buglwIP - A Lightweight TCP/IP stack - Bugs: bug #59349, TCP Memory Issues

 
 

You are not allowed to post comments on this tracker with your current authentication level.

bug #59349: TCP Memory Issues

Submitted by:  Dave Smith <dave_smith_inst>
Submitted on:  Mon 26 Oct 2020 10:39:03 AM UTC  
 
Category:  TCP Severity:  3 - Normal
Item Group:  Crash Error Status:  None
Privacy:  Public Assigned to:  None
Open/Closed:  Open Planned Release:  None
lwIP version:  git head

Tue 03 Nov 2020 10:44:34 AM UTC, comment #1: 

After doing some further investigation work I have found the issue appears to be linked to TCP_WRITE_FLAG_MORE and TCP_WRITE_FLAG_COPY.

In my original project I chose to use the TCP_WRITE_FLAG_MORE as I wish to control the devices memory (I would like to keep MEM_SIZE as low as possible.)

If I change my tcp_write function to do a mem copy with the TCP_WRITE_FLAG_COPY and assign addition memory, I no longer get tcp_write memory errors. I also notice my queue len only increased by 1 when I call tcp_write with the copy flag as oppose to the increase by 2 noted in my original bug report.

Dave Smith <dave_smith_inst>
Mon 26 Oct 2020 10:39:03 AM UTC, original submission:  

Hello LwIP dev team,

I am using the LwIP 2.1.2 version on the TI ARM M4 chipsets (LM3S and TM4C). This is running with no OS where the TCP is used to send large quantities of data. I have attached my LwIP trace below which shows the issue.

In this trace, the connection poll function is called, where I take my ring buffer which holds the data to be transmitted, and loop over to create the tcp_write calls for up to 1388 bytes at a time. From here you can see the following segments are queued,
tcp_write: queueing 76191:77579
tcp_write: queueing 77579:78967
tcp_write: queueing 78967:80355

however, when tcp_output is called, only the final segment is output.

tcp_output_segment: 78967:80355

I have added some additional tracing inside memp.c (memp_malloc: and memp_free:) to log when TCP_segments are allocated and deallocated. Because the tcp_segs are never output, the memory is never deallocated and thus the system will eventually fail once all the tcp segs have been allocated.

The TCP still recognizes there are packets in the queue to be sent as the trace line '_tcp_buff_space: 1388, queue len: 6' reveals after the fault occurs.

This is a call from my own tcp_api where the code first checks there is space in the TCP_SEND_BUF and TCP_QUEUE_LEN prior to making any calls to tcp_write.

1603543340,tcp_output: nothing to send (00000000)
1603543340,GEN: Total bytes read: 62460, Offset: 62460, Data Read: 4164
1603543341,TCP: conn_poll
1603543341,TCP: tcp_chunks
1603543341,TCP: _tcp_buff_space: 5552, queue len: 0
1603543341,TCP: tcp_write. index: 0, len: 1388
1603543341,tcp_write(pcb=200025d4, data=2000c134, len=1388, apiflags=2)
1603543341,memp_malloc: TCP_SEG size: 20, number: 8, base: 20002778, free: 20002804
1603543341,memp_malloc: avail: 8, used: 0, max: 3, err: 0
1603543341,tcp_write: queueing 76191:77579
1603543341,TCP: _tcp_buff_space: 4164, queue len: 2
1603543341,TCP: tcp_write. index: 1388, len: 1388
1603543341,tcp_write(pcb=200025d4, data=2000c6a0, len=1388, apiflags=2)
1603543341,memp_malloc: TCP_SEG size: 20, number: 8, base: 20002778, free: 200027DC
1603543341,memp_malloc: avail: 8, used: 1, max: 3, err: 0
1603543341,tcp_write: queueing 77579:78967
1603543341,TCP: _tcp_buff_space: 2776, queue len: 4
1603543341,TCP: tcp_write. index: 2776, len: 1388
1603543341,tcp_write(pcb=200025d4, data=2000cc0c, len=1388, apiflags=2)
1603543341,memp_malloc: TCP_SEG size: 20, number: 8, base: 20002778, free: 20002804
1603543341,memp_malloc: avail: 8, used: 2, max: 3, err: 0
1603543341,tcp_write: queueing 78967:80355
1603543341,TCP: tcp_chunks END
1603543341,TCP: conn_poll END
1603543341,tcp_output_segment: 78967:80355
1603543341,tcp_output: nothing to send (00000000)

