Am 2014-07-25 15:33, schrieb Marc Kleine-Budde:

On 07/25/2014 12:50 PM, Stefan Agner wrote:

quoted

Am 2014-07-16 08:43, schrieb Stefan Agner:

quoted

Am 2014-07-15 16:24, schrieb Marc Kleine-Budde:
<snip>

quoted

quoted

@@ -150,18 +171,20 @@
  * FLEXCAN hardware feature flags
  *
  * Below is some version info we got:
- *    SOC   Version   IP-Version  Glitch-  [TR]WRN_INT
- *                                Filter?   connected?
- *   MX25  FlexCAN2  03.00.00.00     no         no
- *   MX28  FlexCAN2  03.00.04.00    yes        yes
- *   MX35  FlexCAN2  03.00.00.00     no         no
- *   MX53  FlexCAN2  03.00.00.00    yes         no
- *   MX6s  FlexCAN3  10.00.12.00    yes        yes
+ *    SOC   Version   IP-Version  Glitch-  [TR]WRN_INT  Memory err
+ *                                Filter?   connected?  detection
+ *   MX25  FlexCAN2  03.00.00.00     no         no         no
+ *   MX28  FlexCAN2  03.00.04.00    yes        yes         no
+ *   MX35  FlexCAN2  03.00.00.00     no         no         no
+ *   MX53  FlexCAN2  03.00.00.00    yes         no         no
+ *   MX6s  FlexCAN3  10.00.12.00    yes        yes         no
+ *   VF610 FlexCAN3  ?               no         no        yes

                                        ^^         ^^

Can you check the datasheet if the flexcan core has a "Glitch Filter
Width Register (FLEXCANx_GFWR)"

There is no such register called GFWR/Glitch Filter or similar.

quoted

Can you check if the core generates a warning interrupt with the current
setup, if you don't switch on bus error reporting? This means internally
the [TR]WRN_INT is connected and works as specified.

Ok, so I disabled TWRNMSK (Bit 11) in the control register and printed
out the error and status register (ESR1), this is what I get:
[  191.285295] flexcan_irq, esr=00040080
[  191.288996] flexcan_irq, ctrl=17092051

Bit 17 (TWRNINT) is not set while TWRNMSK is disabled. Hence [TR]WRN_INT
is not connected?

Ping. Anything open/to do from my side?

Please keep the printing of esr and ctrl in the interrupt handler, add a
#define DEBUG in the driver, but do not change anything else. Then:

Ok, my changes look like this:

diff --git a/drivers/net/can/flexcan.c b/drivers/net/can/flexcan.c
index 1c31a5d..fe8b81c 100644
--- a/drivers/net/can/flexcan.c
+++ b/drivers/net/can/flexcan.c

@@ -19,6 +19,7 @@
  *
  */
 
+#define DEBUG
 #include <linux/netdevice.h>
 #include <linux/can.h>
 #include <linux/can/dev.h>

@@ -743,6 +744,9 @@ static irqreturn_t flexcan_irq(int irq, void

*dev_id)
 
        reg_iflag1 = flexcan_read(&regs->iflag1);
        reg_esr = flexcan_read(&regs->esr);
+
+       printk("flexcan_irq, esr=%08x\n", reg_esr);
+       printk("flexcan_irq, ctrl=%08x\n", flexcan_read(&regs->ctrl));
        /* ACK all bus error and state change IRQ sources */
        if (reg_esr & FLEXCAN_ESR_ALL_INT)
                flexcan_write(reg_esr & FLEXCAN_ESR_ALL_INT,
&regs->esr);

@@ -885,8 +889,8 @@ static int flexcan_chip_start(struct net_device

*dev)
         */
        reg_ctrl = flexcan_read(&regs->ctrl);
        reg_ctrl &= ~FLEXCAN_CTRL_TSYN;
-       reg_ctrl |= FLEXCAN_CTRL_BOFF_REC | FLEXCAN_CTRL_LBUF |
-               FLEXCAN_CTRL_ERR_STATE;
+       reg_ctrl |= FLEXCAN_CTRL_BOFF_REC | FLEXCAN_CTRL_LBUF;// |
+               //FLEXCAN_CTRL_ERR_STATE;
        /*
         * enable the "error interrupt" (FLEXCAN_CTRL_ERR_MSK),
         * on most Flexcan cores, too. Otherwise we don't get

