Skip to contentSkip to conversation

Checking the replacement of two specific function calls

Cocci Superseded Markus Elfring 2 participants 26 messages Apr 15, 2020 – Jan 6, 2023 4 messages not in the archive

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hello,

I noticed the suggestion “i2c: busses: convert to devm_platform_get_and_ioremap_resource”.
https://lore.kernel.org/linux-i2c/[email protected]/
https://lore.kernel.org/patchwork/patch/1223734/

I got into the mood to try another script out for the semantic patch language.

@replacement@
expression* base, device, resource;
@@
-resource = platform_get_resource(device, IORESOURCE_MEM, 0);
base =
- devm_ioremap_resource(&device->dev, resource)
+ devm_platform_get_and_ioremap_resource(device, 0, &resource)
;

Source file example:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-rcar.c?id=8632e9b5645bbc2331d21d892b0d6961c1a08429#n921
https://elixir.bootlin.com/linux/v5.7-rc1/source/drivers/i2c/busses/i2c-rcar.c#L921

// SPDX-License-Identifier: GPL-2.0
// deleted part
static int rcar_i2c_probe(struct platform_device *pdev)

{
	struct rcar_i2c_priv *priv;
	struct i2c_adapter *adap;
	struct device *dev = &pdev->dev;
// deleted part
	priv->res = platform_get_resource(pdev, IORESOURCE_MEM, 0);

	priv->io = devm_ioremap_resource(dev, priv->res);
	if (IS_ERR(priv->io))
		return PTR_ERR(priv->io);

// deleted part
}
// deleted part

Now I wonder why the expected patch is not generated by the software “Coccinelle 1.0.8-00029-ga549b9f0” at the moment.
How should this situation be improved?

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

maintainer
#1 lore raw
Quoted 36 lines · Markus Elfring

Hello,

I noticed the suggestion “i2c: busses: convert to devm_platform_get_and_ioremap_resource”.
https://lore.kernel.org/linux-i2c/[email protected]/
https://lore.kernel.org/patchwork/patch/1223734/

I got into the mood to try another script out for the semantic patch language.

@replacement@
expression* base, device, resource;
@@

(quoted diff) +1 -2
-resource = platform_get_resource(device, IORESOURCE_MEM, 0);
 base =
-       devm_ioremap_resource(&device->dev, resource)
+       devm_platform_get_and_ioremap_resource(device, 0, &resource)
 ;

Source file example:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-rcar.c?id=8632e9b5645bbc2331d21d892b0d6961c1a08429#n921
https://elixir.bootlin.com/linux/v5.7-rc1/source/drivers/i2c/busses/i2c-rcar.c#L921

// SPDX-License-Identifier: GPL-2.0
// deleted part
static int rcar_i2c_probe(struct platform_device *pdev)

{
	struct rcar_i2c_priv *priv;
	struct i2c_adapter *adap;
	struct device *dev = &pdev->dev;
// deleted part
	priv->res = platform_get_resource(pdev, IORESOURCE_MEM, 0);

	priv->io = devm_ioremap_resource(dev, priv->res);
	if (IS_ERR(priv->io))
		return PTR_ERR(priv->io);
Quoted 8 lines

// deleted part
}
// deleted part

Now I wonder why the expected patch is not generated by the software “Coccinelle 1.0.8-00029-ga549b9f0” at the moment.
How should this situation be improved?

Probably Coccinelle is not able to determine that priv->res and priv->io are pointers, if you have not given it enough encouragement to look for header files to find the defintion of the type of the priv structure.

I don't think that there is any need for the * in the metavariable declaration. If the various expressions are not pointers, the C compiler would have complained.

julia

Quoted 8 lines

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#2 lore raw
Quoted 3 lines
Quoted 3 lines

static int rcar_i2c_probe(struct platform_device *pdev)

{
	struct rcar_i2c_priv *priv;

…

Quoted 5 lines
Quoted 5 lines
	struct device *dev = &pdev->dev;
// deleted part
	priv->res = platform_get_resource(pdev, IORESOURCE_MEM, 0);

	priv->io = devm_ioremap_resource(dev, priv->res);

…

Quoted 1 line
Quoted 1 line

}

…

Quoted 3 lines

Probably Coccinelle is not able to determine that priv->res and priv->io are pointers, if you have not given it enough encouragement to look for header files to find the defintion of the type of the priv structure.

The system configuration might need another bit of fine-tuning for my test approach.

Quoted 1 line

I don't think that there is any need for the * in the metavariable declaration.

I tried out to specify the desired change by a specification which can be a bit more precise.

Quoted 2 lines

If the various expressions are not pointers, the C compiler would have complained.

The omission of the asterisk did also not trigger a known patch for the mentioned function implementation while patches are generated for code variations from other source files as expected.

