Skip to contentSkip to conversation

coccinnelle: Remove ptr_ret script

PATCH Superseded Maxime Ripard 6 participants 9 messages Jan 7, 2020 – Dec 15, 2020 Coccinelle/Semantic Patches (SmPL)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 coccinnelle: Remove ptr_ret script

The ptr_ret script script addresses a number of situations where we end up testing an error pointer, and if it's an error returning it, or return 0 otherwise to transform it into a PTR_ERR_OR_ZERO call.

So it will convert a block like this:

if (IS_ERR(err))

    return PTR_ERR(err);

return 0;

into

return PTR_ERR_OR_ZERO(err);

While this is technically correct, it has a number of drawbacks. First, it merges the error and success path, which will make it harder for a reviewer or reader to grasp.

It's also more difficult to extend if we were to add some code between the error check and the function return, making the author essentially revert that patch before adding new lines, while it would have been a trivial addition otherwise for the rewiever.

Therefore, since that script is only about cosmetic in the first place, let's remove it since it's not worth it.

Cc Jani Nikula <[email protected]>
Cc Thierry Reding <[email protected]>
Cc Tomi Valkeinen <[email protected]>
Cc Mark Brown <[email protected]>
---
 scripts/coccinelle/api/ptr_ret.cocci | 97 ----------------------------
 1 file changed, 97 deletions(-)
 delete mode 100644 scripts/coccinelle/api/ptr_ret.cocci
scripts/coccinelle/api/ptr_ret.cocci deleted +0 -97
deleted file mode 100644
index e76cd5d90a8a..000000000000
@@ -1,97 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-only
-///
-/// Use PTR_ERR_OR_ZERO rather than if(IS_ERR(...)) + PTR_ERR
-///
-// Confidence: High
-// Copyright: (C) 2012 Julia Lawall, INRIA/LIP6.
-// Copyright: (C) 2012 Gilles Muller, INRIA/LiP6.
-// URL: http://coccinelle.lip6.fr/
-// Options: --no-includes --include-headers
-//
-// Keywords: ERR_PTR, PTR_ERR, PTR_ERR_OR_ZERO
-// Version min: 2.6.39
-//
-
-virtual context
-virtual patch
-virtual org
-virtual report
-
-@depends on patch@
-expression ptr;
-@@
-
-- if (IS_ERR(ptr)) return PTR_ERR(ptr); else return 0;
-+ return PTR_ERR_OR_ZERO(ptr);
-
-@depends on patch@
-expression ptr;
-@@
-
-- if (IS_ERR(ptr)) return PTR_ERR(ptr); return 0;
-+ return PTR_ERR_OR_ZERO(ptr);
-
-@depends on patch@
-expression ptr;
-@@
-
-- (IS_ERR(ptr) ? PTR_ERR(ptr) : 0)
-+ PTR_ERR_OR_ZERO(ptr)
-
-@r1 depends on !patch@
-expression ptr;
-position p1;
-@@
-
-* if@p1 (IS_ERR(ptr)) return PTR_ERR(ptr); else return 0;
-
-@r2 depends on !patch@
-expression ptr;
-position p2;
-@@
-
-* if@p2 (IS_ERR(ptr)) return PTR_ERR(ptr); return 0;
-
-@r3 depends on !patch@
-expression ptr;
-position p3;
-@@
-
-* IS_ERR@p3(ptr) ? PTR_ERR(ptr) : 0
-
-@script:python depends on org@
-p << r1.p1;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-
-@script:python depends on org@
-p << r2.p2;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on org@
-p << r3.p3;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r1.p1;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r2.p2;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r3.p3;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
Signature · 1 line
2.24.1
Acked-by on v1 1/1
#1 lore raw
Quoted 19 lines · Maxime Ripard

The ptr_ret script script addresses a number of situations where we end up testing an error pointer, and if it's an error returning it, or return 0 otherwise to transform it into a PTR_ERR_OR_ZERO call.

So it will convert a block like this:

if (IS_ERR(err))

    return PTR_ERR(err);

return 0;

into

return PTR_ERR_OR_ZERO(err);

While this is technically correct, it has a number of drawbacks. First, it merges the error and success path, which will make it harder for a reviewer or reader to grasp.

Quoted 7 lines

It's also more difficult to extend if we were to add some code between the error check and the function return, making the author essentially revert that patch before adding new lines, while it would have been a trivial addition otherwise for the rewiever.

