ERR<n>FR Error Record <n> Feature Register 0 65534 RAS 0x000 + (64 * n) ERR<n>FR RO Defines whether error record <n> is the first record owned by a node: If error record <n> is the first error record owned by a node, then ERR<n>FR.ED is not 0b00. If error record <n> is not the first error record owned by a node, then ERR<n>FR.ED is 0b00. If error record <n> is the first record owned by the node, defines which of the common architecturally-defined features are implemented by the node and, of the implemented features, which are software programmable. RAS ERR<n>FR is a 64-bit register. When error record n is not implemented or error record n is not the first error record in the node 63 56 63:56 Reserved, RES0. NCE 55 55 0 No countable errors. Describes whether this error record supports recording countable errors. When ERRFR[FirstRecordOfNode(n)].CEC != 0b000, at least one error record owned by the node records countable errors. 0b0 Records countable errors. 0b1 Does not record countable errors. RO When ERR<n>FR.FRX == '1' and ERRFR[FirstRecordOfNode(n)].CEC != '000' 55 55 55 Reserved, RES0. Otherwise CE 54 53 1:0 Corrected Error recording. Describes the types of Corrected errors the error record can record, if any. 0b00 Does not record Corrected errors. 0b01 Records only transient or persistent Corrected errors. That is, Corrected errors recorded by setting ERR<n>STATUS.CE to either 0b01 or 0b11. 0b10 Records only non-specific Corrected errors. That is, Corrected errors recorded by setting ERR<n>STATUS.CE to 0b10. 0b11 Records all types of Corrected error. RO When ERR<n>FR.FRX == '1' 54 53 54:53 Reserved, RES0. Otherwise DE 52 52 0 Deferred Error recording. Describes whether the error record supports recording Deferred errors. 0b0 Does not record Deferred errors. 0b1 Records Deferred errors. RO When ERR<n>FR.FRX == '1' 52 52 52 Reserved, RES0. Otherwise UEO 51 51 0 Latent or Restartable Error recording. Describes whether the error record supports recording Latent or Restartable errors. 0b0 Does not record Latent or Restartable errors. 0b1 Records Latent or Restartable errors. RO When ERR<n>FR.FRX == '1' 51 51 51 Reserved, RES0. Otherwise UER 50 50 0 Signaled or Recoverable Error recording. Describes whether the error record supports recording Signaled or Recoverable errors. 0b0 Does not record Signaled or Recoverable errors. 0b1 Records Signaled or Recoverable errors. RO When ERR<n>FR.FRX == '1' 50 50 50 Reserved, RES0. Otherwise UEU 49 49 0 Unrecoverable Error recording. Describes whether the error record supports recording Unrecoverable errors. 0b0 Does not record Unrecoverable errors. 0b1 Records Unrecoverable errors. RO When ERR<n>FR.FRX == '1' 49 49 49 Reserved, RES0. Otherwise UC 48 48 0 Uncontainable Error recording. Describes whether the error record supports recording Uncontainable errors. 0b0 Does not record Uncontainable errors. 0b1 Records Uncontainable errors. RO When ERR<n>FR.FRX == '1' 48 48 48 Reserved, RES0. Otherwise 47 32 47:32 Reserved, RES0. FRX 31 31 0 Feature Register extension. Defines whether ERR<n>FR[63:48] describe architecturally-defined properties of this error record, including the supported error types. If ERR<n>FR.FRX is 0, error record <n> is implemented, and ERRFR[FirstRecordOfNode(n)].FRX is 1, then ERRFR[FirstRecordOfNode(n)][63:48] describe the architecturally-defined properties of all error records owned by the node. 0b0 ERR<n>FR[63:48] are RES0. 0b1 ERR<n>FR[63:48] are defined by the architecture. RO When error record n is implemented, RAS System Architecture v2 is implemented, and ERR<n>FR.ERT == '00' 31 31 31 Reserved, RES0. Otherwise 30 4 30:4 Reserved, RES0. ERT 3 2 1:0 Error Record Type. Defines the type of error record. All other values are reserved. 0b00 Error record <n> not implemented or is a normal record that is not the first error record of the node. 0b01 Error record <n> is a continuation record of the previous error record, <n-1>. RO When RAS System Architecture v2 is implemented 3 2 3:2 Reserved, RES0. Otherwise ED 1 0 1:0 Error reporting and logging. Indicates error record <n> is not the first error record owned the node. 0b00 Error record <n> is not implemented or is not the first error record owned by the node. RO When error record n is the first error record in the node 63 56 7:0 Reserved, RES0. When ERR<n>FR.FRX == '1' IMPLEMENTATION DEFINED 63 56 7:0 Reserved for identifying IMPLEMENTATION DEFINED controls. Otherwise NCE 55 55 0 No countable errors. Describes whether this error record supports recording countable errors. When ERR<n>FR.CEC != 0b000, at least one error record owned by the node records countable errors. 0b0 Records countable errors. 0b1 Does not record countable errors. RO When RAS System Architecture v2 is implemented, ERR<n>FR.FRX == '1', and ERR<n>FR.CEC != '000' 55 55 0 Reserved, RES0. When ERR<n>FR.FRX == '1' NCE 55 55 0 Reserved for identifying IMPLEMENTATION DEFINED controls. Otherwise CE 54 53 1:0 Corrected Error recording. Describes the types of Corrected errors the node can record, if any. 0b00 Does not record Corrected errors. 0b01 Records only transient or persistent Corrected errors. That is, Corrected errors recorded by setting ERR<n>STATUS.CE to either 0b01 or 0b11. 0b10 Records only non-specific Corrected errors. That is, Corrected errors recorded by setting ERR<n>STATUS.CE to 0b10. 0b11 Records all types of Corrected error. RO When ERR<n>FR.FRX == '1' CE 54 53 1:0 Reserved for identifying IMPLEMENTATION DEFINED controls. Otherwise DE 52 52 0 Deferred Error recording. Describes whether the node supports recording Deferred errors. 0b0 Does not record Deferred errors. 0b1 Records Deferred errors. RO When ERR<n>FR.FRX == '1' DE 52 52 0 Reserved for identifying IMPLEMENTATION DEFINED controls. Otherwise UEO 51 51 0 Latent or Restartable Error recording. Describes whether the node supports recording Latent or Restartable errors. 0b0 Does not record Latent or Restartable errors. 0b1 Records Latent or Restartable errors. RO When ERR<n>FR.FRX == '1' UEO 51 51 0 Reserved for identifying IMPLEMENTATION DEFINED controls. Otherwise UER 50 50 0 Signaled or Recoverable Error recording. Describes whether the node supports recording Signaled or Recoverable errors. 0b0 Does not record Signaled or Recoverable errors. 0b1 Records Signaled or Recoverable errors. RO When ERR<n>FR.FRX == '1' UER 50 50 0 Reserved for identifying IMPLEMENTATION DEFINED controls. Otherwise UEU 49 49 0 Unrecoverable Error recording. Describes whether the node supports recording Unrecoverable errors. 0b0 Does not record Unrecoverable errors. 0b1 Records Unrecoverable errors. RO When ERR<n>FR.FRX == '1' UEU 49 49 0 Reserved for identifying IMPLEMENTATION DEFINED controls. Otherwise UC 48 48 0 Uncontainable Error recording. Describes whether the node supports recording Uncontainable errors. 0b0 Does not record Uncontainable errors. 0b1 Records Uncontainable errors. RO When ERR<n>FR.FRX == '1' UC 48 48 0 Reserved for identifying IMPLEMENTATION DEFINED controls. Otherwise IMPLEMENTATION DEFINED 47 32 47:32 Reserved for identifying IMPLEMENTATION DEFINED controls. FRX 31 31 0 Feature Register extension. Defines whether ERR<n>FR[63:48] describe architecturally-defined properties of this error record or node, including the supported error types, or describe IMPLEMENTATION DEFINED properties. When ERR<n>FR.FRX is 1: If RAS System Architecture v2 is implemented and ERR<m>FR.FRX is 1 for other error records <m> owned by the same node, then ERR<n>FR[63:48] describe the architecturally-defined properties of error record <n> only, and ERR<m>FR[63:48] describe the properties for error record <m>. Otherwise, ERR<n>FR[63:48] describe the architecturally-defined properties of all error records owned by the node. 0b0 ERR<n>FR[63:48] are IMPLEMENTATION DEFINED. 