tjh.dev
Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
selftests/powerpc: Hoist helper code out of eeh-basic
Hoist some of the useful test environment checking and prep code into
eeh-functions.sh so they can be reused in other tests.
Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20201103044503.917128-1-oohall@gmail.com
authored by
Oliver O'Halloran
and committed by
Michael Ellerman
db82f709
718aae91
+51
-36
+3
-36
tools/testing/selftests/powerpc/eeh/eeh-basic.sh
+3
-36
tools/testing/selftests/powerpc/eeh/eeh-basic.sh
···
1
1
#!/bin/sh
2
2
# SPDX-License-Identifier: GPL-2.0-only
3
3
4
-
KSELFTESTS_SKIP=4
5
-
6
4
. ./eeh-functions.sh
7
5
8
-
if ! eeh_supported ; then
9
-
echo "EEH not supported on this system, skipping"
10
-
exit $KSELFTESTS_SKIP;
11
-
fi
12
-
13
-
if [ ! -e "/sys/kernel/debug/powerpc/eeh_dev_check" ] && \
14
-
[ ! -e "/sys/kernel/debug/powerpc/eeh_dev_break" ] ; then
15
-
echo "debugfs EEH testing files are missing. Is debugfs mounted?"
16
-
exit $KSELFTESTS_SKIP;
17
-
fi
6
+
eeh_test_prep # NB: may exit
18
7
19
8
pre_lspci=`mktemp`
20
9
lspci > $pre_lspci
21
-
22
-
# Bump the max freeze count to something absurd so we don't
23
-
# trip over it while breaking things.
24
-
echo 5000 > /sys/kernel/debug/powerpc/eeh_max_freezes
25
10
26
11
# record the devices that we break in here. Assuming everything
27
12
# goes to plan we should get them back once the recover process
···
15
30
16
31
# Build up a list of candidate devices.
17
32
for dev in `ls -1 /sys/bus/pci/devices/ | grep '\.0$'` ; do
18
-
# skip bridges since we can't recover them (yet...)
19
-
if [ -e "/sys/bus/pci/devices/$dev/pci_bus" ] ; then
20
-
echo "$dev, Skipped: bridge"
33
+
if ! eeh_can_break $dev ; then
21
34
continue;
22
35
fi
23
36
24
-
# Skip VFs for now since we don't have a reliable way
25
-
# to break them.
37
+
# Skip VFs for now since we don't have a reliable way to break them.
26
38
if [ -e "/sys/bus/pci/devices/$dev/physfn" ] ; then
27
39
echo "$dev, Skipped: virtfn"
28
-
continue;
29
-
fi
30
-
31
-
if [ "ahci" = "$(basename $(realpath /sys/bus/pci/devices/$dev/driver))" ] ; then
32
-
echo "$dev, Skipped: ahci doesn't support recovery"
33
-
continue
34
-
fi
35
-
36
-
# Don't inject errosr into an already-frozen PE. This happens with
37
-
# PEs that contain multiple PCI devices (e.g. multi-function cards)
38
-
# and injecting new errors during the recovery process will probably
39
-
# result in the recovery failing and the device being marked as
40
-
# failed.
41
-
if ! pe_ok $dev ; then
42
-
echo "$dev, Skipped: Bad initial PE state"
43
40
continue;
44
41
fi
45
42
+48
tools/testing/selftests/powerpc/eeh/eeh-functions.sh
+48
tools/testing/selftests/powerpc/eeh/eeh-functions.sh
···
1
1
#!/bin/sh
2
2
# SPDX-License-Identifier: GPL-2.0-only
3
3
4
+
export KSELFTESTS_SKIP=4
5
+
4
6
pe_ok() {
5
7
local dev="$1"
6
8
local path="/sys/bus/pci/devices/$dev/eeh_pe_state"
···
39
37
eeh_supported() {
40
38
test -e /proc/powerpc/eeh && \
41
39
grep -q 'EEH Subsystem is enabled' /proc/powerpc/eeh
40
+
}
41
+
42
+
eeh_test_prep() {
43
+
if ! eeh_supported ; then
44
+
echo "EEH not supported on this system, skipping"
45
+
exit $KSELFTESTS_SKIP;
46
+
fi
47
+
48
+
if [ ! -e "/sys/kernel/debug/powerpc/eeh_dev_check" ] && \
49
+
[ ! -e "/sys/kernel/debug/powerpc/eeh_dev_break" ] ; then
50
+
echo "debugfs EEH testing files are missing. Is debugfs mounted?"
51
+
exit $KSELFTESTS_SKIP;
52
+
fi
53
+
54
+
# Bump the max freeze count to something absurd so we don't
55
+
# trip over it while breaking things.
56
+
echo 5000 > /sys/kernel/debug/powerpc/eeh_max_freezes
57
+
}
58
+
59
+
eeh_can_break() {
60
+
# skip bridges since we can't recover them (yet...)
61
+
if [ -e "/sys/bus/pci/devices/$dev/pci_bus" ] ; then
62
+
echo "$dev, Skipped: bridge"
63
+
return 1;
64
+
fi
65
+
66
+
# The ahci driver doesn't support error recovery. If the ahci device
67
+
# happens to be hosting the root filesystem, and then we go and break
68
+
# it the system will generally go down. We should probably fix that
69
+
# at some point
70
+
if [ "ahci" = "$(basename $(realpath /sys/bus/pci/devices/$dev/driver))" ] ; then
71
+
echo "$dev, Skipped: ahci doesn't support recovery"
72
+
return 1;
73
+
fi
74
+
75
+
# Don't inject errosr into an already-frozen PE. This happens with
76
+
# PEs that contain multiple PCI devices (e.g. multi-function cards)
77
+
# and injecting new errors during the recovery process will probably
78
+
# result in the recovery failing and the device being marked as
79
+
# failed.
80
+
if ! pe_ok $dev ; then
81
+
echo "$dev, Skipped: Bad initial PE state"
82
+
return 1;
83
+
fi
84
+
85
+
return 0
42
86
}
43
87
44
88
eeh_one_dev() {