Therefore, since that script is only about cosmetic in the first place, let's remove it since it's not worth it.

I guess Julia's patch [1] never got merged then?

Reaffirming my ack on the removal, regardless of whose patch it is,

[1] http://lore.kernel.org/r/[email protected]

Quoted 114 lines
Cc Jani Nikula <[email protected]>
Cc Thierry Reding <[email protected]>
Cc Tomi Valkeinen <[email protected]>
Cc Mark Brown <[email protected]>
---
 scripts/coccinelle/api/ptr_ret.cocci | 97 ----------------------------
 1 file changed, 97 deletions(-)
 delete mode 100644 scripts/coccinelle/api/ptr_ret.cocci
scripts/coccinelle/api/ptr_ret.cocci deleted +0 -97
deleted file mode 100644
index e76cd5d90a8a..000000000000
@@ -1,97 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-only
-///
-/// Use PTR_ERR_OR_ZERO rather than if(IS_ERR(...)) + PTR_ERR
-///
-// Confidence: High
-// Copyright: (C) 2012 Julia Lawall, INRIA/LIP6.
-// Copyright: (C) 2012 Gilles Muller, INRIA/LiP6.
-// URL: http://coccinelle.lip6.fr/
-// Options: --no-includes --include-headers
-//
-// Keywords: ERR_PTR, PTR_ERR, PTR_ERR_OR_ZERO
-// Version min: 2.6.39
-//
-
-virtual context
-virtual patch
-virtual org
-virtual report
-
-@depends on patch@
-expression ptr;
-@@
-
-- if (IS_ERR(ptr)) return PTR_ERR(ptr); else return 0;
-+ return PTR_ERR_OR_ZERO(ptr);
-
-@depends on patch@
-expression ptr;
-@@
-
-- if (IS_ERR(ptr)) return PTR_ERR(ptr); return 0;
-+ return PTR_ERR_OR_ZERO(ptr);
-
-@depends on patch@
-expression ptr;
-@@
-
-- (IS_ERR(ptr) ? PTR_ERR(ptr) : 0)
-+ PTR_ERR_OR_ZERO(ptr)
-
-@r1 depends on !patch@
-expression ptr;
-position p1;
-@@
-
-* if@p1 (IS_ERR(ptr)) return PTR_ERR(ptr); else return 0;
-
-@r2 depends on !patch@
-expression ptr;
-position p2;
-@@
-
-* if@p2 (IS_ERR(ptr)) return PTR_ERR(ptr); return 0;
-
-@r3 depends on !patch@
-expression ptr;
-position p3;
-@@
-
-* IS_ERR@p3(ptr) ? PTR_ERR(ptr) : 0
-
-@script:python depends on org@
-p << r1.p1;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-
-@script:python depends on org@
-p << r2.p2;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on org@
-p << r3.p3;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r1.p1;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r2.p2;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r3.p3;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
Acked-by on v1 1/1
#2 lore raw
In reply to Maxime Ripard
Quoted 32 lines · Maxime Ripard

The ptr_ret script script addresses a number of situations where we end up testing an error pointer, and if it's an error returning it, or return 0 otherwise to transform it into a PTR_ERR_OR_ZERO call.

So it will convert a block like this:

if (IS_ERR(err))

    return PTR_ERR(err);

return 0;

into

return PTR_ERR_OR_ZERO(err);

While this is technically correct, it has a number of drawbacks. First, it merges the error and success path, which will make it harder for a reviewer or reader to grasp.

It's also more difficult to extend if we were to add some code between the error check and the function return, making the author essentially revert that patch before adding new lines, while it would have been a trivial addition otherwise for the rewiever.

Therefore, since that script is only about cosmetic in the first place, let's remove it since it's not worth it.

Cc Jani Nikula <[email protected]>
Cc Thierry Reding <[email protected]>
Cc Tomi Valkeinen <[email protected]>
Cc Mark Brown <[email protected]>
Quoted 4 lines
---
 scripts/coccinelle/api/ptr_ret.cocci | 97 ----------------------------
 1 file changed, 97 deletions(-)
 delete mode 100644 scripts/coccinelle/api/ptr_ret.cocci
maintainer Acked-by on v1 1/1
#3 lore raw
In reply to Maxime Ripard
Quoted 24 lines · Maxime Ripard

The ptr_ret script script addresses a number of situations where we end up testing an error pointer, and if it's an error returning it, or return 0 otherwise to transform it into a PTR_ERR_OR_ZERO call.