0b1 ERR<n>FR[63:48] are defined by the architecture. RO When RAS System Architecture v1p1 is implemented 31 31 31 Reserved, RES0. Otherwise CED 30 30 0 Error counter disable. Indicates whether the node implements a control to disable any implemented Corrected error counters. 0b0 Error counter disable control is not implemented and the error counter(s) are always enabled. ERR<n>CTLR.CED is RES0. 0b1 Enabling and disabling of error counter(s) is supported and controlled by ERR<n>CTLR.CED. RO When RAS System Architecture v1p1 is implemented and ERR<n>FR.CEC != '000' 30 30 30 Reserved, RES0. Otherwise SRV 29 29 0 Status Reset Value. Indicates how node <n> and each error record <m> owned by node <n> is reset. All other values are reserved. 0b0 Node <n> and each error record <m> owned by node <n> are reset as follows: ERR<m>STATUS.{AV, V, MV} are set to {0, 0, 0} on a Cold reset and preserved on Error Recovery reset. ERR<n>CTLR.ED is set to an IMPLEMENTATION DEFINED value on a Cold reset and preserved on Error Recovery reset. 0b1 Node <n> and each error record <m> owned by node <n> are reset as follows: ERR<m>STATUS.{AV, V, MV} are set to architecturally UNKNOWN values on a Cold reset and preserved on Error Recovery reset. ERR<n>CTLR.ED is set to 0 on both Cold reset and Error Recovery reset. RO When RAS System Architecture v2 is implemented 29 29 29 Reserved, RES0. Otherwise RV 28 28 0 Reset Valid. Indicates whether each error record <m> implemented by the node includes the Reset Valid flags, ERR<m>STATUS.{RV, RV2}. All other values are reserved. 0b0 ERR<m>STATUS.{RV, RV2} are RES0. 0b1 ERR<m>STATUS.{RV, RV2} are R/W1C bits. See ERR<m>STATUS.{RV, RV2} for more information. RO When RAS System Architecture v1p1 is implemented 28 28 28 Reserved, RES0. Otherwise DFI 27 26 1:0 Fault handling interrupt for deferred errors control. Indicates whether the enabling and disabling of fault handling interrupts on deferred errors is supported by the node. All other values are reserved. 0b00 Does not support the enabling and disabling of fault handling interrupts on deferred errors. ERR<n>CTLR.DFI is RES0. 0b10 Enabling and disabling of fault handling interrupts on deferred errors is supported and controllable using ERR<n>CTLR.DFI. 0b11 Enabling and disabling of fault handling interrupts on deferred errors is supported, and controllable using ERR<n>CTLR.WDFI for writes and ERR<n>CTLR.RDFI for reads. RO When RAS System Architecture v1p1 is implemented and !(ERR<n>FR.FI IN {'0x'}) 27 26 27:26 Reserved, RES0. Otherwise TS 25 24 25:24 Timestamp Extension. Indicates whether, for each error record <m> owned by this node, ERR<m>MISC3 is used as the timestamp register, and, if it is, the timebase used by the timestamp. All other values are reserved. 0b00 Does not support a timestamp register. 0b01 Implements a timestamp register in ERR<n>MISC3 for each error record <m> owned by the node. The timestamp uses the same timebase as the system Generic Timer. For an error record that has an affinity to a PE, this is the same timer that is visible through CNTPCT_EL0 at the highest Exception level on that PE. 0b10 Implements a timestamp register in ERR<m>MISC3 for each error record <m> owned by the node. The timestamp uses an IMPLEMENTATION DEFINED timebase. RO CI 23 22 23:22 Critical error interrupt. Indicates whether the critical error interrupt and associated controls are implemented by the node. All other values are reserved. 