blob: 07f3ed4e1f208681a93d11bfa2c2b26a1cba4b8e [file] [log] [blame]
Kdump Test Description
----------------------
The test suite consists of two parts
* crasher: This is a basic kdump functionality testing.
* lkdtm : These tests use kprobes / jprobes to put probes in different
function for invoking a dump.
Only one of the above is used for testing at a time. While building the scripts,
select the one you want to run. It is suggested to use the lkdtm tests (which is
the default) for more elaborate testing.
The basic crash points supported by LKDTM are same as supported by LKDTT.
These are as follows :
IRQ Handling with IRQs disabled (INT_HARDWARE_ENTRY)
IRQ Handling with IRQs enabled (INT_HW_IRQ_EN)
Tasklet with IRQs disabled (TASKLET)
Block I/O (FS_DEVRW)
Swap-out (MEM_SWAPOUT)
Timer processing (TIMERADD)
SCSI command (SCSI_DISPATCH_CMD)
IDE command (IDE_CORE_CP)
The lkdtm module consists of following tests
* ALIDP: panic in __do_IRQ
* ALIDB: Bug in __do_IRQ
* ALIDE: Exception in __do_IRQ
* ALIDL: Hang in __do_IRQ
* ALIDO: Overflow in __do_IRQ
The same tests are repeated for Other crash points as mentioned above.
* ALIEP
* ALIEB
* ALIEE
* ALIEL
* ALIEO
* ALTDP
* ALTDB
* ALTDE
* ALTDL
* ALTDO
* ALTEP
* ALTEB
* ALTEE
* ALTEL
* ALTEO
* ALBP
* ALBB
* ALBE
* ALBL
* ALBO
* ALMP
* ALMB
* ALME
* ALML
* ALMO
* ALTP
* ALTB
* ALTE
* ALTL
* ALTO
* ALSP
* ALSB
* ALSE
* ALSL
* ALSO
* ALIP
* ALIB
* ALIE
* ALIL
* ALIO
The crasher module tests kdump in following senarios.
* ACS: invoke dump using sysrq C.
* ACP: invoke dump using panic [ panic test 0 ].
* ACB: invoke dump using BUG [ panic test 1 ].
* ACE: invoke dump using panic_on_oops [ panic test 2 ].
* ACL: invoke dump in a hang situation.
On certain type of hardware [ eg IBM system p and system i machines] tests with
names ending in "B" and "E" might require manual intervention. After running
these tests machine will drop into debugger if one is configured. User will
have to type X to get out of the debugger and continue with the tests. For
tests with name ending in L [loop test] , user will have to invoke a dump using
the Hardware Management Console via soft reset.
While executing some loop testcases after invoking a dump via Hardware
Management Console you might get following message
Sending IPI to other cpus...
done waiting: 1 cpu(s) not responding
Activate soft-reset to stop other cpu(s)
In such case you will have to re-trigger dump via Hardware Management Console
one more time. This is a normal senario and not a bug.