Re: [ros-dev] [ros-diffs] [hbelusca] 75133: [NTOS]: Revert the pool tag change from r75128, and explain in comment why we cannot actually use it.

[Thomas Faber]

On 2017-06-19 18:29, [e-mail] wrote:
>    /*
>     * Free it from the pool.
>     *
>     * We cannot use here ExFreePoolWithTag(..., OB_NAME_TAG); , because
>     * the object name may have been massaged during operation by different
>     * object parse routines. If the latter ones have to resolve a symbolic
>     * link (e.g. as is done by CmpParseKey() and CmpGetSymbolicLink()),
>     * the original object name is freed and re-allocated from the pool,
>     * possibly with a different pool tag. At the end of the day, the new
>     * object name can be reallocated and completely different, but we
>     * should still be able to free it!
>     */
>    ExFreePool(Buffer);

I feel like

ExFreePoolWithTag(Buffer, 0)

conveys that same message without needing a huge comment?

[Javier Agustìn Fernàndez Arroyo]

imho the more documented the code is, the easier to read and understand

El 19/06/2017 19:48, "Thomas Faber" [e-mail] escribió:
> On 2017-06-19 18:29, [e-mail] wrote:
> >    /*
> >     * Free it from the pool.
> >     *
> >     * We cannot use here ExFreePoolWithTag(..., OB_NAME_TAG); ,
> >
> > because
> >     * the object name may have been massaged during operation by
> >
> > different
> >     * object parse routines. If the latter ones have to resolve a
> >
> > symbolic
> >     * link (e.g. as is done by CmpParseKey() and
> >
> > CmpGetSymbolicLink()),
> >     * the original object name is freed and re-allocated from the
> >
> > pool,
> >     * possibly with a different pool tag. At the end of the day,
> >
> > the new
> >     * object name can be reallocated and completely different, but
> >
> > we
> >     * should still be able to free it!
> >     */
> >    ExFreePool(Buffer);
>
> I feel like
>
> ExFreePoolWithTag(Buffer, 0)
>
> conveys that same message without needing a huge comment?
>
> Ros-dev mailing list
> [e-mail]
> http://www.reactos.org/mailman/listinfo/ros-dev