| Message ID | 97ede9bba67f0848fc0b706d757170d7dfacb7fd.1663825728.git.naohiro.aota@wdc.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | btrfs: test active zone tracking | expand |
On Thu, Sep 22, 2022 at 02:54:58PM +0900, Naohiro Aota wrote: > Introduce _zone_capacity() to return a zone capacity of the given address > in the given device (optional). Rewrite btrfs/237 with it. > > Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com> > --- > common/zbd | 17 +++++++++++++++++ > tests/btrfs/237 | 8 ++------ > 2 files changed, 19 insertions(+), 6 deletions(-) > create mode 100644 common/zbd > > diff --git a/common/zbd b/common/zbd > new file mode 100644 > index 000000000000..329bb7be6b7b > --- /dev/null > +++ b/common/zbd I don't like this abbreviation :-P If others don't open this file and read the comment in it, they nearly no chance to guess what's this file for. > @@ -0,0 +1,17 @@ > +# > +# Common zoned block device specific functions > +# > + > +. common/filter > + > +_zone_capacity() { > + local phy=$1 > + local dev=$2 > + > + [ -z "$dev" ] && dev=$SCRATCH_DEV > + > + size=$($BLKZONE_PROG report -o $phy -l 1 $dev |\ > + _filter_blkzone_report |\ > + grep -Po "cap 0x[[:xdigit:]]+" | cut -d ' ' -f 2) > + echo $((size << 9)) > +} Do you have more zone related helpers are going to provide? If only this single one helper, I think it's not worth adding a new common file. You can put it into common/rc simply, as there're already several *zone* related helpers in it. We can pick up these *zone* related function into a separate include file in one day we feel it's time (too many related helpers, and better to be maintained individually). Thanks, Zorro > diff --git a/tests/btrfs/237 b/tests/btrfs/237 > index bc6522e2200a..101094b5ce70 100755 > --- a/tests/btrfs/237 > +++ b/tests/btrfs/237 > @@ -13,7 +13,7 @@ > _begin_fstest auto quick zone balance > > # Import common functions. > -. ./common/filter > +. ./common/zbd > > # real QA test starts here > > @@ -56,11 +56,7 @@ fi > > start_data_bg_phy=$(get_data_bg_physical) > start_data_bg_phy=$((start_data_bg_phy >> 9)) > - > -size=$($BLKZONE_PROG report -o $start_data_bg_phy -l 1 $SCRATCH_DEV |\ > - _filter_blkzone_report |\ > - grep -Po "cap 0x[[:xdigit:]]+" | cut -d ' ' -f 2) > -size=$((size << 9)) > +size=$(_zone_capacity $start_data_bg_phy) > > reclaim_threshold=75 > echo $reclaim_threshold > /sys/fs/btrfs/"$uuid"/bg_reclaim_threshold > -- > 2.37.3 >
On 22.09.22 17:42, Zorro Lang wrote: >> --- /dev/null >> +++ b/common/zbd > I don't like this abbreviation :-P If others don't open this file and read the > comment in it, they nearly no chance to guess what's this file for. > zbd is a well known abbreviation for zoned block devices. I think most people in storage and filesystems know it.
On Fri, Sep 23, 2022 at 08:02:10AM +0000, Johannes Thumshirn wrote: > On 22.09.22 17:42, Zorro Lang wrote: > >> --- /dev/null > >> +++ b/common/zbd > > I don't like this abbreviation :-P If others don't open this file and read the > > comment in it, they nearly no chance to guess what's this file for. > > > > zbd is a well known abbreviation for zoned block devices. I think most > people in storage and filesystems know it. OK, but we haven't been that "a single character is worth a thousand pieces of gold", so we can use a longer name, likes common/zone, common/zoned, common/zoned_block, common/zoned_device or something likes that. Anyway, that's just my personal opinion, if most of people prefer using "common/zbd", I'm fine to have that :) But I hope you can move all zoned block device related helpers to the new common file if you'd like to bring in this file, likes what Darrick did in: commit 67afd5c742464607994316acb2c6e8303b8af4c5 Author: Darrick J. Wong <djwong@kernel.org> Date: Tue Aug 9 14:00:46 2022 -0700 common/rc: move ext4-specific helpers into a separate common/ext4 file Thanks, Zorro > >
