From: Paul Harrison on
I am having problems finding the Base address and socket controller
information for the !exca command. I presume that the socket number is just
the register number, and the Base is one of the allocated IO addresses for
the Cardbus controller. Nealry all of the data that is reported consists of
ffh, I presume that I am not accessing the right range of IO space, any
thoughts.

Paul Harrison

"Pavel A." wrote:

> OK. Now, can you flip this bit in a kernel debugger and
> see if you get 16-bit cycles?
> Set a breakpoint in your driver just before it starts to access the card.
>
> --PA
>
> "Paul Harrison" wrote:
> >
> > I have examined the ExCA regsiters with a utility supllied by TI for there
> > controllers, and the ExCA Register 11h is set to 0Bh. If I have read this
> > correctly we are running with a memory window width of 8 bits as bit 7 of
> > this regsiter should be set for 16 bit memory accesses. Can anyone help me as
> > to why we are running in 8 bit memory access mode. Could the CIS
> > configuartion affect this. I should mention that the PCMCIA card is a
> > multifunction device.
> >
> >
> > "Paul Harrison" wrote:
> >
> > > Thanks I will look into that.
> > >
> > > Paul
> > >
> > >
> > > "Pavel A." wrote:
> > >
> > > > Try the !exca windbg command to check the socket config registers.
> > > > Is the memory width set correctly?
> > > >
> > > > --PA
> > > >
> > > > "Paul Harrison" wrote:
> > > > > Can anyone confirm whether the PcCardConfig (M) option in the ConfigOverride
> > > > > section of an inf file for a PCMCIA card running on a cardbus (Texas
> > > > > Instruments PCI-1510) controller actually does anything under WinXP SP2. I am
> > > > > trying to enable 16 bit memory access to common memory on the card. I have
> > > > > checked the memory acess to the device using a logic analyzer, and the 32 bit
> > > > > memory requests are being broken down into 8 bit accesses. I have tested that
> > > > > the override section is actually being processed by adding extra resource
> > > > > requests, which are dully passed in to the driver, and also by adding an
> > > > > invalid option to the PcCardConfig line which resulted in the driver failing
> > > > > to install with an invalid inf section error. In clonclusion I know that the
> > > > > inf file override section is valid and being precessed.
> > > > > Has anyone managed to get 16 bit memory access to a PCMCIA card to work
> > > > > succesfully undr Win XP SP2. The hardware has been tested and succesfully
> > > > > produces 16 bit memory access under WinCE.
> > > > >
> > > > >
From: Paul Harrison on
I have included below the data from the !cfreg comamnd. The Area appears
memory mapped from the physical address. The data corresponds to the entry
for the device. I have cross checked the asssigned memory and IO addresses. I
need to know how to toggle the appropriate entry as the !dd command does not
work.


kd> !cbreg %%FEBFC000

CardBus Registers:
00: SktEvent 8 PowerCycle
04: SktMask 6 CSCEnabled
08: SktState 30000559 CSTSCHG PowerCycle R2Card OptiDock CardInt
DataLoss 5VCard 5VSkt 3VSkt CardPresent
0c: SktForce 00000000
10: SktCtrl 22 /CCD1

ExCA Registers:
00: IDRev 84 IO&Mem Rev: 04
01: IFStatus 7f BVD1 BVD2 CD1 CD2 WP Ready PowerActive
02: PwrCtrl 91 Vpp1=Vcc Vpp2=Off PwrEnable OutputEnable
03: IntGenCtrl 77 INTREnable IOCard ResetOff CardIRQ:7
04: CardStatChange 00
05: IntConfig 00 CSCIRQ:0
06: WinEnable c1 Mem0Enable IO0Enable IO1Enable
07: IOWinCtrl 00
08: IOWin0Start fe b0
0a: IOWin0Stop fe bf
0c: IOWin1Start fe c0
0e: IOWin1Stop fe ff
10: MemWin0Start 0bfd
12: MemWin0Stop 0bfd, WaitState:c000
14: MemWin0Offset 3403
18: MemWin1Start 0fff
1a: MemWin1Stop 0fff, WaitState:0
1c: MemWin1Offset 0000
20: MemWin2Start 0fff
22: MemWin2Stop 0fff, WaitState:0
24: MemWin2Offset 0000
28: MemWin3Start 0fff
2a: MemWin3Stop 0fff, WaitState:0
2c: MemWin3Offset 0000
30: MemWin4Start 0fff
32: MemWin4Stop 0fff, WaitState:0
34: MemWin4Offset 0000
40: MemWin0High fe
41: MemWin1High 00
42: MemWin2High 00
43: MemWin3High 00
44: MemWin4High 00
45: CLIOWin0High 00
46: CLIOWin1High 00
kd> !dd FEBFC000
Physical memory read at febfc000 failed
If you know the caching attributes used for the memory,
try specifying [c], [uc] or [wc], as in !dd [c] <params>.
WARNING: Incorrect use of these flags will cause unpredictable
processor corruption. This may immediately (or at any time in
the future until reboot) result in a system hang, incorrect data
being displayed or other strange crashes and corruption.

