fix atomic limits check

[alan.adamson at oracle.com]

On 6/10/25 10:54 PM, Christoph Hellwig wrote:

> Hi all,
>
> this series tries to fix the atomics limit check to limit it to
> the per-controller values and to the controller probing.
>
> I think this should solve the root cause of the report from Yi Zhang,
> but needs new verification.
>
> Diffstat:
>   core.c |   84 ++++++++++++++++++++++++++++++-----------------------------------
>   nvme.h |    3 --
>   2 files changed, 40 insertions(+), 47 deletions(-)

Some testing with my qemu-nvme atomic write setup.  My qemu includes ns 
atomic parameters that isn't upstream yet.

-device nvme-subsys,id=subsys0 \
-device 
nvme,serial=deadbeef,id=nvme0,subsys=subsys0,atomic.dn=off,atomic.awun=31,atomic.awupf=15 
\
-drive id=ns1,file=/dev/nullb1,if=none \
-device nvme-ns,drive=ns1,bus=nvme0,nsid=1,zoned=false,shared=false \
-device 
nvme,serial=deadbeef,id=nvme1,subsys=subsys0,atomic.dn=off,atomic.awun=63,atomic.awupf=31 
\
-drive id=ns2,file=/dev/nullb2,if=none \
-device nvme-ns,drive=ns2,bus=nvme1,nsid=2,zoned=false,shared=false \
-device 
nvme,serial=deadbeef,id=nvme2,subsys=subsys0,atomic.dn=off,atomic.awun=15,atomic.awupf=7 
\
-drive id=ns3,file=/dev/nullb3,if=none \
-device 
nvme-ns,drive=ns3,bus=nvme2,nsid=3,zoned=false,shared=false,atomic.nawun=63,atomic.nawupf=31,atomic.nsfeat=true 
\
-device 
nvme,serial=deadbeef,id=nvme3,subsys=subsys0,atomic.dn=off,atomic.awun=15,atomic.awupf=7 
\
-drive id=ns4,file=/dev/nullb4,if=none \
-device 
nvme-ns,drive=ns4,bus=nvme3,nsid=4,zoned=false,shared=false,atomic.nawun=31,atomic.nawupf=15,atomic.nsfeat=true 
\
-drive id=ns5,file=/dev/nullb5,if=none \
-device 
nvme-ns,drive=ns5,bus=nvme3,nsid=5,zoned=false,shared=false,atomic.nawun=127,atomic.nawupf=63,atomic.nsfeat=true 
\
--nographic

A single subsystem with 4 controllers, 1 of the controllers has 2 
namespaces.

[root at localhost ~]# lsblk
NAME        MAJ:MIN RM  SIZE RO TYPE MOUNTPOINTS
sda           8:0    0   40G  0 disk
├─sda1        8:1    0    1G  0 part /boot
└─sda2        8:2    0   39G  0 part
   └─ol-root 252:0    0   39G  0 lvm  /
sr0          11:0    1 1024M  0 rom
nvme1n1     259:1    0  250G  0 disk
nvme1n2     259:5    0  250G  0 disk
nvme1n3     259:6    0  250G  0 disk
nvme1n4     259:7    0  250G  0 disk
nvme1n5     259:9    0  250G  0 disk


[root at localhost ~]# cat testxx.sh
set -x
nvme id-ctrl /dev/$1 | grep cmic
nvme id-ctrl /dev/$1 | grep awupf
nvme id-ns /dev/$1 | grep nawupf
cat /sys/block/$1/queue/atomic_write_max_bytes


[root at localhost ~]# sh testxx.sh nvme1n1
+ nvme id-ctrl /dev/nvme1n1
+ grep cmic
cmic      : 0x2
+ nvme id-ctrl /dev/nvme1n1
+ grep awupf
awupf     : 31
+ nvme id-ns /dev/nvme1n1
+ grep nawupf
nawupf  : 0
+ cat /sys/block/nvme1n1/queue/atomic_write_max_bytes
4096

[root at localhost ~]# sh testxx.sh nvme1n2
+ nvme id-ctrl /dev/nvme1n2
+ grep cmic
cmic      : 0x2
+ nvme id-ctrl /dev/nvme1n2
+ grep awupf
awupf     : 15
+ nvme id-ns /dev/nvme1n2
+ grep nawupf
nawupf  : 0
+ cat /sys/block/nvme1n2/queue/atomic_write_max_bytes
4096

[root at localhost ~]# sh testxx.sh nvme1n3
+ nvme id-ctrl /dev/nvme1n3
+ grep cmic
cmic      : 0x2
+ nvme id-ctrl /dev/nvme1n3
+ grep awupf
awupf     : 7
+ nvme id-ns /dev/nvme1n3
+ grep nawupf
nawupf  : 15
+ cat /sys/block/nvme1n3/queue/atomic_write_max_bytes
8192

[root at localhost ~]# sh testxx.sh nvme1n4
+ nvme id-ctrl /dev/nvme1n4
+ grep cmic
cmic      : 0x2
+ nvme id-ctrl /dev/nvme1n4
+ grep awupf
awupf     : 7
+ nvme id-ns /dev/nvme1n4
+ grep nawupf
nawupf  : 63
+ cat /sys/block/nvme1n4/queue/atomic_write_max_bytes
32768

[root at localhost ~]# sh testxx.sh nvme1n5
+ nvme id-ctrl /dev/nvme1n5
+ grep cmic
cmic      : 0x2
+ nvme id-ctrl /dev/nvme1n5
+ grep awupf
awupf     : 7
+ nvme id-ns /dev/nvme1n5
+ grep nawupf
nawupf  : 31
+ cat /sys/block/nvme1n5/queue/atomic_write_max_bytes
16384

Does this look right?

Before, we had a single atomic_write_max_bytes value per subsystem.  
That's not the case now.