Making a hypercall with the hypercall page absent should not have a happy ending. At best it's just weird, but at worst it might allow an otherwise failing test to pass.
Suggested-by: Roger Pau Monné <[email protected]> Signed-off-by: Alejandro Vallejo <[email protected]> --- `./xtf-runner -a` passes v3: * New patch based on Roger's suggestion --- arch/x86/hypercall_page.S | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/x86/hypercall_page.S b/arch/x86/hypercall_page.S index cc6ddc2..de0a991 100644 --- a/arch/x86/hypercall_page.S +++ b/arch/x86/hypercall_page.S @@ -6,8 +6,8 @@ .p2align PAGE_SHIFT GLOBAL(hypercall_page) - /* Poisoned with `ret` for safety before hypercalls are set up. */ - .fill PAGE_SIZE, 1, 0xc3 + /* Poisoned with `int3` so a stray hypercall is patently clear */ + .fill PAGE_SIZE, 1, 0xcc .type hypercall_page, STT_OBJECT .size hypercall_page, PAGE_SIZE -- 2.43.0
