[PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Stephen Boyd]

Failure to terminate an of_device_id table can lead to confusing
failures depending on where the compiler places the array. Add a
check to make sure these tables are terminated. Thanks to Mitchel
Humpherys for coming up with the pattern initially.

Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
Cc: Julia Lawall <Julia.Lawall@lip6.fr>
Cc: Gilles Muller <Gilles.Muller@lip6.fr>
Cc: Nicolas Palix <nicolas.palix@imag.fr>
Cc: Grant Likely <grant.likely@linaro.org>
Cc: Rob Herring <robh+dt@kernel.org>
Cc: devicetree@vger.kernel.org
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
---
 scripts/coccinelle/misc/of_table.cocci | 73 ++++++++++++++++++++++++++++++++++
 1 file changed, 73 insertions(+)
 create mode 100644 scripts/coccinelle/misc/of_table.cocci

diff --git a/scripts/coccinelle/misc/of_table.cocci b/scripts/coccinelle/misc/of_table.cocci
new file mode 100644
index 000000000000..838c74c418c9
--- /dev/null
+++ b/scripts/coccinelle/misc/of_table.cocci
@@ -0,0 +1,73 @@
+/// Make sure of_device_id tables are NULL terminated
+//
+// Keywords: of_table
+// Confidence: Medium
+// Options: --include-headers
+
+virtual patch
+virtual context
+virtual org
+virtual report
+
+@bad_of_table exists@
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+	}
+};
+
+@depends on bad_of_table && context@
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+*	.var = E,
+	}
+};
+
+@depends on bad_of_table && patch@
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+-	}
++	},
++	{ }
+};
+
+@r depends on bad_of_table exists@
+position p1;
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+	}
+	@p1
+};
+
+@script:python depends on org@
+p1 << r.p1;
+arr << r.arr;
+@@
+
+cocci.print_main(arr,p1)
+
+@script:python depends on report@
+p1 << r.p1;
+arr << r.arr;
+@@
+
+msg = "%s is not NULL terminated at line %s" % (arr, p1[0].line)
+coccilib.report.print_report(p1[0],msg)
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
hosted by The Linux Foundation

Re: [PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Grant Likely]

On Tue, 27 May 2014 11:04:18 -0700, Stephen Boyd <sboyd@codeaurora.org> wrote:
> Failure to terminate an of_device_id table can lead to confusing
> failures depending on where the compiler places the array. Add a
> check to make sure these tables are terminated. Thanks to Mitchel
> Humpherys for coming up with the pattern initially.
> 
> Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
> Cc: Julia Lawall <Julia.Lawall@lip6.fr>
> Cc: Gilles Muller <Gilles.Muller@lip6.fr>
> Cc: Nicolas Palix <nicolas.palix@imag.fr>
> Cc: Grant Likely <grant.likely@linaro.org>
> Cc: Rob Herring <robh+dt@kernel.org>
> Cc: devicetree@vger.kernel.org
> Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>

Looks good to me:

Acked-by: Grant Likely <grant.likely@linaro.org>

> ---
>  scripts/coccinelle/misc/of_table.cocci | 73 ++++++++++++++++++++++++++++++++++
>  1 file changed, 73 insertions(+)
>  create mode 100644 scripts/coccinelle/misc/of_table.cocci
> 
> diff --git a/scripts/coccinelle/misc/of_table.cocci b/scripts/coccinelle/misc/of_table.cocci
> new file mode 100644
> index 000000000000..838c74c418c9
> --- /dev/null
> +++ b/scripts/coccinelle/misc/of_table.cocci
> @@ -0,0 +1,73 @@
> +/// Make sure of_device_id tables are NULL terminated
> +//
> +// Keywords: of_table
> +// Confidence: Medium
> +// Options: --include-headers
> +
> +virtual patch
> +virtual context
> +virtual org
> +virtual report
> +
> +@bad_of_table exists@
> +identifier var, arr;
> +expression E;
> +@@
> +struct of_device_id arr[] = {
> +	...,
> +	{
> +	.var = E,
> +	}
> +};
> +
> +@depends on bad_of_table && context@
> +identifier var, arr;
> +expression E;
> +@@
> +struct of_device_id arr[] = {
> +	...,
> +	{
> +*	.var = E,
> +	}
> +};
> +
> +@depends on bad_of_table && patch@
> +identifier var, arr;
> +expression E;
> +@@
> +struct of_device_id arr[] = {
> +	...,
> +	{
> +	.var = E,
> +-	}
> ++	},
> ++	{ }
> +};
> +
> +@r depends on bad_of_table exists@
> +position p1;
> +identifier var, arr;
> +expression E;
> +@@
> +struct of_device_id arr[] = {
> +	...,
> +	{
> +	.var = E,
> +	}
> +	@p1
> +};
> +
> +@script:python depends on org@
> +p1 << r.p1;
> +arr << r.arr;
> +@@
> +
> +cocci.print_main(arr,p1)
> +
> +@script:python depends on report@
> +p1 << r.p1;
> +arr << r.arr;
> +@@
> +
> +msg = "%s is not NULL terminated at line %s" % (arr, p1[0].line)
> +coccilib.report.print_report(p1[0],msg)
> -- 
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
> hosted by The Linux Foundation
> 

