x86/traps: Improve code generation for set_ist()
The IST field in an IDT entry is a 3 bit field, with 5 adjacent reserved bits
which we always write as zero. By expressing this as a byte field in a union,
we turn an invocation of enable_each_ist() from
4b 8b 14 d3 mov (%r11,%r10,8),%rdx
48 b8 ff ff ff ff f8 ff ff ff movabs $0xfffffff8ffffffff,%rax
48 be 00 00 00 00 01 00 00 00 movabs $0x100000000,%rsi
48 8b 8a 80 00 00 00 mov 0x80(%rdx),%rcx
48 21 c1 and %rax,%rcx
48 09 f1 or %rsi,%rcx
48 be 00 00 00 00 02 00 00 00 movabs $0x200000000,%rsi
48 89 8a 80 00 00 00 mov %rcx,0x80(%rdx)
48 8b 4a 20 mov 0x20(%rdx),%rcx
48 21 c1 and %rax,%rcx
48 23 82 20 01 00 00 and 0x120(%rdx),%rax
48 09 f1 or %rsi,%rcx
48 89 4a 20 mov %rcx,0x20(%rdx)
48 b9 00 00 00 00 03 00 00 00 movabs $0x300000000,%rcx
48 09 c8 or %rcx,%rax
48 89 82 20 01 00 00 mov %rax,0x120(%rdx)
into
4b 8b 04 d3 mov (%r11,%r10,8),%rax
c6 80 84 00 00 00 01 movb $0x1,0x84(%rax)
c6 40 24 02 movb $0x2,0x24(%rax)
c6 80 24 01 00 00 03 movb $0x3,0x124(%rax)
which is far more simple. As the IDT is typically live, this is more
obviously safe.
The net delta for this change is:
add/remove: 0/0 grow/shrink: 0/7 up/down: 0/-334 (-334)
While making changes here, tidy up the set_ist() declaration. Drop the
always_inline (I don't recall why I wrote it like that originally) and the ist
parameter need not be unsigned long (although it will be const-propagated in
practice).
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Release-acked-by: Juergen Gross <jgross@suse.com>