Re: [PATCH 1/3] perf events: fix WARN_ON_ONCE for 64-bit raw data, SW events

[David Ahern <daahern@cisco.com>]

On 02/18/11 07:55, Peter Zijlstra wrote:
>> Another gotcha is in perf_output_sample there is:
>>             struct {
>>                 u32 size;
>>                 u32 data;
>>             } raw = {
>>                 .size = sizeof(u32),
>>                 .data = 0,
>>             };
>>
>> While that meets the 8-byte stride I was concerned that changing the
>> size field to u64 breaks ABI.
> 
> It would.
> 
> No, what you need to do is provide data that is sized such that it
> matches the 8b stride adding padding where needed.

I want to push a 64-bit counter to userspace. By definition it meets the
8-byte stride. Now how do I get sizeof(size field) + sizeof(data) to
meet an 8-byte stride? I can't when sizeof(size field) = 4. Ok, I can
pad the data with 4 bytes, but then I an api to tell the user to strip
the 4-byte pad. Seems like an odd way to go.

Why is an 8-byte stride preferred over a 4-bytes?

David

> 
> Anyway, I don't think you need RAW at all.
>