> On Jun 21, 2025, at 01:38, Zorro Lang <zlang@xxxxxxxxxx> wrote:
>
> On Tue, Jun 03, 2025 at 12:10:09PM +0800, Su Yue wrote:
>> gfs2 supports mount options 'quota=[off/account/quiet/on]' but
>> 'usrquota' and 'grpquota' are unsupported.
>> So in _qmount_option, transform $OPTS to 'quota=on' first.
>>
>> Signed-off-by: Su Yue <glass.su@xxxxxxxx>
>> ---
>> common/quota | 7 +++++++
>> 1 file changed, 7 insertions(+)
>>
>> diff --git a/common/quota b/common/quota
>> index a51386b1dd24..4b8052e07dde 100644
>> --- a/common/quota
>> +++ b/common/quota
>> @@ -283,6 +283,12 @@ _qmount_option()
>> {
>> OPTS=$1
>>
>> + if [[ "$FSTYP" == gfs2 ]]; then
>> + OPTS=`echo $OPTS \
>> + | sed -e 's/usrquota/quota=on/g' \
>> + -e 's/grpquota/quota=on/g'`
>> + fi
>
> There's not response from gfs list. I'm not familiar with gfs2, so I have to
> ask some questions about your patch...
Feel free to communicate : )
>
> From the manual of gfs2:
> quota=[off/account/on]
> Turns quotas on or off for a filesystem. Setting the quotas to be in the "account" state causes the per UID/GID usage statistics to be correctly maintained by the filesystem, limit and warn values are ignored. The default value is "off".
>
> Looks like there's only one mount option "quota=..." for gfs2. And I didn't
> find a description about project quota, so
> 1. What will gfs2 do if there's a "prjquota" or "pquota"?
The mount will fail if prjquota/pquota is in mount options.
_require_prjquota() should be modified.
> 2. Do you need to replace "noquota" with "quota=off"?
Yes.. I forgot the case.
> 3. And "quota" to "quota=on" ?
The only case "_qmount_option ‘quota’” is tests/xfs/820, not in generic group.
> 4. Have you tried your patch on gfs2, can all quota related generic cases pass/notrun
> on gfs2?
>
TBH, not all quota tests run well on gfs2
There was a bug which can cause kernel hang in gfs2 fixed by kernel commit 5a90f8d49922 ("gfs2: Don't start unnecessary transactions during log flush”)
In v6.16-rc1.
> BTW, in this function you can see:
> # ext4 doesn't _do_ "-o pquota/prjquota" because reasons
> # Switch it to "quota" to enable mkfs-time pquota
The comment is confusing to me. ext[234] support ‘-o prjquota’ if MKFS_OPTIONS="-O quota -E quotatype=usrquota:grpquota:prjquota”.
> if [[ "$FSTYP" == ext[234] ]]; then
> OPTS=`echo $OPTS \
> | sed -e 's/\bpquota/quota/g' \
> -e 's/prjquota/quota/g'`
> # ocfs2 doesn't support "-o projquota"
> elif [[ "$FSTYP" == ocfs2 ]]; then
> OPTS=`echo $OPTS | sed -e 's/prjquota//g'`
> fi
>
> please move your gfs2 specific codes here ^^, and give it enough comments.
There is one question from me about the _qmount_option() and _qmount()
Not all tests of qgroup use _qmount, some of them call _scratch_mount() directly.
Is this intentional to avoid influenced by other mount options? If It’s not IMO it’s feasible
to convert them to _qmount?
>
>> +
>> # Replace any user defined quota options
>> # with the quota option that we want.
>> # Simplest to do this rather than delete existing ones first because
>> @@ -300,6 +306,7 @@ _qmount_option()
>> -e 's/\bpquota/QUOTA/g' \
>> -e 's/prjquota/QUOTA/g' \
>> -e 's/noquota/QUOTA/g' \
>> + -e 's/quota=[^, ]*/QUOTA/g' \
>
> I think you might need a word boundary, e.g. "\bquota=..."
Right.
—
Su
>
> Thanks,
> Zorro
>
>> -e 's/quota/QUOTA/g' \
>> -e 's/uqnoenforce/QUOTA/g' \
>> -e 's/gqnoenforce/QUOTA/g' \
>> --
>> 2.48.1
