Re: comm_connect_addr broke in squid3/trunk?

Alex Rousskov wrote:
> Hello,
>
> I am getting the following "Invalid argument" error on any
> connect(2) after updating my trunk copy:
>
>> 00:16:24.833| commConnectStart: FD 10, cb 0x8b60780*1, 10.0.0.104:1344
>> 00:16:24.833| comm_add_close_handler: FD 10, handler=1, data=0x8c64010
>> 00:16:24.833| cbdataLock: 0x8c64010=1
>> 00:16:24.833| cbdataLock: 0x8c64010=2
>> 00:16:24.833| The AsyncCall SomeCloseHandler constructed, this=0x8b5e7c0 [call4
>> 00:16:24.833| cbdataLock: 0x8c64010=3
>> 00:16:24.833| cbdataUnlock: 0x8c64010=2
>> 00:16:24.833| cbdataUnlock: 0x8c64010=1
>> 00:16:24.833| comm_add_close_handler: FD 10, AsyncCall=0x8b5e7c0*1
>> 00:16:24.833| ipcache_nbgethostbyname: Name '10.0.0.104'.
>> 00:16:24.833| ipcacheCheckNumeric: HIT_BYPASS for '10.0.0.104' == 10.0.0.104
>> 00:16:24.833| ipcache_nbgethostbyname: BYPASS for '10.0.0.104' (already numeric
>> 00:16:24.833| ConnectSateData::connect: to 10.0.0.104:1344
>> 00:16:24.833| comm_connect_addr: connecting socket 10 to 10.0.0.104:1344 (want
>> 00:16:24.833| comm_connect_addr: sock=10, addrinfo( , family=2, addrlen=128 )
>> 00:16:24.833| connect FD 10: (-1) (22) Invalid argument
>> 00:16:24.833| connecting to: 10.0.0.104:1344
>> 00:16:24.833| ConnectStateData::connect: FD 10: * - try again
>> 00:16:24.833| ConnectStateData::connect: FD 10: * - ERR tried too many times al
>> 00:16:24.833| commConnectCallback: FD 10
>> 00:16:24.833| comm_remove_close_handler: FD 10, handler=1, data=0x8c64010
>> 00:16:24.833| commSetTimeout: FD 10 timeout -1
>
> The following changes are suspect, based on "bzr log comm.cc" output
> alone, but it could be a change in another source file, of course.
>
>> revno: 8906
>> committer: Henrik Nordstrom <henrik@henriknordstrom.net>
>> branch nick: trunk
>> timestamp: Tue 2008-03-25 16:40:36 +0100
>> message:
>> Fix memset NULL warning in latest comm.cc changes
>>
>> NULL is a NULL pointer, not 0.
>> ------------------------------------------------------------
>> revno: 8901
>> committer: Amos Jeffries <squid3@treenet.co.nz>
>> branch nick: trunk
>> timestamp: Sun 2008-03-23 23:47:50 +1200
>> message:
>> Fix memory leak in Linux builds.
>
> Any clues about what went wrong?
>
> Thank you,
>
> Alex.
>
>

Probably a muckup by me in the code setting the sockaddr_storage content.
Can you get a trace with output of the sas variable fields?

Amos

-- 
Please use Squid 2.6STABLE19 or 3.0STABLE3