- 22 May, 2015 13 commits
-
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
This works (for test-case 1576 at least) on a 16k stack and by default we have a 48k stack. Augment the param description with some helpfull pointers.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
the necessary #ifdef for the sparc64/32bit_userland thing.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
Gcc is really, really weird...
-
Poul-Henning Kamp authored
We will not be able to put the more complex and compound parameters into this table, but it will help keep types in struct param and tweak_functions in sync. Feel free to help get through the list.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
- 21 May, 2015 18 commits
-
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Nils Goroll authored
(even if we exec env python, we should check early if we got one)
-
Nils Goroll authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
Instead of trying to steal vbc's away from the waiter, send the request and wait for the waiter to hand the vbc back to us. All waiters but poll still disabled ... coming up next.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
sparse fd-sets efficiently.
-
Poul-Henning Kamp authored
to ensure this order
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
Don't worry, they'll will come back later.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
and the ones which deal with fd's being waited on (Wait_*)
-
Nils Goroll authored
-
- 20 May, 2015 1 commit
-
-
Lasse Karstensen authored
-
- 18 May, 2015 8 commits
-
-
Poul-Henning Kamp authored
timestamps, when the Bereq timestamp is taken after our write(2) to the socket, which may have caused us to loose the CPU.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
webserver" and dispatches a thread for each incomming connection. This is an example use: server s0 { loop 10 { rxreq txresp -body "foo1" } rxreq txresp -hdr "Connection: close" -body "foo1" expect_close } -dispatch Each connection will spawn a dynamically created s%d instance starting with s1, s2 ... Each of these will respond to 11 requests on the accepted connection, the last response will have "Connection: close" and they will expect the other end (varnish) to do that. The main limitation on using this for bulk traffic tests is the finite and limited size of the vtc_log which varnishtest will collect (half a megabyte). Test b00048 is about as much traffic as will fit in the vtc_log.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Nils Goroll authored
Seen on a solaris vintage edition (snv_111b) as SIGSEGV caused by strlen(NULL) in r01036.vtc and r01037.vtc: a z_stream msg can be NULL.
-