Hi Kurt, tried to give it a try. However, after starting a few (4 - 5) cangen instances to increase bus load it panics: [ 207.259295] Unable to handle kernel NULL pointer dereference at virtual address 00000000 [ 207.267433] pgd = d2bd4866 [ 207.270153] [00000000] *pgd=8ca25831, *pte=00000000, *ppte=00000000 [ 207.276469] Internal error: Oops: 817 [#1] PREEMPT ARM [ 207.281633] Modules linked in: cfg80211 rfkill lm75 musb_dsps musb_hdrc usbcore phy_am335x phy_generic usb_common phy_am335x_control gpio_keys xt_owner iptable_filter ip_tables can_raw [ 207.298189] CPU: 0 PID: 782 Comm: cangen Not tainted 5.5.11-00013-gcdf175eb0ffc #7 [ 207.305793] Hardware name: Generic AM33XX (Flattened Device Tree) [ 207.311942] PC is at can_rx_offload_irq_receive_skb+0x54/0x78 [ 207.317722] LR is at c_can_isr+0x52c/0x5b0 [ 207.321838] pc : [<c04856f4>] lr : [<c04875f8>] psr: 80070193 [ 207.328133] sp : cde6fd30 ip : 00000008 fp : cf4ec5c8 [ 207.333383] r10: 00000000 r9 : 00000000 r8 : cf4ec4e0 [ 207.338633] r7 : 00000001 r6 : 00000000 r5 : 00000039 r4 : cf4ec000 [ 207.345192] r3 : cf4ec5c8 r2 : cf4ec5dc r1 : 00000000 r0 : cf4ec5dc [ 207.351752] Flags: Nzcv IRQs off FIQs on Mode SVC_32 ISA ARM Segment none [ 207.359010] Control: 10c5387d Table: 8cb50019 DAC: 00000051 [ 207.364785] Process cangen (pid: 782, stack limit = 0x7ef64b0d) [ 207.370734] Stack: (0xcde6fd30 to 0xcde70000) [ 207.375116] fd20: 00000000 cf4ec000 00000000 00000000 [ 207.383341] fd40: cde6fd7f 8d12c45f cd82ed80 cc864c80 cf3a9310 00000030 00000001 ffffe000 [ 207.391566] fd60: cf3a9300 cde6fda0 00000000 c0165450 cf3a9300 cde6fda0 00010002 c0a04248 [ 207.399790] fd80: cf3a9310 cf3a9300 00000001 00000000 cde6e000 8ec8018f 000000e8 c01655f4 [ 207.408014] fda0: 00000000 8d12c45f cf3a9300 cf3a9310 cf01f400 c01656a4 cf3a9300 cf3a9310 [ 207.416239] fdc0: cf01f400 c0169520 00000000 00000000 cf01f400 c0164720 00000000 c0164dac [ 207.424463] fde0: c01128c4 20070113 ffffffff cde6fe2c 00000000 c0101a8c 00020837 cfecd200 [ 207.432687] fe00: cfcb9000 00214200 8ec8018f 0000ec80 00000000 cfecd200 00000000 00000000 [ 207.440911] fe20: 8ec8018f 000000e8 20070113 cde6fe48 00020837 c01128c4 20070113 ffffffff [ 207.449136] fe40: 00000051 bf000000 8ec8018f cca261ac 00000000 c0203830 cde6fee4 cf79e0b0 [ 207.457359] fe60: 00000000 c0206358 cfecd200 00000001 cde6fee4 cdcba300 ce38e188 000000ec [ 207.465584] fe80: c0a04248 c01de0d0 00023b20 ce38e18c 000000e8 00280000 ce37e720 00000000 [ 207.473807] fea0: 00000000 8d12c45f 00000011 00000000 000000ec c0a04248 b6e61000 00000040 [ 207.482032] fec0: 000000dd cc8b7000 cc8b703c c0206b80 c0a04248 00000000 cdcba840 00000010 [ 207.490256] fee0: 00000000 cf79e0b0 00000054 00100cca 000000de b6e6b000 ccb52db8 ccb52db8 [ 207.498479] ff00: 00000000 00000000 00000000 00000000 cca261ac cc8b703c 00000000 8d12c45f [ 207.506704] ff20: ce694570 cde6ffb0 b6e61de4 cf79e0b0 00000054 cd944000 cc8b7000 80000007 [ 207.514928] ff40: cc8b703c c0110c44 cdcba840 00000010 00000010 c022c9f4 00000000 00000010 [ 207.523152] ff60: cdcba840 00000007 c0a08e38 b6e61de4 cde6ffb0 10c53c7d 00000000 00023b20 [ 207.531375] ff80: 00023b20 c0110f94 00000000 00000010 00000000 00000004 c0101204 b6e61de4 [ 207.539599] ffa0: 60070010 ffffffff 10c5387d