Skip to contentSkip to conversation

null_blk: allow REQ_OP_ZONE_RESET_ALL to configure

PATCH In review Chaitanya Kulkarni 5 participants 11 messages Oct 6, 2022 – Nov 5, 2022 Block layer

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 null_blk: allow REQ_OP_ZONE_RESET_ALL to configure

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

---

# git log -1 
commit 8ca0bd53a9c9e2f58c4fc9e38e3f5f82d26d3851 (HEAD -> for-next)
Author: Chaitanya Kulkarni <[email protected]>
Date:   Wed Oct 5 21:57:00 2022 -0700

null_blk: allow REQ_OP_ZONE_RESET_ALL to configure

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

# ./compile_nullb.sh 
+ umount /mnt/nullb0
umount: /mnt/nullb0: not mounted.
+ rmdir 'config/nullb/nullb*'
rmdir: failed to remove 'config/nullb/nullb*': No such file or directory
+ dmesg -c
+ modprobe -r null_blk
+ lsmod
+ grep null_blk
++ nproc
+ make -j 48 M=drivers/block modules
+ HOST=drivers/block/null_blk/
++ uname -r
+ HOST_DEST=/lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk/
+ cp drivers/block/null_blk//null_blk.ko /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//
+ ls -lrth /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
-rw-r--r--. 1 root root 1.2M Oct  5 21:57 /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
+ sleep 1
+ dmesg -c
# modprobe null_blk gb=1 zoned=1 zone_size=128 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk 
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom  
nullb0  250:0    0    1G  0 disk 
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk 
nvme0n1 259:0    0    1G  0 disk 
# blkzone reset /dev/nullb0
# dmesg  -c
[  397.079221] null_blk: disk nullb0 created
[  397.079226] null_blk: module loaded
[  406.626500] blkdev_zone_reset_all 237 <---
# modprobe -r null_blk
# 
# 
# 
# modprobe null_blk gb=1 zoned=1 zone_size=128 zone_reset_all=0 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk 
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom  
nullb0  250:0    0    1G  0 disk 
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk 
nvme0n1 259:0    0    1G  0 disk 
# blkzone reset /dev/nullb0
# dmesg  -c
[  425.456187] null_blk: disk nullb0 created
[  425.456192] null_blk: module loaded
[  438.419529] blkdev_zone_reset_all_emulated 197 <---
# modprobe -r null_blk
# 
---
 drivers/block/null_blk/main.c     | 5 +++++
 drivers/block/null_blk/null_blk.h | 1 +
 drivers/block/null_blk/zoned.c    | 3 ++-
 3 files changed, 8 insertions(+), 1 deletion(-)
drivers/block/null_blk/main.c +5 -0
index 8b7f42024f14..a0572e6c28ce 100644
@@ -260,6 +260,10 @@ static unsigned int g_zone_max_active;
 module_param_named(zone_max_active, g_zone_max_active, uint, 0444);
 MODULE_PARM_DESC(zone_max_active, "Maximum number of active zones when block device is zoned. Default: 0 (no limit)");
 
+static bool g_zone_reset_all = true;
+module_param_named(zone_reset_all, g_zone_reset_all, bool, 0444);
+MODULE_PARM_DESC(zone_reset_all, "Allow REQ_OP_ZONE_RESET_ALL. Default: true");
+
 static struct nullb_device *null_alloc_dev(void);
 static void null_free_dev(struct nullb_device *dev);
 static void null_del_dev(struct nullb *nullb);
@@ -715,6 +719,7 @@ static struct nullb_device *null_alloc_dev(void)
 	dev->zone_nr_conv = g_zone_nr_conv;
 	dev->zone_max_open = g_zone_max_open;
 	dev->zone_max_active = g_zone_max_active;
+	dev->zone_reset_all = g_zone_reset_all;
 	dev->virt_boundary = g_virt_boundary;
 	dev->no_sched = g_no_sched;
 	dev->shared_tag_bitmap = g_shared_tag_bitmap;
drivers/block/null_blk/null_blk.h +1 -0
index e692c2a7369e..e7efe8de4ebf 100644
@@ -115,6 +115,7 @@ struct nullb_device {
 	bool discard; /* if support discard */
 	bool write_zeroes; /* if support write_zeroes */
 	bool zoned; /* if device is zoned */
+	bool zone_reset_all; /* if support REQ_OP_ZONE_RESET_ALL */
 	bool virt_boundary; /* virtual boundary on/off for the device */
 	bool no_sched; /* no IO scheduler for the device */
 	bool shared_tag_bitmap; /* use hostwide shared tags */
drivers/block/null_blk/zoned.c +2 -1
index 55a69e48ef8b..7310d1c3f9ec 100644
@@ -160,7 +160,8 @@ int null_register_zoned_dev(struct nullb *nullb)
 	struct request_queue *q = nullb->q;
 
