Re: [PATCH v2] Make sure packed structs follow the gcc memory layout

[Mark Wielaard <mark at klomp dot org>]

Hi,

Looks like you sent this as HTML-only email, which are rejected by the
mailinglist.

On Mon, 2017-08-21 at 12:57 +0200, Ulf Hermann wrote:
> > Wouldn't this be a general problem for 
> > any struct that might be layed out differently but that we might
> > map to some on-disk data structure? 
>
> We should generally not rely on the layout of non-packed structs. The
> compiler can insert padding in various places. I didn't find any
> places where we do, though.

I don't think the compiler can do that without breaking ABI. I believe
we are very careful about the in-memory/on-disk struct layout matching.

I am online now and found the documentation:

https://gcc.gnu.org/onlinedocs/gcc/x86-Type-Attributes.html
https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html#index-mms-bitfields

> If packed is used on a structure, or if bit-fields are used, it may
> be that the Microsoft ABI lays out the structure differently than the
> way GCC normally does. Particularly when moving packed data between
> functions compiled with GCC and the native Microsoft compiler (either
> via function call or as data in a file), it may be necessary to
> access either format.
> 
> This option is enabled by default for Microsoft Windows targets.

OK, so it is x86-only and it only affects packed structs or if the
struct contains bitfields. Otherwise the layout should be the same.

So I do think your patch is correct. I don't think we ever use structs
that contain bit-fields to map from disk to memory.

Applied to master now.

Thanks,

Mark