Message ID | 20100128153643.0fca3c51@nehalam |
---|---|
State | RFC, archived |
Delegated to: | David Miller |
Headers | show |
On 1/28/2010 6:36 PM, Stephen Hemminger wrote: > Please try this patch (and only this patch), on 2.6.33-rc5[*]; > none of the other patches that did not make it upstream because that > confuses things too much. > > The code that checks for DMA mapping errors on receive buffers would > not handle errors correctly. I doubt you have these errors, but if you > did then it would explain the problems. The code has to be a little > tricky and build mapping for new rx buffer before releasing old one, > that way if new mapping fails, the old one can be reused. > > If it works for you, I will resubmit with signed-off. > > --- > * If you want to use DMA debugging, then you will also need the match patch. > Ok - I'll also be running with the recent sched fork vs. hotplug vs. cpuset namespaces patch (commit fabf318e5e4bda0aca2b0d617b191884fda62703) from tip. Without that I get an rcu hang. My plan then is to run with your patch, the rcu patch & the dma debug patch, but disable dma debug for now and see of the problem recurs. If it works, I'll know in a couple of days. If not, perhaps sooner :(. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 1/28/2010 6:36 PM, Stephen Hemminger wrote: > Please try this patch (and only this patch), on 2.6.33-rc5[*]; > none of the other patches that did not make it upstream because that > confuses things too much. > > The code that checks for DMA mapping errors on receive buffers would > not handle errors correctly. I doubt you have these errors, but if you > did then it would explain the problems. The code has to be a little > tricky and build mapping for new rx buffer before releasing old one, > that way if new mapping fails, the old one can be reused. > > If it works for you, I will resubmit with signed-off. > Nope - tx crash again. This time the system stayed up (but hosed) for a few hours. When I tried to recover eth0, the system crashed. Brief summary of events (log extract below): System start Jan 28 19:29 Everything seemed good (load and all) until 17:13:11 the following day when I got rx errors: Jan 29 17:13:11 mail kernel: sky2 eth0: rx error, status 0x6230010 length 1518 Jan 29 17:13:11 mail kernel: sky2 eth0: rx error, status 0x7f40010 length 1518 Jan 29 17:13:11 mail kernel: sky2 eth0: rx error, status 0x8180010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x7f40010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x6230010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x8180010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x6230010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x7f40010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x8180010 length 1518 Jan 29 17:13:14 mail kernel: sky2 eth0: rx error, status 0x5f60010 length 1518 The system continued running normally after this until this morning (Jan 30) at 0:44:55: Jan 30 05:44:55 mail kernel: DRHD: handling fault status reg 2 Jan 30 05:44:55 mail kernel: DMAR:[DMA Read] Request device [06:00.0] fault addr ffc4331ff000 Jan 30 05:44:55 mail kernel: DMAR:[fault reason 06] PTE Read access is not set Jan 30 05:44:55 mail kernel: net_ratelimit: 2 callbacks suppressed Jan 30 05:44:55 mail kernel: sky2 0000:06:00.0: error interrupt status=0xc0000000 