Enhancing the Efficiency of Memory Management in a Super-Paging Environment by AMSQM - Advanced Operating Systems and Kernel Applications - page 283

▪
Else

δ=1 • Do until size i base-pages are released:

r=max ◦ ◦ rank i

Remove ▪ δ=|B

i

1 |/|B i 2 |

◦ LRU among T r 1 , T r 2 and Q r

◦ P i =max( P i -δ,0)

If count(

◦ c r = c r -1

If ( ◦ SP(x j ))≤2·γ· bound i

Call

◦ c r < s r )

▪ Release ( x j ,i)

Fetch x

▪ s r = s r -1

▪

j to the cache.

• Allocate ( x j ,i)
Release ( x j ,i)
If ((|T

Call

▪ Move SP(x j ) to be the MRU of

T i 2
If count(

•

1 |> P i ) or (|T i 1 |= P i and x j is in B i 2 )
Take the

i

◦ SP(x j ))>2·γ· bound i

If 0≤

◦ LRU page between the LRU

of T i 1 and the LRU of Q i and put it as
the MRU of Bi2. i 1 .

▪ C -∑ c i < size i

▪ Call IncreaseBuffer ( x j ,i)

If we couldn't allocate a con- ▪ • Else

tinuous space of size i
Call

◦ LRU page between the LRU
of T i 2 and the LRU of Q i and put it as
the MRU of Bi2. i 2 .
Allocate ( x n ,i)
If there is a contiguous empty space of

Take the

▪ Release ( x j ,i)

▪ Else

▪ Call Allocate ( x j ,i)

Count( ▪ x j )=γ⋅ bound i • size i

▪ s i = s i +1 in the cache

Fetch x • If SP(x j ) is not in T i 1 , T i 2 , B i 1 or B i 2

If (i>1) and (

◦

j to the cache.

◦ SP(x j ) has ever been in

lists B i 1 or B i 2 )
Call Demote (

◦ SP(x j ) to be the MRU of T i 2
◦ c i = c i +1
Demote ( x n ,i)
Cancel

Move

▪ x j ,i)

◦ o Else • SP(x j )

▪ If 0≤ C -∑ c i < size i • If(i>1)

▪ If (| Q i |+|T i 1 |+|B i 1 |= c i ) ◦ Dsize = size i-1

Else ▪ If (| Q i |+|T i 1 |< c i )

Remove the

•

▪ LRU of Bi2 i 1 ◦ Dsize =1

▪ Call Release ( x j ,i) • free =The biggest available continuous

empty space of maximum Dsize.
if (

▪ Else

▪ Remove the LRU among Q i and • free >0)

o Create superpage x' T i 1 .

Else

◦

j of size free

▪ which must contain x j

o Move x' ▪ If (| Q i |+|T i 1 |+|B i 1 |+|T i 2 | +|B i 2 |> c i ) ◦

j to the MRU of Q free .

▪ If (| Q i |+|T i 1 |+|B i 1 |+|T i 2 | • Else

+|B i 2 |=2⋅ c i )
Remove the

▪ Call Release (x j ,1).

Fetch x ▪ LRU of Bi2 i 2 ▪

j to the cache.

▪ Call Release ( x j ,i)

Fetch x

▪ move x

j to the MRU of Q 1 .

▪

j to the cache.

▪ Move SP(x j ) to be the MRU of

eValuation and reSultS T i 1

Else ▪

▪ Allocate ( x j ,i)
IncreaseBuffer ( x j ,i)

Call Actually, the best way to evaluate the AMSQM

page replacement algorithm is by considering its