Re: [cocci] Checking data structure member accesses after function return value assignments?

On Mon, 19 Jan 2026, Markus Elfring wrote:

> …
> >> May I expect that the same data processing result can be achieved together with
> >> the application of an SmPL nest construct?
> …
> > Since you haven't told me what happens when you run this, I don't know
> > what the issue is.
>
> I assume that you can try to reproduce a questionable test result also with
> your current development software version.
>
>
> > But it may work better with () around the <+... ...+>, to avoid a parsing conflict.
>
> Why are you looking for extra parentheses there?

As I already said it is to avoid a parser conflict.  Both statements and
expressions can start with <+... ...+>.  But if there is a (), it can only
be an expression.

>
> Another SmPL script example:
> // See also:
> // [PATCH] regulator: fp9931: Add missing memory allocation check
> // https://lore.kernel.org/lkml/[email protected]/
> // https://lkml.org/lkml/2026/1/18/129
> @display@
> expression action, pointer;
> identifier member;
> @@
> *pointer = action(...);
> *<+... (pointer)->member ...+>
>  ;
>
>
> Questionable test result:
> Markus_Elfring@Sonne:…/Projekte/Linux/next-analyses> time /usr/bin/spatch --no-loops /home/altes_Heim2/elfring/Projekte/Coccinelle/janitor/show_questionable_statement_combination3.cocci drivers/regulator/fp9931.c
> [no diff output]

OK, then I don't know.  It seems like it was possible to parse trhe
semantic patch.

julia

>
> real    0m0,108s
> user    0m0,086s
> sys     0m0,015s
>
>
> Regards,
> Markus
>