Re: [Discuss-gnuradio] memory leak with basic block

[CEL]

Hi Francesco,

interesting!

Can you tell us how that memory leak manifests, maybe even whether you
can pinpoint the reason?

Best regards,
Marcus

On Tue, 2019-07-30 at 11:24 +0000, Francesco Costa wrote:
> Hi,
> we are designing a block that receives multiple streams of bursts of data and 
> produces in return other bursts of data. there is no correlation on the 
> amount of data at the input and at the output and it's not possible to tell 
> in advance how much data it will be arriving.
> the input stream is tagged with rx_eob as it consists in variable length 
> packets each of which is tag terminated
> we are seeing a number of odd behaviours:
> 1) sometimes the general_work will receive, at the end of the buffer, an 
> incomplete packet. the issue with this is that the packet that has been 
> broken in two pieces ends up having a rx_eob tag at the interruption point, 
> thus splitting it effectively in two packets
> 2) if we don't consume all the input data immediately in a call to 
> general_work we correctly see data we left being resent with next buffer 
> however we see a memory leak
> 3) if we consume ninput_items passed to general_work for each stream we see 
> memory leak. odd enough memory leak does not show up if we consume 
> noutput_items which may be bigger than ninput_items
> 4) if we produce less than noutput_items by calling the produce function and 
> returning WORK_CALLED_PRODUCE we see a memory leak
> 
> we are currently using Release 3.7.13.3 (git hash 
> 3fef6dc5d9174c172f75d97b0e332171b5a53f18).
> 
> any help on this would be greatly appreciated...
> thanks,
> Francesco.
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

smime.p7s
Description: S/MIME cryptographic signature