Experimental Physics and
Industrial Control System

Michael Davidsaver via Core-talk <core-talk at aps.anl.gov> · Wed, 12 Feb 2020 11:13:13 +0000

On 2/12/20 3:10 AM, Michael Davidsaver wrote:
> Hi Andrew,
> 
> It think there was some confusion.  Could you test all of the commits on
> the 'reuseport' branch in my github repo.

Scratch this.  Some of the confusion is mine.  I had already pushed this
to the main 7.0 branch.  So just update and test that.

> https://github.com/mdavidsaver/epics-base/pull/new/reuseport
> 
> 
> On 2/6/20 3:10 PM, Andrew Johnson via Core-talk wrote:
>> On 2/6/20 12:28 PM, Michael Davidsaver via Core-talk wrote:
>>> My attempt to fix epicsSocketEnableAddressUseForDatagramFanout() on vxWorks (which still needs confirmation)
>>> and RTEMS had a side-effect on Linux where it seems that SO_REUSEADDR
>>> and SO_REUSEPORT have the same effect, but are not interchangeable.
>> Sorry.
>>
>> I cherry-picked 8dc131dc4cc04eef3df513a978576442af9405b1 from your branch:
>>> iocanj> osiSockTest
>>> 1..18
>>> ok  1 - osiSockAttach
>>> ok  2 - epicsSocketCreate INET, DGRAM, 0
>>> ok  3 - setsockopt BROADCAST := 1
>>> ok  4 - getsockopt BROADCAST => 1
>>> ok  5 - setsockopt BROADCAST := 0
>>> ok  6 - getsockopt BROADCAST => 0
>>> ok  7 - setsockopt MULTICAST_LOOP := 1
>>> ok  8 - getsockopt MULTICAST_LOOP => 1
>>> ok  9 - setsockopt MULTICAST_LOOP := 0
>>> ok 10 - getsockopt MULTICAST_LOOP => 0
>>> ok 11 - setsockopt IP_MULTICAST_TTL := 1
>>> ok 12 - getsockopt IP_MULTICAST_TTL => 1
>>> ok 13 - setsockopt IP_MULTICAST_TTL := 0
>>> not ok 14 - getsockopt IP_MULTICAST_TTL => 1
>>> # First test if epicsSocketEnableAddressUseForDatagramFanout() is necessary
>>> ok 15 - bind() to port 50463
>>> ok 16 - bind() to 50463 error -1, 48
>>> # Now the real test
>>> ok 17 - bind() to port 50463
>>> not ok 18 - bind() to 50463 error -1, 48
>>>
>>>     Results
>>>     =======
>>>        Tests: 18
>>>       Passed:  16 = 88.89%
>>>       Failed:   2 = 11.11%
>>> iocanj> puts strerror(48)
>>> S_errno_EADDRINUSE
>>> value = 10 = 0xa
>>
>> Test 14 always fails on VxWorks, I just haven't had time to chase the reason.
>>
>> - Andrew
>>
>> -- 
>> Complexity comes for free, Simplicity you have to work for.
>>
> 

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System