Re: Evaluating the c_can driver for EG20T

[Wolfgang Grandegger <wg@grandegger.com>]

On 04/25/2013 12:11 PM, j.p.lammertink wrote:
> 
> <><> Hi Wolfgang,
> <><>< Hi Jeroen,
> <><><> Hi Wolfgang,
> <><><> I think I have a patch for the "Loosing CAN messages problem"
> <><><> See at the bottom Appendix 7 Sugested patch.
> <><><> I realize that this patch has a performance penalty, but I didn't dare
> <><><> to make more rigorous changes.
> <><><> I've put the patch on top of Patch 1, of the patches that you sugested (see
> <><><> text below)
> <><><> I'm interested in all feedback.
> <><>< Well, we want to get rid of the pch_can driver sooner than later.
> <><>< Therefore I would appreciate if you could test and improve the c_can
> <><>< driver instead.
> <><>< Wolfgang.
> <><> I tried once before but I did not succeed. I probably did something
> <><> wrong. Can you tell me how I can replace the pch_can driver by the
> <><> c_can driver?
> <><> We have reduced confidence in the pch_can driver and therefor we
> <><> are also intrested in evaluating the c_can driver.
> <>< Good, anyway, could you sent me your patches pch_can.c to understand
> <>< better your lates problems.
> <><> Note, after applying patch 1/7 to the pch_can driver, I also
> <><> attemped to
> <><> apply the patches 2/7-7/7 which mainly/only(?) effect the c_can
> <><> driver.
> <><> 1/7 went succesfull, 2/7-7/7 not. Do I still need to apply those
> <><> patches?
> <>< What exact kernel disto and kernel version are you using. With that
> <>< knowledge I could provid new patches.
> <>< Wolfgang.
> <> Hi Wolfgang,
> <> I've included the patch in the E-mail:
> <> "Losing CAN messages with socket-CAN" of 24 apr 2013 11:11. (Appendix 7)
> <> It also contains kernel disto and kernel version and also
> <> the problem description.
> < OK. could you send the file "pch_can.c" you used (sorry for my bad
> < English). Your kernel is obviously 3.5.0-27-generic from Ubuntu. I'm
> < going to rebase the patches on top of mainline Linux 3.5.0. Any chance
> < for you to use a recent mailing Linux version?
> <> You replied to above mentioned E-mail:
> <> "Therefore I would appreciate if you could test and improve the c_can
> <> driver instead."
> <> Can you please indicate how to activate the c_can driver
> <> (instead of the pch_driver)?
> < You need to deselect the PCH_CAN and select the C_CAN PCI driver in your
> < kernel config. I use "make menuconfig" for that purpose. But you need
> < the complete patch stack.
> < Wolfgang.
> 
> Hi Wolfgang,
> 
> I've attached pch_can.c.
> I also enlarged the scope of the spinlock in the pch_xmit() function.

Ah! Could you please revert your other fix and retry?

> With this version of the pch_can driver I do no longer loose any messages.
> Seems that the problem is solved. I have to admit that I do not
> thoroughly understand this driver (yet) and that I just made educated
> guesses based on testing and debugging.

There is obviously a race which I also have not yet fully understood.

Still the driver has other issues, unfortunately.

Wolfgang.