On Fri, Sep 23, 2022 at 07:51:26PM +0800, Zorro Lang wrote: > On Fri, Sep 23, 2022 at 08:02:10AM +0000, Johannes Thumshirn wrote: > > On 22.09.22 17:42, Zorro Lang wrote: > > >> --- /dev/null > > >> +++ b/common/zbd > > > I don't like this abbreviation :-P If others don't open this file and read the > > > comment in it, they nearly no chance to guess what's this file for. > > > > > > > zbd is a well known abbreviation for zoned block devices. I think most > > people in storage and filesystems know it. > > OK, but we haven't been that "a single character is worth a thousand > pieces of gold", so we can use a longer name, likes common/zone, > common/zoned, common/zoned_block, common/zoned_device or something likes > that. Anyway, that's just my personal opinion, if most of people prefer > using "common/zbd", I'm fine to have that :) Sure. I'll use "zoned" as it is more common in the kernel code. > But I hope you can move all zoned block device related helpers to the new > common file if you'd like to bring in this file, likes what Darrick did in: > > commit 67afd5c742464607994316acb2c6e8303b8af4c5 > Author: Darrick J. Wong <djwong@kernel.org> > Date: Tue Aug 9 14:00:46 2022 -0700 > > common/rc: move ext4-specific helpers into a separate common/ext4 file Yes, that will be better to have things in common/zoned. I considered moving zoned functions (_zone_type, _require_{,non_}zoned_device), but _require_loop() and _require_dm_target() use _require_non_zoned_device() in them. So, moving _require_non_zoned_device() will make a dependency from common/rc to common/zoned, which I considered not much clean. How do you think of it? Moving _filter_blkzone_report() would be fine, though. > Thanks, > Zorro > > > > > >
On Wed, Sep 28, 2022 at 03:57:08AM +0000, Naohiro Aota wrote: > On Fri, Sep 23, 2022 at 07:51:26PM +0800, Zorro Lang wrote: > > On Fri, Sep 23, 2022 at 08:02:10AM +0000, Johannes Thumshirn wrote: > > > On 22.09.22 17:42, Zorro Lang wrote: > > > >> --- /dev/null > > > >> +++ b/common/zbd > > > > I don't like this abbreviation :-P If others don't open this file and read the > > > > comment in it, they nearly no chance to guess what's this file for. > > > > > > > > > > zbd is a well known abbreviation for zoned block devices. I think most > > > people in storage and filesystems know it. > > > > OK, but we haven't been that "a single character is worth a thousand > > pieces of gold", so we can use a longer name, likes common/zone, > > common/zoned, common/zoned_block, common/zoned_device or something likes > > that. Anyway, that's just my personal opinion, if most of people prefer > > using "common/zbd", I'm fine to have that :) > > Sure. I'll use "zoned" as it is more common in the kernel code. > > > But I hope you can move all zoned block device related helpers to the new > > common file if you'd like to bring in this file, likes what Darrick did in: > > > > commit 67afd5c742464607994316acb2c6e8303b8af4c5 > > Author: Darrick J. Wong <djwong@kernel.org> > > Date: Tue Aug 9 14:00:46 2022 -0700 > > > > common/rc: move ext4-specific helpers into a separate common/ext4 file > > Yes, that will be better to have things in common/zoned. I considered > moving zoned functions (_zone_type, _require_{,non_}zoned_device), but > _require_loop() and _require_dm_target() use _require_non_zoned_device() in > them. So, moving _require_non_zoned_device() will make a dependency from > common/rc to common/zoned, which I considered not much clean. How do you > think of it? Oh, below commit [1] brought in the coupling of common/rc and zoned helpers. Hmm... that cause all the 3 helpers (_zone_type, _require_{,non_}zoned_device) have to be in common/rc or be imported in common/rc. Looks like we have to keep them in common/rc, except we make a bigger refactor to common/rc, or you'd like to make your 2 new helpers in common/rc too (likes these 3 old ones:) BTW I doubt if we might need to use more zoned related helpers in common/rc, due to we deal with test devices in common/rc mostly, likes dax. Someone might want a seperated common/dax or common/pmem file one day. The common/rc imports specific fs helpers according to $FSTYP (common/config: _source_specific_fs()). If we need to deal with different kind of device types in common/rc one day, is there a better idea to determine which one should be imported? Welcome any suggestions if anyone has :) [1] commit 952310a57d9323ae0bb174b50be93107a8895e0c Author: Naohiro Aota <naohiro.aota@wdc.com> Date: Mon Aug 16 20:35:08 2021 +0900 common: add zoned block device checks > > Moving _filter_blkzone_report() would be fine, though. Yeah, moving this is fine. Thanks, Zorro > > > Thanks, > > Zorro > > > > > > > > > > >
On Wed, Sep 28, 2022 at 01:47:15PM +0800, Zorro Lang wrote: > On Wed, Sep 28, 2022 at 03:57:08AM +0000, Naohiro Aota wrote: > > On Fri, Sep 23, 2022 at 07:51:26PM +0800, Zorro Lang wrote: > > > On Fri, Sep 23, 2022 at 08:02:10AM +0000, Johannes Thumshirn wrote: > > > > On 22.09.22 17:42, Zorro Lang wrote: > > > > >> --- /dev/null > > > > >> +++ b/common/zbd > > > > > I don't like this abbreviation :-P If others don't open this file and read the > > > > > comment in it, they nearly no chance to guess what's this file for. > > > > > > > > > > > > > zbd is a well known abbreviation for zoned block devices. I think most > > > > people in storage and filesystems know it. > > > > > > OK, but we haven't been that "a single character is worth a thousand > > > pieces of gold", so we can use a longer name, likes common/zone, > > > common/zoned, common/zoned_block, common/zoned_device or something likes > > > that. Anyway, that's just my personal opinion, if most of people prefer > > > using "common/zbd", I'm fine to have that :) > > > > Sure. I'll use "zoned" as it is more common in the kernel code. > > > > > But I hope you can move all zoned block device related helpers to the new > > > common file if you'd like to bring in this file, likes what Darrick did in: > > > > > > commit 67afd5c742464607994316acb2c6e8303b8af4c5 > > > Author: Darrick J. Wong <djwong@kernel.org> > > > Date: Tue Aug 9 14:00:46 2022 -0700 > > > > > > common/rc: move ext4-specific helpers into a separate common/ext4 file > > > > Yes, that will be better to have things in common/zoned. I considered > > moving zoned functions (_zone_type, _require_{,non_}zoned_device), but > > _require_loop() and _require_dm_target() use _require_non_zoned_device() in > > them. So, moving _require_non_zoned_device() will make a dependency from > > common/rc to common/zoned, which I considered not much clean. How do you > > think of it? > > Oh, below commit [1] brought in the coupling of common/rc and zoned helpers. > Hmm... that cause all the 3 helpers (_zone_type, _require_{,non_}zoned_device) > have to be in common/rc or be imported in common/rc. Looks like we have to > keep them in common/rc, except we make a bigger refactor to common/rc, or you'd > like to make your 2 new helpers in common/rc too (likes these 3 old ones:) > > BTW I doubt if we might need to use more zoned related helpers in common/rc, due > to we deal with test devices in common/rc mostly, likes dax. Someone might want > a seperated common/dax or common/pmem file one day. The common/rc imports > specific fs helpers according to $FSTYP (common/config: _source_specific_fs()). > If we need to deal with different kind of device types in common/rc one day, is > there a better idea to determine which one should be imported? Welcome any > suggestions if anyone has :) Leave those three in common/rc and put/move the rest to common/zoned ? I think it's fine for common/rc to have helpers that *detect* the presence of a blockdev feature, and require tests to source common/$feature if they want to do anything clever with that feature. After all, the _require_non_zoned_device tests don't care about _zone_capacity, right? --D > [1] > commit 952310a57d9323ae0bb174b50be93107a8895e0c > Author: Naohiro Aota <naohiro.aota@wdc.com> > Date: Mon Aug 16 20:35:08 2021 +0900 > > common: add zoned block device checks > > > > > Moving _filter_blkzone_report() would be fine, though. > > Yeah, moving this is fine. > > Thanks, > Zorro > > > > > > Thanks, > > > Zorro > > > > > > > > > > > > > > > > >
On Wed, Sep 28, 2022 at 08:32:47AM -0700, Darrick J. Wong wrote: > On Wed, Sep 28, 2022 at 01:47:15PM +0800, Zorro Lang wrote: > > On Wed, Sep 28, 2022 at 03:57:08AM +0000, Naohiro Aota wrote: > > > On Fri, Sep 23, 2022 at 07:51:26PM +0800, Zorro Lang wrote: > > > > On Fri, Sep 23, 2022 at 08:02:10AM +0000, Johannes Thumshirn wrote: > > > > > On 22.09.22 17:42, Zorro Lang wrote: > > > > > >> --- /dev/null > > > > > >> +++ b/common/zbd > > > > > > I don't like this abbreviation :-P If others don't open this file and read the > > > > > > comment in it, they nearly no chance to guess what's this file for. > > > > > > > > > > > > > > > > zbd is a well known abbreviation for zoned block devices. I think most > > > > > people in storage and filesystems know it. > > > > > > > > OK, but we haven't been that "a single character is worth a thousand > > > > pieces of gold", so we can use a longer name, likes common/zone, > > > > common/zoned, common/zoned_block, common/zoned_device or something likes > > > > that. Anyway, that's just my personal opinion, if most of