. This flicker is the same as it occurs with
current code base.
|
|
BugID: 1737132
|
Corruption
is dislayed momentarily after resuming from
Hibernation state (S4).
|
MINIPORT
|
Windows*XP,Windows*2000
|
Intel(R) 915GM Express Chipset, Intel(R) 945GM
Express Chipset, Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: When the adapter comes out of hibernation,
it again maps (temporarily) the GTT so that it points to contiguous pages
in the Stolen Memory. By writing the
pre-hibernate values to the GMADR it restores the contents of the Stolen
Memory. It also restores back the
pre-hibernate entries in the GTT Table. But at the time the driver is
called to come out of hibernate, the VBIOS has already put the adapter and
the child devices to PowerON state – so as to
display the POST screen. At this time the VBIOS is using the Stolen Memory
as the frame buffer. It also sets up displays in VGA Mode. The Displays are
enabled and hence restored Stolen Memory values are displayed onto the
screen –the display corruption.
Fix Description: The fix provided will turn off the display
data fetches and disable video output, before the stolen memory is
restored. Since the display devices are in VGA mode, the “VGA registers”
can be used to achieve this. The VGA’s sequencer register – ‘SR01 Clocking
Mode’ will be used to turn the Screen Off.
|
|
BugID: 1738090
|
Cannot
play WinDVD when setting color depth as 8 bit.
|
RESOURCE
MANAGEMENT
|
Windows*XP
|
Intel(R) 915G Express Chipset
|
|
Resolution Description:
Root Cause: The render core clock and display
core clock values on GDG were both 333MHz. However, since the values passed
to BEMP for display core clock were 200, and the dotclock
of 1600 X 1200 @ 75 Hz was falling within in the 10% range, BEMP was
invalidating the request for overlay creation.
Fix Description: (1) Modified the initialization section of
GDG only to read the display core clock and render core clock frequency
from GCFGC and assign the values to BEMP appropriately. (2) Also corrected
the way #define for render clock position (RNDRCLK_BIT_POS and RNDRCLK_BWG_BIT_POS)
to indicate the appropriate shift. This was wrongly assigned a different
value earlier.
|
|
BugID: 1942935
|
During
3D test of Windbag,
SOD occurs.
|
D3D
|
Windows*XP
|
Intel(R) 915GM Express Chipset, Intel(R) 915G
Express Chipset, Intel(R) 945GM Express Chipset, Intel(R) 945G Express
Chipset
|
|
Resolution Description:
Root Cause: After a context switch, stale indirect
state pointers were being accessed.
Fix Description: Bug fix to make sure the indirect state
block is not deleted until the new indirect state command + primitive +
Stored WORD is complete. This ensures the hardware has a valid pointer in
case a MI_SETCONTEXT is pending.
|
|
BugID: 1946475
|
Multiple
Test Failures.
|
D3D
|
Windows*XP, Windows*XP-64
|
Intel(R) 915GM Express Chipset, Intel(R)
915G Express Chipset, Intel(R) 855GM CHIPSET, Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: There was an "if"
statement using the wrong operator to do the comparison.
Fix Description: Change from the "=" to
"==".
|
|
BugID: 1702259
|
Invalid
label in GfxBugcheckCallback dump data.
|
MINIPORT
|
Windows*XP, Windows Media* Center, TABLET PC
|
Intel(R) 945GM Express Chipset, Intel(R) 945G
Express Chipset, Intel(R) 915GM Express Chipset, Intel(R) 915G Express
Chipset, Intel(R) 955GM Express Chipset
|
|
Resolution Description:
Root Cause: Currently in debug.c
we have two “IGF7” labels one pointing to Power Management and another
pointing to Memory Buffer, which causes problem to automate the formatting
of .enumtag data.
