[Bug 1227] KNI : memory-leaks, each time kni_rx_burst it is allocates new mbufs even if no needs.

[Bug 1227] KNI : memory-leaks, each time kni_rx_burst it is allocates new mbufs even if no needs.

[bugzilla]

[-- Attachment #1: Type: text/plain, Size: 2040 bytes --]

https://bugs.dpdk.org/show_bug.cgi?id=1227

            Bug ID: 1227
           Summary: KNI : memory-leaks, each time kni_rx_burst it is
                    allocates new mbufs even if no needs.
           Product: DPDK
           Version: 22.11
          Hardware: All
                OS: All
            Status: UNCONFIRMED
          Severity: normal
          Priority: Normal
         Component: ethdev
          Assignee: dev@dpdk.org
          Reporter: yasinncaner@gmail.com
  Target Milestone: ---

Created attachment 251
  --> https://bugs.dpdk.org/attachment.cgi?id=251&action=edit
flow_kni_rx_burst

Hello

I tried to explain memory-leak via graph 

https://pastebin.ubuntu.com/p/s4h5psqtgZ/


scenario : 

1. kni_alloc allocates 32 units (kni->alloc_q) then completes initialization
2. rte_kni_rx_burst receive 1 packet from kernel. 1 of 31 units is filled by
message, 31 units is empty but it is allocated in kni->alloc_q.
3. rte_kni_rx_burst then allocates another 32 units i dont know why but there
is a comment like this."If buffers removed, allocate mbufs and then put them
into alloc_q" ( How to tests buffers are removed or not? there is not any query
in code)
4. 1 message mbuf is transmited via rte_eth_tx_burst . 
5. 1 message mbuf never back to freed and back to original memory pool. 31+32
units memory act like in use but they are not. never back to original memory
pool
6. this cycle goes on to 2 .

? Why there is newer allocation in rte_kni_rx_burst i dont know.
? I tried to free 1 packet via rte_pktmbuf_free but sometimes corrupt to pool
like below. sometimes 1 packet/unit back to original memory pool
[UseCount_mpool:65][avail_mpool:10334]
[UseCount_mpool:4294967135][avail_mpool:10560]
? In my view, after transmiting message to mbufs should be in to kni->free_q.
? in my view, there is missing query in func kni_allocate_mbufs to test alloc_q
is empty or not.
Best regards.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[-- Attachment #2: Type: text/html, Size: 4135 bytes --]

[Bug 1227] KNI : memory-leaks, each time kni_rx_burst it is allocates new mbufs even if no needs.

[bugzilla]

[-- Attachment #1: Type: text/plain, Size: 630 bytes --]

https://bugs.dpdk.org/show_bug.cgi?id=1227

Yasin CANER (yasinncaner@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |INVALID
             Status|IN_PROGRESS                 |RESOLVED

--- Comment #3 from Yasin CANER (yasinncaner@gmail.com) ---
it is not possible to pin-point the BUG/Leak so it may be not possible to fix
this way. Also, this patch has side-effect against performance.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[-- Attachment #2: Type: text/html, Size: 2750 bytes --]