Re: [PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Julia Lawall]

Failure to terminate an of_device_id table can lead to confusing
failures depending on where the compiler places the array. Add a
check to make sure these tables are terminated. Thanks to Mitchel
Humpherys for coming up with the pattern initially.

Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
Cc: Julia Lawall <Julia.Lawall@lip6.fr>
Cc: Gilles Muller <Gilles.Muller@lip6.fr>
Cc: Nicolas Palix <nicolas.palix@imag.fr>
Cc: Grant Likely <grant.likely@linaro.org>
Cc: Rob Herring <robh+dt@kernel.org>
Cc: devicetree@vger.kernel.org
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Grant Likely <grant.likely@linaro.org>
Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>

---

v3: Removed unneeded rule bad_of_table.  Placed * in context rule on the 
closing brace; putting it on the field caused all fields to be marked.

 scripts/coccinelle/misc/of_table.cocci |   62 +++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)

diff --git a/scripts/coccinelle/misc/of_table.cocci b/scripts/coccinelle/misc/of_table.cocci
new file mode 100644
index 0000000..3c93404
--- /dev/null
+++ b/scripts/coccinelle/misc/of_table.cocci
@@ -0,0 +1,62 @@
+/// Make sure of_device_id tables are NULL terminated
+//
+// Keywords: of_table
+// Confidence: Medium
+// Options: --include-headers
+
+virtual patch
+virtual context
+virtual org
+virtual report
+
+@depends on context@
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+*	}
+};
+
+@depends on patch@
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+-	}
++	},
++	{ }
+};
+
+@r depends on org || report@
+position p1;
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+	}
+	@p1
+};
+
+@script:python depends on org@
+p1 << r.p1;
+arr << r.arr;
+@@
+
+cocci.print_main(arr,p1)
+
+@script:python depends on report@
+p1 << r.p1;
+arr << r.arr;
+@@
+
+msg = "%s is not NULL terminated at line %s" % (arr, p1[0].line)
+coccilib.report.print_report(p1[0],msg)

Re: [PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Stephen Boyd]

On 06/03/14 03:15, Julia Lawall wrote:
> Failure to terminate an of_device_id table can lead to confusing
> failures depending on where the compiler places the array. Add a
> check to make sure these tables are terminated. Thanks to Mitchel
> Humpherys for coming up with the pattern initially.
>
> Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
> Cc: Julia Lawall <Julia.Lawall@lip6.fr>
> Cc: Gilles Muller <Gilles.Muller@lip6.fr>
> Cc: Nicolas Palix <nicolas.palix@imag.fr>
> Cc: Grant Likely <grant.likely@linaro.org>
> Cc: Rob Herring <robh+dt@kernel.org>
> Cc: devicetree@vger.kernel.org
> Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
> Acked-by: Grant Likely <grant.likely@linaro.org>
> Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>
>
> ---
>
> v3: Removed unneeded rule bad_of_table.  Placed * in context rule on the 
> closing brace; putting it on the field caused all fields to be marked.
>


These seem like only minor changes. Why was my authorship removed?

