A check is made in the logout area for validity of the CSW fields and
bits are set in the channel check record's ECSW field to indicate bad
fields.
The channel logout is then checked and sequence codes are set based
on the presence of a channel control check, or an interface control
check. If a channel control check is present, the codes set are
determined through parity. The count determines if parity is good and
sets a resultant condition code.
The logout area is examined to ensure that the unit address has valid
parity and is the same address passed by DftKIOS. If so, the
unit-address-valid bit in the ECSW is set. If the unit address is not
valid, the unit-address-valid bit is reset to indicate the invalid
condition.
The ECSW field in the channel check record is moved to the IOERBLOK, if one exists.
After completing the ECSW the 2680 routine moves the 2860 I/O extended logout into the channel check record, set the I/O extended
logout area to ones, and returns to the channel control subroutine.
If the channel failed to log out completely, at least part of the logout
area is all ones. If a fullword of ones is found, a CP termination condition exists. A check is made in the logout area for CSW and bits are
set in the channel check record's ECSW field to indicate bad fields.
The termination and sequence codes are set depending on the presence
of an interface control check or channel control check. If a channel
control check is present, the codes set are determined through parity,
count, and/or data transfer checks. For the 2870, parity can be
determined directly from the channel logout.
The logout area is also examined to ensure valid parity in the unit
address and to ensure that the address is the same as that passed to DftKCCH by DMKIOS. If so, the unit-address-valid bit in the ECSW is set.
The third word of the logout area is also analyzed for type II
errors. If a type II error is found, a CP termination condition exists.
The ECSW field in the channel check record is moved to the IOERBLOK, if one exists.
Before returning to the channel control subroutine, the 2870 routine
.oves the 2870 I/O extended logout into the channel check record and
sets the I/O extended logout area to ones.
This routine analyzes 9 words of the 28-word logout. 1-160 IBM VM/370 System Logic and Problem Determination--Vo1ume 1
The 2880 analysis routine handles channel data checks, interface
control checks, and channel control checks.
Termination code 3 (system reset) is not set in the ECSW because the 2880 channel does not issue system reset to the devices. Retry codes of
o to 5 are possible •. Note: There are several catastrophic conditions under
termination flag can be set, in the 2880 analysis routine. • The channel did not complete the logout. • The CSW is not reliable.
which the CP
They are: • The unit address in the I/O interruption device address field is not correct. Only a channel check record is needed if the channel has recognized
an internal error and has recovered from it without any daaage. 10 recovery action is necessary in these cases. If the channel address in the I/O interruption device address field
does not aatch the channel address in the logout, a CP termination
condition exists.
If the channel vas doing a scan and the unit control word had a
parity check a CP termination condition exists. If there was no parity
check, there vas no damage during the scan and only a channel check
record is required.
Depending on the sequence the channel has entered, the termination
and sequence codes are set; command address, unit address, and unit
status validity is determined; and the sequence code is set valid. The ECSW field in the channel check record is moved into the IOERBLOK, if
one exists.
Before returning to the channel control subroutine, the 2880 routine moves the I/O extended logout into the channel check record and sets the I/O extended logout area to ones. ERROR RECORDING INTERFACE FOR VIRTUAL MACHINES The error recording interface provides a means of recording errors
encountered by operating systems running in a virtual aachine under V8/370. If the virtual operating system is V8/370, it aust be the
Release 2.0 version or later. An SVC 76 issued by a virtual machine is
used to signal VM/370 that error recording is required. The SVC interruption handler in DMKPSA examines general registers 0 and 1 to
determine if valid parameters have been passed. If valid paraaeters are
not found, the SVC is reflected back to the virtual machine and no
recording takes place. If valid parameters are passed, a pageable
routine (DMKVER) processes the error record. DMKVER validates the record passed by the virtual aachine. If
invalid conditions are found, no recording takes place. Control is
returned to the SVC interruption routine in D8KPSA to reflect the SVC to
the virtual machine as an SVC interruption. The action taken by the
virtual machine is dependent on the operating system running in the
virtual machine, not V8/370. If the record is valid, it is aodified by
changing virtual information to real. The actual recording is accomplished by using existing aodules in DMKIOE and DMKIOF. CP Introduction 1-161
Previous Page Next Page