Jan 30 05:44:55 mail kernel: sky2 0000:06:00.0: PCI hardware error (0x2010) Jan 30 05:45:01 mail kernel: ------------[ cut here ]------------ Jan 30 05:45:01 mail kernel: WARNING: at net/sched/sch_generic.c:255 dev_watchdog+0xf3/0x161() Jan 30 05:45:01 mail kernel: Hardware name: System Product Name Jan 30 05:45:01 mail kernel: NETDEV WATCHDOG: eth0 (sky2): transmit queue 0 timed out Jan 30 05:45:01 mail kernel: Modules linked in: iptable_raw iptable_mangle ipt_MASQUERADE iptable_nat nf_nat cpufreq_stats ip6table_filter ip6table_mangle ip6_tables bridge stp appletalk psnap llc nfsd lockd nfs_acl auth_rpcgss exportfs hwmon_vid coretemp sunrpc acpi_cpufreq sit tunnel4 ipt_LOG nf_conntrack_netbios_ns nf_conntrack_ftp xt_DSCP xt_dscp xt_MARK nf_conntrack_ipv6 xt_multiport ipv6 dm_multipath kvm_intel kvm snd_hda_codec_analog snd_hda_intel snd_ens1371 gameport snd_hda_codec snd_rawmidi snd_ac97_codec gspca_spca505 ac97_bus gspca_main snd_hwdep videodev snd_seq snd_seq_device v4l1_compat snd_pcm v4l2_compat_ioctl32 snd_timer snd soundcore snd_page_alloc firewire_ohci pcspkr i2c_i801 firewire_core wmi asus_atk0110 crc_itu_t sky2 hwmon iTCO_wdt iTCO_vendor_support fbcon tileblit font bitblit softcursor raid456 async_raid6_recov async_pq raid6_pq async_xor xor async_memcpy async_tx raid1 ata_generic pata_acpi pata_marvell nouveau ttm drm_kms_helper drm agpgart fb i2c_algo_bit cfbcopyarea i2c_core cf Jan 30 05:45:01 mail kernel: bimgblt cfbfillrect [last unloaded: nf_nat] Jan 30 05:45:01 mail kernel: Pid: 0, comm: swapper Tainted: G W 2.6.33-rc5WITHMMAPNODMARFORKTIPSKY2DMAMAP-00283-gd4d37bd-dirty #1 Jan 30 05:45:01 mail kernel: Call Trace: Jan 30 05:45:01 mail kernel: <IRQ> [<ffffffff8104a03d>] warn_slowpath_common+0x7c/0x94 Jan 30 05:45:01 mail kernel: [<ffffffff8104a0ac>] warn_slowpath_fmt+0x41/0x43 Jan 30 05:45:01 mail kernel: [<ffffffff813d2f43>] ? netif_tx_lock+0x44/0x6c Jan 30 05:45:01 mail kernel: [<ffffffff813d30ab>] dev_watchdog+0xf3/0x161 Jan 30 05:45:01 mail kernel: [<ffffffff8106a31f>] ? sched_clock_cpu+0x44/0xce Jan 30 05:45:01 mail kernel: [<ffffffff8105761a>] run_timer_softirq+0x1c3/0x26b Jan 30 05:45:01 mail kernel: [<ffffffff8105060c>] __do_softirq+0xf8/0x1cd Jan 30 05:45:01 mail kernel: [<ffffffff8107192b>] ? tick_program_event+0x2a/0x2c Jan 30 05:45:01 mail kernel: [<ffffffff8100ab1c>] call_softirq+0x1c/0x30 Jan 30 05:45:01 mail kernel: [<ffffffff8100c2b3>] do_softirq+0x4b/0xa3 Jan 30 05:45:01 mail kernel: [<ffffffff810501f8>] irq_exit+0x4a/0x8c Jan 30 05:45:01 mail kernel: [<ffffffff81461859>] smp_apic_timer_interrupt+0x86/0x94 Jan 30 05:45:01 mail kernel: [<ffffffff8100a5d3>] apic_timer_interrupt+0x13/0x20 Jan 30 05:45:01 mail kernel: <EOI> [<ffffffff812afbd4>] ? acpi_idle_enter_bm+0x256/0x28a Jan 30 05:45:01 mail kernel: [<ffffffff812afbcd>] ? acpi_idle_enter_bm+0x24f/0x28a Jan 30 05:45:01 mail kernel: [<ffffffff8139574c>] cpuidle_idle_call+0x9e/0xfa Jan 30 05:45:01 mail kernel: [<ffffffff81008c05>] cpu_idle+0xb4/0xf6 Jan 30 05:45:01 mail kernel: [<ffffffff81455d48>] start_secondary+0x201/0x242 Jan 30 05:45:01 mail kernel: ---[ end trace 57f7151f6a5def07 ]--- Jan 30 05:45:01 mail kernel: sky2 eth0: tx timeout Jan 30 05:45:01 mail kernel: sky2 eth0: transmit ring 14 .. 