 	disk_set_zoned(nullb->disk, BLK_ZONED_HM);
-	blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
+	if (dev->zone_reset_all)
+		blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
 	blk_queue_required_elevator_features(q, ELEVATOR_F_ZBD_SEQ_WRITE);
 
 	if (queue_is_mq(q)) {
Signature · 1 line
2.29.0
#1 lore raw
Quoted 106 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

---

# git log -1 
commit 8ca0bd53a9c9e2f58c4fc9e38e3f5f82d26d3851 (HEAD -> for-next)
Author: Chaitanya Kulkarni <[email protected]>
Date:   Wed Oct 5 21:57:00 2022 -0700

null_blk: allow REQ_OP_ZONE_RESET_ALL to configure

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

# ./compile_nullb.sh 
+ umount /mnt/nullb0
umount: /mnt/nullb0: not mounted.
+ rmdir 'config/nullb/nullb*'
rmdir: failed to remove 'config/nullb/nullb*': No such file or directory
+ dmesg -c
+ modprobe -r null_blk
+ lsmod
+ grep null_blk
++ nproc
+ make -j 48 M=drivers/block modules
+ HOST=drivers/block/null_blk/
++ uname -r
+ HOST_DEST=/lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk/
+ cp drivers/block/null_blk//null_blk.ko /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//
+ ls -lrth /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
-rw-r--r--. 1 root root 1.2M Oct  5 21:57 /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
+ sleep 1
+ dmesg -c
# modprobe null_blk gb=1 zoned=1 zone_size=128 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk 
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom  
nullb0  250:0    0    1G  0 disk 
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk 
nvme0n1 259:0    0    1G  0 disk 
# blkzone reset /dev/nullb0
# dmesg  -c
[  397.079221] null_blk: disk nullb0 created
[  397.079226] null_blk: module loaded
[  406.626500] blkdev_zone_reset_all 237 <---
# modprobe -r null_blk
# 
# 
# 
# modprobe null_blk gb=1 zoned=1 zone_size=128 zone_reset_all=0 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk 
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom  
nullb0  250:0    0    1G  0 disk 
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk 
nvme0n1 259:0    0    1G  0 disk 
# blkzone reset /dev/nullb0
# dmesg  -c
[  425.456187] null_blk: disk nullb0 created
[  425.456192] null_blk: module loaded
[  438.419529] blkdev_zone_reset_all_emulated 197 <---
# modprobe -r null_blk
# 
---
 drivers/block/null_blk/main.c     | 5 +++++
 drivers/block/null_blk/null_blk.h | 1 +
 drivers/block/null_blk/zoned.c    | 3 ++-
 3 files changed, 8 insertions(+), 1 deletion(-)
drivers/block/null_blk/main.c +0 -0
index 8b7f42024f14..a0572e6c28ce 100644

drivers/block/null_blk/main.c static unsigned int g_zone_max_active; · line 264

Quoted 5 lines
drivers/block/null_blk/main.c +2 -0
@@ -260,6 +260,10 @@ static unsigned int g_zone_max_active;
 module_param_named(zone_max_active, g_zone_max_active, uint, 0444);
 MODULE_PARM_DESC(zone_max_active, "Maximum number of active zones when block device is zoned. Default: 0 (no limit)");
 
+static bool g_zone_reset_all = true;
+module_param_named(zone_reset_all, g_zone_reset_all, bool, 0444);

Nit: Why read-only ? You can make it writable without any issue, no ?

