2019-08-01 20:16:55

by Kees Cook

Subject: Re: [PATCH] Documentation patch (gcc-plugins kernel)

On Thu, Aug 01, 2019 at 09:30:58AM +0200, Robin Lindner wrote:
> Cleaned documentation comment up. I removed the "TODO" because it was very old.

Hi, please send these patches "normally" (cc maintainers, include lkml,
inline not attachments, etc):
https://www.kernel.org/doc/html/latest/process/submitting-patches.html

> ---
> scripts/gcc-plugins/stackleak_plugin.c | 2 --
> 1 file changed, 2 deletions(-)
>
> diff --git a/scripts/gcc-plugins/stackleak_plugin.c b/scripts/gcc-plugins/stackleak_plugin.c
> index dbd37460c573e..d8ba12c3bb238 100644
> --- a/scripts/gcc-plugins/stackleak_plugin.c
> +++ b/scripts/gcc-plugins/stackleak_plugin.c
> @@ -144,8 +144,6 @@ static unsigned int stackleak_instrument_execute(void)
> *
> * Case in point: native_save_fl on amd64 when optimized for size
> * clobbers rdx if it were instrumented here.
> - *
> - * TODO: any more special cases?
> */
> if (is_leaf &&
> !TREE_PUBLIC(current_function_decl) &&

As to the content of the patch, let's also CC Alexander...

Are there no more special cases?

--
Kees Cook

2019-08-14 11:30:43

by Alexander Popov

Subject: Re: [PATCH] Documentation patch (gcc-plugins kernel)

Hello Kees and Robin,

On 01.08.2019 23:14, Kees Cook wrote:
> On Thu, Aug 01, 2019 at 09:30:58AM +0200, Robin Lindner wrote:
>> Cleaned documentation comment up. I removed the "TODO" because it was very old.
>> ---
>> scripts/gcc-plugins/stackleak_plugin.c | 2 --
>> 1 file changed, 2 deletions(-)
>>
>> diff --git a/scripts/gcc-plugins/stackleak_plugin.c b/scripts/gcc-plugins/stackleak_plugin.c
>> index dbd37460c573e..d8ba12c3bb238 100644
>> --- a/scripts/gcc-plugins/stackleak_plugin.c
>> +++ b/scripts/gcc-plugins/stackleak_plugin.c
>> @@ -144,8 +144,6 @@ static unsigned int stackleak_instrument_execute(void)
>> *
>> * Case in point: native_save_fl on amd64 when optimized for size
>> * clobbers rdx if it were instrumented here.
>> - *
>> - * TODO: any more special cases?
>> */
>> if (is_leaf &&
>> !TREE_PUBLIC(current_function_decl) &&
>
> As to the content of the patch, let's also CC Alexander...
>
> Are there no more special cases?

I don't know other special cases when we should avoid the instrumentation.

And I can't imagine a method of finding such cases except runtime testing.

Robin, I would better save this comment only removing "TODO" if it matters.

Best regards,
Alexander