Re: [BUG] Linux-2.6.17-rt3 on arm ixdp465

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 





On Thu, 29 Jun 2006, Milan Svoboda wrote:


Thank you for your answer, I look at it too...


eepro100 seems to be SMP safe, so it shouldn't be there.
Have anyone else used eepro100 with preempt-realtime?

I use to use it a while back ago, when e100 would screw up my network
card. But that has been fixed so I don't use eepro100 and I would
recommend anyone else to switch to e100.

I have been using it because the same reason as you, but simply didn't
noticed that e100 works now ;-)

I switched to e100 and turned debug messages on and got many of these:

BUG: scheduling with irqs disabled: softirq-net-rx//0x00000000/6
caller is schedule+0x10/0x114
[<c0024e24>] (dump_stack+0x0/0x28) from [<c01ae528>] (schedule+0xf8/0x114)
[<c01ae430>] (schedule+0x0/0x114) from [<c01afb60>]
(rt_lock_slowlock+0x100/0x240)
r5 = C01F070C  r4 = C4150000
[<c01afa60>] (rt_lock_slowlock+0x0/0x240) from [<c01aff28>]
(__lock_text_start+0x18/0x1c)
[<c01aff10>] (__lock_text_start+0x0/0x1c) from [<c0078b08>]
(kfree+0x2c/0x84)
[<c0078adc>] (kfree+0x0/0x84) from [<c002aab0>]
(dma_unmap_single+0x110/0x1a8)
r5 = C4124BE0  r4 = C7C4B6E0
[<c002a9a0>] (dma_unmap_single+0x0/0x1a8) from [<c012766c>]
(e100_poll+0x2e0/0x59c)
r8 = C432A3A0  r7 = C41C9BA0  r6 = C41C9B60  r5 = 00000001
r4 = FFC881C0
[<c012738c>] (e100_poll+0x0/0x59c) from [<c0148280>]
(net_rx_action+0xa0/0x1a4)
[<c01481e0>] (net_rx_action+0x0/0x1a4) from [<c0039020>]
(ksoftirqd+0x110/0x1b0)
[<c0038f10>] (ksoftirqd+0x0/0x1b0) from [<c00490d8>] (kthread+0x110/0x13c)
[<c0048fc8>] (kthread+0x0/0x13c) from [<c0035054>] (do_exit+0x0/0x998)
r8 = 00000000  r7 = 00000000  r6 = 00000000  r5 = 00000000
r4 = 00000000
---------------------------
| preempt count: 00000000 ]
| 0-level deep critical section nesting:
----------------------------------------


It seems that dma_unmap_single() on arm contains
	local_irq_save(flags);

	unmap_single(dev, dma_addr, size, dir);

	local_irq_restore(flags);

I don't know the dma code on arm. It doesn't look like a per-cpu code but it
seems to me that it is not SMP safe and therefore not preempt-realtime safe, either.

The hard thing is to figure out which datastructures exactly is protected by those irq-disable and put in a spinlock..

I added Deepak Saxena on CC as he seems to be the last one who touched the file.


Esben

These messages are different as their source seems to be softirq-net-rx. I
cannot reproduce
the original bug now...


PS: Is latency tracing working on arm platform? I'm unable to get this
statistic...

Milan


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[Index of Archives]     [Kernel Newbies]     [Netfilter]     [Bugtraq]     [Photo]     [Stuff]     [Gimp]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Video 4 Linux]     [Linux for the blind]     [Linux Resources]
  Powered by Linux