80 - Bcache Tag Address Parity Error
Indicator:EI_STAT<bc_tperr> = 1
Description:A Bcache read transaction encountered bad parity in the tag address RAM.
Recovery:No Recovery
Analysis:EI_ADDR - contains the physical address associated with this error.
BC_TAG_ADDR - contains the results of the Bcache tag read.
Faulting FRU:MLB, Bcache

82 - Bcache Tag Control Parity Error
Indicator:EI_STAT<bc_tc_perr> = 1
Description:A Bcache read transaction encountered bad parity in the tag control RAM.
Recovery:No Recovery
Analysis:EI_ADDR - contains the physical address associated with this error.
BC_TAG_ADDR - contains the results of the Bcache tag read.
Faulting FRU:MLB, Bcache

86 - CPU-Detected Correctable ECC Error
Indicator: ISR<crd> = 1
(note that EI_STAT<unc_cor_err> is not used)
Description:The CPU detected a correctable ECC error.
Recovery:None required. The operating system may choose to log this error.
Analysis:EI_ADDR - Contains the physical address of the octaword associated with this error.
BC_TAG_ADDR - contains the results of the Bcache tag read:
  • If BC_TAG_ADDR<12> =0, the error is from a memory read
  • If BC_TAG_ADDR<12> =1, the error is from a Bcache read
FILL_SYN - contains the ECC syndrome bits for the two quadwords at this address.
Faulting FRU:MLB, Bcache, DIMM

88 - CPU-Detected Noncorrectable ECC Error
Indicator:EI_STAT<unc_ecc_err> = 1
Description:The CPU detected an uncorrectable ECC error.
Recovery:TBD
Analysis:EI_ADDR - Contains the physical address of the octaword associated with this error.
BC_TAG_ADDR - contains the results of the Bcache tag read; if BC_TAG_ADDR<hit> = 1, the error is from a Bcache read, otherwise it is from a memory read.
FILL_SYN - contains the ECC syndrome bits for the two quadwords at this address.
Faulting FRU:MLB, Bcache, DIMM

8A - Unknown Error
Indicator: 
Description:If none of the errors contain a bit that is set to indicate an error, an unknown error machine check will result.
Recovery: 
Analysis: 
Faulting FRU: 

8E - Bugcheck
Indicator:CALL_PAL BUGCHK
Description:The operating system requested a BUGCHECK.
Recovery:No recovery.
Analysis: 
Faulting FRU:Software.

90 - OS Bugcheck
Indicator: 
Description:This is the result of a CALL_PAL to perform queue manipulation and catastrophe strikes resulting in an operating system bugcheck.
Recovery:No recovery.
Analysis: 
Faulting FRU:Software, PAL code

92 - Dcache Parity Error
Indicator: DC_PERR_STAT<5:2> not = 0
Description:A parity error was received from Dcache.
Recovery:No Recovery
Analysis:VA - contains the effective virtual address associated with the error.
VA_FORM - contains the virtual page table entry address associated with the error.
DC_PERR_STAT - indicates the type of error:
  • if dc_perr_stat<dp0> = 1, bank 0 data parity error
  • if dc_perr_stat<dp1> = 1, bank 1 data parity error
  • if dc_perr_stat<tp0> = 1, bank 0 tag parity error
  • if dc_perr_stat<tp1> = 1, bank 1 tag parity error
MMSTAT - contains more information on this error.
Faulting FRU:MLB

94 - Icache Parity Error
Indicator:ICPERR_STAT<12:11> not = 0
Description:A parity error was received from Icache.
Recovery:No recovery.
Analysis:ICPERR_STAT - indicates the type of error:
  • if ICPERR_STAT<dpe> = 1, data parity error
  • if ICPERR_STAT<tpe> = 1, tag parity error
Faulting FRU:MLB

96 - Istream Retryable Error
Indicator:EI_STAT<fil_ird> = 1
Description:TDB
Recovery:TDB
Analysis:TBD
Faulting FRU:TBD