I'm not sure whether you really want to keep the FLEXCAN_CTRL_ERR_STATE
commented out...

start the driver as usual:
- configure bitrate
- ifconfig can0 up
- _do_not_ enable bit error reporint
- start candump from gitorious with:
  candump -e any,0:0,#FFFFFFFF
- short circuit CAN-High and CAN-Low
- send a CAN frame

root at colibri-vf:~# ip link set can0 type can bitrate 500000
[  146.140862] flexcan 40020000.flexcan can0: bitrate error 0.7%
root at colibri-vf:~# ip link set can0 up
[  146.661430] flexcan 40020000.flexcan can0: writing ctrl=0x17092001
[  146.667902] flexcan 40020000.flexcan can0: flexcan_set_bittiming:
mcr=0x5980000f ctrl=0x17092001
[  146.676790] flexcan 40020000.flexcan can0: flexcan_chip_start:
writing mcr=0x79a20208
[  146.684709] flexcan 40020000.flexcan can0: flexcan_chip_start:
writing ctrl=0x17092051
[  146.698228] flexcan 40020000.flexcan can0: flexcan_chip_start:
reading mcr=0x60a20208 ctrl=0x1709205

# cansend can0 1F334455#1122334455667788
interface = can0, family = 29, type = 3, proto = 

Nothing happens on candump.
 

Another test is to setup a proper CAN bus, but configure the a second
CAN node with a different bitrate. Start the can0 on vf610 as usual,
then candump -e any,0:0,#FFFFFFFF, but do not send any CAN frames on the
vf610. Then on the other system keep sending CAN frames, you might have
to restart the CAN on the second system....

I'm using a PCAN-USB from Peak System for this test. When running with
the same bitrates on both sides (500000, things work smoothly as
expected:

# ip link set can0 type can bitrate 500000
# ip link set can0 up
# cansend can0 1F334455#1122334455667788

root at colibri-vf:~# ./candump -e any,0:0,#FFFFFFFF
  can0  1F334455   [8]  11 22 33 44 55 66 77 88

dmesg:
[  541.772502] flexcan_irq, esr=00040180
[  541.776207] flexcan_irq, ctrl=17092051

Then I reconfigure the system on my host:
# ip link set can0 down
# ip link set can0 type can bitrate 1000000
# ip link set can0 up
# cansend can0 1F334455#1122334455667788

Nothing happens on Vybrid side.


However, I got once this Kernel panic after I reconfigured to normal
mode. But I could not reproduce this.

