Re: [cocci] Searching for special function implementations with SmPL

[Julia Lawall <julia.lawall@inria.fr>]

On Mon, 30 Jan 2023, Luca Coelho wrote:

> On Mon, 2023-01-30 at 09:37 +0100, Julia Lawall wrote:
> > > Sorry, false alarm.  I was running spatch by itself on an inner
> > > directory (.../i915/display), but when I used find, I was using
> > > .../i915, so there was one more match happening there.
> > >
> > > So, it seems that the latest head _does_ work fine! And I'm gettign
> > > all
> > > the matches and changes that I expect.  Thanks!
> >
> > Great news :)
>
> Maybe I celebrated a bit too early.  There is still one instance that
> is not matching...

Probably it's a parse error.

Try putting the following in your semantic patch, egin the metavariable
list of macros_noargs:

iterator name with_intel_display_power_if_enabled;

If that is still not sufficient, try running

spatch --parse-c file.c

for the file of interest and see if it says BAD or bad on the lines of the
function that you expect should be matched.

julia

>
> I currently have these rules:
>
> @macros_noargs@
> identifier m;
> expression e =~ "dev_priv";
> @@
> #define m <+...e...+>
>
> @nested_macros@
> identifier macros_noargs.m;
> identifier nm;
> identifier list il;
> @@
> #define nm(il) <+...m...+>
>
> @@
> identifier nested_macros.nm;
> identifier dev_priv, f;
> expression list el;
> @@
> f(...) {
> 	...
> 	struct drm_i915_private *dev_priv = ...;
>
> 	<+...
> 	nm(
> +	dev_priv,
> 	el)
> 	...+>
> }
>
> @@
> identifier nested_macros.nm;
> identifier dev_priv, f;
> expression list el;
> @@
> f(..., struct drm_i915_private *dev_priv, ...) {
> 	<+...
> 	nm(
> +	dev_priv,
> 	el)
> 	...+>
> }
>
> And then we have this function in
> drivers/gpu/drm/i915/display/intel_display.c:
>
> void vlv_wait_port_ready(struct drm_i915_private *dev_priv,
> 			 struct intel_digital_port *dig_port,
> 			 unsigned int expected_mask)
> {
> [...]
> 	switch (dig_port->base.port) {
> 	default:
> 		MISSING_CASE(dig_port->base.port);
> 		fallthrough;
> 	case PORT_B:
> 		port_mask = DPLL_PORTB_READY_MASK;
> 		dpll_reg = DPLL(0);
> 		break;
> 	case PORT_C:
> 		port_mask = DPLL_PORTC_READY_MASK;
> 		dpll_reg = DPLL(0);
> 		expected_mask <<= 4;
> 		break;
> [...]
> }
>
> The DPLL macro is defined like this (in
> drivers/gpu/drm/i915/display/i915_reg.h):
>
> #define _DPLL_A (DISPLAY_MMIO_BASE(dev_priv) + 0x6014)
> #define _DPLL_B (DISPLAY_MMIO_BASE(dev_priv) + 0x6018)
> #define _CHV_DPLL_C (DISPLAY_MMIO_BASE(dev_priv) + 0x6030)
> #define DPLL(pipe) _MMIO_PIPE3((pipe), _DPLL_A, _DPLL_B, _CHV_DPLL_C)
>
>
> So it should match like the other instances, but it doesn't.  I copied
> the macro definition and the function to stand-alone files (i.e.
> without the rest of the code) and it matches, but in the real code,
> running cocci like this, it doesn't:
>
> ~/public/coccinelle/spatch.opt --sp-file ~/dev_priv_i915.spatch --all-includes intel_display.c
>
> I noticed some parsing errors when using --debug (output attached), so
> maybe there's some parsing issues with the real file that prevents the
> rule from applying?
>
> Any clues?
>
> --
> Cheers,
> Luca.
>