On 5/5/23 11:37, Peter Maydell wrote:
On Wed, 3 May 2023 at 08:08, Richard Henderson
<[email protected]> wrote:
Hosts using Intel and AMD AVX cpus are quite common.
Add fast paths through ldst_atomicity using this.
Only enable with CONFIG_INT128; some older clang versions do not
support __int128_t, and the inline assembly won't work on structures.
Signed-off-by: Richard Henderson <[email protected]>
---
accel/tcg/ldst_atomicity.c.inc | 76 +++++++++++++++++++++++++++-------
1 file changed, 60 insertions(+), 16 deletions(-)
Inline x86 asm in a bit of generic code seems rather awkward.
Ideally the compiler should be doing this for us; failing
that can we at least abstract out the operations to a
set of functions that we can provide (or not provide)
implementations of?
This was the best I could come up with, because of the fall through into other generic
code. Specific suggestions welcome.
r~
