In specific cases, header files will be included incorrectly
during the configuration phase, leading to erroneous failure
of affected configure step.
This patch fixes this issue.
Signed-off-by: Krzysztof Majzerowicz-Jaszcz <krzysztof.majzerowicz-jaszcz@intel.com>
ALRU does not have a configurable sleep time between active awakenings
of the cleaning thread. Once it starts cleaning, it does not back off
unless staleness time and/or activity threshold constraints force it
to do so. Wake up time parameter only controls sleep time after
an idle wake up.
Updating casadm help and man page accordingly.
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
ALRU does not have a configurable sleep time between active awakenings
of the cleaning thread. Once it starts cleaning, it does not back off
unless staleness time and/or activity threshold constraints force it
to do so.
Changing test_cleaning_policy_config assumptions so that it only checks
whether minimum expected amount of data is cleaned rather than assume
exact amount of data to be cleaned.
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
There are multiple issues with this test, but regardless
of them it seems that it duplicates existing trim tests.
The intention of the test is to verify that there is no
eviction when inserting to cache after a trim. This
is equivalent to checking that a trim reduced cache
occupancy, and this is done in other trim test
(test_trim_device_discard_support).
Signed-off-by: Rutkowski, Adam J <adam.j.rutkowski@intel.com>
This check will always return 0, as none of kernel versions supported
by Open CAS sets a barrier flag in for bio. Those flags are exclusive to
struct request, which is no longer used by Open CAS.
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
Kmemleak is a tool which allows to detect memory leaks in kernel. It requires
`CONFIG_DEBUG_KMEMLEAK` to be enabled during kernel compilation. Otherwise all
the kmemleak_* functions are NOPs.
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
This test has multiple assumptions that do not hold for WB/WO:
1. it assumes primary storage errors would propagate to user I/O
2. cache is stopped with metadata flush, which would fail
3. after stopping CAS core content is verified with fio
4. cache is so small that it overfills, resulting in PT I/O further
complicating error accounting.
Signed-off-by: Rutkowski, Adam J <adam.j.rutkowski@intel.com>
Change dd flag specification so that "direct" is specified for
the tested device (CAS exp object or core), regardless of I/O
direction (read vs write).
Signed-off-by: Rutkowski, Adam J <adam.j.rutkowski@intel.com>
