xen/arm: introduce vwfi parameter

Introduce new Xen command line parameter called "vwfi", which stands for
virtual wfi. The default is "trap": Xen traps guest wfi and wfe
instructions. In the case of wfi, Xen calls vcpu_block on the guest
vcpu; in the case of guest wfe, Xen calls vcpu_yield on the guest vcpu.
The behavior can be changed by setting vwfi to "native", in that case
Xen doesn't trap neither wfi nor wfe, running them in guest context.

The result is strong reduction in irq latency (from 5000ns to 2000ns,
measured using https://github.com/edgarigl/tbm, the physical timer, and
1 pcpu dedicated to 1 vcpu). The downside is that the scheduler thinks
that the guest is busy when actually is sleeping, leading to suboptimal
scheduling decisions.

Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: Julien Grall <julien.grall@arm.com>

docs/misc/xen-command-line.markdown

@@ -1491,6 +1491,21 @@ Note that if **watchdog** option is also specified vpmu will be turned off.
 As the virtualisation is not 100% safe, don't use the vpmu flag on
 production systems (see http://xenbits.xen.org/xsa/advisory-163.html)!
 
+### vwfi
+> `= trap | native
+
+> Default: `trap`
+
+WFI is the ARM instruction to "wait for interrupt". WFE is similar and
+means "wait for event". This option, which is ARM specific, changes the
+way guest WFI and WFE are implemented in Xen. By default, Xen traps both
+instructions. In the case of WFI, Xen blocks the guest vcpu; in the case
+of WFE, Xen yield the guest vcpu. When setting vwfi to `native`, Xen
+doesn't trap either instruction, running them in guest context. Setting
+vwfi to `native` reduces irq latency significantly. It can also lead to
+suboptimal scheduling decisions, but only when the system is
+oversubscribed (i.e., in total there are more vCPUs than pCPUs).
+
 ### watchdog
 > `= force | <boolean>`
 

xen/arch/arm/traps.c

@@ -98,6 +98,19 @@ static int debug_stack_lines = 40;
 
 integer_param("debug_stack_lines", debug_stack_lines);
 
+static enum {
+	TRAP,
+	NATIVE,
+} vwfi;
+
+static void __init parse_vwfi(const char *s)
+{
+	if ( !strcmp(s, "native") )
+		vwfi = NATIVE;
+	else
+		vwfi = TRAP;
+}
+custom_param("vwfi", parse_vwfi);
 
 void __cpuinit init_traps(void)
 {
@@ -124,8 +137,8 @@ void __cpuinit init_traps(void)
 
     /* Setup hypervisor traps */
     WRITE_SYSREG(HCR_PTW|HCR_BSU_INNER|HCR_AMO|HCR_IMO|HCR_FMO|HCR_VM|
-                 HCR_TWE|HCR_TWI|HCR_TSC|HCR_TAC|HCR_SWIO|HCR_TIDCP|HCR_FB,
-                 HCR_EL2);
+                 (vwfi != NATIVE ? (HCR_TWI|HCR_TWE) : 0) |
+                 HCR_TSC|HCR_TAC|HCR_SWIO|HCR_TIDCP|HCR_FB,HCR_EL2);
     isb();
 }