Fix Description: Renaming the second “IGF7” label to “IGF8”
label and updating the version number of the header as mentioned. Things to
be considered while modifying Bug Check data: (1) Every Label in Bug Check
data has to be unique. Duplication
of labels leads to problem in separating out data through tools. (2) On
every update to the Bug Check data the version number of the header should
be incremented by 1.
|
|
BugID: 1941300
|
Garbage
is displayed on HDTV and LFP in DDC.
|
MINIPORT
|
DOS 6.22
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: When we open lid, Event Manager chooses
LFP+TV (Clone) as the next configuration. Since we are in FSDOS we can
drive only SPSD or twin and so we convert LFP + TV (Clone) to LFP+TV(Twin) as the next configuration. However, LFP+TV (Twin) is an invalid
combination and leading to garbage on TV.
(In 14.18 code base (old SB), Softbios
used to take care of this by a WA which will drive only LFP even if
Miniport sends LFP+TV(Twin) as the configuration to be driven since LFP+TV
is invalid. But in new SB, this WA
is missing and hence we see the problem.)
Fix Description: During FSDOS switching, after converting
clone combination to twin, validate the Twin Configuration before sending
it to SB. If the (Twin)
configuration is not valid, call SB with only primary device in SPSD.
|
|
BugID: 1738708
|
The
incorrect information is detected in the IGT page after switching the
display mode to TV mode after pressing ACPI hotkey twice on LFP+ CRT DDT
mode.
|
MINIPORT
|
Windows*XP
|
Intel(R) 915GM Express Chipset, Intel(R) 945GM
Express Chipset, Intel(R) 945G Express Chipset, Intel(R) 855GM CHIPSET
|
|
Resolution Description:
Root Cause: We do not clear the registry for
DISPLAY1_UID2 (i.e. Twin Registry) when switching from Twin to SPSD using
hotkey. While going to SPSD, the registry of primary display and the
configuration registry are cleared but not apply to the registry for
secondary. Now CUI gets confused because it sees DISPLAY1_UID2 not equal to
zero in SPSD configuration.
Fix Description: During registry update in EMSetNextConfiguration, updating the values of relevant
registries and making all other registries values equal to zero. By doing
this, CUI will not be confused since while switching from Twin to SPSD
since we will clear DISPLAY1_UID2. To fix the problem of registry of
secondary getting overwritten in SET MODE PRIMARY HANDLER during log
off/log on and fast user switching, if the next configuration received from
Event Manager is the same as current configuration then do not update the
registry.
|
|
BugID: 1944238
|
Display
will exchange after undocking then hot docking with extended mode enabled.
|
MINIPORT
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: Miniport Driver notifies CUI
about the Dock vent in VALIDATE_CHILD_IOCTL. But during physical dock/undock
we are getting some SET MODE calls before VALIDATE_CHILD_IOCTL. During
these SET MODE calls, CUI persistence registries get overwritten with the
new configuration (LFP+CRT in case-1). So, now when the actual dock
notification comes through VALIDATE_CHILD_IOTCL, driver will notify CUI
about dock event and but since CUI persistence is already overwritten with
(LFP+CRT) in case-1 the display will go to wrong configuration i.e. LFP+CRT
in Extended Desktop.
Fix Description: We need a mechanism to handle dock/undock
event irrespective of the order in which SET_MODE call or
VALIDATE_CHILD_IOCTL is received. To do this we will notify CUI about the
dock event if it is found during SET_MODE call. The flowchart below
explains the mechanism in detail.
|
|
BugID: 1818717
|
Monitor
device suspend issues with Calistoga GFX.
|
SOFTBIOS
|
Windows*XP,
Windows*2000
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: There is a pixel stall coming
from dpiounit which is being shut off when the
Port is turned OFF. Note that the Display Plane and Pipe are still turned
ON and the display fifo is being drainied in an accelerated fashion as the port is
turned OFF. This causes an underrun (real underrun) finally. And this happens even with FBC shut
off. But Display h/w has hooks to
recover, but the Decompressor does not recover if
FBC is turned.
