From: Maxime Ripard <maxime@cerno.tech>
To: Julia Lawall <julia.lawall@inria.fr>
Cc: cocci@systeme.lip6.fr
Subject: Re: [Cocci] Performance issue with quite simple patch?
Date: Thu, 14 Jan 2021 14:24:53 +0100 [thread overview]
Message-ID: <20210114132453.trvqsjhx4j6gqbmj@gilmour> (raw)
In-Reply-To: <alpine.DEB.2.22.394.2101131152470.2930@hadrien>
[-- Attachment #1.1: Type: text/plain, Size: 3374 bytes --]
Hi Julia,
On Wed, Jan 13, 2021 at 11:58:41AM +0100, Julia Lawall wrote:
> On Wed, 13 Jan 2021, Maxime Ripard wrote:
>
> > Hi!
> >
> > I've been trying to get a patch to rename any variable called "state" in
> > a given set of callbacks.
> >
> > This is the patch that I've come up with:
> >
> > @ plane_atomic_func @
> > identifier helpers;
> > identifier func;
> > @@
> >
> > (
> > static const struct drm_plane_helper_funcs helpers = {
> > ...,
> > .atomic_check = func,
> > ...,
> > };
> > |
> > static const struct drm_plane_helper_funcs helpers = {
> > ...,
> > .atomic_disable = func,
> > ...,
> > };
> > |
> > static const struct drm_plane_helper_funcs helpers = {
> > ...,
> > .atomic_update = func,
> > ...,
> > };
> > )
>
> You don't need the ...s in the above. For structure declarations
> Coccinelle is happy as long as what you specify is a subset of what is
> present. The static and const aren't essential either. If you remove
> them, the pattern will match whethe thy are present or not.
Oh, that's good to know, thanks!
> >
> > @@
> > identifier plane_atomic_func.func;
> > symbol state;
> > expression e;
> > type T;
> > @@
> >
> > func(...)
> > {
> > ...
> > - T state = e;
> > + T plane_state = e;
> > <+...
> > - state
> > + plane_state
> > ...+>
> > }
> >
> > However, it seems like at least on a file (in Linux,
> > drivers/gpu/drm/atmel-hlcdc/atmel_hlcdc_plane.c), it takes quite big
> > performance hit with one CPU running at 100% until the timeout is hit.
> >
> > Replacing <+... by ... makes it work instantly, but doesn't really do
> > what I'm expecting, so I guess it's a matter of the patch being
> > subobtimal?
> >
> > Is there a more optimal way of doing it?
>
> In your rule, I donkt think that there is really any essential connection
> between the declaration and the use? You just want to change state to
> plane_state when it occurs in one of the functions that you detected.
I'm sorry, I forgot to mention that I only want the change to occur if
it's a variable declared in the function, not an argument to it
> So you could at least try the following and see if it gives any false positives:
>
> @@
> identifier plane_atomic_func.func;
> symbol state;
> expression e;
> type T;
> @@
>
> func(...)
> {
> <...
> (
> - T state = e;
> + T plane_state = e;
> |
> - state
> + plane_state
> )
> ...>
> }
I would never have thought of that, thanks :)
I tried to adjust it to match only on functions that have such a
variable defined:
@ plane_atomic_func @
identifier helpers;
identifier func;
@@
struct drm_plane_helper_funcs helpers = {
.atomic_check = func,
};
@ has_variable_state @
identifier plane_atomic_func.func;
symbol state;
expression e;
type T;
@@
func(...)
{
...
T state = e;
...
}
@ depends on has_variable_state @
identifier plane_atomic_func.func;
symbol state;
expression e;
type T;
@@
func(...)
{
<...
(
- T state = e;
+ T plane_state = e;
|
- state
+ plane_state
)
...>
}
But it's still choking on that same file
(drivers/gpu/drm//atmel-hlcdc/atmel_hlcdc_plane.c). Every other file
that matches takes at most a few seconds though, so it seems a bit weird
Maxime
[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]
[-- Attachment #2: Type: text/plain, Size: 136 bytes --]
_______________________________________________
Cocci mailing list
Cocci@systeme.lip6.fr
https://systeme.lip6.fr/mailman/listinfo/cocci
prev parent reply other threads:[~2021-01-14 13:25 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-13 9:48 [Cocci] Performance issue with quite simple patch? Maxime Ripard
2021-01-13 10:58 ` Julia Lawall
2021-01-14 13:24 ` Maxime Ripard [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20210114132453.trvqsjhx4j6gqbmj@gilmour \
--to=maxime@cerno.tech \
--cc=cocci@systeme.lip6.fr \
--cc=julia.lawall@inria.fr \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).