98 - CPU-Detected Machine Check
Indicator: ICPERR_STAT<tmr> = 1
DC_PERR_STAT<lock> = 1
EI_STAT<bc_tperr> = 1
EI_STAT<bc_tc_perr> = 1
EI_STAT<ei_par_err> = 1
EI_STAT<seo_hrd_err> = 1
EI_STAT<unc_ecc_err> = 1 .and. EI_STAT<fil_ird> = 0
SC_STAT<2:0> != 0 (sc_tperr)
SC_STAT<10:3> != 0 .and. SC_STAT<15:11> != 00001b
SC_STAT<scnd_err> = 1
Description:Machine check trap.
Recovery:No recovery.
Analysis: If ICPERR_STAT<tmr> is set, there was a timeout reset error or a cfail_h/no cack_h error
If DC_PERR_STAT<lock> is set if a parity error was detected in the Dcache. Bits <5:2> are locked against further updates.
If EI_STAT<bc_tperr> is set, there was a parity error reading Bcache tag address RAM. If EI_STAT<bc_tc_perr> is set, there was a parity error reading Bcache tag control RAM. If EI_STAT<ei_par_err> is set, there was a parity error in an address and command received by the CPU.
If EI_STAT<seo_hrd_err> = 1, a fill from Bcache or main memory, or a system address/command received by the CPU has a hard error while one of the hard error bits is already set.
If EI_STAT<unc_ecc_err> = 1 AND EI_STAT<fil_ird> = 0, an uncorrectable ECC error occured during a Dstream fill.
If SC_STAT<2:0> not = 0 (sc_tperr), there was a parity error reading Scache tags.
If SC_STAT<10:3> not = 0 AND SC_STAT<15:11> not = 00001b, there was a parity error resulting from an Scache read, but not during an IREAD command.
If SC_STAT<scnd_err> = 1, an Scache transaction resulted in an error while SC_TPERR or SC_DPERR was already set.
Faulting FRU:MLB, Bcache

201 - PYXIS-Detected Correctable ECC Error
Indicator:PYXIS_ERR<cor_err> = 1
Description:The PYXIS detected a correctable single-bit error on a read from memory.
Recovery:Pyxis returns the correct data. The operating system may choose to log this error.
Analysis: The octaword address of the error, EA, can be determined from various registers:
EA<33:32> = MESR<01:00>
EA<31:06> = MEAR<31:06>
EA<05:04> = (MEAR<05:04> + MESR<17:16>) MOD 4
The quadword in error can be determined, note that both quadwords can incur simultaneous single-bit errors.
PYXIS_SYN<24> set ==> low quadword (EA<03>=0)
PYXIS_SYN<25> set ==> high quadword (EA<03>=1)
PYXIS_SYN<07:00> is the syndrome for the low quadword.
PYXIS_SYN<15:08> is the syndrome for the high quadword.
MESR<24:20> is the DATA_CYCLE_TYPE in progress at the time of the error.
Faulting FRU: MLB, Bcache, DIMM

203 - PYXIS-Detected Noncorrectable ECC Error
Indicator:PYXIS_ERR<un_cor_err> = 1
Description:PYXIS detected an uncorrectable error on a read from memory
Recovery:No recovery
Analysis:The octaword address of the error, EA, can be determined from various registers:
EA<33:32> = MESR<01:00>
EA<31:06> = MEAR<31:06>
EA<05:04> = (MEAR<05:04> + MESR<17:16>) MOD 4

The quadword in error can be determined, note that both quadwords can incur simultaneous uncorrectable errors:
PYXIS_SYN<26> set ==> low quadword (EA<03>=0)
PYXIS_SYN<27> set ==> high quadword (EA<03>=1)
MESR<24:20> is the DATA_CYCLE_TYPE in progress at the time of the error
Faulting FRU:MLB, Bcache, DIMM

205 - PYXIS-Detected Command Bus Parity Error
Indicator:PYXIS_ERR<cpu_pe> = 1
Description:Parity error on the CPU/PYXIS command bus.
Recovery:No recovery
Analysis:PYXIS must not check parity on the CPU command bus, so either firmware failed to disable the check or PYXIS is broken.
Faulting FRU:Firmware, MLB

207 - PYXIS-Detected Nonexistent Memory Reference
Indicator:PYXIS_ERR<mem_nem> = 1
Description:Indicates that a read or write occurred to an invalid address which does not map to any memory bank, CSR or I/O quadrant.
Recovery:No recovery
Analysis:The octaword address of the error, EA, can be determined from various registers:
EA<33:32> = MESR<01:00>
EA<31:06> = MEAR<31:06>
EA<05:04> = (MEAR<05:04> + MESR<17:16>) MOD 4

The type of memory access which caused the failure can be determined from the MESR register:
  • if MESR<8> = 1, PCI DMA read
  • if MESR<9> = 1, PCI DMA write
  • if MESR<10> = 1, CPU memory read
  • if MESR<11> = 1, CPU memory write
  • if MESR<12> = 1, I/O read
  • if MESR<13> = 1, I/O write
  • if MESR<14> = 1, victim eject
  • if MESR<15> = 1, TLB read
Faulting FRU:MLB, software, firmware, PCI option, ISA option

209 - PCI System Error (SERR)
Indicator:PYXIS_ERR<pci_serr> = 1
Description:A PCI SERR (system error) was reported.
Recovery:No recovery
Analysis:SERR is reported by any PCI device when it detects a parity error on the PCI bus during an address cycle. (PYXIS reports an address parity error through PYXIS_ERR<6>.) These errors are hard to trace, but the following registers may have relevant information:
  • PYXIS_STAT
  • PCI_ERR0
  • PCI_ERR1
  • PCI_ERR2
Faulting FRU:MLB, riser, PCI option