people prefer > > > > using "common/zbd", I'm fine to have that :) > > > > > > Sure. I'll use "zoned" as it is more common in the kernel code. > > > > > > > But I hope you can move all zoned block device related helpers to the new > > > > common file if you'd like to bring in this file, likes what Darrick did in: > > > > > > > > commit 67afd5c742464607994316acb2c6e8303b8af4c5 > > > > Author: Darrick J. Wong <djwong@kernel.org> > > > > Date: Tue Aug 9 14:00:46 2022 -0700 > > > > > > > > common/rc: move ext4-specific helpers into a separate common/ext4 file > > > > > > Yes, that will be better to have things in common/zoned. I considered > > > moving zoned functions (_zone_type, _require_{,non_}zoned_device), but > > > _require_loop() and _require_dm_target() use _require_non_zoned_device() in > > > them. So, moving _require_non_zoned_device() will make a dependency from > > > common/rc to common/zoned, which I considered not much clean. How do you > > > think of it? > > > > Oh, below commit [1] brought in the coupling of common/rc and zoned helpers. > > Hmm... that cause all the 3 helpers (_zone_type, _require_{,non_}zoned_device) > > have to be in common/rc or be imported in common/rc. Looks like we have to > > keep them in common/rc, except we make a bigger refactor to common/rc, or you'd > > like to make your 2 new helpers in common/rc too (likes these 3 old ones:) > > > > BTW I doubt if we might need to use more zoned related helpers in common/rc, due > > to we deal with test devices in common/rc mostly, likes dax. Someone might want > > a seperated common/dax or common/pmem file one day. The common/rc imports > > specific fs helpers according to $FSTYP (common/config: _source_specific_fs()). > > If we need to deal with different kind of device types in common/rc one day, is > > there a better idea to determine which one should be imported? Welcome any > > suggestions if anyone has :) > > Leave those three in common/rc and put/move the rest to common/zoned ? > > I think it's fine for common/rc to have helpers that *detect* the > presence of a blockdev feature, and require tests to source > common/$feature if they want to do anything clever with that feature. > After all, the _require_non_zoned_device tests don't care about > _zone_capacity, right? Thank you for your suggestions. Yes, _require_non_zoned_device() just read /sys/block/${sdev}/queue/zoned. I'll do so to leave the three functions in common/rc and move/create other helpers in common/zoned. Thanks, > --D > > > [1] > > commit 952310a57d9323ae0bb174b50be93107a8895e0c > > Author: Naohiro Aota <naohiro.aota@wdc.com> > > Date: Mon Aug 16 20:35:08 2021 +0900 > > > > common: add zoned block device checks > > > > > > > > Moving _filter_blkzone_report() would be fine, though. > > > > Yeah, moving this is fine. > > > > Thanks, > > Zorro > > > > > > > > > Thanks, > > > > Zorro > > > > > > > > > > > > > > > > > > > > > > >
diff --git a/common/zbd b/common/zbd new file mode 100644 index 000000000000..329bb7be6b7b --- /dev/null +++ b/common/zbd @@ -0,0 +1,17 @@ +# +# Common zoned block device specific functions +# + +. common/filter + +_zone_capacity() { + local phy=$1 + local dev=$2 + + [ -z "$dev" ] && dev=$SCRATCH_DEV + + size=$($BLKZONE_PROG report -o $phy -l 1 $dev |\ + _filter_blkzone_report |\ + grep -Po "cap 0x[[:xdigit:]]+" | cut -d ' ' -f 2) + echo $((size << 9)) +} diff --git a/tests/btrfs/237 b/tests/btrfs/237 index bc6522e2200a..101094b5ce70 100755 --- a/tests/btrfs/237 +++ b/tests/btrfs/237 @@ -13,7 +13,7 @@ _begin_fstest auto quick zone balance # Import common functions. -. ./common/filter +. ./common/zbd # real QA test starts here @@ -56,11 +56,7 @@ fi start_data_bg_phy=$(get_data_bg_physical) start_data_bg_phy=$((start_data_bg_phy >> 9)) - -size=$($BLKZONE_PROG report -o $start_data_bg_phy -l 1 $SCRATCH_DEV |\ - _filter_blkzone_report |\ - grep -Po "cap 0x[[:xdigit:]]+" | cut -d ' ' -f 2) -size=$((size << 9)) +size=$(_zone_capacity $start_data_bg_phy) reclaim_threshold=75 echo $reclaim_threshold > /sys/fs/btrfs/"$uuid"/bg_reclaim_threshold
Introduce _zone_capacity() to return a zone capacity of the given address in the given device (optional). Rewrite btrfs/237 with it. Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com> --- common/zbd | 17 +++++++++++++++++ tests/btrfs/237 | 8 ++------ 2 files changed, 19 insertions(+), 6 deletions(-) create mode 100644 common/zbd