Fujitsu Fujitsu SPARC64 V manual Table P-2Action Upon Detection of an Error 2, Ideal specification

Models: Fujitsu SPARC64 V

1 255
Download 255 pages 53.5 Kb
Page 166
Image 166

TABLE P-2Action Upon Detection of an Error (2 of 4)

 

 

Error State Transition

 

 

 

Fatal Error (FE)

Error (EE)

Urgent Error (UGE)

Restrainable Error (RE)

 

 

 

 

 

 

 

 

 

 

Action upon the

1. CPU enters

1. CPU enters

Detection of I_UGE

Ideal specification

error detection

CPU fatal

error_state.

When

1. The error detection is kept

 

state.

2. Watchdog reset

ASI_ECR.UGE_HANDLER = 0,

pending in one bit of

 

2. CPU informs

(WDR) is caused

a single-ADEtrap is caused.

ASI_AFSR.

 

the system of

on the CPU.

Otherwise, when

2. When the trap condition

 

fatal error

 

ASI_ECR.UGE_HANDLER = 1,

for the pending error

 

occurrence.

 

a multiple-ADEtrap is caused.

detection is enabled, the

 

3. The FATAL

 

Detection of A_UGE

ECC_error exception is

 

reset (which is

 

generated.

 

 

When the trap is enabled, a

 

a form of POR

 

Deviation in SPARC64 V

 

 

single-ADEtrap is caused.

 

reset) is issued

 

An ECC_error trap can occur

 

 

When the trap is disabled, the

 

to the whole

 

 

 

trap condition is kept pending

even though ASI_AFSR

 

system.

 

 

 

in hardware.

does not indicate any

 

4. POR reset is

 

 

 

detected error(s)

 

 

Detection of IAE

 

caused to all

 

 

 

corresponding to any trap-

 

CPUs in the

 

When

 

 

enable bit (RTE_UE or

 

system.

 

ASI_ECR.UGE_HANDLER = 0,

 

 

RTE_CEDG) set to 1 in

 

 

 

an IAE trap is caused. Other-

ASI_ECR, in the following

 

 

 

wise, a multiple-ADEtrap is

 

 

 

cases:

 

 

 

caused.

 

 

 

1. A pending detected error

 

 

 

Detection of DAE

 

 

 

is erased from ASI_ASFR

 

 

 

When

(by writing 1 to

 

 

 

ASI_ECR.UGE_HANDLER = 0,

ASI_AFSR) after the error

 

 

 

a DAE trap is caused. Other-

is detected but before the

 

 

 

wise, a multiple-ADEtrap is

ECC_error trap is

 

 

 

caused.

generated.

 

 

 

 

2. A pending CE or DG is

 

 

 

 

erased by writing 1 to

 

 

 

 

ASI_AFSR after the

 

 

 

 

ECC_error trap is caused

 

 

 

 

by the UE error detection.

 

 

 

 

3. A pending UE is erased by

 

 

 

 

writing 1 to ASI_AFSR

 

 

 

 

after the ECC_error trap is

 

 

 

 

caused by CE or DG

 

 

 

 

detection.

 

 

 

 

Privileged software should

 

 

 

 

ignore an ECC_error trap

 

 

 

 

when the AFSR contains no

 

 

 

 

errors corresponding to

 

 

 

 

those enabled in ASI_ECR

 

 

 

 

to cause a trap.

 

 

 

 

 

Priority of

1 — CPU fatal

2 — error_state

3 — ADE trap

6 — ECC_error trap

action when

state

 

4 — DAE trap

 

multiple types

 

 

5 — IAE trap

 

of errors are

 

 

 

 

 

 

 

simultaneously

 

 

 

 

detected

 

 

 

 

 

 

 

 

 

Release 1.0, 1 July 2002

F. Chapter P Error Handling 155

Page 166
Image 166
Fujitsu Fujitsu SPARC64 V manual Table P-2Action Upon Detection of an Error 2, Ideal specification, Deviation in SPARC64