[Bug tree-optimization/107952] tree-object-size: inconsistent size for flexible arrays nested in structs

["qinzhao at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107952

qinzhao at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |qinzhao at gcc dot gnu.org

--- Comment #6 from qinzhao at gcc dot gnu.org ---
(In reply to Siddhesh Poyarekar from comment #2)
> The standard does not allow the nesting, but gcc supports it as an extension.

when GCC supports it as an extension, I see two possible situations:

A. the structure with the flexible array member will be the last field of the
outer structure;

B. the structure with the flexible array member will be the middle field of the
outer structure;

I see GCC compile the above 2 cases without any complain (i.e, GCC extension
accepts both A and B) and Adding -Wpedantic issues warnings for both.

My questions:

1. Should GCC extension support the above case B? (it should NOT, right? what's
the point to support it)
2. If GCC extension support the above case A (looks like this is the case, and
some application use this case extensively, for example, Linux Kernel uses this
a lot, See bug https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101832), what's the
clear definition for it? Does it still treat the flexible array member in the
inner structure as a flexible array member in the outer structure? If so, we
might need to clearly document this in GCC's extension, and then user will have
consistent expectation on this.