So it will convert a block like this:

if (IS_ERR(err))

    return PTR_ERR(err);

return 0;

into

return PTR_ERR_OR_ZERO(err);

While this is technically correct, it has a number of drawbacks. First, it merges the error and success path, which will make it harder for a reviewer or reader to grasp.

It's also more difficult to extend if we were to add some code between the error check and the function return, making the author essentially revert that patch before adding new lines, while it would have been a trivial addition otherwise for the rewiever.

Quoted 8 lines

Therefore, since that script is only about cosmetic in the first place, let's remove it since it's not worth it.

Cc Jani Nikula <[email protected]>
Cc Thierry Reding <[email protected]>
Cc Tomi Valkeinen <[email protected]>
Cc Mark Brown <[email protected]>
Quoted 112 lines
---
 scripts/coccinelle/api/ptr_ret.cocci | 97 ----------------------------
 1 file changed, 97 deletions(-)
 delete mode 100644 scripts/coccinelle/api/ptr_ret.cocci
scripts/coccinelle/api/ptr_ret.cocci deleted +0 -98
deleted file mode 100644
index e76cd5d90a8a..000000000000
@@ -1,97 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-only
-///
-/// Use PTR_ERR_OR_ZERO rather than if(IS_ERR(...)) + PTR_ERR
-///
-// Confidence: High
-// Copyright: (C) 2012 Julia Lawall, INRIA/LIP6.
-// Copyright: (C) 2012 Gilles Muller, INRIA/LiP6.
-// URL: http://coccinelle.lip6.fr/
-// Options: --no-includes --include-headers
-//
-// Keywords: ERR_PTR, PTR_ERR, PTR_ERR_OR_ZERO
-// Version min: 2.6.39
-//
-
-virtual context
-virtual patch
-virtual org
-virtual report
-
-@depends on patch@
-expression ptr;
-@@
-
-- if (IS_ERR(ptr)) return PTR_ERR(ptr); else return 0;
-+ return PTR_ERR_OR_ZERO(ptr);
-
-@depends on patch@
-expression ptr;
-@@
-
-- if (IS_ERR(ptr)) return PTR_ERR(ptr); return 0;
-+ return PTR_ERR_OR_ZERO(ptr);
-
-@depends on patch@
-expression ptr;
-@@
-
-- (IS_ERR(ptr) ? PTR_ERR(ptr) : 0)
-+ PTR_ERR_OR_ZERO(ptr)
-
-@r1 depends on !patch@
-expression ptr;
-position p1;
-@@
-
-* if@p1 (IS_ERR(ptr)) return PTR_ERR(ptr); else return 0;
-
-@r2 depends on !patch@
-expression ptr;
-position p2;
-@@
-
-* if@p2 (IS_ERR(ptr)) return PTR_ERR(ptr); return 0;
-
-@r3 depends on !patch@
-expression ptr;
-position p3;
-@@
-
-* IS_ERR@p3(ptr) ? PTR_ERR(ptr) : 0
-
-@script:python depends on org@
-p << r1.p1;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-
-@script:python depends on org@
-p << r2.p2;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on org@
-p << r3.p3;
-@@
-
-coccilib.org.print_todo(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r1.p1;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r2.p2;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
-
-@script:python depends on report@
-p << r3.p3;
-@@
-
-coccilib.report.print_report(p[0], "WARNING: PTR_ERR_OR_ZERO can be used")
--

2.24.1

Reviewed-by
#4 lore raw
Quoted 31 lines · Julia Lawall
Quoted 27 lines · Maxime Ripard

The ptr_ret script script addresses a number of situations where we end up testing an error pointer, and if it's an error returning it, or return 0 otherwise to transform it into a PTR_ERR_OR_ZERO call.

So it will convert a block like this:

if (IS_ERR(err))

    return PTR_ERR(err);

return 0;

into

return PTR_ERR_OR_ZERO(err);

While this is technically correct, it has a number of drawbacks. First, it merges the error and success path, which will make it harder for a reviewer or reader to grasp.

It's also more difficult to extend if we were to add some code between the error check and the function return, making the author essentially revert that patch before adding new lines, while it would have been a trivial addition otherwise for the rewiever.

Therefore, since that script is only about cosmetic in the first place, let's remove it since it's not worth it.