Fix Description: Due to the underlined condition, this was
considered as a WA. So a WA flag (WaDisablePlaneForMonitorOff) was defined in
the WA table which will be set only for Calistoga. Issue is not validated/reported on Alviso
& hence WA will not be enabled on platforms other than Calistoga. If
this WA flag is set, SetPortPowerState() interface of GMCH will be overloaded by GMCHNAPA_SetPortPowerState(). This method in addition
to inform port controller of power state change (default implementation)
will turn off display planes assigned to the pipe driving the input port, ulPort. This will be done iff
there are no enabled ports on the pipe. If there is one, then we don’t need
this WA as per the statement above. Similarly on monitor turn on, it will
enable planes as required.
|
|
BugID: 1943044
|
MV signals
are seen for 720p-standard when you run MV enabled DVD on HDTV.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause:
Issue happens because TV std from the registry was assigned to dwTVStandard due to incorrect check while updating the dwTVStandard.
Fix Description:
|
|
BugID: 1943462
|
The video
standard is changed from HDTV_480i59 to NTSC_M on HDTV after resuming from
Suspend state.
|
TVOUT
|
Windows*2000
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: This issue happens only after fresh
installation OS because DAC state is initialized to s-video by default
during resume from s3/s4.
Fix Description: Active detection is required
during resume from standby/hibernation becausebefore
detection set mode call will enable the encoder with s-video and never go
to active detection and it will detect as a s-video instead of component.
|
|
BugID: 1684292
|
Fixes
for Issues 1738922 and 1684292.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 915GM Express Chipset, Intel(R) 945GM
Express Chipset
|
|
Resolution Description:
Root Cause: Int10 functioality
in new SB was not implemented.
Fix Description: The following functions are added (1) 5F14:078Fh
and (2) 5F14:078Dh in the new SoftBios.
|
|
BugID: 1945711
|
In
extended mode, if you set TV as primary and LCD monitor as secondary, when
changing TV standard from NTSC to PAL, the LCD monitor can not display.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: Details: Plane swap happens
while setting the configuration to extended desktop with TV(p)
and LFP(s) and CSL is unaware of the plane swap because call is getting
into GAL directly. If we try to
change a TV standard when it was in ED and set Video parameter in iTvout calls TV set timing and its uses the previously
used info such as mode info and pipe and plane.
Fix Description: Issue happens because plane swap
happened during set display configuration and it was not updated in the
CSL.
|
|
BugID: 1939671
|
TV
format was not saved to CMOS.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945GM Express Chipset, Intel(R) 945G
Express Chipset, Intel(R) 915GM Express Chipset, Intel(R) 915G Express
Chipset, Intel(R) 955GM Express Chipset
|
|
Resolution Description:
Root Cause: When TV standard is changed, SMI
call to “set TV standard and connector” should be called. This function
call is missing. Therefore, the TV standard value was not updated in CMOS.
Fix Description: 1) Modified data structure
TVSTANDARDANDVIDEOCONNECTOR_SET to contain only those fields that are
required. 2) Added the call to the function that will generate SMI call to
update TV standard and video connector. 3) Made updates to the “setTVstandardandVideoConnector” interface to ensure
compliance with SMI specs (1.3, 1.3A, 1.3C).
|
|
BugID: 1940035
|
Vsync pulse width at 1080i is set to 3 lines. The correct
width is 5 lines.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: The Vsync
pulse width as calculated from DTD in the EDIDPARSER_GetTimingFromDTD
functions is not doubled for interlaced modes. This was resulting in the Vsync pulse width getting halved for 1080i mode. Also,
the Vsync end value in the static timing info for
1080i was not programmed correctly to output a pulse width of 5 Hsync pulses.
Fix Description: 1) Added code to double the Vsync pulse width for interlaced modes in the EDIDPARSER_GetTimingFromDTD method. 2) Changed the V
sync End value from 0x1088 to 0x1093 for 1080i_@50 and 1080i_@60 modes in GlobalTimings.c.
|
|
BugID: 1964177
|
Conexant TV-OUT switch failure
|
AIM
|
Windows*XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: A settiming
call was missing for Conexant TV-OUT which caused
the issue.
