Forum Discussion

WFitt's avatar
WFitt
Icon for New Contributor rankNew Contributor
3 years ago

eSPI Agent Core, bug in WAIT_STATE response

Hello, We are using Intel eSPI Agent Core in Intel FPGA MAX 10. Now we possibly found a bug in the Intel eSPI Agent: The bits 15:12 ("Maximum WAIT STATE Allowed") of the General Capabilites and...
WFitt's avatar
WFitt
Icon for New Contributor rankNew Contributor
3 years ago

Hi,

we are using an own mainboard with an Intel Elkhart Lake CPU module. The Elkhart Lake CPU is communicating via eSPI with an Intel FPGA on our mainboard. At the moment we use eSPI with 20 MHz eSpi clock and Single I/O mode. The IP clock is 130 MHz.

During CPU BIOS booting we can see many transactions between Host (CPU) and Agent (FPGA):

21 00 08 10 GET CONFIGURATION
21 00 20 C8 GET CONFIGURATION
22 00 20 01 00 07 00 01 SET CONFIGURATION
21 00 20 C8 GET CONFIGURATION
05 1B GET_VWIRE
21 00 30 B8 GET CONFIGURATION
22 00 30 00 01 00 00 25 SET CONFIGURATION
22 00 08 03 00 20 00 95 SET CONFIGURATION
22 00 20 01 00 07 00 01 SET CONFIGURATION
22 00 30 00 01 00 00 25 SET CONFIGURATION
21 00 40 EF GET CONFIGURATION
22 00 40 0C 11 00 00 F1 SET CONFIGURATION
04 03 02 70 03 30 41 BA 42 30 11 PUT_VWIRE
04 01 43 F0 44 F0 DA PUT_VWIRE
04 03 02 70 03 30 41 BB 42 30 7A PUT_VWIRE
25 FB GET_STATUS
05 1B GET_VWIRE
04 03 02 70 03 30 41 B9 42 30 AC PUT_VWIRE
04 03 02 74 03 30 41 B9 42 30 D9 PUT_VWIRE
04 03 02 76 03 30 41 B9 42 30 60 PUT_VWIRE
04 03 02 77 03 30 41 B9 42 30 BF PUT_VWIRE
04 03 02 77 03 30 41 B9 42 32 B1 PUT_VWIRE
04 03 02 77 03 31 41 B9 42 32 D3 PUT_VWIRE
21 00 10 58 GET CONFIGURATION
04 03 02 77 03 33 41 B9 42 32 17 PUT_VWIRE
21 00 10 58 GET CONFIGURATION
22 00 10 01 11 00 00 F5 SET CONFIGURATION
45 00 80 15 9B F0 IOWR
160 other Port 80h I/O writes
21 00 10 58 GET CONFIGURATION
22 00 10 17 11 00 00 E6 SET CONFIGURATION
21 00 10 58 GET CONFIGURATION
4F FE 0B 00 00 78 56 34 12 FF PUT_MEMWR32_SHORT

The Agent is responding always with ACCEPT. But always before the ACCEPT response 16 WAIT_STATE are sent by Agent. We detected this issue with oscilloscope and with Intel Quartus SignalTap.

Regards

Wolfram