On 02/02/18 14:37, Andre Przywara wrote:
On 02/02/18 10:14, Julien Grall wrote:
domain_crash_synchronous() should only be used when something went wrong
in Xen. It is better to inject to the guest as it will be in a better
position to provide helpful information (stack trace...).
Signed-off-by: Julien Grall <julien.gr...@arm.com>
Reviewed-by: Stefano Stabellini <sstabell...@kernel.org>
We potentially want to return -1 instead. This would make Xen more
future-proof if we decide to implement the other HVC immediate.
Changes in v2:
- Add Stefano's reviewed-by
xen/arch/arm/traps.c | 13 ++++++++-----
1 file changed, 8 insertions(+), 5 deletions(-)
diff --git a/xen/arch/arm/traps.c b/xen/arch/arm/traps.c
index 1e85f99ec1..1cba7e584d 100644
@@ -1471,14 +1471,17 @@ static void do_debug_trap(struct cpu_user_regs *regs,
unsigned int code)
static void do_trap_hypercall(struct cpu_user_regs *regs, register_t *nr,
- unsigned long iss)
+ const union hsr hsr)
arm_hypercall_fn_t call = NULL;
BUILD_BUG_ON(NR_hypercalls < ARRAY_SIZE(arm_hypercall_table) );
- if ( iss != XEN_HYPERCALL_TAG )
+ if ( hsr.iss != XEN_HYPERCALL_TAG )
+ gprintk(XENLOG_WARNING, "Invalid HVC imm 0x%x\n", hsr.iss);
+ return inject_undef_exception(regs, hsr);
Why is this UNDEF?
UNDEF is the exception used when either HVC is disabled or if it's
called from EL0.
I couldn't find anything that talks about how non-implemented immediates
should be handled, I guess they would just be ignored?
Do you have any sources that recommend UNDEF?
Per the SMCCC, all nonzero immediate for HVC are designated for use by
the hypervisors. So I take as we can implement the way we want.
We have few solutions here:
1) Ignore. This means that a guest may wrongly call an HVC and will
never notice it until implemented. This would lead to some trouble with
introduce new ABI.
2) Return a value. We could return -1 (or -ENOSYS). But that would need
to be defined.
3) Undefined. The guest should not be allow to touch it, so it makes
1# is not a solution because catching them wrong usage on old Xen would
be a nightmare.
2# might be doable, thought I am not sure we want to commit on a return
value now. But old Xen would see crash the domain.
So I choose #3 which still crash the domain but by injecting an undefined.
Xen-devel mailing list