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#3 lore raw
In reply to Julia Lawall
Quoted 8 lines
Quoted 8 lines

@replacement@
expression* base, device, resource;
@@

(quoted diff) +1 -2
-resource = platform_get_resource(device, IORESOURCE_MEM, 0);
 base =
-       devm_ioremap_resource(&device->dev, resource)
+       devm_platform_get_and_ioremap_resource(device, 0, &resource)
 ;

…

Quoted 3 lines
Quoted 3 lines

static int rcar_i2c_probe(struct platform_device *pdev)

{
	struct rcar_i2c_priv *priv;

…

Quoted 1 line
Quoted 1 line
	struct device *dev = &pdev->dev;

…

Quoted 5 lines
Quoted 5 lines
	priv->res = platform_get_resource(pdev, IORESOURCE_MEM, 0);

	priv->io = devm_ioremap_resource(dev, priv->res);
	if (IS_ERR(priv->io))
		return PTR_ERR(priv->io);

…

Quoted 1 line
Quoted 1 line

}

…

Quoted 3 lines

Probably Coccinelle is not able to determine that priv->res and priv->io are pointers, if you have not given it enough encouragement to look for header files to find the defintion of the type of the priv structure.

Would you find the following test command encouraging enough?

elfring@Sonne:~/Projekte/Linux/next-patched> spatch --include-headers-for-types ~/Projekte/Coccinelle/janitor/use_devm_platform_get_and_ioremap_resource3.cocci drivers/i2c/busses/i2c-rcar.c

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

maintainer
#4 lore raw
Quoted 22 lines · Markus Elfring
Quoted 8 lines
Quoted 8 lines

@replacement@
expression* base, device, resource;
@@

(quoted diff) +1 -2
-resource = platform_get_resource(device, IORESOURCE_MEM, 0);
 base =
-       devm_ioremap_resource(&device->dev, resource)
+       devm_platform_get_and_ioremap_resource(device, 0, &resource)
 ;

…

Quoted 3 lines
Quoted 3 lines

static int rcar_i2c_probe(struct platform_device *pdev)

{
	struct rcar_i2c_priv *priv;

…

Quoted 1 line
Quoted 1 line
	struct device *dev = &pdev->dev;

…

Quoted 5 lines
Quoted 5 lines
	priv->res = platform_get_resource(pdev, IORESOURCE_MEM, 0);

	priv->io = devm_ioremap_resource(dev, priv->res);
	if (IS_ERR(priv->io))
		return PTR_ERR(priv->io);

…

Quoted 1 line
Quoted 1 line

}

Quoted 8 lines

…

Quoted 3 lines

Probably Coccinelle is not able to determine that priv->res and priv->io are pointers, if you have not given it enough encouragement to look for header files to find the defintion of the type of the priv structure.

Would you find the following test command encouraging enough?

elfring@Sonne:~/Projekte/Linux/next-patched> spatch --include-headers-for-types ~/Projekte/Coccinelle/janitor/use_devm_platform_get_and_ioremap_resource3.cocci drivers/i2c/busses/i2c-rcar.c

No, include headers for types doesn't have any impact on how many header files are included. You need options like --all-includes or --recursive-includes. Include headers for types means only look at the included files to collect type information, but ignore these files in the matching and transformation process.

julia

_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#5 lore raw
Quoted 5 lines
Quoted 1 line

elfring@Sonne:~/Projekte/Linux/next-patched> spatch --include-headers-for-types ~/Projekte/Coccinelle/janitor/use_devm_platform_get_and_ioremap_resource3.cocci drivers/i2c/busses/i2c-rcar.c

No, include headers for types doesn't have any impact on how many header files are included. You need options like --all-includes or --recursive-includes.

I still observe that the known patch hunk is not generated even with the addition of one of these command options.
https://lore.kernel.org/patchwork/patch/1223734/
https://lore.kernel.org/linux-i2c/[email protected]/

Would I need any extra parameters here?

Will the Coccinelle software need any further improvements also around the desired processing for pointers and corresponding data structures?

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

maintainer
#6 lore raw
Quoted 6 lines · Markus Elfring
Quoted 5 lines
Quoted 1 line

elfring@Sonne:~/Projekte/Linux/next-patched> spatch --include-headers-for-types ~/Projekte/Coccinelle/janitor/use_devm_platform_get_and_ioremap_resource3.cocci drivers/i2c/busses/i2c-rcar.c

No, include headers for types doesn't have any impact on how many header files are included. You need options like --all-includes or --recursive-includes.

Quoted 6 lines

I still observe that the known patch hunk is not generated even with the addition of one of these command options.
https://lore.kernel.org/patchwork/patch/1223734/
https://lore.kernel.org/linux-i2c/[email protected]/

Would I need any extra parameters here?

Perhaps -I options to help it find the relevant .h file.

julia

Quoted 8 lines

Will the Coccinelle software need any further improvements also around the desired processing for pointers and corresponding data structures?

Regards,
Markus

_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#7 lore raw
Quoted 3 lines
Quoted 1 line

Would I need any extra parameters here?

Perhaps -I options to help it find the relevant .h file.

I wonder why the relevant header files would not be found directly from the corresponding Git repository.

Can a file inclusion failure become an immediate SmPL script execution error when a special program parameter would be passed?

Can it be that the pointer variable “priv” to the data structure “rcar_i2c_priv” distracts from the desired understanding of a questionable software situation when the relevant declaration is provided in the same source file?
https://elixir.bootlin.com/linux/v5.7-rc1/source/drivers/i2c/busses/i2c-rcar.c#L119
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-rcar.c?id=8632e9b5645bbc2331d21d892b0d6961c1a08429#n119

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

maintainer
#8 lore raw
Quoted 6 lines · Markus Elfring
Quoted 3 lines
Quoted 1 line

Would I need any extra parameters here?

Perhaps -I options to help it find the relevant .h file.

I wonder why the relevant header files would not be found directly from the corresponding Git repository.

OK, it seems that the structure is defined in the same file so no include options should be needed. But I have now lost track of what the semantic patch actually is, so I don't know what is the problem.

julia

Quoted 5 lines

Can a file inclusion failure become an immediate SmPL script execution error when a special program parameter would be passed?

Quoted 9 lines

Can it be that the pointer variable “priv” to the data structure “rcar_i2c_priv” distracts from the desired understanding of a questionable software situation when the relevant declaration is provided in the same source file?
https://elixir.bootlin.com/linux/v5.7-rc1/source/drivers/i2c/busses/i2c-rcar.c#L119
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-rcar.c?id=8632e9b5645bbc2331d21d892b0d6961c1a08429#n119

Regards,
Markus

_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#9 lore raw
Quoted 3 lines

OK, it seems that the structure is defined in the same file so no include options should be needed. But I have now lost track of what the semantic patch actually is, so I don't know what is the problem.

1. Unfortunately, I needed a bit longer to become more aware of relevant differences in source code variations.
Another case distinction would be needed. Thus a corresponding SmPL disjunction is probably required.
I have tried out to express the desired change by a SmPL specification which can be a bit more precise.

@replacement@
expression base, device, private, resource;
@@
-resource = platform_get_resource(device, IORESOURCE_MEM, 0);
base =
- devm_ioremap_resource
+ devm_platform_get_and_ioremap_resource

                             (

(
- &
device
- ->dev

                              ,

+ 0, &
resource
|
device,
+ 0, &

                                      private->res
)
                             );

elfring@Sonne:~/Projekte/Coccinelle/janitor> spatch --parse-cocci use_devm_platform_get_and_ioremap_resource6.cocci …
warning: iso ptr_to_array does not match the code below on line 11
device->dev

the following code matched is not uniformly minus or context, or contains a disjunction:
Exp:
device->dev
…

2. Would it be occasionally nice to intentionally express restrictions on pointer data types by using the asterisk in metavariable declarations?

3. If we would require information from header files:
Is a file inclusion failure treated as an immediate SmPL script execution error?

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#10 lore raw
In reply to Julia Lawall
Quoted 2 lines

But I have now lost track of what the semantic patch actually is, so I don't know what is the problem.

Do you find the following transformation approach easier to clarify?

SmPL script:
@replacement@
expression base, device, index, private;
@@
-private->res = platform_get_resource(device, IORESOURCE_MEM, index);
base =
- devm_ioremap_resource
+ devm_platform_get_and_ioremap_resource
(device,
+ index, &

                                              private->res
                             );

Command example:
elfring@Sonne:~/Projekte/Coccinelle/janitor> spatch use_devm_platform_get_and_ioremap_resource8.cocci ../Probe/i2c-rcar-excerpt-20200415.c

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#11 lore raw
Quoted 2 lines
Quoted 2 lines

But I have now lost track of what the semantic patch actually is, so I don't know what is the problem.

Another possibility should be taken into account for source code variations.

SmPL script example:
@replacement@
expression base, device1, device2, index, private;
@@
-private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
base =
- devm_ioremap_resource
+ devm_platform_get_and_ioremap_resource

                             (device2,
+                                     index, &
                                              private->res
                             );

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#12 lore raw
Quoted 6 lines

Another possibility should be taken into account for source code variations.

SmPL script example:
@replacement@
expression base, device1, device2, index, private; @@

Review

Quoted 8 lines
(quoted diff) +2 -2
-private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
 base =
-       devm_ioremap_resource
+       devm_platform_get_and_ioremap_resource
                             (device2,
+                                     index, &
                                              private->res
                             );

Such a transformation approach can work to some degree.
Thus I would expect that it can be reused for a case distinction in the way of a SmPL disjunction like the following.

