|
Hello Wensong,
I do, because there is some changes accumulated in my ipvs-1.0 tree.
Ok.
Well, maybe you need read the wrr code for the algorithm. ;) After weight
is introduced, the algorithm is a little bit complicated. Servers with
higher weights receive new and get more jobs than servers with lower
weights. Servers with equal weights get an equal distribution of new jobs.
I see the cw, mw, di relation now and the added complexity. The
update_service() callback is called for every destination addition from
user space which in case of the wrr scheduler will update the max
weight, correct? Quite intelligent :)
Why do we keep the struct iphdr *iph as second parameter in
ip_vs_{w}rr_schedule()?
Much faster :) :)
:) Maybe I miss something in the assemble code of gcd_new, where is the
mod (%) operation?
Ohh shouwt!! Forgot half of the assembly. Well, it's just reducing the
stack usage anyway and it's not that this function would be called on a
jiffy basis :):
gcd_lvs:
pushl %ebp
movl %esp, %ebp
pushl %esi
pushl %ebx
subl $4, %esp
movl 8(%ebp), %ebx
movl 12(%ebp), %esi
movl %ebx, %eax
cltd
idivl %esi
testl %edx, %edx
movl %edx, %ecx
je .L7
.p2align 4,,15
.L5:
movl %esi, %ebx
movl %ecx, %esi
movl %ebx, %eax
cltd
idivl %ecx
testl %edx, %edx
movl %edx, %ecx
jne .L5
.L7:
popl %edx
movl %esi, %eax
popl %ebx
popl %esi
popl %ebp
ret
gcd_new:
pushl %ebp
movl %esp, %ebp
pushl %ebx
subl $4, %esp
movl 12(%ebp), %ebx
movl 8(%ebp), %ecx
testl %ebx, %ebx
jle .L14
.p2align 4,,15
.L12:
movl %ecx, %eax
cltd
idivl %ebx
movl %edx, %ecx
xchgl %ebx, %ecx
testl %ebx, %ebx
jg .L12
.L14:
movl %ecx, %eax
popl %ecx
popl %ebx
popl %ebp
ret
Cheers,
Roberto Nibali, ratz
--
echo '[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq'|dc
|
