drm/xe: Use fault injection infrastructure to find issues at probe time
authorFrancois Dugast <francois.dugast@intel.com>
Fri, 27 Sep 2024 15:12:06 +0000 (17:12 +0200)
committerRodrigo Vivi <rodrigo.vivi@intel.com>
Thu, 3 Oct 2024 12:58:26 +0000 (08:58 -0400)
commit91b2c42c214f570efaff80a666e30b8f6ce4f12b
tree2178eccc2db6b31f8d9ecaf3848d943c61a2add3
parent11bfc4a2cfeaa012113d9b64fc30a5e6e742fc19
drm/xe: Use fault injection infrastructure to find issues at probe time

The kernel fault injection infrastructure is used to test proper error
handling during probe. The return code of the functions using
ALLOW_ERROR_INJECTION() can be conditionnally modified at runtime by
tuning some debugfs entries. This requires CONFIG_FUNCTION_ERROR_INJECTION
(among others).

One way to use fault injection at probe time by making each of those
functions fail one at a time is:

    FAILTYPE=fail_function
    DEVICE="0000:00:08.0" # depends on the system
    ERRNO=-12 # -ENOMEM, can depend on the function

    echo N > /sys/kernel/debug/$FAILTYPE/task-filter
    echo 100 > /sys/kernel/debug/$FAILTYPE/probability
    echo 0 > /sys/kernel/debug/$FAILTYPE/interval
    echo -1 > /sys/kernel/debug/$FAILTYPE/times
    echo 0 > /sys/kernel/debug/$FAILTYPE/space
    echo 1 > /sys/kernel/debug/$FAILTYPE/verbose

    modprobe xe
    echo $DEVICE > /sys/bus/pci/drivers/xe/unbind

    grep -oP "^.* \[xe\]" /sys/kernel/debug/$FAILTYPE/injectable | \
    cut -d ' ' -f 1 | while read -r FUNCTION ; do
        echo "Injecting fault in $FUNCTION"
        echo "" > /sys/kernel/debug/$FAILTYPE/inject
        echo $FUNCTION > /sys/kernel/debug/$FAILTYPE/inject
        printf %#x $ERRNO > /sys/kernel/debug/$FAILTYPE/$FUNCTION/retval
        echo $DEVICE > /sys/bus/pci/drivers/xe/bind
    done

    rmmod xe

It will also be integrated into IGT for systematic execution by CI.

v2: Wrappers are not needed in the cases covered by this patch, so
    remove them and use ALLOW_ERROR_INJECTION() directly.

v3: Document the use of fault injection at probe time in xe_pci_probe
    and refer to it where ALLOW_ERROR_INJECTION() is used.

Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240927151207.399354-1-francois.dugast@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
13 files changed:
drivers/gpu/drm/xe/xe_device.c
drivers/gpu/drm/xe/xe_ggtt.c
drivers/gpu/drm/xe/xe_guc_ads.c
drivers/gpu/drm/xe/xe_guc_ct.c
drivers/gpu/drm/xe/xe_guc_log.c
drivers/gpu/drm/xe/xe_guc_relay.c
drivers/gpu/drm/xe/xe_pci.c
drivers/gpu/drm/xe/xe_pm.c
drivers/gpu/drm/xe/xe_sriov.c
drivers/gpu/drm/xe/xe_tile.c
drivers/gpu/drm/xe/xe_uc_fw.c
drivers/gpu/drm/xe/xe_wa.c
drivers/gpu/drm/xe/xe_wopcm.c