EEH report when Network card iperf test

[lili-lilili]

When we do the netcard iperf test(on P10), the EEH error always report.

Here is the register dump from OPAL.
It looks like the PCIe link get some link problems.
"phbRegbErrorStatus" register records a link down error.

I also had problems with parsing the registers.
I only have a pcie spec for P9, bu it has no explanation for "phbRegbErrorStatus bit31".

Any suggestions? @fbarrat

[ 1005.489389250,3] PHB#006f[6:3]: PHB Freeze/Fence detected !
[ 1005.489438608,3] PHB#006f[6:3]: PCI FIR=0000000000000000
[ 1005.489474501,3] PHB#006f[6:3]: PCI FIR WOF=0000000000000000
[ 1005.489510147,3] PHB#006f[6:3]: NEST FIR=0800000000000000
[ 1005.489542000,3] PHB#006f[6:3]: NEST FIR WOF=0800000000000000
[ 1005.489577729,3] PHB#006f[6:3]: ERR RPT0=0010000000000000
[ 1005.489607921,3] PHB#006f[6:3]: ERR RPT1=0000000000000000
[ 1005.489642726,3] PHB#006f[6:3]: AIB ERR=00cc100000000000
[ 1005.490223446,3] PHB#006f[6:3]: brdgCtl = 00000002
[ 1005.490262691,3] PHB#006f[6:3]: deviceStatus = 00000140
[ 1005.490295542,3] PHB#006f[6:3]: slotStatus = 00402000
[ 1005.490335762,3] PHB#006f[6:3]: linkStatus = c1010008
[ 1005.490366070,3] PHB#006f[6:3]: devCmdStatus = 00100107
[ 1005.490397781,3] PHB#006f[6:3]: devSecStatus = 00000000
[ 1005.490428166,3] PHB#006f[6:3]: rootErrorStatus = 00000000
[ 1005.490463922,3] PHB#006f[6:3]: corrErrorStatus = 00000000
[ 1005.490497253,3] PHB#006f[6:3]: uncorrErrorStatus = 00000000
[ 1005.490530530,3] PHB#006f[6:3]: devctl = 00000140
[ 1005.490562582,3] PHB#006f[6:3]: devStat = 00000000
[ 1005.490593271,3] PHB#006f[6:3]: tlpHdr1 = 00000000
[ 1005.490623159,3] PHB#006f[6:3]: tlpHdr2 = 00000000
[ 1005.490654178,3] PHB#006f[6:3]: tlpHdr3 = 00000000
[ 1005.490687329,3] PHB#006f[6:3]: tlpHdr4 = 00000000
[ 1005.490721063,3] PHB#006f[6:3]: sourceId = 00000000
[ 1005.490751859,3] PHB#006f[6:3]: nFir = 0800000000000000
[ 1005.490786389,3] PHB#006f[6:3]: nFirMask = 003001d000000000
[ 1005.490820066,3] PHB#006f[6:3]: nFirWOF = 0800000000000000
[ 1005.490853079,3] PHB#006f[6:3]: phbPlssr = 0000001c00000000
[ 1005.490888389,3] PHB#006f[6:3]: phbCsr = 0000001c00000000
[ 1005.490926839,3] PHB#006f[6:3]: lemFir = 0000000100000100
[ 1005.490964803,3] PHB#006f[6:3]: lemErrorMask = 0000000000000000
[ 1005.491002854,3] PHB#006f[6:3]: lemWOF = 0000000000000100
[ 1005.491036256,3] PHB#006f[6:3]: phbErrorStatus = 000004e000000000
[ 1005.491069266,3] PHB#006f[6:3]: phbFirstErrorStatus = 0000040000000000
[ 1005.491102274,3] PHB#006f[6:3]: phbErrorLog0 = 0000000000000000
[ 1005.491140034,3] PHB#006f[6:3]: phbErrorLog1 = 0000000000000000
[ 1005.491177123,3] PHB#006f[6:3]: phbTxeErrorStatus = 0000000000000000
[ 1005.491213447,3] PHB#006f[6:3]: phbTxeFirstErrorStatus = 0000000000000000
[ 1005.491248051,3] PHB#006f[6:3]: phbTxeErrorLog0 = 0000000000000000
[ 1005.491281223,3] PHB#006f[6:3]: phbTxeErrorLog1 = 0000000000000000
[ 1005.491314903,3] PHB#006f[6:3]: phbRxeArbErrorStatus = 0000000000000000
[ 1005.491349975,3] PHB#006f[6:3]: phbRxeArbFrstErrorStatus = 0000000000000000
[ 1005.491388215,3] PHB#006f[6:3]: phbRxeArbErrorLog0 = 0000000000000000
[ 1005.491425458,3] PHB#006f[6:3]: phbRxeArbErrorLog1 = 0000000000000000
[ 1005.491459174,3] PHB#006f[6:3]: phbRxeMrgErrorStatus = 0000000000000001
[ 1005.491492464,3] PHB#006f[6:3]: phbRxeMrgFrstErrorStatus = 0000000000000001
[ 1005.491524353,3] PHB#006f[6:3]: phbRxeMrgErrorLog0 = 0000000000000000
[ 1005.491557762,3] PHB#006f[6:3]: phbRxeMrgErrorLog1 = 0000000000000000
[ 1005.491594088,3] PHB#006f[6:3]: phbRxeTceErrorStatus = 0000000000000000
[ 1005.491632865,3] PHB#006f[6:3]: phbRxeTceFrstErrorStatus = 0000000000000000
[ 1005.491671010,3] PHB#006f[6:3]: phbRxeTceErrorLog0 = 0000000000000000
[ 1005.491705600,3] PHB#006f[6:3]: phbRxeTceErrorLog1 = 0000000000000000
[ 1005.491738612,3] PHB#006f[6:3]: phbPblErrorStatus = 0100000000000000
[ 1005.491772417,3] PHB#006f[6:3]: phbPblFirstErrorStatus = 0100000000000000
[ 1005.491810821,3] PHB#006f[6:3]: phbPblErrorLog0 = 0000000000000000
[ 1005.491846866,3] PHB#006f[6:3]: phbPblErrorLog1 = 0000000000000000
[ 1005.491884228,3] PHB#006f[6:3]: phbPcieDlpErrorLog1 = 0000000000000000
[ 1005.491921107,3] PHB#006f[6:3]: phbPcieDlpErrorLog2 = 0000000000000000
[ 1005.491954581,3] PHB#006f[6:3]: phbPcieDlpErrorStatus = 0000000000000000
[ 1005.491988376,3] PHB#006f[6:3]: phbRegbErrorStatus = 0090001100000000
[ 1005.492022294,3] PHB#006f[6:3]: phbRegbFirstErrorStatus = 0000001000000000
[ 1005.492058404,3] PHB#006f[6:3]: phbRegbErrorLog0 = 2480005800000000
[ 1005.492094757,3] PHB#006f[6:3]: phbRegbErrorLog1 = 0000000000000000

[fbarrat]

REGB Error Status Register, bit 31 is:

31 pdlo_dl_target_speed_not_reached_err
Error bit to detect HW not training to negotiated speed due to EQ timeouts.

I'm getting mixed messages from some of those errors. I think it needs to be analyzed by the hw team. You must have some official support channel for hw, right?