Quoted 7 lines
Quoted 5 lines
Cc Jani Nikula <[email protected]>
Cc Thierry Reding <[email protected]>
Cc Tomi Valkeinen <[email protected]>
Cc Mark Brown <[email protected]>

Convincing patch description, good catch!

#5 lore raw

Hi,

Quoted 38 lines · Wolfram Sang
Quoted 37 lines · Julia Lawall
Quoted 32 lines · Maxime Ripard

The ptr_ret script script addresses a number of situations where we end up testing an error pointer, and if it's an error returning it, or return 0 otherwise to transform it into a PTR_ERR_OR_ZERO call.

So it will convert a block like this:

if (IS_ERR(err))

    return PTR_ERR(err);

return 0;

into

return PTR_ERR_OR_ZERO(err);

While this is technically correct, it has a number of drawbacks. First, it merges the error and success path, which will make it harder for a reviewer or reader to grasp.

It's also more difficult to extend if we were to add some code between the error check and the function return, making the author essentially revert that patch before adding new lines, while it would have been a trivial addition otherwise for the rewiever.

Therefore, since that script is only about cosmetic in the first place, let's remove it since it's not worth it.

Cc Jani Nikula <[email protected]>
Cc Thierry Reding <[email protected]>
Cc Tomi Valkeinen <[email protected]>
Cc Mark Brown <[email protected]>
Quoted 5 lines
Quoted 1 line

Convincing patch description, good catch!

It looks like this patch was never applied, whose tree should it go through?

Thanks!
Maxime

maintainer
#6 lore raw
Quoted 41 lines · Maxime Ripard

Hi,

Quoted 38 lines · Wolfram Sang
Quoted 37 lines · Julia Lawall
Quoted 32 lines · Maxime Ripard

The ptr_ret script script addresses a number of situations where we end up testing an error pointer, and if it's an error returning it, or return 0 otherwise to transform it into a PTR_ERR_OR_ZERO call.

So it will convert a block like this:

if (IS_ERR(err))

    return PTR_ERR(err);

return 0;

into

return PTR_ERR_OR_ZERO(err);

While this is technically correct, it has a number of drawbacks. First, it merges the error and success path, which will make it harder for a reviewer or reader to grasp.

It's also more difficult to extend if we were to add some code between the error check and the function return, making the author essentially revert that patch before adding new lines, while it would have been a trivial addition otherwise for the rewiever.

Therefore, since that script is only about cosmetic in the first place, let's remove it since it's not worth it.

Cc Jani Nikula <[email protected]>
Cc Thierry Reding <[email protected]>
Cc Tomi Valkeinen <[email protected]>
Cc Mark Brown <[email protected]>
Quoted 8 lines
Quoted 5 lines
Quoted 1 line

Convincing patch description, good catch!

It looks like this patch was never applied, whose tree should it go through?

Sorry. I can take it. I'm not sure that I still have the original message, though. If you have it handy, that would be helpful.

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

#7 lore raw
Quoted 49 lines · Julia Lawall
Quoted 45 lines · Maxime Ripard

Hi,

Quoted 42 lines · Wolfram Sang
Quoted 31 lines · Julia Lawall
Quoted 27 lines · Maxime Ripard

The ptr_ret script script addresses a number of situations where we end up testing an error pointer, and if it's an error returning it, or return 0 otherwise to transform it into a PTR_ERR_OR_ZERO call.

So it will convert a block like this:

if (IS_ERR(err))

    return PTR_ERR(err);

return 0;

into

return PTR_ERR_OR_ZERO(err);

While this is technically correct, it has a number of drawbacks. First, it merges the error and success path, which will make it harder for a reviewer or reader to grasp.

It's also more difficult to extend if we were to add some code between the error check and the function return, making the author essentially revert that patch before adding new lines, while it would have been a trivial addition otherwise for the rewiever.

Therefore, since that script is only about cosmetic in the first place, let's remove it since it's not worth it.

Quoted 7 lines
Quoted 5 lines
Cc Jani Nikula <[email protected]>
Cc Thierry Reding <[email protected]>
Cc Tomi Valkeinen <[email protected]>
Cc Mark Brown <[email protected]>

Convincing patch description, good catch!

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

It looks like this patch was never applied, whose tree should it go through?

Sorry. I can take it. I'm not sure that I still have the original message, though. If you have it handy, that would be helpful.

Sure, I just sent a new version with all the tags

Thanks!
Maxime