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Re: [PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Julia Lawall]

On Tue, 3 Jun 2014, Stephen Boyd wrote:

> On 06/03/14 03:15, Julia Lawall wrote:
> > Failure to terminate an of_device_id table can lead to confusing
> > failures depending on where the compiler places the array. Add a
> > check to make sure these tables are terminated. Thanks to Mitchel
> > Humpherys for coming up with the pattern initially.
> >
> > Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
> > Cc: Julia Lawall <Julia.Lawall@lip6.fr>
> > Cc: Gilles Muller <Gilles.Muller@lip6.fr>
> > Cc: Nicolas Palix <nicolas.palix@imag.fr>
> > Cc: Grant Likely <grant.likely@linaro.org>
> > Cc: Rob Herring <robh+dt@kernel.org>
> > Cc: devicetree@vger.kernel.org
> > Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
> > Acked-by: Grant Likely <grant.likely@linaro.org>
> > Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>
> >
> > ---
> >
> > v3: Removed unneeded rule bad_of_table.  Placed * in context rule on the 
> > closing brace; putting it on the field caused all fields to be marked.
> >
> 
> 
> These seem like only minor changes. Why was my authorship removed?

Probably because I don't know how to provide a new vesion of the patch 
properly.  What should I have done?  There were quite a lot of 
modifications.  It seemed easier to just send a complete new version.

julia

Re: [PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Stephen Boyd]

On 06/03/14 11:14, Julia Lawall wrote:
> On Tue, 3 Jun 2014, Stephen Boyd wrote:
>
>> On 06/03/14 03:15, Julia Lawall wrote:
>>> Failure to terminate an of_device_id table can lead to confusing
>>> failures depending on where the compiler places the array. Add a
>>> check to make sure these tables are terminated. Thanks to Mitchel
>>> Humpherys for coming up with the pattern initially.
>>>
>>> Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
>>> Cc: Julia Lawall <Julia.Lawall@lip6.fr>
>>> Cc: Gilles Muller <Gilles.Muller@lip6.fr>
>>> Cc: Nicolas Palix <nicolas.palix@imag.fr>
>>> Cc: Grant Likely <grant.likely@linaro.org>
>>> Cc: Rob Herring <robh+dt@kernel.org>
>>> Cc: devicetree@vger.kernel.org
>>> Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
>>> Acked-by: Grant Likely <grant.likely@linaro.org>
>>> Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>
>>>
>>> ---
>>>
>>> v3: Removed unneeded rule bad_of_table.  Placed * in context rule on the 
>>> closing brace; putting it on the field caused all fields to be marked.
>>>
>>
>> These seem like only minor changes. Why was my authorship removed?
> Probably because I don't know how to provide a new vesion of the patch 
> properly.  What should I have done? 

If you use git format-patch it should insert a "From: <original author>"
line at the beginning of the commit text. This ensures that when git am
is used the correct authorship is retained. It looks like you're using
alpine so I suspect you could just paste this information at the
beginning of the message.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Re: [PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Julia Lawall]

From: Stephen Boyd <sboyd@codeaurora.org>

Failure to terminate an of_device_id table can lead to confusing
failures depending on where the compiler places the array. Add a
check to make sure these tables are terminated. Thanks to Mitchel
Humpherys for coming up with the pattern initially.

Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
Cc: Julia Lawall <Julia.Lawall@lip6.fr>
Cc: Gilles Muller <Gilles.Muller@lip6.fr>
Cc: Nicolas Palix <nicolas.palix@imag.fr>
Cc: Grant Likely <grant.likely@linaro.org>
Cc: Rob Herring <robh+dt@kernel.org>
Cc: devicetree@vger.kernel.org
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Grant Likely <grant.likely@linaro.org>
Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>

---

v3: Removed unneeded rule bad_of_table.  Placed * in context rule on the 
closing brace; putting it on the field caused all fields to be marked.

