Re: [PATCH net-next] net: hns3: debugfs add dump tm info of nodes, priority and qset

[Jakub Kicinski <kuba@kernel.org>]

On Tue, 19 Jan 2021 11:14:51 +0800 Huazhong Tan wrote:
> On 2021/1/19 3:48, Jakub Kicinski wrote:
> > On Mon, 18 Jan 2021 19:20:23 +0800 Huazhong Tan wrote:  
> >> On 2021/1/17 10:23, Jakub Kicinski wrote:  
> >>> On Fri, 15 Jan 2021 15:09:29 +0800 Huazhong Tan wrote:  
> >>>> From: Guangbin Huang <huangguangbin2@huawei.com>
> >>>>
> >>>> To increase methods to dump more tm info, adds three debugfs commands
> >>>> to dump tm info of nodes, priority and qset. And a new tm file of debugfs
> >>>> is created for only dumping tm info.
> >>>>
> >>>> Unlike previous debugfs commands, to dump each tm information, user needs
> >>>> to enter two commands now. The first command writes parameters to tm and
> >>>> the second command reads info from tm. For examples, to dump tm info of
> >>>> priority 0, user needs to enter follow two commands:
> >>>> 1. echo dump priority 0 > tm
> >>>> 2. cat tm
> >>>>
> >>>> The reason for adding new tm file is because we accepted Jakub Kicinski's
> >>>> opinion as link https://lkml.org/lkml/2020/9/29/2101. And in order to
> >>>> avoid generating too many files, we implement write ops to allow user to
> >>>> input parameters.  
> >>> Why are you trying to avoid generating too many files? How many files
> >>> would it be? What's the size of each dump/file?  
> >> The maximum number of tm node, priority and qset are 8, 256,
> >> 1280, if we create a file for each one, then there are 8 node
> >> files, 256 priority files, 1280 qset files. It seems a little
> >> bit hard for using as well.  
> > Would the information not fit in one file per type with multiple rows?  
> 
> What you means is as below ?
> 
> estuary:/debugfs/hns3/0000:7d:00.0$ cat qset
> 
> qset id: 0
> QS map pri id: 0
> QS map link_vld: 1
> QS schedule mode: dwrr
> QS dwrr: 100
> 
> qset id: 1
> QS map pri id: 0
> QS map link_vld: 0
> QS schedule mode: sp
> QS dwrr: 0
> 
> ...

I was thinking more like:

ID	PRI	LINK_VLD	MODE	DWRR
0	0	1		dwrr	0
1	0	0		sp	0
...


but the exact format is up to you.

> For example, user want to query qset 1, then all qset info will be output,

I hope you don't mean end user when you say _user_.  This is debugfs,
it's intended for developers to debug issues and system dump to gather 
info at customer site.

> there are too many useless info.
> 
> So we add an interface to designage the specified id for node, priority 
> or qset.
> 
> > Can you show example outputs?  
> 
> 
> here is the output of this patch.
> 
> estuary:/debugfs/hns3/0000:7d:00.0$ echo dump qset 0 > tm
> estuary:/debugfs/hns3/0000:7d:00.0$ cat tm
> qset id: 0
> QS map pri id: 0
> QS map link_vld: 1
> QS schedule mode: dwrr
> QS dwrr: 100

Thanks. Not that much info per entry, as expected. Single file should
do nicely.