Quoted 39 lines
drivers/block/null_blk/main.c +3 -0
+MODULE_PARM_DESC(zone_reset_all, "Allow REQ_OP_ZONE_RESET_ALL. Default: true");
+
 static struct nullb_device *null_alloc_dev(void);
 static void null_free_dev(struct nullb_device *dev);
 static void null_del_dev(struct nullb *nullb);
@@ -715,6 +719,7 @@ static struct nullb_device *null_alloc_dev(void)
 	dev->zone_nr_conv = g_zone_nr_conv;
 	dev->zone_max_open = g_zone_max_open;
 	dev->zone_max_active = g_zone_max_active;
+	dev->zone_reset_all = g_zone_reset_all;
 	dev->virt_boundary = g_virt_boundary;
 	dev->no_sched = g_no_sched;
 	dev->shared_tag_bitmap = g_shared_tag_bitmap;
drivers/block/null_blk/null_blk.h +1 -0
index e692c2a7369e..e7efe8de4ebf 100644
@@ -115,6 +115,7 @@ struct nullb_device {
 	bool discard; /* if support discard */
 	bool write_zeroes; /* if support write_zeroes */
 	bool zoned; /* if device is zoned */
+	bool zone_reset_all; /* if support REQ_OP_ZONE_RESET_ALL */
 	bool virt_boundary; /* virtual boundary on/off for the device */
 	bool no_sched; /* no IO scheduler for the device */
 	bool shared_tag_bitmap; /* use hostwide shared tags */
drivers/block/null_blk/zoned.c +2 -1
index 55a69e48ef8b..7310d1c3f9ec 100644
@@ -160,7 +160,8 @@ int null_register_zoned_dev(struct nullb *nullb)
 	struct request_queue *q = nullb->q;
 
 	disk_set_zoned(nullb->disk, BLK_ZONED_HM);
-	blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
+	if (dev->zone_reset_all)
+		blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
 	blk_queue_required_elevator_features(q, ELEVATOR_F_ZBD_SEQ_WRITE);
 
 	if (queue_is_mq(q)) {
Signature · 2 lines
Damien Le Moal
Western Digital Research
#2 lore raw
Quoted 110 lines · Damien Le Moal
Quoted 109 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

---

# git log -1
commit 8ca0bd53a9c9e2f58c4fc9e38e3f5f82d26d3851 (HEAD -> for-next)
Author: Chaitanya Kulkarni <[email protected]>
Date:   Wed Oct 5 21:57:00 2022 -0700

