On Tue, Dec 21, 2010 at 11:36 PM, Harvey Harrison <[email protected]> wrote: > On Tue, Dec 21, 2010 at 2:43 AM, Will Newton <[email protected]> wrote: >> On Tue, Dec 21, 2010 at 5:44 AM, Andrew Morton >>>> #include <linux/kernel.h> > >>>> >>>> -struct __una_u16 { u16 x __attribute__((packed)); }; >>>> -struct __una_u32 { u32 x __attribute__((packed)); }; >>>> -struct __una_u64 { u64 x __attribute__((packed)); }; >>>> +struct __una_u16 { u16 x; } __attribute__((packed)); >>>> +struct __una_u32 { u32 x; } __attribute__((packed)); >>>> +struct __una_u64 { u64 x; } __attribute__((packed)); >>>> >>> >>> Yes, that was wrong. >>> >>> Do you think this bug affects 2.6.36 or earlier? > > Interesting, I thought about this when modifying this awhile ago, and > was relying on this from the gcc manual: > > packed > <snip> > > Specifying this attribute for struct and union types is equivalent > to specifying the packed attribute on each of the > > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > structure or union members. Specifying the -fshort-enums flag on the > line is equivalent to specifying the packed > attribute on all enum definitions.
I think that's possibly a gcc documentation deficiency, it is equivalent to adding the packed attribute to all the struct members but *also* affects the assumed alignment of the struct. Looking at the gcc svn it seems the documentation has been improved a bit in this area. > What version of gcc was this? 4.2.4 _______________________________________________ stable mailing list [email protected] http://linux.kernel.org/mailman/listinfo/stable
