This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Fix build with compiler defining __i686 (bug 411)


> There is exactly one such inline asm instance.  You mean add a macro 
> expanding to a string version of LOAD_PIC_REG (for that is what 
> check_consistency in sysdeps/unix/sysv/linux/i386/sysdep.h is doing) and 
> use it in that one place?

I guess so, yes.

> I haven't done that because given Carlos's question when I suggested it 
> I'm waiting on confirmation from bug-standards regarding the rules for 
> collapsing ranges.

Fair enough.

> The increased alignment on thunks means you get differences (LOAD_PIC_REG 
> was already used in sysdeps/i386/fpu, for example, and there were 
> previously no instances of the dx thunk with the extra alignment).  It's 
> fully in accordance with the alignment used in the ENTRY macro in sysdep.h 
> - a benefit of using the macros more consistently is that such unnecessary 
> differences between files using the thunks are eliminated - but it does 
> mean that you can't just compare the linked libraries.

Yeah, I guess nobody has written a good labelifier for objdump so that
diffs stay small in the face of simple address changes.


Thanks,
Roland


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]