Another SmPL script example:
@replacement@
expression base, device1, device2, index, private, resource; @@
-resource = platform_get_resource(device1, IORESOURCE_MEM, index);
base =
- devm_ioremap_resource
+ devm_platform_get_and_ioremap_resource

                             (

(
- &
device1
- ->dev

                              ,

+ index, &
resource
|
device2,
+ index, &

                                               private->res
)
                             );

Test command:
elfring@Sonne:~/Projekte/Coccinelle/janitor> spatch use_devm_platform_get_and_ioremap_resource10.cocci ../Probe/i2c-rcar-excerpt-20200415.c

Now I wonder again why the known patch is not generated as expected so far. How would you like to clarify this software situation?

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#13 lore raw
In reply to Julia Lawall
Quoted 3 lines
Quoted 1 line

Would I need any extra parameters here?

Perhaps -I options to help it find the relevant .h file.

The dependencies for the data structure “device” can become also more interesting.
https://elixir.bootlin.com/linux/v5.7-rc1/source/include/linux/device.h#L451
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/linux/device.h?id=00086336a8d96a04aa960f912287692a258f6cf5#n451

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#14 lore raw
In reply to Julia Lawall
Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
Quoted 1 line

elfring@Sonne:~/Projekte/Linux/next-patched> spatch --include-headers-for-types ~/Projekte/Coccinelle/janitor/use_devm_platform_get_and_ioremap_resource3.cocci drivers/i2c/busses/i2c-rcar.c

No, include headers for types doesn't have any impact on how many header files are included. You need options like --all-includes or --recursive-includes.

Quoted 8 lines
Quoted 6 lines

I still observe that the known patch hunk is not generated even with the addition of one of these command options.
https://lore.kernel.org/patchwork/patch/1223734/
https://lore.kernel.org/linux-i2c/[email protected]/

Would I need any extra parameters here?

Perhaps -I options to help it find the relevant .h file.

I became curious if the software situation can be clarified better.

Thus I have tested my transformation approach again with the commit “i2c: altera: use proper variable to hold errno” before the referenced change.

@replacement@
expression base, device, resource;
@@
-resource = platform_get_resource(device, IORESOURCE_MEM, 0);
base =
- devm_ioremap_resource(&device->dev, resource)
+ devm_platform_get_and_ioremap_resource(device, 0, &resource)
;

elfring@Sonne:~/Projekte/Linux/next-patched> git checkout edb2c9dd3948738ef030c32b948543e84f4d3f81 && LANG=C make MODE=patch COCCI=~/Projekte/Coccinelle/janitor/use_devm_platform_get_and_ioremap_resource3.cocci M=drivers/i2c/busses V=1 coccicheck
Running (4 in parallel): /usr/local/bin/spatch -D patch --very-quiet --cocci-file /home/elfring/Projekte/Coccinelle/janitor/use_devm_platform_get_and_ioremap_resource3.cocci --patch . --dir drivers/i2c/busses -I ./arch/x86/include -I ./arch/x86/include/generated -I ./include -I ./arch/x86/include/uapi -I ./arch/x86/include/generated/uapi -I ./include/uapi -I ./include/generated/uapi --include ./include/linux/kconfig.h --jobs 4 --chunksize 1 coccicheck failed
make: *** [Makefile:1781: coccicheck] Error 255

I can try another customised command out with the shown parameters.

elfring@Sonne:~/Projekte/Linux/next-patched> spatch -I ./arch/x86/include -I ./arch/x86/include/generated -I ./include -I ./arch/x86/include/uapi -I ./arch/x86/include/generated/uapi -I ./include/uapi -I ./include/generated/uapi --include ./include/linux/kconfig.h ~/Projekte/Coccinelle/janitor/use_devm_platform_get_and_ioremap_resource3.cocci drivers/i2c/busses/i2c-rcar.c
init_defs_builtins: /usr/local/bin/../lib/coccinelle/standard.h
HANDLING: drivers/i2c/busses/i2c-rcar.c

May I expect the generation of a patch by the Coccinelle software here?

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#15 lore raw
In reply to Julia Lawall
Quoted 1 line

I don't think that there is any need for the * in the metavariable declaration.

I can try another SmPL script variant out also without the specification of an extra asterisk. If I test with the following source file example, a patch is generated by the Coccinelle software in the expected way.

// SPDX-License-Identifier: GPL-2.0
static int rcar_i2c_probe(struct platform_device *pdev)

{
	struct rcar_i2c_priv *priv;

	priv->res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
	priv->io = devm_ioremap_resource(&pdev->dev, priv->res);
	if (IS_ERR(priv->io))
		return PTR_ERR(priv->io);
}

Command example:
elfring@Sonne:~/Projekte/Coccinelle/janitor> spatch use_devm_platform_get_and_ioremap_resource3.cocci ../Probe/i2c-rcar-variant-20200416.c

Implementation details from the original source file trigger some software development challenges.
https://elixir.bootlin.com/linux/v5.7-rc1/source/drivers/i2c/busses/i2c-rcar.c#L921
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-rcar.c?id=8632e9b5645bbc2331d21d892b0d6961c1a08429#n921

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#16 lore raw

PATCH 1/1 Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

In reply to Markus Elfring

From: Markus Elfring <[email protected]>
Date: Mon, 7 Sep 2020 13:14:44 +0200

Another wrapper function is available since the commit 890cc39a879906b63912482dfc41944579df2dc6 ("drivers: provide devm_platform_get_and_ioremap_resource()").
Provide design options for the adjustment of affected source code
by the means of the semantic patch language (Coccinelle software).

