Hi Salvatore,
On 20/05/26 22:07, Salvatore Bonaccorso wrote:
On Thu, May 14, 2026 at 01:29:37AM +0200, Gioele Barabucci wrote:
It is believed that nfs-utils uses `pidof` due to the following
code snippets:
```
path: nfs-utils_1:2.8.6-1/debian/tests/v3-mount
grep "${hello}" /mnt/hello.txt
pidof rpc.statd > /dev/null 2>&1 || {
echo "rpc.statd isn't running after the NVSv3 mount, and it should..."
exit 1
path: nfs-utils_1:2.8.6-1/debian/tests/v3-mount
# stop statd if it's running, because we will check later if it
# was started automatically for the NFSv3 mount as it should
if pidof rpc.statd > /dev/null 2>&1; then
systemctl stop rpc-statd.service > /dev/null 2>&1 || :
pidof rpc.statd > /dev/null 2>&1 && {
```
Feel free to close this issue if this is a false positive (for example
if this code is in an unreachable code path).
I addressed the case for the autopkgtest adding a Depends there.
Thanks!
But I won't add a Depends for nfs-common binary package providing
still the init script as well. The idea is to either drop the status
there or ask people who will still use the init scripts instead of the
systemd units to please provide a patch moving away of the pidof
usage there.
That's correct. The use of pidof in initscripts has been explicitly left
out of this MBF because it will be addressed separately (see for
instance https://bugs.debian.org/810018#327>).
Regards,
--
Gioele Barabucci
