|
Hi Wensong,
I promised to clean up the code a bit but this week was once again a desaster
time-wise. First a family member passed away and I thus had to organise the
funeral and other things and then I had some issues with the military (I finally
got released from duty, after 7 years of pain and anger in the army!!!). So
there is only little coming from me and you might as well just release the next
drop:
This patch addresses the locking in the est scheduler (I wasn't aware of it :)).
It is then compliant to 2.5.x locking. Please consider applying if correct.
Questions:
o What about syncing the OVERLOADED code from 2.5.x to 2.4.x?
o What do you think about introducing the sltimer changes from 2.5.x to 2.4.x?
o Why do we have two different services list traversals for rr and wrr?
Best regards,
Roberto Nibali, ratz
--
echo '[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq' | dc
diff -ur ipvs-1.0.9-misc/ipvs/ip_vs_est.c ipvs-1.0.9-cleanup/ipvs/ip_vs_est.c
--- ipvs-1.0.9-misc/ipvs/ip_vs_est.c 2002-07-11 16:26:41.000000000 +0200
+++ ipvs-1.0.9-cleanup/ipvs/ip_vs_est.c 2003-06-26 11:36:38.000000000 +0200
@@ -139,6 +139,7 @@
est->last_outbytes = stats->outbytes;
est->outbps = stats->outbps<<5;
+ write_lock_bh(&est_lock);
est->next = est_list;
if (est->next == NULL) {
init_timer(&est_timer);
@@ -146,7 +147,6 @@
est_timer.function = estimation_timer;
add_timer(&est_timer);
}
- write_lock_bh(&est_lock);
est_list = est;
write_unlock_bh(&est_lock);
return 0;
|