---
 ...vm_platform_get_and_ioremap_resource.cocci | 71 +++++++++++++++++++
 1 file changed, 71 insertions(+)
 create mode 100644 scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +71 -0
new file mode 100644
index 000000000000..8e67359f6b76
@@ -0,0 +1,71 @@
+// SPDX-License-Identifier: GPL-2.0
+/// Simplify a function call combination by using a known wrapper function.
+//
+// Keywords: wrapper function conversion ioremap resources
+// Confidence: High
+
+virtual context, patch, report, org
+
+@display depends on context@
+expression base, device1, device2, index, private, resource;
+@@
+(
+*resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+*       devm_ioremap_resource
+                             (&device1->dev, resource);
+|
+*private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+*       devm_ioremap_resource
+                             (device2, private->res);
+)
+
+@replacement depends on patch@
+expression base, device1, device2, index, private, resource;
+@@
+(
+-resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (
+-                             &
+                               device1
+-                                     ->dev
+                              ,
+-                             resource
++                             index, &resource
+                             );
+|
+-private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (device2,
+-                             private->res
++                             index, &private->res
+                             );
+)
+
+@or depends on org || report@
+expression base, device1, device2, index, private, resource;
+position p;
+@@
+(
+ resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base = devm_ioremap_resource@p(&device1->dev, resource);
+|
+ private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base = devm_ioremap_resource@p(device2, private->res);
+)
+
+@script:python to_do depends on org@
+p << or.p;
+@@
+coccilib.org.print_todo(p[0], "WARNING: opportunity for devm_platform_get_and_ioremap_resource()")
+
+@script:python reporting depends on report@
+p << or.p;
+@@
+coccilib.report.print_report(p[0], "WARNING: opportunity for devm_platform_get_and_ioremap_resource()")

--
2.28.0

_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

maintainer
#17 lore raw

Re: [Cocci] [PATCH] Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

Quoted 25 lines · Markus Elfring

From: Markus Elfring <[email protected]>
Date: Mon, 7 Sep 2020 13:14:44 +0200

Another wrapper function is available since the commit 890cc39a879906b63912482dfc41944579df2dc6 ("drivers: provide devm_platform_get_and_ioremap_resource()").
Provide design options for the adjustment of affected source code
by the means of the semantic patch language (Coccinelle software).

---
 ...vm_platform_get_and_ioremap_resource.cocci | 71 +++++++++++++++++++
 1 file changed, 71 insertions(+)
 create mode 100644 scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +5 -0
new file mode 100644
index 000000000000..8e67359f6b76
@@ -0,0 +1,71 @@
+// SPDX-License-Identifier: GPL-2.0
+/// Simplify a function call combination by using a known wrapper function.
+//
+// Keywords: wrapper function conversion ioremap resources
+// Confidence: High

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 15

Quoted 10 lines
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +10 -0
+
+virtual context, patch, report, org
+
+@display depends on context@
+expression base, device1, device2, index, private, resource;
+@@
+(
+*resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base > +*       devm_ioremap_resource
+                             (&device1->dev, resource);

Why do you require these statements to be next to each other?

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 21

Quoted 4 lines
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +4 -0
+|
+*private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base > +*       devm_ioremap_resource
+                             (device2, private->res);

Why do you have this special case?

Quoted 12 lines
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +12 -0
+)
+
+@replacement depends on patch@
+expression base, device1, device2, index, private, resource;
+@@
+(
+-resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base > +-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (
+-                             &
+                               device1

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 45

Quoted 10 lines
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +10 -0
+-                                     ->dev
+                              ,
+-                             resource
++                             index, &resource
+                             );
+|
+-private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base > +-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (device2,

It is very suspicious that in one case you change the first argument of devm_platform_get_and_ioremap_resource and in one case you don't. If you don't know how to make the change in some cases, it would be better to do nothing at all.

julia

Quoted 34 lines
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +26 -1
+-                             private->res
++                             index, &private->res
+                             );
+)
+
+@or depends on org || report@
+expression base, device1, device2, index, private, resource;
+position p;
+@@
+(
+ resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base = devm_ioremap_resource@p(&device1->dev, resource);
+|
+ private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base = devm_ioremap_resource@p(device2, private->res);
+)
+
+@script:python to_do depends on org@
+p << or.p;
+@@
+coccilib.org.print_todo(p[0], "WARNING: opportunity for devm_platform_get_and_ioremap_resource()")
+
+@script:python reporting depends on report@
+p << or.p;
+@@
+coccilib.report.print_report(p[0], "WARNING: opportunity for devm_platform_get_and_ioremap_resource()")
--

2.28.0

_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

#18 lore raw

Re: [Cocci] [PATCH] Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

Quoted 10 lines

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 16

Quoted 8 lines

+@display depends on context@

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +7 -0
+expression base, device1, device2, index, private, resource;
+@@
+(
+*resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+*       devm_ioremap_resource
+                             (&device1->dev, resource);

Why do you require these statements to be next to each other?