Fix Description: Made appropriate changes for it.
|
|
BugID: 1964033
|
A green
band is shown in the upper part of TV screen.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: When GDI viewport
is different than the mode to be set, the image should be centered. This
was taken care in old SB. But in new SB, this was not done for IntTV encoder.
Fix Description: The mode needs to be centered if
the GDI view port X or Y is less than the Hactive
or Vactive of the timing to be set. But INTTVOUTENCODER_PreSetControllerTiming returns TRUE
without doing any check.
|
|
BugID: 1950107
|
When
repeatedly changing the resolution, the selectable resolution is lost.
|
CUICOM
|
Windows*XP, Windows Media* Center
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: The issue happens because GDI
returns only 800 x 600 resolution in DrvGetmodes call from OS. Since we have rotation
disabled in INF, OS calls DrvGetmodes with buffer
that can accommodate only non-rotated modes. In DrvGetmodes
fn while checking for whether rotation is enabled, we have rotation enabled
in registry. So GDI reduces the buffer size four times inside DrvGetmodes.
The
Display1_EnableRotation registry key is enabled temporarily by CUI during
the resolution change in the function PersistRotationEnabledStatus(). Whenever CUI wants to get the current rotation
angle, it enables rotation temporarily. Within that period if GDI looks at
the registry, it find rotation enabled.
Fix Description: While checking current rotation angle, if
the rotation capabilities is 0, do not check for
current rotation angle and send the default value, which is Rotation_0
degree.
|
|
BugID: 1950632
|
Running
Hardware overlay like Mosquito.exe in panning mode gives rise to BSOD.
|
DD
|
Windows*XP
|
Intel(R) 945GM Express Chipset, Intel(R) 945G Express
Chipset, Intel(R) 915GM Express Chipset, Intel(R) 915G Express Chipset,
Intel(R) 955GM Express Chipset
|
|
Resolution Description:
Root Cause: The problem is from the
inconsistent coding mechanisms between view port rect
and overlay rect to calculate the right position
through the addition of the left position and the width. The same issue
also happens to the bottom position by the addition of the top and the
height.
Fix Description: The codes to add one to the overlay
destination in DCN 447071 will be removed to fix this issue. The one adding
to the viewport will be removed to fix the purple
line bug in DCN 447071 and to make the comparison to be consistent in ClipRectsOutsideViewport().
|
|
BugID: 1700061
|
Panning
display on CRT after Docking and Undocking of CRT while running SS.
|
RESOURCE
MANAGER
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: This issue happens because, when
we plug in the CRT, the modetable is getting
rebuilt. Here when we query RM for Trim modes, RM takes the current config into consideration(Twin
on PipeB). Thus it removes hi-res modes. Thus the
CRT modecaps get changed to 12x10 since that is
the highest available EDID mode in the modetable
after trimming. So display goes to panning.
Fix Description:
|
|
BugID: 1765410
|
Unsupported
LFP refresh rate modes are available.
|
GDI
|
Windows*XP
|
Intel(R) 915GM Express Chipset
|
|
Resolution Description:
Root Cause: If one of the device
in DDC is a legacy or if the hides modes check box is un checked for DDC,
then all the combined modes are reported to either of the devices. Hence
primary and Secondary mode table will have same X,Y,BPP
and RR. Hence it may have those refresh rates that are supported on the
other display and not supported on the current.
Fix Description: In DDC of it is hides modes un-checked
behavior for both Primary and Secondary report all the modes having X, Y,
and BPP in the combined mode list, but with a refresh rate common between
the Single Display mode and Combined mode list.
|
|
BugID: 1943405
|
Can not
execute Fn+F7 function after suspend/resume on extended desktop mode.
|
GDI,
MINIPORT
|
TABLET
PC
|
Intel(R) 915GM Express Chipset
|
|
Resolution Description:
Root Cause: After a suspend resume operation
in extended desktop mode, when we try to go to single display, DisableSurface() call does not come . The expected call sequence in
this case is DrvAssertMode false and DisableSurface() for secondary. This is a tablet PC OS issue, so
driver does not come out of extended desktop and switch does not happen.