102 report=14 done=14 Jan 30 05:45:01 mail kernel: sky2 eth0: disabling interface Jan 30 05:45:01 mail kernel: sky2 eth0: enabling interface This down/up continued for several hours until I intervened at about 10:05. I saw that there was no eth0 connectivity, eth1 was ok. It appeard that eth0 was receiving traffic but unable to send. arpwatch was reporting bogons, DHCP showed many DISCOVER/OFFER pairs, no REQUEST/ACK. Pings to any system failed; arp showed incomplete for anything hanging off of eth0. arping also failed. I manually stopped and started eth0 (ifconfig) and reset iptables (although eth0 has no filters). As I started looking at logs, the system hung and rebooted. I'm up now with dma debug enabled, however as with 2.6.32.4 num_entries is dropping and I don't think that dma debug will remain enabled long enough to catch a crash. So, as I see things, there are two issues here: 1) the TX hang post DMAR error and 2) the inability to recover the interface and subsequent system instability. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 01/28/2010 06:36 PM, Stephen Hemminger wrote: > Please try this patch (and only this patch), on 2.6.33-rc5[*]; > none of the other patches that did not make it upstream because that > confuses things too much. > > The code that checks for DMA mapping errors on receive buffers would > not handle errors correctly. I doubt you have these errors, but if you > did then it would explain the problems. The code has to be a little > tricky and build mapping for new rx buffer before releasing old one, > that way if new mapping fails, the old one can be reused. > > If it works for you, I will resubmit with signed-off. > > - > Nope - tx crash again. This time the system stayed up (but hosed) for a few hours. When I tried to recover eth0 the system then crashed. Brief summary of events (log extract below): System start Jan 28 19:29 Everything seemed good (load and all) until 17:13:11 the following day when I got rx errors: Jan 29 17:13:11 mail kernel: sky2 eth0: rx error, status 0x6230010 length 1518 Jan 29 17:13:11 mail kernel: sky2 eth0: rx error, status 0x7f40010 length 1518 Jan 29 17:13:11 mail kernel: sky2 eth0: rx error, status 0x8180010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x7f40010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x6230010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x8180010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x6230010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x7f40010 length 1518 Jan 29 17:13:12 mail kernel: sky2 eth0: rx error, status 0x8180010 length 1518 Jan 29 17:13:14 mail kernel: sky2 eth0: rx error, status 0x5f60010 length 1518 The system continued running normally after this until this morning (Jan 30) at 0:44:55: Jan 30 05:44:55 mail kernel: DRHD: handling fault status reg 2 Jan 30 05:44:55 mail kernel: DMAR:[DMA Read] Request device [06:00.0] fault addr ffc4331ff000 Jan 30 05:44:55 mail kernel: DMAR:[fault reason 06] PTE Read access is not set Jan 30 05:44:55 mail kernel: net_ratelimit: 2 callbacks suppressed Jan 30 05:44:55 mail kernel: sky2 0000:06:00.0: error interrupt status=0xc0000000 Jan 30 05:44:55 mail kernel: sky2 0000:06:00.0: PCI hardware error (0x2010) Jan 30 05:45:01 mail kernel: ------------[ cut here ]------------ Jan 30 05:45:01 mail kernel: WARNING: at net/sched/sch_generic.c:255 dev_watchdog+0xf3/0x161() Jan 30 05:45:01 mail kernel: Hardware