null_blk: allow REQ_OP_ZONE_RESET_ALL to configure

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

# ./compile_nullb.sh
+ umount /mnt/nullb0
umount: /mnt/nullb0: not mounted.
+ rmdir 'config/nullb/nullb*'
rmdir: failed to remove 'config/nullb/nullb*': No such file or directory
+ dmesg -c
+ modprobe -r null_blk
+ lsmod
+ grep null_blk
++ nproc
+ make -j 48 M=drivers/block modules
+ HOST=drivers/block/null_blk/
++ uname -r
+ HOST_DEST=/lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk/
+ cp drivers/block/null_blk//null_blk.ko /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//
+ ls -lrth /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
-rw-r--r--. 1 root root 1.2M Oct  5 21:57 /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
+ sleep 1
+ dmesg -c
# modprobe null_blk gb=1 zoned=1 zone_size=128 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom
nullb0  250:0    0    1G  0 disk
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk
nvme0n1 259:0    0    1G  0 disk
# blkzone reset /dev/nullb0
# dmesg  -c
[  397.079221] null_blk: disk nullb0 created
[  397.079226] null_blk: module loaded
[  406.626500] blkdev_zone_reset_all 237 <---
# modprobe -r null_blk
#
#
#
# modprobe null_blk gb=1 zoned=1 zone_size=128 zone_reset_all=0 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom
nullb0  250:0    0    1G  0 disk
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk
nvme0n1 259:0    0    1G  0 disk
# blkzone reset /dev/nullb0
# dmesg  -c
[  425.456187] null_blk: disk nullb0 created
[  425.456192] null_blk: module loaded
[  438.419529] blkdev_zone_reset_all_emulated 197 <---
# modprobe -r null_blk
#
---
  drivers/block/null_blk/main.c     | 5 +++++
  drivers/block/null_blk/null_blk.h | 1 +
  drivers/block/null_blk/zoned.c    | 3 ++-
  3 files changed, 8 insertions(+), 1 deletion(-)
drivers/block/null_blk/main.c +0 -0
index 8b7f42024f14..a0572e6c28ce 100644
@@ -260,6 +260,10 @@ static unsigned int g_zone_max_active;
  module_param_named(zone_max_active, g_zone_max_active, uint, 0444);
  MODULE_PARM_DESC(zone_max_active, "Maximum number of active zones when block device is zoned. Default: 0 (no limit)");
 
Quoted 5 lines

drivers/block/null_blk/main.c static unsigned int g_zone_max_active; · line 264

Quoted 2 lines
drivers/block/null_blk/main.c +2 -0
+static bool g_zone_reset_all = true;
+module_param_named(zone_reset_all, g_zone_reset_all, bool, 0444);

Nit: Why read-only ? You can make it writable without any issue, no ?

Sorry I didn't understand this comment. In case this comment is about new zone_reset_all modpram being readonly and not allowed to set via command line? The test log shows I was able to set it to 0 and trigger call to blkdev_zone_reset_all_emulated() and default behavior ends up calling blkdev_zone_reset_all().

-ck

#3 lore raw
Quoted 117 lines · Chaitanya Kulkarni
Quoted 110 lines · Damien Le Moal
Quoted 109 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

---

# git log -1
commit 8ca0bd53a9c9e2f58c4fc9e38e3f5f82d26d3851 (HEAD -> for-next)
Author: Chaitanya Kulkarni <[email protected]>
Date:   Wed Oct 5 21:57:00 2022 -0700

null_blk: allow REQ_OP_ZONE_RESET_ALL to configure

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

# ./compile_nullb.sh
+ umount /mnt/nullb0
umount: /mnt/nullb0: not mounted.
+ rmdir 'config/nullb/nullb*'
rmdir: failed to remove 'config/nullb/nullb*': No such file or directory
+ dmesg -c
+ modprobe -r null_blk
+ lsmod
+ grep null_blk
++ nproc
+ make -j 48 M=drivers/block modules
+ HOST=drivers/block/null_blk/
++ uname -r
+ HOST_DEST=/lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk/
+ cp drivers/block/null_blk//null_blk.ko /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//
+ ls -lrth /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
-rw-r--r--. 1 root root 1.2M Oct  5 21:57 /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
+ sleep 1
+ dmesg -c
# modprobe null_blk gb=1 zoned=1 zone_size=128 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom
nullb0  250:0    0    1G  0 disk
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk
nvme0n1 259:0    0    1G  0 disk
# blkzone reset /dev/nullb0
# dmesg  -c
[  397.079221] null_blk: disk nullb0 created
[  397.079226] null_blk: module loaded
[  406.626500] blkdev_zone_reset_all 237 <---
# modprobe -r null_blk
#
#
#
# modprobe null_blk gb=1 zoned=1 zone_size=128 zone_reset_all=0 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom
nullb0  250:0    0    1G  0 disk
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk
nvme0n1 259:0    0    1G  0 disk
# blkzone reset /dev/nullb0
# dmesg  -c
[  425.456187] null_blk: disk nullb0 created
[  425.456192] null_blk: module loaded
[  438.419529] blkdev_zone_reset_all_emulated 197 <---
# modprobe -r null_blk
#
---
  drivers/block/null_blk/main.c     | 5 +++++
  drivers/block/null_blk/null_blk.h | 1 +
  drivers/block/null_blk/zoned.c    | 3 ++-
  3 files changed, 8 insertions(+), 1 deletion(-)
drivers/block/null_blk/main.c +0 -0
index 8b7f42024f14..a0572e6c28ce 100644
@@ -260,6 +260,10 @@ static unsigned int g_zone_max_active;
  module_param_named(zone_max_active, g_zone_max_active, uint, 0444);
  MODULE_PARM_DESC(zone_max_active, "Maximum number of active zones when block device is zoned. Default: 0 (no limit)");
 
