Re: [PATCH net-next 1/6] bnxt_en: Check devlink allocation and registration status

On Thu, Sep 23, 2021 at 6:39 PM Jakub Kicinski [off-list ref] wrote:

On Fri, 24 Sep 2021 02:11:19 +0300 Leon Romanovsky wrote:

quoted

quoted

minor nit: There's obviously nothing incorrect about doing this (and
adding the additional error label in the cleanup code above), but bnxt
has generally adopted a style of having cleanup functions being
idempotent. It generally makes error handling simpler and less error
prone.

I would argue that opposite is true. Such "impossible" checks hide unwind
flow errors, missing releases e.t.c.

+1, fwiw

I appreciate that being more explicit can improve visibility, but it
does not make error handling inherently less error prone, nor is it
simpler (ie. the opposite isn't true). Idempotency is orthogonal to
unwind flow or the presence or not of a particular unwind handler (one
can still enforce either in review). But, if release handlers are
independent (most in bnxt are), then permitting other orderings can be
perfectly valid and places less burden on achieving the canonical form
for correctness (ie. usage is simpler and less error prone). That's
not to say we should throw caution to the wind and allow arbitrary
unwind flows, but it does mean certain mistakes don't result in actual
bugs. There are other flexibility benefits too. A single, unwind
everything, handler can be reused in more than one context.

That said, isn't the more important question what style and
assumptions the surrounding code has adopted? In this particular case,
I checked that this change wouldn't introduce the possibility of a
double unwind, but in other contexts in this driver code base,
changing error handling in this piecemeal way might actually introduce
a bug in contexts where the caller has assumed the overall function is
idempotent. Isn't local consistency of style a more important concern,
especially given that you are not predominantly responsible for
maintenance of this driver? Dealing with this exception to the norm in
our driver certainly places an additional burden on us to remember to
treat this particular case with special care. We should either rework
all of bnxt error handling to adopt the more accepted canonical form,
or we should adopt the surrounding conventions. What we shouldn't do
is mix approaches in one driver.

Regards,
Edwin Peer