Re: [PATCH v3 3/3] selftests: net: Fix udpgro_frglist.sh shellcheck warnings and errors

On 23/01/27 06:32AM, Maciej Żenczykowski wrote:

On Fri, Jan 27, 2023 at 6:09 AM Andrei Gherzan andrei.gherzan@canonical.com wrote:

...

This change fixes the following shellcheck warnings and errors:

• SC2155 (warning): Declare and assign separately to avoid masking return values.
• SC2124 (warning): Assigning an array to a string! Assign as array, or use instead of @ to concatenate.
• SC2034 (warning): ipv4_args appears unused. Verify use (or export if used externally).
• SC2242 (error): Can only exit with status 0-255. Other data should be written to stdout/stderr.
• SC2068 (error): Double quote array expansions to avoid re-splitting elements.

Fixes: edae34a3ed92 ("selftests net: add UDP GRO fraglist + bpf self-tests") Signed-off-by: Andrei Gherzan andrei.gherzan@canonical.com

---

tools/testing/selftests/net/udpgro_frglist.sh | 20 +++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-)

diff --git a/tools/testing/selftests/net/udpgro_frglist.sh b/tools/testing/selftests/net/udpgro_frglist.sh index e1ca49de2491..97bf20e9afd8 100755 --- a/tools/testing/selftests/net/udpgro_frglist.sh +++ b/tools/testing/selftests/net/udpgro_frglist.sh @@ -3,7 +3,8 @@ # # Run a series of udpgro benchmarks

-readonly PEER_NS="ns-peer-$(mktemp -u XXXXXX)" +PEER_NS="ns-peer-$(mktemp -u XXXXXX)" +readonly PEER_NS

BPF_FILE="../bpf/xdp_dummy.bpf.o" BPF_NAT6TO4_FILE="nat6to4.o" @@ -19,7 +20,7 @@ trap cleanup EXIT

run_one() { # use 'rx' as separator between sender args and receiver args

•   local -r all="$@"
  

•   local -r all="$*"
  

this should technically use arrays, something like

local -a -r args=("$@")

but perhaps just get rid of args and just use "$@" directly below

Using it directly would not work with the substitutions later because it would try to apply it to each positional parameter in turn. Same when using an intermediate array.

...

    local -r tx_args=${all%rx*}
    local rx_args=${all#*rx}

@@ -56,13 +57,13 @@ run_one() { }

run_in_netns() {

•   local -r args=$@
  

•   local -r args="$*"
  

  echo ${args} ./in_netns.sh $0 __subprocess ${args}

ie. here could just use "$@" directly twice instead of defining args. $0 should be doublequoted - though I guess it'll never be empty, and is unlikely to include spaces.

That sounds fine to me. I'll include it in v4.

...

}

run_udp() {

•   local -r args=$@
  

•   local -r args="$*"
  
    echo "udp gso - over veth touching data"
    run_in_netns ${args} -u -S 0 rx -4 -v
  

@@ -72,7 +73,7 @@ run_udp() { }

run_tcp() {

•   local -r args=$@
  

•   local -r args="$*"
  
    echo "tcp - over veth touching data"
    run_in_netns ${args} -t rx -4 -t
  

@@ -80,7 +81,6 @@ run_tcp() {

run_all() { local -r core_args="-l 4"

is this still useful? embed directly in ipv6_args

...

•   local -r ipv4_args="${core_args} -4  -D 192.168.1.1"
  

perhaps this should stay as a comment??

Well the way I see it is one or the other. We either embed and drop ipv4_args or keep ipv4_args as a comment but also core_args. I'll do the former in v4 if no other objections.

...

    local -r ipv6_args="${core_args} -6  -D 2001:db8::1"

    echo "ipv6"

@@ -90,19 +90,19 @@ run_all() {

if [ ! -f ${BPF_FILE} ]; then

double quote "${BPF_FILE}" in case space in file name

This change only targets warning/error issues. There are way more "info" ones, but I didn't want to splash a big patch for all those.

...

    echo "Missing ${BPF_FILE}. Build bpf selftest first"

•   exit -1
  

•   exit 1
  

fi

if [ ! -f "$BPF_NAT6TO4_FILE" ]; then

there seems to be inconsistency around [ vs [[, use [[ if relying on bash anyway

...

    echo "Missing nat6to4 helper. Build bpf nat6to4.o selftest first"

•   exit -1
  

•   exit 1
  

fi

if [[ $# -eq 0 ]]; then run_all elif [[ $1 == "__subprocess" ]]; then

while this does indeed work, imho $1 should be "$1" to be less confusing

Agreed and again, there are a good set of other places where this is needed. Should I just address them all in an extra patch? Again, this one only scoped warnings/errors to avoid impact.

Thanks for the review.