Re: [Intel-wired-lan] [PATCH net-next 1/3] net: provide macros for commonly copied lockless queue stop/wake code

Re: [Intel-wired-lan] [PATCH net-next 1/3] net: provide macros for commonly copied lockless queue stop/wake code

[Jakub Kicinski]

CC: maintainers, in case there isn't a repost
https://lore.kernel.org/all/20230322233028.269410-1-kuba@kernel.org/

On Thu, 23 Mar 2023 01:35:34 +0100 Andrew Lunn wrote:
> On Wed, Mar 22, 2023 at 04:30:26PM -0700, Jakub Kicinski wrote:
> > A lot of drivers follow the same scheme to stop / start queues
> > without introducing locks between xmit and NAPI tx completions.
> > I'm guessing they all copy'n'paste each other's code.
> >
> > Smaller drivers shy away from the scheme and introduce a lock
> > which may cause deadlocks in netpoll.  
> 
> I notice there is no patch 0/X. Seems like the above would be good
> material for it, along with a comment that a few drivers are converted
> to make use of the new macros.

Then do I repeat the same text in the commit? Or cut the commit down?
Doesn't that just take away information from the commit which will
show up in git blame?

Having a cover letter is a good default, and required if the series 
is a larger change decomposed into steps. But here there is a major
change and a bunch of loose conversions. More sample users than
meaningful part.

LMK what your preference for splitting this info is, I'm unsure.
_______________________________________________
Intel-wired-lan mailing list
Intel-wired-lan@osuosl.org
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

Re: [Intel-wired-lan] [PATCH net-next 1/3] net: provide macros for commonly copied lockless queue stop/wake code

[Andrew Lunn]

On Wed, Mar 22, 2023 at 06:04:06PM -0700, Jakub Kicinski wrote:
> CC: maintainers, in case there isn't a repost
> https://lore.kernel.org/all/20230322233028.269410-1-kuba@kernel.org/
> 
> On Thu, 23 Mar 2023 01:35:34 +0100 Andrew Lunn wrote:
> > On Wed, Mar 22, 2023 at 04:30:26PM -0700, Jakub Kicinski wrote:
> > > A lot of drivers follow the same scheme to stop / start queues
> > > without introducing locks between xmit and NAPI tx completions.
> > > I'm guessing they all copy'n'paste each other's code.
> > >
> > > Smaller drivers shy away from the scheme and introduce a lock
> > > which may cause deadlocks in netpoll.  
> > 
> > I notice there is no patch 0/X. Seems like the above would be good
> > material for it, along with a comment that a few drivers are converted
> > to make use of the new macros.
> 
> Then do I repeat the same text in the commit? Or cut the commit down?
> Doesn't that just take away information from the commit which will
> show up in git blame?
> 
> Having a cover letter is a good default, and required if the series 
> is a larger change decomposed into steps. But here there is a major
> change and a bunch of loose conversions. More sample users than
> meaningful part.
> 
> LMK what your preference for splitting this info is, I'm unsure.

We do seem to have a policy of asking for a 0/X. And it is used for
the merge commit. That is my real point. And i don't see why the text
can be repeated in the merge commit and the individual commits.

    Andrew
_______________________________________________
Intel-wired-lan mailing list
Intel-wired-lan@osuosl.org
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

Re: [Intel-wired-lan] [PATCH net-next 1/3] net: provide macros for commonly copied lockless queue stop/wake code

[Jakub Kicinski]

On Thu, 23 Mar 2023 22:02:29 +0100 Andrew Lunn wrote:
> We do seem to have a policy of asking for a 0/X. And it is used for
> the merge commit. That is my real point. And i don't see why the text
> can be repeated in the merge commit and the individual commits.

Alright, I'll respin and add a cover. But I think we should be critical
of the rules. The rules are just a mental shortcut, we shouldn't lose
sight of why they are in place and maintain some flexibility.
_______________________________________________
Intel-wired-lan mailing list
Intel-wired-lan@osuosl.org
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan