在 2023/11/1 1:22, Verma, Vishal L 写道:
On Mon, 2023-10-30 at 14:41 +0800, Cao, Quanquan/曹 全全 wrote:
[..]
After investigation, it was found that when disabling the region and
attempting to disable the same region again, the message "cxl region:
cmd_disable_region: disabled 1 region" is still returned.
I consider this to be unreasonable.
Test Example:
[root@fedora-37-client memory]# cxl list
[
{
"memdevs":[
{
"memdev":"mem0",
"ram_size":1073741824,
"serial":0,
"host":"0000:0d:00.0"
}
]
},
{
"regions":[
{
"region":"region0",
"resource":27111981056,
"size":1073741824,
"type":"ram",
"interleave_ways":1,
"interleave_granularity":256,
"decode_state":"commit"
}
]
}
]
[root@fedora-37-client ~]# cxl disable-region region0
cxl region: cmd_disable_region: disabled 1 region
[root@fedora-37-client ~]# cxl disable-region region0
cxl region: cmd_disable_region: disabled 1 region
expectation:cmd_disable_region: disabled 0 region
This is by design, I think it would be more confusing if the user asks
to disable-region, the response is "disabled 0 regions", and then finds
that the region is actually in the disabled state.
There is also precedent for this, as all disable-<foo> commands in
ndctl and cxl-cli behave the same way.
Perhaps a clarification in the man page makes sense noting this
behavior?
About this,i would like to ask if the design adopts the "idempotency"
pattern