0b00 Does not support the critical error interrupt. ERR<n>CTLR.CI is RES0. 0b01 Critical error interrupt is supported and always enabled. ERR<n>CTLR.CI is RES0. 0b10 Critical error interrupt is supported and controllable using ERR<n>CTLR.CI. RO INJ 21 20 21:20 Fault Injection Extension. Indicates whether the Common Fault Injection Model Extension is implemented by the node. All other values are reserved. 0b00 Does not support the Common Fault Injection Model Extension. 0b01 Supports the Common Fault Injection Model Extension. See ERR<n>PFGF for more information. RO CEO 19 18 1:0 Corrected Error overwrite. Indicates the behavior of the node when a second or subsequent Corrected error is recorded and a first Corrected error has previously been recorded by an error record <m> owned by the node. All other values are reserved. The second or subsequent Corrected error is counted by the Corrected error counter, regardless of the value of this field. If counting the error causes unsigned overflow of the counter, then ERR<m>STATUS.OF is set to 1. This means that, if no other error is subsequently recorded that overwrites the syndrome: If ERR<n>FR.CEO is 0b00, the error record holds the syndrome for the first recorded Corrected error. If ERR<n>FR.CEO is 0b01, the error record holds the syndrome for the most recently recorded Corrected error before the counter overflows. 0b00 Keeps the previous error syndrome. 0b01 If ERR<m>STATUS.OF is 1 before the Corrected error is counted, then the error record keeps the previous syndrome. Otherwise the previous syndrome is overwritten. RO When ERR<n>FR.CEC != '000' 19 18 19:18 Reserved, RES0. Otherwise DUI 17 16 1:0 Error recovery interrupt for deferred errors control. Indicates whether the enabling and disabling of error recovery interrupts on deferred errors is supported by the node. All other values are reserved. 0b00 Does not support the enabling and disabling of error recovery interrupts on deferred errors. ERR<n>CTLR.DUI is RES0. 0b10 Enabling and disabling of error recovery interrupts on deferred errors is supported and controllable using ERR<n>CTLR.DUI. 0b11 Enabling and disabling of error recovery interrupts on deferred errors is supported, and controllable using ERR<n>CTLR.WDUI for writes and ERR<n>CTLR.RDUI for reads. RO When ERR<n>FR.UI != '00' 17 16 17:16 Reserved, RES0. Otherwise RP 15 15 0 Repeat counter. Indicates whether the node implements a second Corrected error counter in ERR<m>MISC0 for each error record <m> owned by the node that can record countable errors. 0b0 Implements a single Corrected error counter in ERR<m>MISC0 for each error record <m> owned by the node that can record countable errors. 0b1 Implements a first (repeat) counter and a second (other) counter in ERR<m>MISC0 for each error record <m> owned by the node that can record countable errors. The repeat counter is the same size as the primary error counter. RO When ERR<n>FR.CEC != '000' 15 15 15 Reserved, RES0. Otherwise CEC 14 12 14:12 Corrected Error Counter. Indicates whether the node implements the standard format Corrected error counter mechanisms in ERR<m>MISC0 for each error record <m> owned by the node that can record countable errors. All other values are reserved. Implementations might include other error counter models, or might include the standard format model and not indicate this in ERR<n>FR. 0b000 Does not implement the standard format Corrected error counter model. 0b010 Implements an 8-bit Corrected error counter in ERR<m>MISC0[39:32] for each error record <m> owned by the node that can record countable errors. 