I would appreciate indications for a general change acceptance also according to such a simple transformation approach. I imagine that it will become more challenging to tolerate extra source code between these function calls (by the specification of special SmPL filters).

Quoted 7 lines

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 21

Quoted 5 lines

+|

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +4 -0
+*private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+*       devm_ioremap_resource
+                             (device2, private->res);

Why do you have this special case?

The usage of the data structure member “res” triggers corresponding software design consequences.
The expressions which are passed as the first function call parameters can be different.

Quoted 17 lines
Quoted 17 lines

+@replacement depends on patch@

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +16 -0
+expression base, device1, device2, index, private, resource;
+@@
+(
+-resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (
+-                             &
+                               device1
+-                                     ->dev
+                              ,
+-                             resource
++                             index, &resource
+                             );
+|
Quoted 8 lines
Quoted 5 lines

+-private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =

+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (device2,

It is very suspicious that in one case you change the first argument of devm_platform_get_and_ioremap_resource and in one case you don't.

I noticed a few special cases during my source code analysis approach.

Quoted 2 lines

If you don't know how to make the change in some cases, it would be better to do nothing at all.

How do you think about to take another look at any update candidates?

Examples:
* mvebu_sei_probe
https://elixir.bootlin.com/linux/v5.9-rc4/source/drivers/irqchip/irq-mvebu-sei.c#L368
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/irqchip/irq-mvebu-sei.c?id=f4d51dffc6c01a9e94650d95ce0104964f8ae822#n368

* hi655x_pmic_probe
https://elixir.bootlin.com/linux/v5.9-rc4/source/drivers/mfd/hi655x-pmic.c#L92
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/mfd/hi655x-pmic.c?id=f4d51dffc6c01a9e94650d95ce0104964f8ae822#n92

Regards,
Markus
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

maintainer
#19 lore raw

Re: [PATCH] Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

Quoted 26 lines · Markus Elfring
Quoted 10 lines

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 16

Quoted 8 lines

+@display depends on context@

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +7 -0
+expression base, device1, device2, index, private, resource;
+@@
+(
+*resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+*       devm_ioremap_resource
+                             (&device1->dev, resource);

Why do you require these statements to be next to each other?

I would appreciate indications for a general change acceptance also according to such a simple transformation approach. I imagine that it will become more challenging to tolerate extra source code between these function calls (by the specification of special SmPL filters).

Quoted 7 lines

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 21

Quoted 5 lines

+|

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +4 -0
+*private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+*       devm_ioremap_resource
+                             (device2, private->res);

Why do you have this special case?

Quoted 2 lines

The usage of the data structure member “res” triggers corresponding software design consequences.

I don't think this is a reliable rule. You included two examples below. They involve structures of different types. It seems unlikely that there is a guarantee that the res field of all structures is used in the same way. Furthermore, this is the context case. What is the purpose of making the distinction? The user will have to figure out what to do by hand in any case.

Quoted 30 lines

The expressions which are passed as the first function call parameters can be different.

Quoted 17 lines
Quoted 17 lines

+@replacement depends on patch@

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +16 -0
+expression base, device1, device2, index, private, resource;
+@@
+(
+-resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (
+-                             &
+                               device1
+-                                     ->dev
+                              ,
+-                             resource
++                             index, &resource
+                             );
+|
Quoted 8 lines
Quoted 5 lines

+-private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =

+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (device2,

It is very suspicious that in one case you change the first argument of devm_platform_get_and_ioremap_resource and in one case you don't.

Quoted 1 line

I noticed a few special cases during my source code analysis approach.

This is not a reasonable answer. Does the rule work correctly or not? If it doesn't work correctly, it needs to be removed.

Quoted 9 lines
Quoted 2 lines

If you don't know how to make the change in some cases, it would be better to do nothing at all.

How do you think about to take another look at any update candidates?

