[PATCH -next] x86/events:Use struct_size() helper in kzalloc()

[PATCH -next] x86/events:Use struct_size() helper in kzalloc()

[Lin Yujun]

Replace sizeof() with struct_size() to avoid potential integer
or heap overflow.

Also, address the following sparse warnings:
arch/x86/events/rapl.c:685:16: warning: using sizeof on a
flexible structure

Signed-off-by: Lin Yujun <linyujun809@huawei.com>
---
 arch/x86/events/rapl.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
index 77e3a47af5ad..c7e79f0ac04f 100644
--- a/arch/x86/events/rapl.c
+++ b/arch/x86/events/rapl.c
@@ -685,7 +685,7 @@ static int __init init_rapl_pmus(void)
 	int maxdie = topology_max_packages() * topology_max_die_per_package();
 	size_t size;
 
-	size = sizeof(*rapl_pmus) + maxdie * sizeof(struct rapl_pmu *);
+	size = struct_size(rapl_pmus, pmus, maxdie);
 	rapl_pmus = kzalloc(size, GFP_KERNEL);
 	if (!rapl_pmus)
 		return -ENOMEM;
-- 
2.17.1

Re: [PATCH -next] x86/events:Use struct_size() helper in kzalloc()

[Gustavo A. R. Silva]

On Wed, May 18, 2022 at 09:16:26PM +0800, Lin Yujun wrote:
> Replace sizeof() with struct_size() to avoid potential integer
> or heap overflow.

This is not exactly what the patch does.

Your patch is replacing the open-coded calculation of the total amount
of memory to be allocated for rapl_pmus with struct_size().

> 
> Also, address the following sparse warnings:
> arch/x86/events/rapl.c:685:16: warning: using sizeof on a
> flexible structure

This does not address this warning because struct_size()
still make use of sizeof on the struct-with-flex-array,
internally.

So, the mention to the sparse warning should be removed from
this changelog text.

> 
> Signed-off-by: Lin Yujun <linyujun809@huawei.com>
> ---
>  arch/x86/events/rapl.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
> index 77e3a47af5ad..c7e79f0ac04f 100644
> --- a/arch/x86/events/rapl.c
> +++ b/arch/x86/events/rapl.c
> @@ -685,7 +685,7 @@ static int __init init_rapl_pmus(void)
>  	int maxdie = topology_max_packages() * topology_max_die_per_package();
>  	size_t size;
>  
> -	size = sizeof(*rapl_pmus) + maxdie * sizeof(struct rapl_pmu *);
> +	size = struct_size(rapl_pmus, pmus, maxdie);
>  	rapl_pmus = kzalloc(size, GFP_KERNEL);

It seems that in this case, size could be entirely replaced by
struct_size().

Thanks
--
Gustavo

Re: [PATCH -next] x86/events:Use struct_size() helper in kzalloc()

[Peter Zijlstra]

On Wed, May 18, 2022 at 08:49:20AM -0500, Gustavo A. R. Silva wrote:

> > Signed-off-by: Lin Yujun <linyujun809@huawei.com>
> > ---
> >  arch/x86/events/rapl.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
> > index 77e3a47af5ad..c7e79f0ac04f 100644
> > --- a/arch/x86/events/rapl.c
> > +++ b/arch/x86/events/rapl.c
> > @@ -685,7 +685,7 @@ static int __init init_rapl_pmus(void)
> >  	int maxdie = topology_max_packages() * topology_max_die_per_package();
> >  	size_t size;
> >  
> > -	size = sizeof(*rapl_pmus) + maxdie * sizeof(struct rapl_pmu *);
> > +	size = struct_size(rapl_pmus, pmus, maxdie);
> >  	rapl_pmus = kzalloc(size, GFP_KERNEL);
> 
> It seems that in this case, size could be entirely replaced by
> struct_size().

Except every time I look at struct_size() I go: WTF does that do; while
the code as-is is crystal clear, no de-obfucstaion required.

Re: [PATCH -next] x86/events:Use struct_size() helper in kzalloc()

[Gustavo A. R. Silva]

On Wed, May 18, 2022 at 03:55:00PM +0200, Peter Zijlstra wrote:
> On Wed, May 18, 2022 at 08:49:20AM -0500, Gustavo A. R. Silva wrote:
> 
> > > Signed-off-by: Lin Yujun <linyujun809@huawei.com>
> > > ---
> > >  arch/x86/events/rapl.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
> > > index 77e3a47af5ad..c7e79f0ac04f 100644
> > > --- a/arch/x86/events/rapl.c
> > > +++ b/arch/x86/events/rapl.c
> > > @@ -685,7 +685,7 @@ static int __init init_rapl_pmus(void)
> > >  	int maxdie = topology_max_packages() * topology_max_die_per_package();
> > >  	size_t size;
> > >  
> > > -	size = sizeof(*rapl_pmus) + maxdie * sizeof(struct rapl_pmu *);
> > > +	size = struct_size(rapl_pmus, pmus, maxdie);
> > >  	rapl_pmus = kzalloc(size, GFP_KERNEL);
> > 
> > It seems that in this case, size could be entirely replaced by
> > struct_size().
> 
> Except every time I look at struct_size() I go: WTF does that do; while
> the code as-is is crystal clear, no de-obfucstaion required.

Naming it flex_struct_size() would be more informative?

--
Gustavo

Re: [PATCH -next] x86/events:Use struct_size() helper in kzalloc()

[Borislav Petkov]

On Wed, May 18, 2022 at 11:03:45AM -0500, Gustavo A. R. Silva wrote:
> Naming it flex_struct_size() would be more informative?

Not use it at all if it makes the code less readable.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [PATCH -next] x86/events:Use struct_size() helper in kzalloc()

[linyujun (C)]

Thank you for your suggestion. I will modify the patch according
to your suggestion and make the v2 version.

Thanks
--
LinYujun

在 2022/5/18 21:49, Gustavo A. R. Silva 写道:
> On Wed, May 18, 2022 at 09:16:26PM +0800, Lin Yujun wrote:
>> Replace sizeof() with struct_size() to avoid potential integer
>> or heap overflow.
> This is not exactly what the patch does.
>
> Your patch is replacing the open-coded calculation of the total amount
> of memory to be allocated for rapl_pmus with struct_size().
>
>> Also, address the following sparse warnings:
>> arch/x86/events/rapl.c:685:16: warning: using sizeof on a
>> flexible structure
> This does not address this warning because struct_size()
> still make use of sizeof on the struct-with-flex-array,
> internally.
>
> So, the mention to the sparse warning should be removed from
> this changelog text.
>
>> Signed-off-by: Lin Yujun <linyujun809@huawei.com>
>> ---
>>   arch/x86/events/rapl.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
>> index 77e3a47af5ad..c7e79f0ac04f 100644
>> --- a/arch/x86/events/rapl.c
>> +++ b/arch/x86/events/rapl.c
>> @@ -685,7 +685,7 @@ static int __init init_rapl_pmus(void)
>>   	int maxdie = topology_max_packages() * topology_max_die_per_package();
>>   	size_t size;
>>   
>> -	size = sizeof(*rapl_pmus) + maxdie * sizeof(struct rapl_pmu *);
>> +	size = struct_size(rapl_pmus, pmus, maxdie);
>>   	rapl_pmus = kzalloc(size, GFP_KERNEL);
> It seems that in this case, size could be entirely replaced by
> struct_size().
>
> Thanks
> --
> Gustavo
> .