[bug #17338] GNU Mach vs. GCC 4.1

[Samuel Thibault]

Follow-up Comment #1, bug #17338 (project hurd):

GCC gets smarter and smarter...  When seeing this:

struct pseudo_descriptor
{
        short pad;
        unsigned short limit;
        unsigned long linear_base;
}
MACH_INLINE void lgdt(struct pseudo_descriptor *pdesc)
{
        __asm volatile("lgdt %0" : : "m" (pdesc->limit));
}
gdt_init() {
        ...
        {
                struct pseudo_descriptor pdesc;

                pdesc.limit = sizeof(gdt)-1;
                pdesc.linear_base = kvtolin(&gdt);
                lgdt(&pdesc);
        }
}

, it understands that filling the linear_base field is not needed, since we
only give the limit field to the asm statement... The result it this:

gcc 4.0:
<gdt_init+102>:      movw   $0x57,0xfffffffa(%ebp)
<gdt_init+108>:      movl   $0xc0000000,0xfffffffc(%ebp)
<gdt_init+115>:      lgdtl  0xfffffffa(%ebp)

gcc 4.1:
<gdt_init+102>:      movw   $0x57,0xfffffffa(%ebp)
<gdt_init+108>:      lgdtl  0xfffffffa(%ebp)

it doesn't fill the linear_base field!

There are two possible fixes, that I'll attach:

- reorder the structure, pack it, and give it as a whole to the asm
statement.
- clobber memory in the asm statement.

My preference goes to the first, since it explains to gcc what we exactly
want. That supposes the support of the "packed" attribute, however.


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?func=detailitem&item_id=17338>

_______________________________________________
  Message posté via/par Savannah
  http://savannah.gnu.org/