| Age | Commit message (Collapse) | Author |
|---|
|
This is still pretty stupid, but greatly reduces my routing-sim time.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
We fill the upper buckets after we've moved on from the first bucket
in this layer. This means two things:
1) If we have to look at an upper layer, we need to look at the next bucket,
unless offset is 0.
2) We need to keep looking up layers in the corner case, too.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
This catches duplicate timer_add() calls, as well as meaning we don't
need to track if the timer is active before calling timer_del().
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
The linked list method was problematic, especially if timers delete
other expired timers (eg. the next one in the expired list: the timer_del
will delete it from expired, but that's a bit unexpected).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
There are a couple of small problems with the includes used in most of
ccan's _info files.
* _info routinely uses printf(), and so should include <stdio.h>, but
only some of them do. We get away with it, because they do include
<string.h>, which apparently includes <stdio.h> indirectly, but we should
be explicit about it.
* Most _info files were including config.h after the system headers.
That _seems_ sensible, but actually causes problems. Because config.h
defines _GNU_SOURCE it can change the behaviour of the system headers.
More specifically it can make them behave differently to how the individual
module headers (which have included config.h) expects them to behave.
This patch adjusts all the existing _info files and, more importantly,
the template constructed by ccanlint.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Add examples and a documentation fix.
Remove unused cascade function (was used in initial always-step-1-bucket
version).
Restore timers_dump() to within CCAN_TIMER_DEBUG.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
There's a sweet spot around 5, without going all the way to 13 (which
would make each level 64/128k large).
TIMER_LEVEL_BITS 2: 100000000 in 5.833399-5.884659(5.85912+/-0.015) (18 levels / 32)
TIMER_LEVEL_BITS 3: 100000000 in 5.267610-5.303589(5.29018+/-0.011) (12-13(12.5+/-0.5) levels / 22)
TIMER_LEVEL_BITS 4: 100000000 in 4.215605-4.417297(4.27771+/-0.062) (9-10(9.1+/-0.3) levels / 16)
TIMER_LEVEL_BITS 5: 100000000 in 3.859340-3.972603(3.90149+/-0.034) (7-8(7.1+/-0.3) levels / 13)
TIMER_LEVEL_BITS 6: 100000000 in 3.976157-4.037230(4.00199+/-0.02) (6 levels / 11)
TIMER_LEVEL_BITS 7: 100000000 in 4.014228-4.082031(4.05024+/-0.022) (5-6(5.4+/-0.49) levels / 10)
TIMER_LEVEL_BITS 8: 100000000 in 3.915615-3.978781(3.94972+/-0.02) (5 levels / 8)
TIMER_LEVEL_BITS 9: 100000000 in 3.859413-4.025842(3.89993+/-0.046) (4-5(4.6+/-0.49) levels / 8)
TIMER_LEVEL_BITS 10: 100000000 in 3.983507-4.170152(4.06743+/-0.055) (4 levels / 7)
TIMER_LEVEL_BITS 11: 100000000 in 3.468756-3.610746(3.52843+/-0.04) (4 levels / 6)
TIMER_LEVEL_BITS 12: 100000000 in 3.274397-3.377530(3.3192+/-0.026) (4 levels / 6)
TIMER_LEVEL_BITS 13: 100000000 in 3.116845-3.178162(3.1398+/-0.017) (3 levels / 5)
TIMER_LEVEL_BITS 14: 100000000 in 3.152599-3.264060(3.20733+/-0.043) (3 levels / 5)
TIMER_LEVEL_BITS 15: 100000000 in 3.186556-3.552432(3.25597+/-0.11) (3 levels / 5)
TIMER_LEVEL_BITS 16: 100000000 in 3.139352-3.485653(3.27263+/-0.11) (3 levels / 4)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
We spend a lot of time searching for the next timer to expire: by caching
the minimum, we can skip most of this work. Even if timers are deleted,
the minimum will be a starting point for searching.
The expected-usage benchmark has to be increased by a factor of 100,
otherwise it's now too short.
Before:
$ ./expected-usage
1000000 in 12.701647935
After:
$ ./expected-usage 1000000
1000000 in 0.061095153
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