1603543341,TCP
1603543341, xmit: 4700
1603543341, recv: 2300
1603543341, fw: 0
1603543341, drop: 0
1603543341, chkerr: 0
1603543341, lenerr: 0
1603543341, memerr: 0
1603543341, rterr: 0
1603543341, proterr: 0
1603543341, opterr: 0
1603543341, err: 0
1603543341, cachehit: 2294

1603543341,MEM HEAP
1603543341, avail: 5840
1603543341, used: 2684
1603543341, max: 2752
1603543341, err: 0

1603543341,MEM UDP_PCB
1603543341, avail: 4
1603543341, used: 2
1603543341, max: 2
1603543341, err: 0

1603543341,MEM TCP_PCB
1603543341, avail: 4
1603543341, used: 1
1603543341, max: 2
1603543341, err: 0

1603543341,MEM TCP_PCB_LISTEN
1603543341, avail: 8
1603543341, used: 1
1603543341, max: 1
1603543341, err: 0

1603543341,MEM TCP_SEG
1603543341, avail: 8
1603543341, used: 3
1603543341, max: 3
1603543341, err: 0

1603543341,MEM SYS_TIMEOUT
1603543341, avail: 8
1603543341, used: 5
1603543341, max: 5
1603543341, err: 0

1603543341,MEM PBUF_REF/ROM
1603543341, avail: 8
1603543341, used: 3
1603543341, max: 3
1603543341, err: 0

1603543341,MEM PBUF_POOL
1603543341, avail: 14
1603543341, used: 0
1603543341, max: 4
1603543341, err: 0
1603543341,TCP: conn_poll
1603543341,TCP: tcp_chunks
1603543341,TCP: _tcp_buff_space: 1388, queue len: 6
1603543341,TCP: tcp_chunks END
1603543341,TCP: conn_poll END
1603543341,tcp_output: nothing to send (00000000)
1603543341,TCP: conn_poll
1603543341,TCP: tcp_chunks
1603543341,TCP: _tcp_buff_space: 1388, queue len: 6
1603543341,TCP: tcp_chunks END
1603543341,TCP: conn_poll END
1603543341,tcp_output: nothing to send (00000000)
1603543341,TCP: conn_poll
1603543341,TCP: tcp_chunks
1603543341,TCP: _tcp_buff_space: 1388, queue len: 6
1603543341,TCP: tcp_chunks END
1603543341,TCP: conn_poll END
1603543342,TCP: conn_poll
1603543342,TCP: tcp_chunks
1603543342,TCP: _tcp_buff_space: 1388, queue len: 6
1603543342,TCP: tcp_chunks END
1603543342,TCP: conn_poll END
1603543342,TCP: conn_poll
1603543342,TCP: tcp_chunks
1603543342,TCP: _tcp_buff_space: 1388, queue len: 6
1603543342,TCP: tcp_chunks END
1603543342,TCP: conn_poll END
1603543342,TCP: conn_poll
1603543342,TCP: tcp_chunks
1603543342,TCP: _tcp_buff_space: 1388, queue len: 6
1603543342,TCP: tcp_chunks END
1603543342,TCP: conn_poll END

If you would like anymore info then please let me know. I can easily recreate this issue so if you would like any additional traces please feel free to ask!

Thank you,

Dave Smith

Dave Smith <dave_smith_inst>

 

No files currently attached

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by dave_smith_inst (Submitted the item)
  • -email is unavailable- added by dave_smith_inst (LWIP TCP Issue)
  •  

    Do you think this task is very important?
    If so, you can add your encouragement to it.
    This task has 0 encouragements so far.

    Only logged-in users can vote.

     

     

     

    Follows 1 latest change.

    Date Changed by Updated Field Previous Value => Replaced by
    2020-10-26 dave_smith_inst Carbon-Copy- => Added -email is unavailable-

    Back to the top


    Powered by Savane 3.7