Paul Harrison



"Paul Harrison" wrote:

> I am having problems finding the Base address and socket controller
> information for the !exca command. I presume that the socket number is just
> the register number, and the Base is one of the allocated IO addresses for
> the Cardbus controller. Nealry all of the data that is reported consists of
> ffh, I presume that I am not accessing the right range of IO space, any
> thoughts.
>
> Paul Harrison
>
> "Pavel A." wrote:
>
> > OK. Now, can you flip this bit in a kernel debugger and
> > see if you get 16-bit cycles?
> > Set a breakpoint in your driver just before it starts to access the card.
> >
> > --PA
> >
> > "Paul Harrison" wrote:
> > >
> > > I have examined the ExCA regsiters with a utility supllied by TI for there
> > > controllers, and the ExCA Register 11h is set to 0Bh. If I have read this
> > > correctly we are running with a memory window width of 8 bits as bit 7 of
> > > this regsiter should be set for 16 bit memory accesses. Can anyone help me as
> > > to why we are running in 8 bit memory access mode. Could the CIS
> > > configuartion affect this. I should mention that the PCMCIA card is a
> > > multifunction device.
> > >
> > >
> > > "Paul Harrison" wrote:
> > >
> > > > Thanks I will look into that.
> > > >
> > > > Paul
> > > >
> > > >
> > > > "Pavel A." wrote:
> > > >
> > > > > Try the !exca windbg command to check the socket config registers.
> > > > > Is the memory width set correctly?
> > > > >
> > > > > --PA
> > > > >
> > > > > "Paul Harrison" wrote:
> > > > > > Can anyone confirm whether the PcCardConfig (M) option in the ConfigOverride
> > > > > > section of an inf file for a PCMCIA card running on a cardbus (Texas
> > > > > > Instruments PCI-1510) controller actually does anything under WinXP SP2. I am
> > > > > > trying to enable 16 bit memory access to common memory on the card. I have
> > > > > > checked the memory acess to the device using a logic analyzer, and the 32 bit
> > > > > > memory requests are being broken down into 8 bit accesses. I have tested that
> > > > > > the override section is actually being processed by adding extra resource
> > > > > > requests, which are dully passed in to the driver, and also by adding an
> > > > > > invalid option to the PcCardConfig line which resulted in the driver failing
> > > > > > to install with an invalid inf section error. In clonclusion I know that the
> > > > > > inf file override section is valid and being precessed.
> > > > > > Has anyone managed to get 16 bit memory access to a PCMCIA card to work
> > > > > > succesfully undr Win XP SP2. The hardware has been tested and succesfully
> > > > > > produces 16 bit memory access under WinCE.
> > > > > >
> > > > > >
From: Paul Harrison on
I have managed to find the appropriate entries in the device memory and
modified the appropriate register via memory with the !ed command and the
logic analyser now indicates that 16 bit memory writes are being successfully
generated. So it works on the controller hardware under XP. All I nned to
know now is why it isn't being enabled from the .inf file. Any thoughts
anyone. Could it be the CIS on the card. Thanks for the help so far.

Paul Harrison

"Paul Harrison" wrote:

> I have included below the data from the !cfreg comamnd. The Area appears
> memory mapped from the physical address. The data corresponds to the entry
> for the device. I have cross checked the asssigned memory and IO addresses. I
> need to know how to toggle the appropriate entry as the !dd command does not
> work.
>
>
> kd> !cbreg %%FEBFC000
>
> CardBus Registers:
> 00: SktEvent 8 PowerCycle
> 04: SktMask 6 CSCEnabled
> 08: SktState 30000559 CSTSCHG PowerCycle R2Card OptiDock CardInt
> DataLoss 5VCard 5VSkt 3VSkt CardPresent
> 0c: SktForce 00000000
> 10: SktCtrl 22 /CCD1
>
> ExCA Registers:
> 00: IDRev 84 IO&Mem Rev: 04
> 01: IFStatus 7f BVD1 BVD2 CD1 CD2 WP Ready PowerActive
> 02: PwrCtrl 91 Vpp1=Vcc Vpp2=Off PwrEnable OutputEnable
> 03: IntGenCtrl 77 INTREnable IOCard ResetOff CardIRQ:7
> 04: CardStatChange 00
> 05: IntConfig 00 CSCIRQ:0
> 06: WinEnable c1 Mem0Enable IO0Enable IO1Enable
> 07: IOWinCtrl 00
> 08: IOWin0Start fe b0
> 0a: IOWin0Stop fe bf
> 0c: IOWin1Start fe c0
> 0e: IOWin1Stop fe ff
> 10: MemWin0Start 0bfd
> 12: MemWin0Stop 0bfd, WaitState:c000
> 14: MemWin0Offset 3403
> 18: MemWin1Start 0fff
> 1a: MemWin1Stop 0fff, WaitState:0
> 1c: MemWin1Offset 0000
> 20: MemWin2Start 0fff
> 22: MemWin2Stop 0fff, WaitState:0
> 24: MemWin2Offset 0000
> 28: MemWin3Start 0fff
> 2a: MemWin3Stop 0fff, WaitState:0
> 2c: MemWin3Offset 0000
> 30: MemWin4Start 0fff
> 32: MemWin4Stop 0fff, WaitState:0
> 34: MemWin4Offset 0000
> 40: MemWin0High fe
> 41: MemWin1High 00
> 42: MemWin2High 00
> 43: MemWin3High 00
> 44: MemWin4High 00
> 45: CLIOWin0High 00
> 46: CLIOWin1High 00
> kd> !dd FEBFC000
> Physical memory read at febfc000 failed
> If you know the caching attributes used for the memory,
> try specifying [c], [uc] or [wc], as in !dd [c] <params>.
> WARNING: Incorrect use of these flags will cause unpredictable
> processor corruption. This may immediately (or at any time in
> the future until reboot) result in a system hang, incorrect data
> being displayed or other strange crashes and corruption.
>
> Paul Harrison
>
>
>
> "Paul Harrison" wrote:
>
> > I am having problems finding the Base address and socket controller
> > information for the !exca command. I presume that the socket number is just
> > the register number, and the Base is one of the allocated IO addresses for
> > the Cardbus controller. Nealry all of the data that is reported consists of
> > ffh, I presume that I am not accessing the right range of IO space, any
> > thoughts.
> >
> > Paul Harrison
> >
> > "Pavel A." wrote:
> >
> > > OK. Now, can you flip this bit in a kernel debugger and
> > > see if you get 16-bit cycles?
> > > Set a breakpoint in your driver just before it starts to access the card.
> > >
> > > --PA
> > >
> > > "Paul Harrison" wrote:
> > > >
> > > > I have examined the ExCA regsiters with a utility supllied by TI for there
> > > > controllers, and the ExCA Register 11h is set to 0Bh. If I have read this
> > > > correctly we are running with a memory window width of 8 bits as bit 7 of
> > > > this regsiter should be set for 16 bit memory accesses. Can anyone help me as
> > > > to why we are running in 8 bit memory access mode. Could the CIS
> > > > configuartion affect this. I should mention that the PCMCIA card is a
> > > > multifunction device.
> > > >
> > > >
> > > > "Paul Harrison" wrote:
> > > >
> > > > > Thanks I will look into that.
> > > > >
> > > > > Paul
> > > > >
> > > > >
> > > > > "Pavel A." wrote:
> > > > >
> > > > > > Try the !exca windbg command to check the socket config registers.
> > > > > > Is the memory width set correctly?
> > > > > >
> > > > > > --PA
> > > > > >
> > > > > > "Paul Harrison" wrote:
> > > > > > > Can anyone confirm whether the PcCardConfig (M) option in the ConfigOverride
> > > > > > > section of an inf file for a PCMCIA card running on a cardbus (Texas
> > > > > > > Instruments PCI-1510) controller actually does anything under WinXP SP2. I am
> > > > > > > trying to enable 16 bit memory access to common memory on the card. I have
> > > > > > > checked the memory acess to the device using a logic analyzer, and the 32 bit
> > > > > > > memory requests are being broken down into 8 bit accesses. I have tested that
> > > > > > > the override section is actually being processed by adding extra resource
> > > > > > > requests, which are dully passed in to the driver, and also by adding an
> > > > > > > invalid option to the PcCardConfig line which resulted in the driver failing
> > > > > > > to install with an invalid inf section error. In clonclusion I know that the
> > > > > > > inf file override section is valid and being precessed.
> > > > > > > Has anyone managed to get 16 bit memory access to a PCMCIA card to work
> > > > > > > succesfully undr Win XP SP2. The hardware has been tested and succesfully
> > > > > > > produces 16 bit memory access under WinCE.
> > > > > > >
> > > > > > >