0b100 Implements a 16-bit Corrected error counter in ERR<m>MISC0[47:32] for each error record <m> owned by the node that can record countable errors. RO CFI 11 10 1:0 Fault handling interrupt for corrected errors control. Indicates whether the enabling and disabling of fault handling interrupts on corrected errors is supported by the node. All other values are reserved. 0b00 Does not support the enabling and disabling of fault handling interrupts on corrected errors. ERR<n>CTLR.CFI is RES0. 0b10 Enabling and disabling of fault handling interrupts on corrected errors is supported and controllable using ERR<n>CTLR.CFI. 0b11 Enabling and disabling of fault handling interrupts on corrected errors is supported, and controllable using ERR<n>CTLR.WCFI for writes and ERR<n>CTLR.RCFI for reads. RO When !(ERR<n>FR.FI IN {'0x'}) 11 10 11:10 Reserved, RES0. Otherwise UE 9 8 9:8 In-band error response (External abort). Indicates whether the in-band error response and associated controls are implemented by the node. It is IMPLEMENTATION DEFINED whether an uncorrected error that is deferred and recorded as Deferred error, but is not deferred to the Requester, will signal an in-band error response to the Requester. 0b00 Does not support the in-band error response. ERR<n>CTLR.UE is RES0. 0b01 In-band error response is supported and always enabled. ERR<n>CTLR.UE is RES0. 0b10 In-band error response is supported and controllable using ERR<n>CTLR.UE. 0b11 In-band error response is supported, and controllable using ERR<n>CTLR.WUE for writes and ERR<n>CTLR.RUE for reads. RO FI 7 6 7:6 Fault handling interrupt. Indicates whether the fault handling interrupt and associated controls are implemented by the node. 0b00 Does not support the fault handling interrupt. ERR<n>CTLR.FI is RES0. 0b01 Fault handling interrupt is supported and always enabled. ERR<n>CTLR.FI is RES0. 0b10 Fault handling interrupt is supported and controllable using ERR<n>CTLR.FI. 0b11 Fault handling interrupt is supported, and controllable using ERR<n>CTLR.WFI for writes and ERR<n>CTLR.RFI for reads. RO UI 5 4 5:4 Error recovery interrupt for uncorrected errors. Indicates whether the error handling interrupt and associated controls are implemented by the node. 0b00 Does not support the error handling interrupt. ERR<n>CTLR.UI is RES0. 0b01 Error handling interrupt is supported and always enabled. ERR<n>CTLR.UI is RES0. 0b10 Error handling interrupt is supported and controllable using ERR<n>CTLR.UI. 0b11 Error handling interrupt is supported, and controllable using ERR<n>CTLR.WUI for writes and ERR<n>CTLR.RUI for reads. RO IMPLEMENTATION DEFINED 3 2 3:2 IMPLEMENTATION DEFINED. ED 1 0 1:0 Error reporting and logging. Indicates error record <n> is a normal record and the first record owned the node, and whether the node implements the controls for enabling and disabling error reporting and logging. All other values are reserved. 0b01 Error reporting and logging always enabled. ERR<n>CTLR.ED is RES0. 0b10 Error reporting and logging is controllable using ERR<n>CTLR.ED. RO When RAS System Architecture v2 is implemented and error record <n> is a proxy for a RAS agent 63 4 63:4 Reserved, RES0. ERT 3 2 3:2 Error Record Type. Defines the type of error record. All other values are reserved. 0b01 Error record is a proxy for a RAS agent. RO ED 1 0 1:0 Error reporting and logging. Indicates error record <n> is not a true error record. 0b11 Error record <n> is not an error record. RO When error record n is not implemented or error record n is not the first error record in the node When error record n is the first error record in the node When RAS System Architecture v2 is implemented and error record <n> is a proxy for a RAS agent 2026-03-26 20:27:25 2026-03_rel