On Sun, Feb 19, 2023, Chao Gao wrote:
On Tue, Feb 14, 2023 at 06:46:01PM +0000, Mingwei Zhang wrote:
After the execution of __tilerelease(), AMX component will be in INIT state. Therefore, execution of XSAVEC saving the AMX state into memory will cause the xstate_bv[18] cleared in xheader. However, the xcomp_bv[18] will remain set. Fix the error in comment. Also, update xsavec() to XSAVEC because xcomp_bv[18] is set due to the instruction, not the function. Finally, use XTILEDATA instead 'bit 18' in comments.
Cc: Jim Mattson jmattson@google.com Cc: Venkatesh Srinivas venkateshs@google.com Cc: Aaron Lewis aaronlewis@google.com Signed-off-by: Mingwei Zhang mizhang@google.com
tools/testing/selftests/kvm/x86_64/amx_test.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/tools/testing/selftests/kvm/x86_64/amx_test.c b/tools/testing/selftests/kvm/x86_64/amx_test.c index d506821a5a26..aac727ff7cf8 100644 --- a/tools/testing/selftests/kvm/x86_64/amx_test.c +++ b/tools/testing/selftests/kvm/x86_64/amx_test.c @@ -190,7 +190,10 @@ static void __attribute__((__flatten__)) guest_code(struct tile_config *amx_cfg, GUEST_SYNC(4); __tilerelease(); GUEST_SYNC(5);
- /* bit 18 not in the XCOMP_BV after xsavec() */
- /*
* After XSAVEC, XTILEDATA is cleared in the xstate_bv but is set in* the xcomp_bv.*/maybe it would be better to add another GUEST_ASSERT() to enforce that XTILEDATA is set in the xcomp_bv.
yeah. The check has been added in the 6th patch of the series, but I think it is hard to see. I will reorder the change next to this one in the next version.
-- 2.39.1.581.gbfd45094c4-goog