[PATCH 2/2] KVM/arm64: Print emulated register table name when it is unsorted

[Alexandru Elisei]

Hi,

On Tue, Apr 26, 2022 at 10:18:27PM +0100, Marc Zyngier wrote:
> On Mon, 25 Apr 2022 17:39:03 +0100,
> Alexandru Elisei <alexandru.elisei at arm.com> wrote:
> > 
> > When a sysreg table entry is out-of-order, KVM attempts to print the
> > address of the table:
> > 
> > [    0.143881] kvm [1]: sys_reg table (____ptrval____) out of order (0)
> > 
> > Printing the name of the table instead of a pointer is more helpful in this
> > case:
> > 
> > [    0.143881] kvm [1]: sys_reg table sys_reg_descs out of order (0)
> > 
> > Signed-off-by: Alexandru Elisei <alexandru.elisei at arm.com>
> > ---
> >  arch/arm64/kvm/sys_regs.c | 20 +++++++++++---------
> >  1 file changed, 11 insertions(+), 9 deletions(-)
> > 
> > diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> > index 57302048afd0..7b62a2daf056 100644
> > --- a/arch/arm64/kvm/sys_regs.c
> > +++ b/arch/arm64/kvm/sys_regs.c
> > @@ -2188,18 +2188,18 @@ static const struct sys_reg_desc cp15_64_regs[] = {
> >  };
> >  
> >  static bool check_sysreg_table(const struct sys_reg_desc *table, unsigned int n,
> > -			       bool is_32)
> > +			       const char *table_name, bool is_32)
> >  {
> >  	unsigned int i;
> >  
> >  	for (i = 0; i < n; i++) {
> >  		if (!is_32 && table[i].reg && !table[i].reset) {
> > -			kvm_err("sys_reg table %p entry %d lacks reset\n", table, i);
> > +			kvm_err("sys_reg table %s entry %d lacks reset\n", table_name, i);
> 
> Instead of passing a table name, could we simply use something like
> %pS? If this works, it would be a good indication of both what table
> and what entry in that table is at fault.

With the change:

-                       kvm_err("sys_reg table %s out of order (%d)\n", table_name, i - 1);
+                       kvm_err("sys_reg table %pS out of order (%d)\n", &table[i - 1], i - 1);


this is what KVM prints with %pS if the second entry is out-of-order:

[    0.143698] kvm [1]: sys_reg table sys_reg_descs+0x50/0x7490 out of order (1)

There's redundant information now, the entry can be calculated from the
table offset, but printing the offset directly is certainly convenient.

I like it more than passing the table name, if you agree I'll send a v2
with this change.

Thanks,
Alex