v4: corrected authorship

 scripts/coccinelle/misc/of_table.cocci |   62 +++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)

diff --git a/scripts/coccinelle/misc/of_table.cocci b/scripts/coccinelle/misc/of_table.cocci
new file mode 100644
index 0000000..3c93404
--- /dev/null
+++ b/scripts/coccinelle/misc/of_table.cocci
@@ -0,0 +1,62 @@
+/// Make sure of_device_id tables are NULL terminated
+//
+// Keywords: of_table
+// Confidence: Medium
+// Options: --include-headers
+
+virtual patch
+virtual context
+virtual org
+virtual report
+
+@depends on context@
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+*	}
+};
+
+@depends on patch@
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+-	}
++	},
++	{ }
+};
+
+@r depends on org || report@
+position p1;
+identifier var, arr;
+expression E;
+@@
+struct of_device_id arr[] = {
+	...,
+	{
+	.var = E,
+	}
+	@p1
+};
+
+@script:python depends on org@
+p1 << r.p1;
+arr << r.arr;
+@@
+
+cocci.print_main(arr,p1)
+
+@script:python depends on report@
+p1 << r.p1;
+arr << r.arr;
+@@
+
+msg = "%s is not NULL terminated at line %s" % (arr, p1[0].line)
+coccilib.report.print_report(p1[0],msg)

Re: [PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Michal Marek]

Dne 3.6.2014 20:25, Julia Lawall napsal(a):
> From: Stephen Boyd <sboyd@codeaurora.org>
> 
> Failure to terminate an of_device_id table can lead to confusing
> failures depending on where the compiler places the array. Add a
> check to make sure these tables are terminated. Thanks to Mitchel
> Humpherys for coming up with the pattern initially.
> 
> Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
> Cc: Julia Lawall <Julia.Lawall@lip6.fr>
> Cc: Gilles Muller <Gilles.Muller@lip6.fr>
> Cc: Nicolas Palix <nicolas.palix@imag.fr>
> Cc: Grant Likely <grant.likely@linaro.org>
> Cc: Rob Herring <robh+dt@kernel.org>
> Cc: devicetree@vger.kernel.org
> Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
> Acked-by: Grant Likely <grant.likely@linaro.org>
> Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>
> 
> ---
> 
> v3: Removed unneeded rule bad_of_table.  Placed * in context rule on the 
> closing brace; putting it on the field caused all fields to be marked.
> 
> v4: corrected authorship
> 
>  scripts/coccinelle/misc/of_table.cocci |   62 +++++++++++++++++++++++++++++++++
>  1 file changed, 62 insertions(+)

Applied to kbuild.git#misc, thanks. BTW, what is so special about of
device drivers? Shouldn't we check for missing sentinels in all kinds of
device tables?

Thanks,
Michal

Re: [PATCH v2] coccinelle: Check for missing NULL terminators in of_device_id tables

[Stephen Boyd]

On 06/09/14 14:47, Michal Marek wrote:
> Dne 3.6.2014 20:25, Julia Lawall napsal(a):
>> From: Stephen Boyd <sboyd@codeaurora.org>
>>
>> Failure to terminate an of_device_id table can lead to confusing
>> failures depending on where the compiler places the array. Add a
>> check to make sure these tables are terminated. Thanks to Mitchel
>> Humpherys for coming up with the pattern initially.
>>
>> Cc: Mitchel Humpherys <mitchelh@codeaurora.org>
>> Cc: Julia Lawall <Julia.Lawall@lip6.fr>
>> Cc: Gilles Muller <Gilles.Muller@lip6.fr>
>> Cc: Nicolas Palix <nicolas.palix@imag.fr>
>> Cc: Grant Likely <grant.likely@linaro.org>
>> Cc: Rob Herring <robh+dt@kernel.org>
>> Cc: devicetree@vger.kernel.org
>> Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
>> Acked-by: Grant Likely <grant.likely@linaro.org>
>> Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>
>>
>> ---
>>
>> v3: Removed unneeded rule bad_of_table.  Placed * in context rule on the 
>> closing brace; putting it on the field caused all fields to be marked.
>>
>> v4: corrected authorship
>>
>>  scripts/coccinelle/misc/of_table.cocci |   62 +++++++++++++++++++++++++++++++++
>>  1 file changed, 62 insertions(+)
> Applied to kbuild.git#misc, thanks. BTW, what is so special about of
> device drivers? Shouldn't we check for missing sentinels in all kinds of
> device tables?
>

Yes the rule should probably be expanded to cover more tables. I guess
this is just more noticeable because a driver may match some device that
it isn't supposed to match and start writing random bits to registers it
shouldn't be, causing the system to hang.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation