- 08 Jan, 2016 7 commits
-
-
Poul-Henning Kamp authored
-
Geoff Simmons authored
-
Geoff Simmons authored
all other methods use rdlock
-
Dridi Boukelmoune authored
-
Poul-Henning Kamp authored
for more clarity, before we start hacking VCC further.
-
Poul-Henning Kamp authored
Submitted by: github::onovy
-
Poul-Henning Kamp authored
-
- 07 Jan, 2016 1 commit
-
-
Poul-Henning Kamp authored
try to delete a objhead while it still has a waiting list, by forcing the last ref holder to rush the WL. Since the hasher owns the refcounts on objhead, we cannot just mingle req and objcore refcounts. Fortunately we don't need to add another refcounter, because all we really care about is the wl being empty when we drop the last ref. The wl/hsh_rush() mechanism will work differently with different thread-scheduling schenarios, and I cannot definitively rule out that we can drop the last ref on an oh, while there are still req's on the waiting list. Given that, and the existence proof in ticket #1823's race, this might have been the indicated memory-trampler.
-
- 06 Jan, 2016 21 commits
-
-
Federico G. Schwindt authored
In the 304 case we were already skipping it for non-zero C-L. For 204 responses this was introduced as a side effect of 271e1c52.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
of one pointer does not amortize the extra CPU overhead and cache misses.
-
Poul-Henning Kamp authored
Fixes: #1823
-
Poul-Henning Kamp authored
for 5 seconds if there is a waiting list. This allows the waiting list to drain (returning 503) quickly, rather than one by one trying to pester the backend and only giving up after the connect_timeout. Related to: #1823
-
Federico G. Schwindt authored
-
Federico G. Schwindt authored
-
Poul-Henning Kamp authored
new improved probability of getting the VSL out.
-
Poul-Henning Kamp authored
That obscured problems in cleanup
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Martin Blix Grydeland authored
The script lacked a logexpect -wait at the end telling it to actually wait for the required log content. Added a new rule to match first on an actual retry backend transaction and match the header set/unset/set tests on that vxid.
-
Federico G. Schwindt authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
Check if system has closefrom(2) and use that where available
-
Poul-Henning Kamp authored
represents the current limit. Also don't call it every iteration of the loop.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
- 05 Jan, 2016 9 commits
-
-
Federico G. Schwindt authored
-
Federico G. Schwindt authored
-
Federico G. Schwindt authored
-
Federico G. Schwindt authored
-
Federico G. Schwindt authored
-
Federico G. Schwindt authored
Avoids memcpy/free if the buffer was dynamic.
-
Federico G. Schwindt authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
- 04 Jan, 2016 2 commits
-
-
Poul-Henning Kamp authored
-
Lasse Karstensen authored
-