c01020c8 bea8fbc8 00000000 00000010 00000000 [ 207.547822] ffc0: 00000000 00000010 00000000 00000003 00000000 00000000 00023b20 00023b20 [ 207.556047] ffe0: b6e61de4 bea8fb78 00011454 b6e61de4 60070010 ffffffff 00000000 00000000 [ 207.564287] [<c04856f4>] (can_rx_offload_irq_receive_skb) from [<cf4ec000>] (0xcf4ec000) [ 207.572428] Code: e58430e8 e8bd8010 e5932018 e2830014 (e5810000) [ 207.578557] ---[ end trace 67a8f60eda872c72 ]--- [ 207.583199] Kernel panic - not syncing: Fatal exception in interrupt I also noticed two things while building the patchset. The GPL export of can_rx_offload_irq_receive_skb uses the wrong name in the second patch and patch 5 prevents the c_can driver from being built as a module as it references the unexported __can_get_echo_skb function. Regards, Felix > -----Original Message----- > From: Kurt Van Dijck <dev.kurt@xxxxxxxxxxxxxxxxxxxxxx> > Sent: Monday, March 30, 2020 9:02 AM > > On vr, 27 mrt 2020 10:09:03 +0000, Felix Riemann wrote: > > Is it this one? > > https://marc.info/?l=linux-can&m=157683579518282 > > yes > > > > > It seems to be missing patch 6 in the series. > > you can get there via > https://github.com/kurt-vd/linux, branch c_can-next > > Kurt > > > > > Regards, > > Felix > > > > -----Original Message----- > > From: Kurt Van Dijck > > Sent: Thursday, March 26, 2020 8:39 PM > > > > Interesting topic. > > > > After the bus-off problem, I also attacked the overflow problems. No > > idea why this patch series got stalled. I activated the complete 64 > > message buffers of the D_CAN hardware (only 64 of the 128 being > > present in the IP for some reason I don't know) and modified the IRQ > > to use rx-offload. That may affect the circumstances under load, I admit. > > > > Whould you mind trying that series, just for reference, in your setup? > > If your hypothesis is right, my observations do not survive your > > busload. > > > > Kind regards, > > Kurt > > > > On do, 26 mrt 2020 14:49:23 +0000, Felix Riemann wrote: > > > Hi Kurt, > > > > > > yes that would be 3cb3eaac52c0f145d895f4b6c22834d5f02b8569. > > > > > > I tried your patch as well, since the description matched the problem pretty > well. However, I could still produce the hang on our board (based on AM3352) > especially if the bus off occurs while the bus load is very high. > > > > > > What I could see was the worker being triggered by the interrupt with the > bus being in passive mode. When the worker then re-enabled the interrupt the > hardware had already entered init mode due to the bus-off but nor would it > trigger again nor would the interrupt register indicate a pending IRQ. The status > register also indicates an error (LEC != 0x7). > > > > > > A first attempt to fix/workaround this placed the check just before the > worker re-enables the interrupts. But even this setup could reproduce the > problem, although it took longer to do so. So the timing around enabling the > interrupts seems to matter here. Actually, adding a printout at the top of the > worker where your patch checks the value of the sie_pending atomic seems to > increase the chances of making it happen on my board, but that's just a feeling I > got while cleaning up my patch. > > > > > > Regards, > > > > > > Felix > > > > > > -----Original Message----- > > > From: Kurt Van Dijck <mailto:dev.kurt@xxxxxxxxxxxxxxxxxxxxxx> > > > Sent: Thursday, March 26, 2020 2:40 PM > > > > > > This patch looks very familiar. > > > I tried this 6 months ago, when I got a setup where I could reproduce this > within minutes. > > > > > > This patch didn't solve my problem, and I fixed my problem with reading the > ISR register only once in the IRQ, and _not_ 2nd time in napi handler. > > > Reading the ISR in C_CAN is consuming the info, so unless the there is more > state to clear (such as received message buffers), the interrupt cause is assumed > to be known, and the C_CAN proceeds to the next interrupt cause. > > > I've not experienced a single problem since, where before the fix, I'd > experience a few occurances every month. > > > We have an am3358 (beaglebone-like hardware). > > > > > > I believe I submitted a patch for that. > > > > > > Re-scheduling the napi is a workaround, IMHO. > > > > > > Kurt > > > > > > On do, 26 mrt 2020 11:43:18 +0100, Felix Riemann wrote: > > > > There are appears to be a race condition where interrupts caused > > > > by bus offs get lost if it occurs while interrupts are disabled or > > > > being re-enabled. > > > > > > > > This tries to avoid the deadlock by rescheduling the NAPI worker > > > > to handle the bus-off condition. > > > > > > > > Signed-off-by: Felix Riemann <mailto:felix.riemann@xxxxxx> > > > > Reviewed-by: Andre Kalb <mailto:andre.kalb@xxxxxx> > > > > --- > > > > drivers/net/can/c_can/c_can.c | 19 ++++++++++++++++++- > > > > 1 file changed, 18 insertions(+), 1 deletion(-) > > > > > > > > diff --git a/drivers/net/can/c_can/c_can.c > > > > b/drivers/net/can/c_can/c_can.c index 8e9f5620c9a2..0952ac0b9123 > > > > 100644 > > > > --- a/drivers/net/can/c_can/c_can.c > > > > +++ b/drivers/net/can/c_can/c_can.c > > > > @@ -1123,8 +1123,25 @@ static int c_can_poll(struct napi_struct > > > > *napi, int quota) if (work_done < quota) { > > > > napi_complete_done(napi, work_done); > > > > /* enable all IRQs if we are not in bus off state */ -if > > > > (priv->can.state != CAN_STATE_BUS_OFF) > > > > +if (priv->can.state != CAN_STATE_BUS_OFF) { > > > > +u32 ctrl; > > > > c_can_irq_control(priv, true); > > > > + > > > > +/* There appears to be a race condition when the device > > > > + * enters bus off while interrupts are off or being > > > > + * re-enabled causing the bus off to get lost. > > > > + * This tries to avoid this condition. > > > > + */ > > > > +ctrl = priv->read_reg(priv, C_CAN_CTRL_REG); > > > > + > > > > +if (ctrl & CONTROL_INIT) { > > > > +netdev_warn(dev, "lost bus off\n"); c_can_irq_control(priv, > > > > +false); > > > > +/* Reschedule worker to handle bus off */ > > > > +atomic_set(&priv->sie_pending, 1); napi_reschedule(napi); } } > > > > } > > > > > > > > return work_done; > > > > -- > > > > 2.26.0 ___________________________________________________ SMA Solar Technology AG Aufsichtsrat: Dr. Erik Ehrentraut (Vorsitzender) Vorstand: Ulrich Hadding, Dr.-Ing. Juergen Reinert Handelsregister: Amtsgericht Kassel HRB 3972 Sitz der Gesellschaft: 34266 Niestetal USt-ID-Nr. DE 113 08 59 54 WEEE-Reg.-Nr. DE 95881150 ___________________________________________________