Two problems here: 1)
We do not see the complete menu for toggle sequence on
pressing Fn + F7, because scratch bit is not set. Scratch bit is used to
inform SoftBios and Bios about the SD and ED
modes that are responsible for displaying the options for Fn + F7. The
scratch bit is set in DrvDisableSurface(). But since DrvDisableSurface
does not come, the scratch bit is not cleared. 2) If we somehow set the scratchbit in Assertmode(false) call, the
complete menu of ACPI devices is displayed with the DDC option, but
miniport does not allow the switch as registry is still MDS.
Fix Description: Since this is an OS issue, the following
are only workarounds. 1) Made changes in DeAssertMode() function
and inform Softbios about mode change by clearing
scratch bit in DrvAssertMode(false). 2) Made
changes to EM_ACPI_HOTKEY_handler. The AttachToDesktop Key is properly set by OS to indicate
single display mode. Here OS thinks that it is out of extended desktop, but
driver registry is not updated. The fix is if the AttachToDesktop
is 0 (SD) and driver’s config shows ED, then the
current configuration is set to SD. This is done in ACPI handler.
|
|
BugID: 1951199
|
On HDTV,
we need to hit restore button twice to restore the default settings.
|
TVOUT
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: The calculation to obtain the
default x and y in IntTV_GetSTDTVParameters is
not consistent with the calculation of position in IntTV_SetPositionByPct.
Fix Description: Modified the calculation to have the two
functions in sync.
|
|
BugID: 1950138
|
Not able
to select desired resolutions and refresh rates on some displays.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: EDID has 780x576@50Hz as the
largest mode and while removing modes with Unsupported RR, none of the
modes match this RR and hence were getting removed. So no modes get shown with
hide mode unchecked. Check for largest mode at same RR fixes this. For hide
mode checked as OS prunes all modes less than 800x600, no modes are shown.
This issue with OS pruning with hide mode checked is a universal problem
and hence no fix is provided in this case.
Fix Description: Code Changes: 1) Added a new parameter –
BOOLEAN bCheckRR - to the MODETABLE_GetLargestEDIDModeEntry
method to get the Largest Mode at the same RR. 2) Added a new parameter –
BOOLEAN bSameRR – to the MODESMANAGER_GetBestTimingAsInEDID
method to get the best timing for given mode at the same RR. The MODETABLE_GetLargestEDIDModeEntry is called with bCheckRR as TRUE if bSameRR
is TRUE. If bSameRR is FALSE or MODETABLE_GetLargestEDIDModeEntry returns no mode at
same RR, get the largestMode at any RR. 3)
Modified the signature of MODETABLE_GetLargestEDIDModeEntry
in Modetable.h to accommodate BOOLEAN bCheckRR.
|
|
BugID: 1950603
|
Blank screen
is displayed on LFP after switching the display to LFP only mode.
|
SOFTBIOS
|
Windows*XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: System BIOS reports lid state as
lid closed on a desktop system (Lakeport). Hence, LVDS port was getting
turned off.
Fix Description:
|
|
BugID: 1766227
|
Multi
monitor mode cannot be supported using CUI menu.
|
CUI2
|
Windows*XP
|
Intel(R) 915GM Express Chipset
|
|
Resolution Description:
Root Cause: Not able to switch to multi
monitor configuration as the modelist returned in
extended desktop/clone/twin was pruned for modes below 800 x 600.
Fix Description: Since the maximum mode for this
panel is 800 x 480, hence we were not getting any modelist. Also required a change in cfg,
wherein on interchnaging the devices in extended
desktop the modelist of the secondary was getting
copied to the primary and vice versa. Issue has been fixed appropriately in
a way which will not affect any system other than Santa Fe.
|
|
BugID: 1950613
|
Advance seting for CRT in DDC mode cannot be hidden by CUI
customization.
|
CUI2
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: Wrong placement of customization
of advance settings in clone.
