Re: [PATCH 26/94] Maple Tree: Add new data structure

[Liam Howlett]

* Peter Zijlstra <peterz@xxxxxxxxxxxxx> [210514 07:27]:
> On Fri, May 14, 2021 at 01:13:51PM +0200, Peter Zijlstra wrote:
> > On Wed, Apr 28, 2021 at 03:36:02PM +0000, Liam Howlett wrote:
> 
> > > +static inline struct maple_node *mte_to_node(const struct maple_enode *entry)
> > > +{
> > > +	return (struct maple_node *)((unsigned long)entry & ~127);
> > > +}
> > 
> > > +static inline struct maple_topiary *mte_to_mat(const struct maple_enode *entry)
> > > +{
> > > +	return (struct maple_topiary *)((unsigned long)entry & ~127);
> > > +}
> > 
> > Can we please write masks as hex, also do they want a pretty name?
> > 
> > 
> > This has more magic mask values, proper names might be good:
> > 
> > > +static inline void mte_set_parent(struct maple_enode *enode,
> > > +				 const struct maple_enode *parent,
> > > +				 unsigned char slot)
> > > +{
> > > +	unsigned long bitmask = 0x78;
> > > +	unsigned long val = (unsigned long) parent;
> > > +	unsigned long type = 0;
> > > +
> > > +	switch (mte_node_type(parent)) {
> > > +	case maple_range_64:
> > > +	case maple_arange_64:
> > > +		type = 6;
> > > +		break;
> > > +	default:
> > > +		break;
> > > +	}
> > > +
> > > +	val &= ~bitmask; // Remove any old slot number.
> > > +	val |= (slot << MAPLE_PARENT_SHIFT); // Set the slot.
> > > +	val |= type;
> > > +	mte_to_node(enode)->parent = ma_parent_ptr(val);
> > > +}
> > 
> > > +static inline unsigned int mte_parent_slot(const struct maple_enode *enode)
> > > +{
> > > +	unsigned long bitmask = 0x7C;
> > > +	unsigned long val = (unsigned long) mte_to_node(enode)->parent;
> > > +
> > > +	if (val & 1)
> > > +		return 0; // Root.
> > > +
> > > +	return (val & bitmask) >> MAPLE_PARENT_SHIFT;
> > 
> > 7c is 1111100, but then you're shifting out the one bit that makes the
> > difference from the above magic 0x78. What gives?
> 
> IMO the more obvious way is something like:
> 
> 	(val >> MAPLE_PARENT_SHIFT) & ((1 << MAPLE_SLOT_BITS)-1);
> 
> And then we also see that 3+4 gives 7, which is that magical 127 above,
> are them the same? Related names would be good in that case.
> 


This is a residual of having changed the mte_parent_slot() calculation
from supporting more node types which had different shifts.

Thank you for pointing this out.  I will clean this up.