Hello, I notice that in some shutdown/reboot scenario, mount-ro is not able to mount root readonly. It is failing with "in use but fuser finds nothing" I suspect on multi-core machines, this issue is more probable. Changing the relevant code as below seems to help: " ") if [ $retry -le 0 ]; then eend 1 "in use but fuser finds nothing" else : $(( retry -= 1 )) sleep 1 fi ;; *) Thanks Reproducible: Sometimes
Created attachment 361732 [details, diff] proposed patch to rc-mount.sh
Created attachment 383168 [details] updated rc-mount.sh
I noticed a couple of things about this update that I am concerned about. - The treeps function defined on line 8, and the call to it on line 46 and the echo on line 47 seem to be debugging code you left in, is this correct? - What about the set -o commands on lines 40 and 41? - The let command on line 53 is definitely not portable; this will fail for any shell other than bash, so we need something posix compliant for this. - Why is bash run directly on line 57? Also, should I close the new bug you opened as a dupe of this one? There isn't a need to duplicate updates on more than one bug. Thanks, William
(In reply to William Hubbs from comment #3) > - The let command on line 53 is definitely not portable; this will fail > for any shell other than bash, so we need something posix > compliant for this. can answer this myself; the posix compliant form is the one that was originally used, so I have added it back in my copy of your changes: : $(( retry -= 1 )) For reference, posix compliant means compatible with this spec: http://pubs.opengroup.org/onlinepubs/009604599/utilities/xcu_chap02.html
(In reply to William Hubbs from comment #3) > I noticed a couple of things about this update that I am concerned about. I put the update more as a guide rather than as patch. Sorry I was not clearer about it. > - The treeps function defined on line 8, and the call to it on line 46 > and the echo on line 47 seem to be debugging code you left in, is this > correct? it is my debugging code, yes > - What about the set -o commands on lines 40 and 41? debugging code. > - The let command on line 53 is definitely not portable; this will fail > for any shell other than bash, so we need something posix > compliant for this. ok, thanks for pointing it out. > - Why is bash run directly on line 57? debugging code > Also, should I close the new bug you opened as a dupe of this one? There > isn't a need to duplicate updates on more than one bug. good idea. I opened the other to show that the case statement does not work as intended. Will close that. Thanks Joe
*** Bug 520266 has been marked as a duplicate of this bug. ***
Created attachment 383372 [details] updated rc-mount.sh
Hey Joe, where are we on this bug? can version 0.21.7 be used with read-only root?
(In reply to William Hubbs from comment #8) > Hey Joe, > > where are we on this bug? can version 0.21.7 be used with read-only root? Yes, 0.21.7 can be used with read-only root. Sorry for the delayed response.