Fix Description: Corrected this and added an else condition
for hiding the button if registry entry is NULL.
|
|
BugID: 1604205
|
Changes
to support Integrated Scalar in Gen4 platforms.
|
GDI,
RESOURCE MANAGER, SOFTBIOS
|
Windows*XP
|
Intel(R) 945GM Express Chipset, Intel(R) 945G Express
Chipset, Intel(R) 915GM Express Chipset, Intel(R) 915G Express Chipset,
Intel(R) 955GM Express Chipset
|
|
Resolution Description:
Root Cause: CUI-Driver Compensation interface
did not comprehend SDVO encoder limitations.
Fix Description: As part of scalar RCR implementation
needed for BWG, CUI-driver interface was re-hauled to comprehend SDVO
encoder limitations. For more details, refer to scalar HLD.
|
|
BugID: 1949389
|
Garbage
appears when moving the mouse on "AutoCAD2006."
|
GDI
|
Windows*XP
|
Intel(R) 945G Express Chipset
|
|
Resolution Description:
Root Cause: This application will send the mix
values to the stroke path function when we erase the lines, it will make
the back ground of the erased line visible. Garbage is shown on the screen
according to the mix values. In stroke path function, we will punt the
entire call if any of the line (in multiple paths) view port is greater
than 2048. If we draw some paths using the hardware and then if we punt the
call because of the above reason, OS will redraw from the first path
onwards above the paths drawn by the hardware. Now if we move the figure,
while erasing the paths, the previously half drawn paths by the hardware
will be shown on the screen according to the mix values and becoming the
garbage.
Fix Description: Do not punt the entire stroke path if we
already had drawn some lines in the paths using hardware. Instead, just
draw that particular line (whose view port is greater than 2048) by calling
EngLineTo function. This will avoid the redrawing
of the paths by OS which were already drawn by the hardware.
|
|
BugID: 1965055
|
Hotkeys
column will flicker when you adjust video overlay settings bar.
|
CUI2
|
Windows*XP
|
Intel(R) 945GM Express Chipset
|
|
Resolution Description:
Root Cause: Refreshing the Hotkey page when overrlay settings are modified.
Fix Description: Added an extra check to see that HK page
is not refreshed for a change in overlay settings.
|
|
Issues Resolved in 4452
|
|
Reference No.
|
Description
|
Affected Component(s)
|
Affected OS(s)
|
Affected Project(s)
|
|
BugID: 1579422
|
Unable
to Apply new color Setting with CRT and DFP in Twin Mode.
|
CUI2
|
Windows* XP
|
INTEL(R)
945G EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: For twin configuration, the
primary/Secondary devices concept is not used, so we should not treat the
devices connected as two seperate devices as they
are connected to a single pipe.
Fix Description: Fixed the issue
by updating the Builder structure only for one device.
|
|
BugID: 1740184
|
The
created schemes disappears in DTCM \ Scheme submenu after setting CRT+EFP Extended
scheme with CRT at 90 degrees view and then setting CRT+EFP DDC 90 degrees
scheme.
|
CUI2
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET, INTEL(R) 945G EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: Check for validity of schemes
failed.
Fix Description: Removed schemes
code based on PRD and updated files.
|
|
BugID: 1701635
|
IGDI
(Intel Graphics Driver Installer) Command Line Option does not work
correctly.
|
MINIPORT
|
Windows* XP
|
INTEL(R)
915GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: 1) The driver is not reading the
installer keys in case of a driver upgrade. 2) The driver is not writing to
proper place in case of driver upgrade as well as fresh install.
Fix Description: 1) The driver
reads the installer keys for driver upgrade as well fresh install. 2) The
registry in “HKLM\System\CurrentControlSet\Hardware
Profiles\Current\System\Current-ControlSet\Services\iAlm\Device0\Mon(display device number)" is written to for all
devices that were active at boot time and the display configuration
registry is also appropriately updated.
|
|
BugID: 1738782
|
75 Hz refresh
rate cannot be applied to CRT when the CRT was in DDC LFP+CRT.
|
CUICOM
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: Primary RR is copied into Secondary
RR. So Primary RR(60hz) is assigned for secondary
also and hence the issue.