20B - PCI Data Parity Error (PERR)
Indicator:PYXIS_ERR<pci_perr> = 1
Description:A PCI parity error occurred in the data phase of an I/O read/write transaction, or a PCI PERR (parity error) was reported.
Recovery:No recovery
Analysis:PERR is reported by any PCI device when it detects a parity error on the PCI bus during a data cycle. PYXIS reports a data parity error which it detects here, too. These errors are hard to trace, but the following registers may have relevant information:
  • PYXIS_STAT
  • PCI_ERR0
  • PCI_ERR1
  • PCI_ERR2
Faulting FRU:PCI option, MLB, riser

20D - PCI Address Parity Error
Indicator:PYXIS_ERR<pci_addr_pe> = 1
Description:PYXIS detected a parity error in the address phase of PCI bus transaction.
Recovery:No recovery
Analysis:PCI_ERR0<3:0> (dma_cmd) holds the C/BE command for this DMA transaction
PCI_ERR0<dac> is set if this DMA transaction is a dual-address cycle
PCI_ERR1 contains the DMA address
PYXIS_STAT might have pertinent data also.
Faulting FRU:PCI option, MLB, riser

20F - PCI Master Abort
Indicator:PYXIS_ERR<rcvd_mas_abt> = 1
Description:PYXIS generated a PCI master abort cycle. Note that PYXIS does not report if a master ends a transaction with a target abort.
Recovery:No recovery
Analysis:TBD
Faulting FRU:PCI option, MLB, riser

211 - PCI Target Abort
Indicator:PYXIS_ERR<rcvd_tar_abt> = 1
Description:A PCI target device ended a transaction using the PCI target abort protocol. Note that Pyxis does not report if it ends a transaction with a target abort.
Recovery:No recovery
Analysis:If PYXIS_STAT<0> = 1, PYXIS PCI master state engine is active and:
  • PCI_ERR0<pci_cmd> holds the C/BE command for this transaction
  • PCI_ERR0<pci_dac> is set if this transaction is a dual-address cycle
  • PCI_ERR2 contains the PCI address
If PYXIS_STAT<1> = 1, PYXIS CI target state engine is active and:
  • PCI_ERR0<dma_cmd> holds the C/BE command for this DMA transaction
  • PCI_ERR0<dma_dac> is set if this DMA transaction is a dual-address cycle
  • PCI_ERR1 contains the DMA address
Read PCI Config Status Register for each PCI device:
  • If STATUS<15> = 1 - device detected parity error
  • If STATUS<14> = 1 - device signalled system error
  • If STATUS<13> = 1 - device received master abort
  • If STATUS<12> = 1 - device received target abort
  • If STATUS<11> = 1 - device signalled target abort
  • If STATUS<8> = 1 - Parity Error checking is enabled plus device asserted PERR or device observed PERR assert by the bus master.
Faulting FRU:PCI option, ISA option

213 - Invalid Scatter/Gather TLB Entry
Indicator:PYXIS_ERR<pa_pte_inv> = 1
Description:Invalid page table entry (PTE) on scatter/gather access. A read error on a TLB fill is reported elsewhere.
Recovery:No recovery
Analysis:PCI_ERR0<dma_cmd> holds the C/BE command for this DMA transaction.
PCI_ERR0<dma_dac> is set if this DMA transaction is a dual-address cycle.
PCI_ERR1 contains the DMA address.
Faulting FRU:MLB, DIMM, PCI option, software

215 - FlashROM Write Error
Indicator:Not used on a-Series
Description: 
Recovery: 
Analysis: 
Faulting FRU: 

217 - PCI I/O Timeout
Indicator:PYXIS_ERR<ioa_timeout> = 1
Description:I/O read/write timed out.
Recovery:No recovery
Analysis:PCI_ERR0<pci_cmd> holds the C/BE command for this transaction.
PCI_ERR0<pci_dac> is set if this transaction is a dual-address cycle.
PCI_ERR2 contains the PCI address.
Faulting FRU:PCI option, ISA option, MLB, riser, software

219 - ISA NMI
Indicator:Not used on a-Series
Description: 
Recovery: 
Analysis: 
Faulting FRU: 

221 - Unexpected Interrupt At IPL23
Indicator: 
Description:The CPU detected an interrupt on a line that is not supposed to be connected in the Personal Workstation a-Series.
Recovery: 
Analysis: 
Faulting FRU:MLB, firmware, software

223 - Fan Failure
Indicator:INT_REQ<0> = 1 (fan fault)
Description:One or both of the system fans are either not properly connected to the riser, or have stopped spinning due to failure or obstruction.
Recovery:No Recovery. The operating system should attempt to log this error to disk, then report the failure to the console firmware which will turn off the power supply to prevent the system from overheating.
Analysis: 
Faulting FRU:Fans, riser, MLB