Examples:
* mvebu_sei_probe
https://elixir.bootlin.com/linux/v5.9-rc4/source/drivers/irqchip/irq-mvebu-sei.c#L368
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/irqchip/irq-mvebu-sei.c?id=f4d51dffc6c01a9e94650d95ce0104964f8ae822#n368

I don't see any purpose to providing two links for everything.

julia

Quoted 8 lines

* hi655x_pmic_probe
https://elixir.bootlin.com/linux/v5.9-rc4/source/drivers/mfd/hi655x-pmic.c#L92
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/mfd/hi655x-pmic.c?id=f4d51dffc6c01a9e94650d95ce0104964f8ae822#n92

Regards,
Markus

A message that is not in the archive (sent off-list or not yet received): [email protected]

maintainer
#21 lore raw

Re: [PATCH] Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

Quoted 24 lines
Quoted 24 lines
Quoted 24 lines
Quoted 23 lines
Quoted 10 lines

+@replacement depends on patch@

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +9 -0
+expression base, device1, device2, index, private, resource;
+@@
+(
+-resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base > >>>> +-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (
+-                             &
+                               device1

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 45

Quoted 10 lines
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +10 -0
+-                                     ->dev
+                              ,
+-                             resource
++                             index, &resource
+                             );
+|
+-private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base > >>>> +-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (device2,

It is very suspicious that in one case you change the first argument of devm_platform_get_and_ioremap_resource and in one case you don't.

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

I noticed a few special cases during my source code analysis approach.

This is not a reasonable answer. Does the rule work correctly or not? If it doesn't work correctly, it needs to be removed.

Both source code change patterns from these branches of a SmPL disjunction work as I would expect it by the current Coccinelle software. Would you like to clarify any remaining related open issues?

In one case the first argument of devm_ioremap_resource has to be changed to produce the first argument of devm_platform_get_and_ioremap_resource and in the other case there is no such change. "work as I would expect" is not a proper explanation of why this is correct. Maybe you can point to some previous commits that have made the change in this way.

julia

A message that is not in the archive (sent off-list or not yet received): [email protected]

maintainer
#23 lore raw

Re: [PATCH] Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

Quoted 32 lines · Markus Elfring
Quoted 22 lines
Quoted 22 lines
Quoted 22 lines
Quoted 22 lines
Quoted 22 lines
Quoted 22 lines

+@replacement depends on patch@

(quoted diff) +21 -0
+expression base, device1, device2, index, private, resource;
+@@
+(
+-resource = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (
+-                             &
+                               device1
+-                                     ->dev
+                              ,
+-                             resource
++                             index, &resource
+                             );
+|
+-private->res = platform_get_resource(device1, IORESOURCE_MEM, index);
+ base =
+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (device2,

…

Quoted 8 lines
Quoted 3 lines

Both source code change patterns from these branches of a SmPL disjunction work as I would expect it by the current Coccinelle software. Would you like to clarify any remaining related open issues?

In one case the first argument of devm_ioremap_resource has to be changed to produce the first argument of devm_platform_get_and_ioremap_resource and in the other case there is no such change. "work as I would expect" is not a proper explanation of why this is correct.

Quoted 7 lines

Would like to achieve that a SmPL script (like the one which I present here) can help to adjust the implementation of the functions “mvebu_sei_probe” and “hi655x_pmic_probe” (for example) for the desired reuse of the wrapper function “devm_platform_get_and_ioremap_resource”?

Quoted 1 line

Maybe you can point to some previous commits that have made the change in this way.

I looked at some of the provided examples, but none of them leaves the first argument of devm_ioremap_resource unchanged.

julia

Quoted 10 lines

Not yet for these source code examples.

Some patches were contributed.
https://lore.kernel.org/patchwork/project/lkml/list/?q=devm_platform_get_and_ioremap_resource&archive=both&state=*
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/log/?qt=grep&q=devm_platform_get_and_ioremap_resource

Regards,
Markus

#24 lore raw

PATCH v2 1/1 Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

In reply to Markus Elfring

From: Markus Elfring <[email protected]>
Date: Mon, 7 Sep 2020 18:38:04 +0200

Another wrapper function is available since the commit 890cc39a879906b63912482dfc41944579df2dc6 ("drivers: provide devm_platform_get_and_ioremap_resource()").
Provide safe design options for the adjustment of affected source code
by the means of the semantic patch language (Coccinelle software).

---

v2:
Julia Lawall requested to omit case distinctions (disjunctions) from the first SmPL script.
The usage of different expression metavariables for the first parameter of function calls was too questionable for the proposed source code transformation.

 ...vm_platform_get_and_ioremap_resource.cocci | 48 +++++++++++++++++++
 1 file changed, 48 insertions(+)
 create mode 100644 scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +48 -0
new file mode 100644
index 000000000000..319583716ac8
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0
+/// Simplify a function call combination by using a known wrapper function.
+//
+// Keywords: wrapper function conversion ioremap resources
+// Confidence: High
+
+virtual context, patch, report, org
+
+@display depends on context@
+expression base, device, index, resource;
+@@
+*resource = platform_get_resource(device, IORESOURCE_MEM, index);
+ base =
+*       devm_ioremap_resource
+                             (&device->dev, resource);
+
+@replacement depends on patch@
+expression base, device, index, resource;
+@@
+-resource = platform_get_resource(device, IORESOURCE_MEM, index);
+ base =
+-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (
+-                             &
+                               device
+-                                    ->dev
+                              ,
+-                             resource
++                             index, &resource
+                             );
+
+@or depends on org || report@
+expression base, device, index, resource;
+position p;
+@@
+ resource = platform_get_resource(device, IORESOURCE_MEM, index);
+ base = devm_ioremap_resource@p(&device->dev, resource);
+
+@script:python to_do depends on org@
+p << or.p;
+@@
+coccilib.org.print_todo(p[0], "WARNING: opportunity for devm_platform_get_and_ioremap_resource()")
+
+@script:python reporting depends on report@
+p << or.p;
+@@
+coccilib.report.print_report(p[0], "WARNING: opportunity for devm_platform_get_and_ioremap_resource()")

