[U-Boot-Users] SM712 + U-Boot
Hello,
Highly possible this is not related to U-Boot but I would like to know if someone has ran into the same problem.
We bought a SM712+ card::LynxEM4+ Rev.BA Eval Board -> DK712GX04PC00 PCI Evaboard... But this card seems problematic. I enabled it in U-Boot and was able to boot normally. There is no error. But display is something strange. The problem is that some pixels are painted, some successor pixels are black, some successor pixels are painted, some successor pixels are black, ... etc. This space seems equal. I tried some resolutions to see what is the problem. It is such that at 640*480*8 res.: 4 pix painted + 4 pix black + 4 pix painted + 4 pix black etc. This problem is in horizantal line. There is no problem in vertical line except same columns are black! I tried to illuminate the black pixels to see what happens and although I could achieve to change painted pixels from U-Boot cmdline, I couldnt manage to change black pixels as expected.
In fact, as we bought the card, I wanted to sure if this card was working or not. So I tried it in Windows. Because It had already got the drivers.(This test doesnt ensure the card working properly under Linux but I think still it was important in hardware point of view) It worked only in WinXP and Vestel monitor combo. It wasnt succesful in some combinations of Winxx and monitor combo.
I think this card is not working properly in the means of hardware point of view. Does anyone come across the same or similar problems?
Hi,
I don't think that your card has a problem, I think it isn't configured properly. The LynxEM4+ Rev. BA has an 32 bits internal memory bus, the Rev. AA has a 64 bits internal memory bus. For the Rev BA, you need to switch ON the SW2:7 configuration switch.
I hope this helps. Best regards
Bora Sahin wrote:
Hello,
Highly possible this is not related to U-Boot but I would like to know if someone has ran into the same problem.
We bought a SM712+ card::LynxEM4+ Rev.BA Eval Board -> DK712GX04PC00 PCI Evaboard... But this card seems problematic. I enabled it in U-Boot and was able to boot normally. There is no error. But display is something strange. The problem is that some pixels are painted, some successor pixels are black, some successor pixels are painted, some successor pixels are black, ... etc. This space seems equal. I tried some resolutions to see what is the problem. It is such that at 640*480*8 res.: 4 pix painted + 4 pix black + 4 pix painted + 4 pix black etc. This problem is in horizantal line. There is no problem in vertical line except same columns are black! I tried to illuminate the black pixels to see what happens and although I could achieve to change painted pixels from U-Boot cmdline, I couldnt manage to change black pixels as expected.
In fact, as we bought the card, I wanted to sure if this card was working or not. So I tried it in Windows. Because It had already got the drivers.(This test doesnt ensure the card working properly under Linux but I think still it was important in hardware point of view) It worked only in WinXP and Vestel monitor combo. It wasnt succesful in some combinations of Winxx and monitor combo.
I think this card is not working properly in the means of hardware point of view. Does anyone come across the same or similar problems?
Hello,
Pierre> I don't think that your card has a problem, I think it isn't configured Pierre> properly. Pierre> The LynxEM4+ Rev. BA has an 32 bits internal memory bus, the Rev. AA has a Pierre> 64 bits internal memory bus. Pierre> For the Rev BA, you need to switch ON the SW2:7 configuration switch.
OK, but SW2 is related to SM722. The info on the backend of the card says SW2 -> SETTING FOR SM722...
Still I tried your suggestion but there is no change as I expected!
AFAIK, SW1 is nothing to do with a non-x86 system. It only boots a x86 system with determined resoluiton. That's it! Then OS changes it to what it wants! Also U-Boot sets its own resoultion thanks to compile time option or videomode env. variable...
Pierre> I hope this helps.
Thanks...
Bora Sahin wrote:
Hello,
Pierre> I don't think that your card has a problem, I think it isn't configured Pierre> properly. Pierre> The LynxEM4+ Rev. BA has an 32 bits internal memory bus, the Rev. AA has a Pierre> 64 bits internal memory bus. Pierre> For the Rev BA, you need to switch ON the SW2:7 configuration switch.
OK, but SW2 is related to SM722. The info on the backend of the card says SW2 -> SETTING FOR SM722...
I don't have such information on my board. I own a DK712GX04PC00-00 400029 board. On the upper right corner of the board, there's two blue dip switch for configuration. The SW1 is used for LCD panel configuration, and SW2 is used for memory configuration. This hardware configuration is read once at boot time and could be overwritten by software. The SW2:7 switch is connected on the MD6 signal of the SMI712 (on my board) which enable the 32 bits memory data bus. See the SMI712 data book at paragraph 3. If your board hasn't any configuration switch on the MD6 signal, you can try to modify the MCR62 register (address 3c5h,62). Best regards
Still I tried your suggestion but there is no change as I expected!
