Re: [avrdude-dev] [patch #7635] Fix avrftdi.c/h Random lockup/moved globals to private data against 5.11.1

[Hannes Weisbach]

Am 16.10.2011 um 06:23 schrieb Doug S:

> URL:
>  <http://savannah.nongnu.org/patch/?7635>
> 
>                 Summary: Fix avrftdi.c/h Random lockup/moved globals to
> private data against 5.11.1
>                 Project: AVR Downloader/UploaDEr
>            Submitted by: rickyrockrat
>            Submitted on: Sun 16 Oct 2011 04:23:53 AM GMT
>                Category: None
>                Priority: 5 - Normal
>                  Status: None
>                 Privacy: Public
>             Assigned to: None
>        Originator Email: 
>             Open/Closed: Open
>         Discussion Lock: Any
> 
>    _______________________________________________________
> 
> Details:
> 
> I found a strange lockup with avrftdi on writes - verifys and reads work with
> no problem.  I had a hardware change (new mother board) so I thought at first
> that was it, but it happens on at least two different computers.
> 
> At random places while writting, it will lock up and loop forever.
> 
> strace gives this (forever, AFAIK):
> 
> ioctl(3, USBDEVFS_REAPURBNDELAY, 0x7fff2a91b1e8) = -1 EAGAIN (Resource
> temporarily unavailable)
> 
> Removing the purge buffers seems to fix the problem. I'm not sure what all
> ramifications the purge has, since I didn't bother to delve into the ftdi
> code, but I found one other example, and I noticed they had found this problem
> and removed the purge call as well.
> 
> This patch also moves the globals to a private structure, as per Joerg's last
> request to do so.
> 
If i remember correctly, this happens if the/a USB-device is re-enumerated when 
it is in use. Re-enumeration may be triggered by a loose connection (bad 
cable/soldering joint). Have you tried different cables/programmers/USB-ports?

As fix, you commented out the flush-routine, but that will trigger problems 
when using high speed transfers, as the comment of avrftdi_flush() states. At 
least I had problems, using an FTDI 2232C. Do you have problems using 1MHz 
clock speed and more, without avrftdi_flush()?

Thanks for taking the initiative on moving the globals variables to a private 
structure.

Best regards,
Hannes