--
2.28.0

_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

maintainer
#25 lore raw

Re: [Cocci] [PATCH v2] Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

Quoted 29 lines · Markus Elfring

From: Markus Elfring <[email protected]>
Date: Mon, 7 Sep 2020 18:38:04 +0200

Another wrapper function is available since the commit 890cc39a879906b63912482dfc41944579df2dc6 ("drivers: provide devm_platform_get_and_ioremap_resource()").
Provide safe design options for the adjustment of affected source code
by the means of the semantic patch language (Coccinelle software).

---

v2:
Julia Lawall requested to omit case distinctions (disjunctions) from the first SmPL script.
The usage of different expression metavariables for the first parameter of function calls was too questionable for the proposed source code transformation.

 ...vm_platform_get_and_ioremap_resource.cocci | 48 +++++++++++++++++++
 1 file changed, 48 insertions(+)
 create mode 100644 scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +0 -0
new file mode 100644
index 000000000000..319583716ac8

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 5

Quoted 5 lines
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +5 -0
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0
+/// Simplify a function call combination by using a known wrapper function.
+//
+// Keywords: wrapper function conversion ioremap resources
+// Confidence: High

Shouldn't there be some options? --no-includes and perhaps --include-headers would seem reasonable.

Please also remove all of the rule names except for "or" which is necessary. The rest just add clutter and make it harder to find the useful information, which is the depends on. Rules only need names if some other rule inherits their metavariables.

julia

Quoted 49 lines
scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +41 -1
+
+virtual context, patch, report, org
+
+@display depends on context@
+expression base, device, index, resource;
+@@
+*resource = platform_get_resource(device, IORESOURCE_MEM, index);
+ base > +*       devm_ioremap_resource
+                             (&device->dev, resource);
+
+@replacement depends on patch@
+expression base, device, index, resource;
+@@
+-resource = platform_get_resource(device, IORESOURCE_MEM, index);
+ base > +-       devm_ioremap_resource
++       devm_platform_get_and_ioremap_resource
+                             (
+-                             &
+                               device
+-                                    ->dev
+                              ,
+-                             resource
++                             index, &resource
+                             );
+
+@or depends on org || report@
+expression base, device, index, resource;
+position p;
+@@
+ resource = platform_get_resource(device, IORESOURCE_MEM, index);
+ base = devm_ioremap_resource@p(&device->dev, resource);
+
+@script:python to_do depends on org@
+p << or.p;
+@@
+coccilib.org.print_todo(p[0], "WARNING: opportunity for devm_platform_get_and_ioremap_resource()")
+
+@script:python reporting depends on report@
+p << or.p;
+@@
+coccilib.report.print_report(p[0], "WARNING: opportunity for devm_platform_get_and_ioremap_resource()")
--

2.28.0

_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

A message that is not in the archive (sent off-list or not yet received): [email protected]

maintainer
#27 lore raw

Re: [PATCH v2] Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

Quoted 11 lines · Markus Elfring
Quoted 10 lines

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci line 5

Quoted 7 lines

+++ b/scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci

scripts/coccinelle/api/use_devm_platform_get_and_ioremap_resource.cocci new file +5 -0
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0
+/// Simplify a function call combination by using a known wrapper function.
+//
+// Keywords: wrapper function conversion ioremap resources
+// Confidence: High

Shouldn't there be some options? --no-includes and perhaps --include-headers would seem reasonable.

Quoted 2 lines

I imagine that the proposed source code transformation can work also without additional semantic patch command options.

It can. But it will be more efficient and more exhaustive if you include them.

julia

A message that is not in the archive (sent off-list or not yet received): [email protected]

#29 lore raw

Re: [cocci] [PATCH v3] Coccinelle: api: Add SmPL script “use_devm_platform_get_and_ioremap_resource.cocci”

Quoted 6 lines

Date: Mon, 7 Sep 2020 20:00:09 +0200

Another wrapper function is available since the commit 890cc39a879906b63912482dfc41944579df2dc6 ("drivers: provide devm_platform_get_and_ioremap_resource()").
Provide safe design options for the adjustment of affected source code
by the means of the semantic patch language (Coccinelle software).

Do you find this transformation approach variant acceptable?

How are the chances for integration of this SmPL script version?

Regards,
Markus