Thread (6 messages) flat view 6 messages, 3 authors, 2021-02-03

Re: [PATCH net-next v6] net: psample: Introduce stubs to remove NIC driver dependency

From: Jakub Kicinski <kuba@kernel.org>
Date: 2021-02-02 21:25:37

On Mon, 01 Feb 2021 23:31:15 -0800 Saeed Mahameed wrote:
On Mon, 2021-02-01 at 17:14 -0800, Jakub Kicinski wrote:
quoted
On Mon,  1 Feb 2021 10:04:12 +0800 Chris Mi wrote:  
quoted
In order to send sampled packets to userspace, NIC driver calls
psample api directly. But it creates a hard dependency on module
psample. Introduce psample_ops to remove the hard dependency.
It is initialized when psample module is loaded and set to NULL
when the module is unloaded.

Reported-by: kernel test robot <redacted>
Signed-off-by: Chris Mi <redacted>
Reviewed-by: Jiri Pirko <redacted>  
The necessity of making this change is not obvious (you can fix the
distro scripts instead), you did not include a clarification in the
commit message even though two people asked you why it's needed and 
on top of that you keep sending code which doesn't build. 

Please consider this change rejected and do not send a v7.  
Jakub, it is not only about installation dependencies, the issue is
more critical than this, 

We had some other issues with similar dependency problem where
mlx5_core had hard dependency with netfilter, firewalld when disabled,
removes netfilter and all its dependencies including mlx5, this is a no
go for our users.

Again, having a hard dependency between a hardware driver and a
peripheral module in the kernel is a bad design.
That is not the point.

The technical problem is minor, and it's a problem for _your_ driver.
Yet, it appears to be my responsibility to make sure the patch even
compiles.

I believe there should be a limit to the ignorance a community
volunteer is expected to put up with when dealing with patches 
from and for the benefit of a commercial vendor.

This is up for discussion, if you disagree let's talk it out. I'm 
not particularly patient (to put it mildly), but I don't understand 
how v5 could have built, and yet v6 gets posted with the same exact
problem :/

So from my perspective it seems like the right step to push back.
If you, Tariq, Jiri, Ido or any other seasoned kernel contributor
reposts this after making sure it's up to snuff themselves I will
most certainly take a look / apply.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help