Re: [Socketcan-users] mscan: RXACT never got cleared

[Wolfgang Grandegger]

Hi Fu,

Wolfgang Grandegger wrote:
> Luotao Fu wrote:
>> Hi,
>>
>> On Wed, Nov 04, 2009 at 03:25:16PM +0100, Wolfgang Grandegger wrote:
>>> Wolfgang Grandegger wrote:
>>>> Luotao Fu wrote:
>>>>> Hi Wolfgang,
>>>>>
>>>>> On Wed, Nov 04, 2009 at 09:30:18AM +0100, Wolfgang Grandegger wrote:
>>>>>> Luotao Fu wrote:
>> ....
>>>> As I already said, I use an old MPC5200 (without *B*). Will try later
>>>> today with a MPC5200B.
>>> I switched to "MPC5200B v2.2, Core v1.4 at 396 MHz" and the problem does
>>> *not* show up any more. Before I used a "MPC5200 v1.2, Core v1.1 at 396
>>> MHz". Now I remember vaguely, that I was indeed fighting with this
>>> strange behavior.
>> Ah, great to get confirmation that I was not hallucinating :-). It
>> indeed took me some time to sort out what was going wrong in this
>> strange issue. We shall probably ask freescale if this issue is konwn.
> 
> Good luck ;-).
> 
>> BTW. I just checked the datasheet of mpc5200. The slprq should always
>> acked, just like in mpc5200b, and otherwise the controller should be in
>> idle mode with slprq set. It looks really strange that your mpc5200
>> board works properly without being able to get the slpack. Could you do
>> some more inspection on this one by occassion (don't have a mpc5200
>> board here)?
> 
> I don't think it's related to my board. I'm quite sure that it's like
> that on all mpc5200a. Oliver has a board with such a processor, IIRC.
> I will try to find another mpc5200a board.

Here are the values of the canctrl0/1 registers after the sleep and init loop:

 mpc52xx_can f0000980.can: setting BTR0=0x2f BTR1=0x4f
 Sleep: ctrl0=0x12 ctrl1=0xc0
 mpc52xx_can f0000900.can: SLPRQ timeout reached, failed to enter 
 Init: ctrl0=0x03 ctrl1=0xc3

Unfortunately, I do not have another mpc5200a board at hand.

>>> The manual instructs to enter sleep mode first and thereafter the init
>>> mode, but as that did'nt work, sleep and init are set in one step.
>> Well, usually it works already. In case of a failed slprq the we now
>> just force the init.
> 
> Well, the mpc5200a obviously does not enter properly the sleep mode.
> Nevertheless, the board still works fine after an "ifconfig up".
> 
>>> Maybe we simply do s/dev_err/dev_dbg/. What do you think?
>>>
>> Fail to enter sleep mode could indeed be a trouble. I personally would
>> prefer a dev_err here, since it's nevertheless a little hacky here. In
>> the end the message would not appear to frequently since it occures only
>> when we (re)initialize the interface, so it should not be too annoying.

I would apply the patch witch a more descriptive warning message.

Just for curiosity. Could you try the patch below waiting to
the TX messages to get aborted. Does that help? 

Wolfgang.


---
 kernel/2.6/drivers/net/can/mscan/mscan.c |   15 +++++++++++++++
 1 file changed, 15 insertions(+)

Index: trunk/kernel/2.6/drivers/net/can/mscan/mscan.c
===================================================================
--- trunk.orig/kernel/2.6/drivers/net/can/mscan/mscan.c 2009-11-04 
20:07:13.000000000 +0100
+++ trunk/kernel/2.6/drivers/net/can/mscan/mscan.c      2009-11-04 
20:31:16.563302686 +0100
@@ -134,10 +134,21 @@
        if (mode != MSCAN_NORMAL_MODE) {
 
                if (priv->tx_active) {
+                       u8 cantflg;
+
                        /* Abort transfers before going to sleep */#
                        out_8(&regs->cantarq, priv->tx_active);
                        /* Suppress TX done interrupts */
                        out_8(&regs->cantier, 0);
+
+                       printk("tflg=0x%02x\n", in_8(&regs->cantflg));
+                       printk("Wait for TX messages to be aborted "
+                              "tflg=0x%02x\n", in_8(&regs->cantflg));
+                       while ((cantflg = in_8(&regs->cantflg)) != MSCAN_TXE) {
+                               printk("tflg=0x%02x\n", cantflg);
+                               mdelay(1);
+                       }
+                       printk("Done: tflg=0x%02x\n", cantflg);
                }
 
                canctl1 = in_8(&regs->canctl1);
@@ -149,6 +160,8 @@
                                        break;
                                udelay(100);
                        }
+                       printk("Sleep: ctrl0=0x%02x ctrl1=0x%02x\n",
+                              in_8(&regs->canctl0), in_8(&regs->canctl1));
                        if (i >= MSCAN_SET_MODE_RETRIES)
                        /* XXX: sometimes while there are irregular activities
                         * on bus, like somebody keeps retransmitting. The mscan
@@ -171,6 +184,8 @@
                                if (in_8(&regs->canctl1) & MSCAN_INITAK)
                                        break;
                        }
+                       printk("Init: ctrl0=0x%02x ctrl1=0x%02x\n",
+                              in_8(&regs->canctl0), in_8(&regs->canctl1));
                        if (i >= MSCAN_SET_MODE_RETRIES)
                                ret = -ENODEV;
                }


_______________________________________________
Socketcan-users mailing list
Socketcan-users@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/socketcan-users