[  461.954394] flexcan_irq, esr=00059d82
[  461.958093] flexcan_irq, ctrl=17092051
[  461.961873] ------------[ cut here ]------------
[  461.966536] WARNING: CPU: 0 PID: 0 at kernel/locking/mutex.c:826
mutex_trylock+0x1b0/0x208()
[  461.974982] DEBUG_LOCKS_WARN_ON(in_interrupt())
[  461.979347] Modules linked in:
[  461.982621] CPU: 0 PID: 0 Comm: swapper/0 Not tainted
3.16.0-rc4-00017-ge9a974f-dirty #215
[  461.990896] Backtrace:
[  461.993412] [<80011e1c>] (dump_backtrace) from [<80011fb8>]
(show_stack+0x18/0x1c)
[  462.000991]  r6:806718ec r5:00000000 r4:00000000 r3:00000000
[  462.006836] [<80011fa0>] (show_stack) from [<8066e160>]
(dump_stack+0x88/0xa4)
[  462.014143] [<8066e0d8>] (dump_stack) from [<80028f48>]
(warn_slowpath_common+0x70/0x94)
[  462.022280]  r5:00000009 r4:808f9d50
[  462.025920] [<80028ed8>] (warn_slowpath_common) from [<80029010>]
(warn_slowpath_fmt+0x38/0x40)
[  462.034666]  r8:808f9e14 r7:8110cf7c r6:804c2fa0 r5:8e9ee000
r4:8094cdcc
[  462.041478] [<80028fdc>] (warn_slowpath_fmt) from [<806718ec>]
(mutex_trylock+0x1b0/0x208)
[  462.049792]  r3:807e3f6c r2:807e1b8c
[  462.053466] [<8067173c>] (mutex_trylock) from [<804c2fa0>]
(clk_prepare_lock+0x14/0xec)
[  462.061479]  r7:90880000 r6:8e81f800 r5:8e9ee000 r4:8e81f800
[  462.067280] [<804c2f8c>] (clk_prepare_lock) from [<804c50d0>]
(clk_prepare+0x14/0x2c)
[  462.075158]  r6:8e81f800 r5:8e9ee000 r4:8e81f800 r3:00000000
[  462.080918] [<804c50bc>] (clk_prepare) from [<803e7a50>]
(flexcan_get_berr_counter+0x24/0xd0)
[  462.089487]  r4:00000001 r3:00000000
[  462.093156] [<803e7a2c>] (flexcan_get_berr_counter) from [<803e895c>]
(flexcan_poll+0x40c/0x58c)
[  462.101992]  r8:0000000a r7:0000000a r6:8e372388 r5:8e172a80
r4:00000001 r3:00000000
[  462.109843] [<803e8550>] (flexcan_poll) from [<80511f90>]
(net_rx_action+0xcc/0x1b4)
[  462.117630]  r10:8e9ee6c0 r9:00000003 r8:0000000a r7:8fdde188
r6:808fa0c0 r5:8fdde180
[  462.125587]  r4:0000012c
[  462.128164] [<80511ec4>] (net_rx_action) from [<8002d290>]
(__do_softirq+0x120/0x26c)
[  462.136038]  r10:808fa080 r9:00000003 r8:00000100 r7:00000003
r6:808f8000 r5:808fa08c
[  462.143994]  r4:00000000
[  462.146566] [<8002d170>] (__do_softirq) from [<8002d6d4>]
(irq_exit+0xb0/0x104)
[  462.153923]  r10:806788ac r9:808f8000 r8:00000000 r7:0000005a
r6:808f8000 r5:808f5e2c
[  462.161877]  r4:808f8000
[  462.164459] [<8002d624>] (irq_exit) from [<8000f4bc>]
(handle_IRQ+0x58/0xb8)
[  462.171520]  r4:80900d2c r3:000000a0
[  462.175200] [<8000f464>] (handle_IRQ) from [<800086c0>]
(gic_handle_irq+0x30/0x68)
[  462.182818]  r8:8095c587 r7:90802100 r6:808f9f28 r5:80900ea0
r4:9080210c r3:000000a0
[  462.190671] [<80008690>] (gic_handle_irq) from [<80012ae4>]
(__irq_svc+0x44/0x5c)
[  462.198203] Exception stack(0x808f9f28 to 0x808f9f70)
[  462.203315] 9f20:                   00000001 00000001 00000000
80904070 8090098c 80900938
[  462.211517] 9f40: 8095c587 00000000 8095c587 808f8000 806788ac
808f9f7c 808f9f40 808f9f70
[  462.219746] 9f60: 80066a98 8000f834 20000013 ffffffff
[  462.224852]  r7:808f9f5c r6:ffffffff r5:20000013 r4:8000f834
[  462.230621] [<8000f80c>] (arch_cpu_idle) from [<8005fba4>]
(cpu_startup_entry+0x104/0x16c)
[  462.238964] [<8005faa0>] (cpu_startup_entry) from [<80668cf8>]
(rest_init+0xb0/0xd8)
[  462.246757]  r7:8ffffcc0 r3:00000000
[  462.250419] [<80668c48>] (rest_init) from [<808a5bf8>]
(start_kernel+0x340/0x3ac)
[  462.257979]  r5:8095c7c0 r4:80900a38
[  462.261620] [<808a58b8>] (start_kernel) from [<80008074>]
(0x80008074)
[  462.268204] ---[ end trace c9c9dee0ce2272a7 ]---
[  462.272899] flexcan 40020000.flexcan can0: Error Warning IRQ

candump showed this:
  can0  20000004   [8]  00 04 00 00 00 00 00 00   ERRORFRAME
	controller-problem{rx-error-warning}

Please send the outputs of candump and demsg to the list. It should
generate a warning, error passive and finally a busoff message.

--
Stefan