[PATCH 00/19] Reworking initializion of data sent to clients

[Michal Suchanek]

On 25 June 2012 16:19, Alan Coopersmith <alan.coopersmith at oracle.com> wrote:
> On 06/25/12 02:38 AM, Michal Suchanek wrote:
>> Hello,
>>
>> thanks for these patches.
>>
>> These will hopefully quiet valgrind complaints about garbage.
>
> Yes, that was another motivation - it should help silence a lot of false
> alarms about reading uninitialized padding values when writing to clients,
> at the cost of silencing a few true alarms about reading uninitialized data
> values from those structures - making it a bit harder to find bugs like
> the recently reported https://bugs.freedesktop.org/show_bug.cgi?id=51375 .
>

As that was only discovered by checking the return code at client side
I don't think properly initializing the structure is any worse than
returning a random value. The return code will be what the application
expects only when set properly by the X server.

Still it is possible to initialize it to an intentionally bogus value
so that the returned value is correct only when set properly.

Thanks

Michal