On Dec 11, 2003, at 11:40 AM, Christos Zoulas wrote:

> In article <59A6FF24-2C04-11D8-A844-000A957020BC@3am-software.com>,
> Matt Thomas <matt@3am-software.com> wrote:
>>
>> On Dec 11, 2003, at 8:55 AM, Christos Zoulas wrote:
>>> BTW, this is the case [leaking kernel memory info] in many of the
>>> siginfo
>>> machdep.c files (sparc,sparc64,vax, etc.) It did not seem important 
>>> at
>>> the time, but maybe we should bring it up for discussion again.
>>
>> The vax implementation (and powerpc) does not leak kernel memory info.
>> However, it will leave portions of the siginfo_t that are not
>> used/overlayed
>> by ksiginfo_t unchanged.  That is not a kernel memory leak.
>
> Yes, that's it. It leaves uninitialized userland memory [stack]. Things
> like purify will bitch at it... I was wrong about the kernel leak.

It's not uninitialized.  Initially, it was set to all-0 and may now 
contain
previous stack contents.  I doubt purify would complain since it was not
able to control its creation.  I see no reason to copy data to the user
stack that the user won't be looking at.  It's just wasteful.
-- 
Matt Thomas                     email: matt@3am-software.com
3am Software Foundry              www: http://3am-software.com/bio/matt/
Cupertino, CA              disclaimer: I avow all knowledge of this 
message.