Fix Description: 1. Use the return
value from GetValidateAdviseCfgMode only if
Validation fails. 2. Copy the Primary values from driver to Primary and
secondary values to secondary (when validation fails and advise
passes).
|
|
BugID: 1703673
|
Divide
by zero exception error.
|
MINIPORT
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: pModeTable->ScreenWidth
and pModeTable->ScreenHeight
were zero which caused a divide by zero condition.
Fix Description: Added checks to
handle such conditions gracefully.
|
|
BugID: 1742058
|
WinDVD 5.0.11.818 - Video corruption in MCE and in WinDVD w/ VLD On.
|
DVD
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: HW cannot accept the slice data
within one DWORD.
Fix Description: Work Around:
Shift the slice data to make it cross DWORD boundary.
|
|
BugID: 1741407
|
DPST
setting changes to default after disable/enable this function.
|
CUI
|
Windows* XP, WINDOWS* 2000
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: Minimum value was updated wrong
in CUI.
Fix Description: Fixed it
appropriately.
|
|
BugID: 1697484
|
After
Resume from the standby, Primary and Secondary display of EDS configuration was swaped.
|
MINIPORT
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: The issue is caused by GDI limitation
in switching from EDS to EDS in certain cases. One such case is when we are in CRT+DFP
(ED) with lid closed and pipe B assigned to DFP. So, now on lid open event if we select
LFP+CRT(ED) as the next configuration, following possible sequence of
SET_MODE calls can cause the problems.
Fix Description: Modifying the
rule for LID_OPEN as follows: ALWAYS->GO TO LFP + X IN Extended Desktop.
PRIMARY LOST->NO RULE (rule has no effect since there exists a valid
rule for ALWAYS case). SECONDARY LOST->NO RULE (rule has no effect since
there exists a valid rule for ALWAYS case). BOTH LOST->NO RULE (rule has
no effect since there exists a valid rule for ALWAYS case).
|
|
BugID: 1706215
|
When resuming
from S3/S4 with CRT connection, the LCD blinks.
|
CUI2
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: When resuming back from S3/S4,
initial configuration is applied. With persistence, configuration is read
from registry and is applied. If the config is
the same, it results in blink.
Fix Description: Added a check to
see if the config read from registry is same as
applied configuration. If so, the read config is
NOT reapplied.
Root Cause:
Fix Description:
|
|
BugID: 1940397
|
Hotkeys
are getting "disabled" before applying in CUI.
|
CUI2
|
Windows* XP
|
INTEL(R)
945G EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: The hotkeys were getting
registered even if you click on cancel.
Fix Description: Made a change in
a condition to make sure hotkeys are not registered when the application is
exiting.
|
|
BugID: 1909158
|
BSOD at
ialmrnt5 when launching TVfunSTUDIO after
power-on.
|
SOFTBIOS
|
Windows* XP
|
INTEL(R)
915G EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: Instead of using ulPipe, we are using pThis->m_ulPipeInUse. Due to this, array becomes out of bound
and system blue screens.
Fix Description: corrected m_ulPipeInUse
to ulPipe.:
|
|
BugID: 1698398
|
System
takes a long time to come up after resume from S3.
|
SOFTBIOS
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: 1) Wait for Vblank
times out a number of times during resume from S3. Wait for Vblank called from various functions in GAL, more often
for register update and during pipe disable. However, this wait is not
required when VGA plane is enabled, since double-buffering is bypassed if
VGA native is enabled, allowing pipe to shut-off immediately. 2) Wait for
pipe disable timeouts during resume from S3. Wait for pipe disable is
called after pipe disable, where we need to wait for the vertical counter
to stop to ensure pipe is completely turned-off. There has been a bit of
uncertainty as to the right logic to ensure pipe is completely turned off.
