Re: [lwip-users] lwIP trouble (tcp_write called from more than one location)

[Slava Zilberfayn]

Hello,

Hopefully  somebody  could  shed  some  light on the things I
observe.

I have

#define LWIP_TCPIP_CORE_LOCKING_INPUT 0
#define LWIP_TCPIP_CORE_LOCKING 0

but  when the RX thread receives the ACK in the middle
of   the   TCP/IP core   thread   serving  my  second  netconn_write
(specifically  being  inside  tcp_write),  it (the rx-thread) ends up calling
tcp_write     itself.     My     understanding     is    that
LWIP_TCPIP_CORE_LOCKING_INPUT==0 should prevent the RX thread
from  trying to call  tcp_write directly. Am I wrong? In this configuration 
ASSERT_CORE_LOCKED()  is empty and
does  not  do  anything,  so  that  seems  to  be  an invalid
situation.

Is  that  a  legal  scenario? Immediately thereafter
things blow up.

I have tried with
#define LWIP_TCPIP_CORE_LOCKING_INPUT 1
#define LWIP_TCPIP_CORE_LOCKING 1

in this case almost  exact  same thing happens, with minor difference that
the  first  tcp_write  is now called from the tcp_echo thread
(via netconn_write).

Looking  through  source code I see that TCP_EVENT_SENT macro
in  tcp_input()  is  dependent  upon LWIP_EVENT_API, which is
supposed  to  make  it  use  message  instead  of calling the
function directly. But when I enable LWIP_EVENT_API == 1 the code
does   not   build   at   all,   because   tcp_recv()  used  in
api_msg.c:520(setup_tcp())  is  only  declared  in tcp.h when
LWIP_CALLBACK_API == 1.

Two  concurrent  calls  to  netconn_write  seems  to be wrong
situation, but I don't see what I do wrong here.

Thank you, Slava



lurno> Message: 3
lurno> Date: Tue, 26 Feb 2019 22:10:24 +0100
lurno> From: Simon Goldschmidt <address@hidden>
lurno> To: address@hidden
lurno> Subject: Re: [lwip-users] lwIP trouble
lurno> Message-ID: <address@hidden>
lurno> Content-Type: text/plain; charset=utf-8; format=flowed


lurno> On 26.02.2019 22:07, Slava Zilberfayn wrote:
>> Hello all,
>>
>> I've  done some more testing with tcpecho example. When I set
>> chunk  size  to  be  less  than the size of the tcpecho
>> packets,  I  get  exactly  the same failure during the second
>> call  to netconn_write. When the packet size small than chunk
>> size  it  crashes fairly quickly. For example I set chunk size
>> to  400  in the tcpecho example and send 400 bytes packets to
>> it. Works fine.
>>
>> If   I  start  sending  401 bytes, it crashes fairly quickly.
>> There   must   me   something  wrong  with  my  settings. Any
>> help would be appreciated.


lurno> If you could get this reproduced with minimal changes on the win32 or
lurno> linux port, it would be *much* easier to see what's going on.


lurno> Regards,

lurno> Simon




lurno> ------------------------------

lurno> Message: 4
lurno> Date: Tue, 26 Feb 2019 18:45:29 -0500
lurno> From: Dave Nadler <address@hidden>
lurno> To: Johan Borkhuis <address@hidden>,     Mailing list for lwIP
lurno>         users <address@hidden>
lurno> Subject: Re: [lwip-users] Throughput benchmark question - nasty pauses
lurno> Message-ID: <address@hidden>
lurno> Content-Type: text/plain; charset="utf-8"; Format="flowed"

lurno> I figured out how to get the wireshark trace,
lurno> but how to get the wireshark GUI to output the summary below in text
lurno> baffles me, hope the pic is OK:
lurno> Everything is going swimmingly until 4316.
lurno> I don't understand the meaning of "previous segment not captured" here -
lurno> something got dropped.
lurno> And then it takes a second to get going again.
lurno> Any pointers appreciated!
lurno> Thanks,
lurno> Best Regards, Dave

lurno> On 2/20/2019 1:45 AM, Johan Borkhuis wrote:
>> Dave,
>>
>> First thing would be to sniff the network using Wireshark, and see 
>> what happens there when the traffic pauses.
>> This would usually give a good indication on what did happen.
>>
>> Regards,
>> Johan
>>
>> On 2019-02-20 00:15, Dave Nadler wrote:
>>> Hi - Newbie here trying to do some basic throughput tests.
>>> LwIP 2.1.2 on FreeRTOS 9, ST32F429, IPv4, TCP.
>>>
>>> I want to see how much I can consistently push through the stack.
>>> Made a simple test server (sockets API) which repeatedly outputs
>>> 101-character lines.
>>> I access the server via PuTTY raw mode on Winbloze over a local
>>> network.
>>> I can usually send 3 lines per msec for a second (3000 lines in 1
>>> second), but...
>>> Sometimes, I get ~ 1-second pauses (as seen in Putty or TeraTerm).
>>>
>>> How should I go about understanding where the pauses come from?
>>>
>>> Thanks in advance for any hints,
>>> Best Regards, Dave
>>>
>>> -- 
>>> Dave Nadler, USA East Coast voice (978) 263-0097, address@hidden,
>>> Skype
>>> ?Dave.Nadler1
>>> _______________________________________________
>>> lwip-users mailing list
>>> address@hidden
>>> https://lists.nongnu.org/mailman/listinfo/lwip-users
>>
>>


-- 
Slava Zilberfayn mailto:address@hidden
Phone: 416 7289367

Home Electronics, www.home-electro.com
100 Drumlin Circle,
Suite 205
Concord, ON, L4K 3E6
CANADA

stack.PNG
Description: PNG image