Quoted 5 lines

drivers/block/null_blk/main.c static unsigned int g_zone_max_active; · line 264

Quoted 2 lines
drivers/block/null_blk/main.c +2 -0
+static bool g_zone_reset_all = true;
+module_param_named(zone_reset_all, g_zone_reset_all, bool, 0444);

Nit: Why read-only ? You can make it writable without any issue, no ?

Quoted 8 lines

Sorry I didn't understand this comment. In case this comment is about new zone_reset_all modpram being readonly and not allowed to set via command line? The test log shows I was able to set it to 0 and trigger call to blkdev_zone_reset_all_emulated() and default behavior ends up calling blkdev_zone_reset_all().

-ck

If you change the param mode to 644, one can do:

echo 0 >/sys/module/null_blk/parameters/zone_reset_all or
echo 1 >/sys/module/null_blk/parameters/zone_reset_all

But given that all null_blk parameters are read-only, not a big deal.

Signature · 2 lines
Damien Le Moal
Western Digital Research
#4 lore raw

Damien,

MODULE_PARM_DESC(zone_max_active, "Maximum number of active zones when
block device is zoned. Default: 0 (no limit)");

Quoted 16 lines
Quoted 7 lines
Quoted 6 lines

drivers/block/null_blk/main.c static unsigned int g_zone_max_active; · line 264

Quoted 3 lines
drivers/block/null_blk/main.c +2 -0
+static bool g_zone_reset_all = true;
+module_param_named(zone_reset_all, g_zone_reset_all, bool, 0444);

Nit: Why read-only ? You can make it writable without any issue, no ?

Quoted 8 lines

Sorry I didn't understand this comment. In case this comment is about new zone_reset_all modpram being readonly and not allowed to set via command line? The test log shows I was able to set it to 0 and trigger call to blkdev_zone_reset_all_emulated() and default behavior ends up calling blkdev_zone_reset_all().

-ck

Quoted 8 lines

If you change the param mode to 644, one can do:

echo 0 >/sys/module/null_blk/parameters/zone_reset_all or
echo 1 >/sys/module/null_blk/parameters/zone_reset_all

But given that all null_blk parameters are read-only, not a big deal.

if you are okay with this can I get your reviewed-by ?

-ck

Reviewed-by on v1 1/1
#5 lore raw
In reply to Chaitanya Kulkarni
Quoted 9 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Quoted 5 lines

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

Quoted 136 lines

---

# git log -1 
commit 8ca0bd53a9c9e2f58c4fc9e38e3f5f82d26d3851 (HEAD -> for-next)
Author: Chaitanya Kulkarni <[email protected]>
Date:   Wed Oct 5 21:57:00 2022 -0700