AFAIK, SW1 is nothing to do with a non-x86 system. It only boots a x86 system with determined resoluiton. That's it! Then OS changes it to what it wants! Also U-Boot sets its own resoultion thanks to compile time option or videomode env. variable...
Pierre> I hope this helps.
Thanks...
Hi,
Pierre> I don't have such information on my board. I own a DK712GX04PC00-00 Pierre> 400029 board.
The problem is here. My eva board number is also DK712GX04PC00 but I think this card is changed! I sent an e-mail to HyLine but they didnt help much! Only they mentioned some SW1 switch settings and said "For the SM712 board, SW2 is don't care, so you can put all SW2 to OFF". I deduced "this card can be changed" from their writings.
Pierre> On the upper right corner of the board, there's two blue dip switch for configuration. The Pierre> SW1 is used for LCD panel configuration, and SW2 is used for memory configuration.
Yes. I think they changed the card to host either a SM712 or SM722 chip. So they allocate SW2 to keep the SM722 settings.
Pierre> This hardware configuration is read once at boot time and could be overwritten by software. Pierre> The SW2:7 switch is connected on the MD6 signal of the SMI712 (on my board) which enable the Pierre> 32 bits memory data bus. See the SMI712 data book at paragraph 3. If your board hasn't any Pierre> configuration switch on the MD6 signal, you can try to modify the MCR62 register (address 3c5h,62).
Thanks for the tip. Your point seems reasonable. I examined that part of the doc.(page: 169) The doc says it's power-on configurable. Also this pins are pulled-up. So their default value is 1 -> 64 bit internal bus, disable external 32 bit mem etc...
I'll come back after trying it...
Hi again,
Bora Sahin wrote:
Hi,
Pierre> I don't have such information on my board. I own a DK712GX04PC00-00 Pierre> 400029 board.
The problem is here. My eva board number is also DK712GX04PC00 but I think this card is changed! I sent an e-mail to HyLine but they didnt help much! Only they mentioned some SW1 switch settings and said "For the SM712 board, SW2 is don't care, so you can put all SW2 to OFF". I deduced "this card can be changed" from their writings.
Pierre> On the upper right corner of the board, there's two blue dip switch for configuration. The Pierre> SW1 is used for LCD panel configuration, and SW2 is used for memory configuration.
Yes. I think they changed the card to host either a SM712 or SM722 chip. So they allocate SW2 to keep the SM722 settings.
Pierre> This hardware configuration is read once at boot time and could be overwritten by software. Pierre> The SW2:7 switch is connected on the MD6 signal of the SMI712 (on my board) which enable the Pierre> 32 bits memory data bus. See the SMI712 data book at paragraph 3. If your board hasn't any Pierre> configuration switch on the MD6 signal, you can try to modify the MCR62 register (address 3c5h,62).
Thanks for the tip. Your point seems reasonable. I examined that part of the doc.(page: 169) The doc says it's power-on configurable. Also this pins are pulled-up. So their default value is 1 -> 64 bit internal bus, disable external 32 bit mem etc...
I'll come back after trying it...
It's possible to identify the SMI712 revision in the CSR08 register (Revision ID and Class Code). Rev AA has 0xa0 as revision Id, and Rev BA has 0xb0 as revision Id.
Best regards.
Hello,
Pierre> It's possible to identify the SMI712 revision in the CSR08 register Pierre> (Revision ID and Class Code). Pierre> Rev AA has 0xa0 as revision Id, and Rev BA has 0xb0 as revision Id.
Thanks for the info Pierre. I changed the related field[s] both in U-Boot and kernel frame buffer driver and it worked! In the meanwhile, kernel frame buffer driver seems needing some cleanup!
As I expressed in the first mail, I have a couple of problems with this card. For example, I had flicking problems in W2K. I tie it to the card changed! Therefore, perhaps X driver also need some little changes. What do you think about this subject?
Hi Bora Sahin wrote:
Hello,
Pierre> It's possible to identify the SMI712 revision in the CSR08 register Pierre> (Revision ID and Class Code). Pierre> Rev AA has 0xa0 as revision Id, and Rev BA has 0xb0 as revision Id.
Thanks for the info Pierre. I changed the related field[s] both in U-Boot and kernel frame buffer driver and it worked! In the meanwhile, kernel frame buffer driver seems needing some cleanup!
Good news ! Feel free to submit a patch.
As I expressed in the first mail, I have a couple of problems with this card. For example, I had flicking problems in W2K. I tie it to the card changed! Therefore, perhaps X driver also need some little changes. What do you think about this subject?
No opinion. I don't use W2K nor X driver.
Best regards
participants (2)
-
Bora Sahin -
Pierre AUBERT