| Message ID | 20230328150104.129486-1-ailiop@suse.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | common: switch _get_available_space accounting to units of 512 bytes | expand |
On Tue, Mar 28, 2023 at 05:01:04PM +0200, Anthony Iliopoulos wrote: > xfs/220 relies on detecting free space changes after truncation of > single filesystem blocks, and this fails when the fs block size is 512b. > > By default df is counting 1024b block units and as such is not reporting > any change in the number of available blocks after freeing just a 512b > block. > > Switch the df reporting block size from units of 1024b blocks to 512b > blocks to accommodate accounting for smaller granularity fs block sizes. > > Signed-off-by: Anthony Iliopoulos <ailiop@suse.com> > --- > common/rc | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/common/rc b/common/rc > index 90749343f3c4..a32bbe3c0b5f 100644 > --- a/common/rc > +++ b/common/rc > @@ -4209,8 +4209,8 @@ _get_available_space() > exit 1 > fi > local avail_kb; > - avail_kb=`$DF_PROG $1 | tail -n1 | awk '{ print $5 }'` > - echo $((avail_kb * 1024)) > + avail_kb=`$DF_PROG --block-size=512 $1 | tail -n1 | awk '{ print $5 }'` > + echo $((avail_kb * 512)) Why not have df report the free space in bytes? $DF_PROG -B 1 $1 | tail -n 1 | awk '{print $5}' --D > } > > # get the total space in bytes > -- > 2.35.3 >
On Tue, Mar 28, 2023 at 09:02:25AM -0700, Darrick J. Wong wrote: > On Tue, Mar 28, 2023 at 05:01:04PM +0200, Anthony Iliopoulos wrote: > > xfs/220 relies on detecting free space changes after truncation of > > single filesystem blocks, and this fails when the fs block size is 512b. > > > > By default df is counting 1024b block units and as such is not reporting > > any change in the number of available blocks after freeing just a 512b > > block. > > > > Switch the df reporting block size from units of 1024b blocks to 512b > > blocks to accommodate accounting for smaller granularity fs block sizes. > > > > Signed-off-by: Anthony Iliopoulos <ailiop@suse.com> > > --- > > common/rc | 4 ++-- > > 1 file changed, 2 insertions(+), 2 deletions(-) > > > > diff --git a/common/rc b/common/rc > > index 90749343f3c4..a32bbe3c0b5f 100644 > > --- a/common/rc > > +++ b/common/rc > > @@ -4209,8 +4209,8 @@ _get_available_space() > > exit 1 > > fi > > local avail_kb; > > - avail_kb=`$DF_PROG $1 | tail -n1 | awk '{ print $5 }'` > > - echo $((avail_kb * 1024)) > > + avail_kb=`$DF_PROG --block-size=512 $1 | tail -n1 | awk '{ print $5 }'` > > + echo $((avail_kb * 512)) > > Why not have df report the free space in bytes? > > $DF_PROG -B 1 $1 | tail -n 1 | awk '{print $5}' Agree, if we finally hope to get "bytes", and we can get it at beginning, why take a long way? If you'd like to send a V2 to do that, you can do the same change on _get_total_space() too. Thanks, Zorro > > --D > > > } > > > > # get the total space in bytes > > -- > > 2.35.3 > > >
diff --git a/common/rc b/common/rc index 90749343f3c4..a32bbe3c0b5f 100644 --- a/common/rc +++ b/common/rc @@ -4209,8 +4209,8 @@ _get_available_space() exit 1 fi local avail_kb; - avail_kb=`$DF_PROG $1 | tail -n1 | awk '{ print $5 }'` - echo $((avail_kb * 1024)) + avail_kb=`$DF_PROG --block-size=512 $1 | tail -n1 | awk '{ print $5 }'` + echo $((avail_kb * 512)) } # get the total space in bytes
xfs/220 relies on detecting free space changes after truncation of single filesystem blocks, and this fails when the fs block size is 512b. By default df is counting 1024b block units and as such is not reporting any change in the number of available blocks after freeing just a 512b block. Switch the df reporting block size from units of 1024b blocks to 512b blocks to accommodate accounting for smaller granularity fs block sizes. Signed-off-by: Anthony Iliopoulos <ailiop@suse.com> --- common/rc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)