The current logic looks for the scanline counter
to be equal to VblankStart. But it has been
observed that this wait times out sometimes and scanline
counter values are not incrementing to reach VblankStart
after pipe is disabled. A much better approach is to check whether scanline counter values are incrementing after each
horizontal refresh time (maximum horizontal refresh time possible) and wait
until it stops. Implementation of this logic has not resulted in any
timeouts and also an improvement in resume time.
Fix Description: 1) Added the check to return TRUE if VGA
plane is enabled on pipe, before waiting for Vblank
in NAPAPIPE_WaitForVBlank. 2) Added the check to
bypass wait for pipe disable if VGA plane is enabled on pipe, after pipe is
disabled in NAPAPIPE_Disable. 3) Modified the NAPAPIPE_WaitForPipeDisable function to implement the
new logic.
|
|
BugID:
1940490
|
Position
icon does not get grayed out even after it reaches the extremes.
|
CUI2
|
Windows* XP
|
INTEL(R)
945G EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: Control ID's was changed.
Fix Description: Corrected the
control ID's to MCCS page ID's.
|
|
BugID:
1939497
|
TV
Settings display will shift right when we hit left.
|
CUI2
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: Default value is not a multiple of the step size
returned, hence we were overshooting the max size or going to a negative
value when moving
Fix Description: Added checks to prevent this from happening.
|
|
BugID: 1777529
|
Fn+F7
function does not work properly after resume from suspend.
|
SOFTBIOS
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: While resuming back from S3/S4
the scratch register 71410 was not getting updated. The system bios depends
on the Bits 26-24(of 71410) to decide the next display combination for
switching; since the register was not getting updated the various display
combination to switch was not consistent with the combinations possible
when we do a fresh boot.
Fix Description: Restoring the
upper 16bits of 71410 only in BASEMVBIOSHANDLER_PostSetAdapterPowerState.
Added a mask based restoration function in MMIOReg.c
file for the same.
|
|
BugID: 1766155, 1777522, 1777529, 1939607, 1940004
|
Unsupported
LCD 60Hz refresh rate appear on Intel CUI and OS user interface.
|
SOFTBIOS
|
Windows* XP
|
INTEL(R)
915GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: The LFP EDID has 1024x768@60Hz
bit set in established timing. The DTD supports 10x7 @50 and 40 Hz. The
expected behavior is to have only DTD timings being displayed. Hence anything
in EDID other that DTD should not be considered as timing supported for
LFP.
Fix Description: 1) Added INTLVDSENCODER_IsTimingSupported to prune modes not
supported in DTD. 2) In MODESMANAGER_BuildPrunedCloneModeTable,
we added check for ModeListNorPrimary also, else
if display 1 supports 60 Hz only and display2 supports 50 Hz only then the
number of modes on clone mode list added were null.
|
|
BugID: 1942663
|
HDTV
standard not getting updated on CUI.
|
CUI2
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: In Clone, when the HDTV is on
secondary and in primary if we change a mode MainDlg::ClonePagesValueSwapper()
function is called and there if the current mode supports the existing
format, we should not do a format switch.
Fix Description: If the current
format supports the selected mode then we retain the same format in MainDlg::ClonePagesValueSwapper(() function.
|
|
BugID: 1634078
|
OpenGL
API failed in resolutions higher than 1400x1050
|
OGL
|
Windows* XP
|
INTEL(R)
915GM EXPRESS CHIPSET
|
|
Resolution Description:
Roor Cause: The buffer region was limited by the viewport
which was not properly set so at higher resolutions it was not large enough
and failed to create the buffer region.
Fix Description: Properly set the viewport size for NAPA
and Gen4 and removed buffer region's dependancy
on the viewport size.
|
|
BugID: 1740003, 1941553
|
Hotkey/CUI
does not work when cold boot without CRT or TV.
|
SOFTBIOS
|
Windows* XP
|
INTEL(R)
945GM EXPRESS CHIPSET
|
|
Resolution Description:
Root Cause: SMI 1.3a and 1.3c spec have inconsistency in specifying
how the supported version is interpreted.
Fix Description:
Check the
difference in the 1.3 & 1.3 Update A versions and include 1.3 Update
‘A’ information in 1.3
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|