null_blk: allow REQ_OP_ZONE_RESET_ALL to configure

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

# ./compile_nullb.sh 
+ umount /mnt/nullb0
umount: /mnt/nullb0: not mounted.
+ rmdir 'config/nullb/nullb*'
rmdir: failed to remove 'config/nullb/nullb*': No such file or directory
+ dmesg -c
+ modprobe -r null_blk
+ lsmod
+ grep null_blk
++ nproc
+ make -j 48 M=drivers/block modules
+ HOST=drivers/block/null_blk/
++ uname -r
+ HOST_DEST=/lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk/
+ cp drivers/block/null_blk//null_blk.ko /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//
+ ls -lrth /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
-rw-r--r--. 1 root root 1.2M Oct  5 21:57 /lib/modules/6.0.0-rc7blk+/kernel/drivers/block/null_blk//null_blk.ko
+ sleep 1
+ dmesg -c
# modprobe null_blk gb=1 zoned=1 zone_size=128 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk 
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom  
nullb0  250:0    0    1G  0 disk 
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk 
nvme0n1 259:0    0    1G  0 disk 
# blkzone reset /dev/nullb0
# dmesg  -c
[  397.079221] null_blk: disk nullb0 created
[  397.079226] null_blk: module loaded
[  406.626500] blkdev_zone_reset_all 237 <---
# modprobe -r null_blk
# 
# 
# 
# modprobe null_blk gb=1 zoned=1 zone_size=128 zone_reset_all=0 <---
# lsblk
NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda       8:0    0   50G  0 disk 
├─sda1    8:1    0    1G  0 part /boot
└─sda2    8:2    0   49G  0 part /home
sdb       8:16   0  100G  0 disk /mnt/data
sr0      11:0    1 1024M  0 rom  
nullb0  250:0    0    1G  0 disk 
zram0   251:0    0    8G  0 disk [SWAP]
vda     252:0    0  512M  0 disk 
nvme0n1 259:0    0    1G  0 disk 
# blkzone reset /dev/nullb0
# dmesg  -c
[  425.456187] null_blk: disk nullb0 created
[  425.456192] null_blk: module loaded
[  438.419529] blkdev_zone_reset_all_emulated 197 <---
# modprobe -r null_blk
# 
---
 drivers/block/null_blk/main.c     | 5 +++++
 drivers/block/null_blk/null_blk.h | 1 +
 drivers/block/null_blk/zoned.c    | 3 ++-
 3 files changed, 8 insertions(+), 1 deletion(-)
drivers/block/null_blk/main.c +5 -0
index 8b7f42024f14..a0572e6c28ce 100644
@@ -260,6 +260,10 @@ static unsigned int g_zone_max_active;
 module_param_named(zone_max_active, g_zone_max_active, uint, 0444);
 MODULE_PARM_DESC(zone_max_active, "Maximum number of active zones when block device is zoned. Default: 0 (no limit)");
 
+static bool g_zone_reset_all = true;
+module_param_named(zone_reset_all, g_zone_reset_all, bool, 0444);
+MODULE_PARM_DESC(zone_reset_all, "Allow REQ_OP_ZONE_RESET_ALL. Default: true");
+
 static struct nullb_device *null_alloc_dev(void);
 static void null_free_dev(struct nullb_device *dev);
 static void null_del_dev(struct nullb *nullb);
@@ -715,6 +719,7 @@ static struct nullb_device *null_alloc_dev(void)
 	dev->zone_nr_conv = g_zone_nr_conv;
 	dev->zone_max_open = g_zone_max_open;
 	dev->zone_max_active = g_zone_max_active;
+	dev->zone_reset_all = g_zone_reset_all;
 	dev->virt_boundary = g_virt_boundary;
 	dev->no_sched = g_no_sched;
 	dev->shared_tag_bitmap = g_shared_tag_bitmap;
drivers/block/null_blk/null_blk.h +1 -0
index e692c2a7369e..e7efe8de4ebf 100644
@@ -115,6 +115,7 @@ struct nullb_device {
 	bool discard; /* if support discard */
 	bool write_zeroes; /* if support write_zeroes */
 	bool zoned; /* if device is zoned */
+	bool zone_reset_all; /* if support REQ_OP_ZONE_RESET_ALL */
 	bool virt_boundary; /* virtual boundary on/off for the device */
 	bool no_sched; /* no IO scheduler for the device */
 	bool shared_tag_bitmap; /* use hostwide shared tags */
drivers/block/null_blk/zoned.c +2 -1
index 55a69e48ef8b..7310d1c3f9ec 100644
@@ -160,7 +160,8 @@ int null_register_zoned_dev(struct nullb *nullb)
 	struct request_queue *q = nullb->q;
 
 	disk_set_zoned(nullb->disk, BLK_ZONED_HM);
-	blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
+	if (dev->zone_reset_all)
+		blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
 	blk_queue_required_elevator_features(q, ELEVATOR_F_ZBD_SEQ_WRITE);
 
 	if (queue_is_mq(q)) {
Signature · 2 lines
Damien Le Moal
Western Digital Research
#6 lore raw
In reply to Chaitanya Kulkarni
Quoted 9 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Quoted 5 lines

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

For the sake of completeness would it be reasonable to also provide a means to set this option via configfs?

Vincent

#7 lore raw
Quoted 14 lines · Vincent Fu
Quoted 13 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

Quoted 6 lines
Quoted 1 line

For the sake of completeness would it be reasonable to also provide a means to set this option via configfs?

Vincent

I deliberately avoided that as I don't see any need for it as of now. but if it is a hard requirement I Can certainly send V2 with configfs param.

-ck

#8 lore raw
Quoted 20 lines · Chaitanya Kulkarni
Quoted 19 lines · Vincent Fu
Quoted 9 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Quoted 5 lines

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

For the sake of completeness would it be reasonable to also provide a means to set this option via configfs?

Quoted 7 lines
Quoted 1 line

Vincent

I deliberately avoided that as I don't see any need for it as of now. but if it is a hard requirement I Can certainly send V2 with configfs param.

-ck

I think the default should be to have features available as both module options and via configfs unless there are good reasons to make an exception.

https://lore.kernel.org/linux-block/[email protected]/
https://lore.kernel.org/linux-block/[email protected]/

Vincent

maintainer
#9 lore raw
Quoted 27 lines · Vincent Fu
Quoted 26 lines · Chaitanya Kulkarni
Quoted 14 lines · Vincent Fu
Quoted 13 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

Quoted 6 lines
Quoted 1 line

For the sake of completeness would it be reasonable to also provide a means to set this option via configfs?

Vincent

I deliberately avoided that as I don't see any need for it as of now. but if it is a hard requirement I Can certainly send V2 with configfs param.

Quoted 8 lines
Quoted 1 line

-ck

I think the default should be to have features available as both module options and via configfs unless there are good reasons to make an exception.

https://lore.kernel.org/linux-block/[email protected]/
https://lore.kernel.org/linux-block/[email protected]/

Yes agree, work was done to bring those closer, would be silly to add something new and NOT have both available.

Signature · 1 line
Jens Axboe
#10 lore raw
Quoted 34 lines · Jens Axboe
Quoted 33 lines · Vincent Fu
Quoted 20 lines · Chaitanya Kulkarni
Quoted 19 lines · Vincent Fu
Quoted 9 lines · Chaitanya Kulkarni

For a Zoned Block Device zone reset all is emulated if underlaying device doesn't support REQ_OP_ZONE_RESET_ALL operation. In null_blk Zoned mode there is no way to test zone reset all emulation present in the block layer since we enable it by default :-

blkdev_zone_mgmt()
blkdev_zone_reset_all_emulation() <---
blkdev_zone_reset_all()

Quoted 5 lines

Add a module parameter zone_reset_all to enable or disable REQ_OP_ZONE_RESET_ALL, enable it by default to retain the existing behaviour.

For the sake of completeness would it be reasonable to also provide a means to set this option via configfs?

Quoted 7 lines
Quoted 1 line

Vincent

I deliberately avoided that as I don't see any need for it as of now. but if it is a hard requirement I Can certainly send V2 with configfs param.

-ck

I think the default should be to have features available as both module options and via configfs unless there are good reasons to make an exception.

Quoted 6 lines
Quoted 2 lines

https://lore.kernel.org/linux-block/[email protected]/
https://lore.kernel.org/linux-block/[email protected]/

Yes agree, work was done to bring those closer, would be silly to add something new and NOT have both available.

I was trying to avoid polluting configfs space with tons of params guess which are only required for blktetsts, will send out V2 with added parameter.

-ck