name: System Product Name Jan 30 05:45:01 mail kernel: NETDEV WATCHDOG: eth0 (sky2): transmit queue 0 timed out Jan 30 05:45:01 mail kernel: Modules linked in: iptable_raw iptable_mangle ipt_MASQUERADE iptable_nat nf_nat cpufreq_stats ip6table_filter ip6table_mangle ip6_tables bridge stp appletalk psnap llc nfsd lockd nfs_acl auth_rpcgss exportfs hwmon_vid coretemp sunrpc acpi_cpufreq sit tunnel4 ipt_LOG nf_conntrack_netbios_ns nf_conntrack_ftp xt_DSCP xt_dscp xt_MARK nf_conntrack_ipv6 xt_multiport ipv6 dm_multipath kvm_intel kvm snd_hda_codec_analog snd_hda_intel snd_ens1371 gameport snd_hda_codec snd_rawmidi snd_ac97_codec gspca_spca505 ac97_bus gspca_main snd_hwdep videodev snd_seq snd_seq_device v4l1_compat snd_pcm v4l2_compat_ioctl32 snd_timer snd soundcore snd_page_alloc firewire_ohci pcspkr i2c_i801 firewire_core wmi asus_atk0110 crc_itu_t sky2 hwmon iTCO_wdt iTCO_vendor_support fbcon tileblit font bitblit softcursor raid456 async_raid6_recov async_pq raid6_pq async_xor xor async_memcpy async_tx raid1 ata_generic pata_acpi pata_marvell nouveau ttm drm_kms_helper drm agpgart fb i2c_algo_bit cfbcopyarea i2c_core cf Jan 30 05:45:01 mail kernel: bimgblt cfbfillrect [last unloaded: nf_nat] Jan 30 05:45:01 mail kernel: Pid: 0, comm: swapper Tainted: G W 2.6.33-rc5WITHMMAPNODMARFORKTIPSKY2DMAMAP-00283-gd4d37bd-dirty #1 Jan 30 05:45:01 mail kernel: Call Trace: Jan 30 05:45:01 mail kernel: <IRQ> [<ffffffff8104a03d>] warn_slowpath_common+0x7c/0x94 Jan 30 05:45:01 mail kernel: [<ffffffff8104a0ac>] warn_slowpath_fmt+0x41/0x43 Jan 30 05:45:01 mail kernel: [<ffffffff813d2f43>] ? netif_tx_lock+0x44/0x6c Jan 30 05:45:01 mail kernel: [<ffffffff813d30ab>] dev_watchdog+0xf3/0x161 Jan 30 05:45:01 mail kernel: [<ffffffff8106a31f>] ? sched_clock_cpu+0x44/0xce Jan 30 05:45:01 mail kernel: [<ffffffff8105761a>] run_timer_softirq+0x1c3/0x26b Jan 30 05:45:01 mail kernel: [<ffffffff8105060c>] __do_softirq+0xf8/0x1cd Jan 30 05:45:01 mail kernel: [<ffffffff8107192b>] ? tick_program_event+0x2a/0x2c Jan 30 05:45:01 mail kernel: [<ffffffff8100ab1c>] call_softirq+0x1c/0x30 Jan 30 05:45:01 mail kernel: [<ffffffff8100c2b3>] do_softirq+0x4b/0xa3 Jan 30 05:45:01 mail kernel: [<ffffffff810501f8>] irq_exit+0x4a/0x8c Jan 30 05:45:01 mail kernel: [<ffffffff81461859>] smp_apic_timer_interrupt+0x86/0x94 Jan 30 05:45:01 mail kernel: [<ffffffff8100a5d3>] apic_timer_interrupt+0x13/0x20 Jan 30 05:45:01 mail kernel: <EOI> [<ffffffff812afbd4>] ? acpi_idle_enter_bm+0x256/0x28a Jan 30 05:45:01 mail kernel: [<ffffffff812afbcd>] ? acpi_idle_enter_bm+0x24f/0x28a Jan 30 05:45:01 mail kernel: [<ffffffff8139574c>] cpuidle_idle_call+0x9e/0xfa Jan 30 05:45:01 mail kernel: [<ffffffff81008c05>] cpu_idle+0xb4/0xf6 Jan 30 05:45:01 mail kernel: [<ffffffff81455d48>] start_secondary+0x201/0x242 Jan 30 05:45:01 mail kernel: ---[ end trace 57f7151f6a5def07 ]--- Jan 30 05:45:01 mail kernel: sky2 eth0: tx timeout Jan 30 05:45:01 mail kernel: sky2 eth0: transmit ring 14 .. 102 report=14 done=14 Jan 30 05:45:01 mail kernel: sky2 eth0: disabling interface Jan 30 05:45:01 mail kernel: sky2 eth0: enabling interface This down/up continued for several hours until I intervened at about 10:05. I saw that there was no eth0 connectivity, eth1 was ok. It appeard that eth0 was receiving traffic but unable to send. arpwatch was reporting bogons, DHCP showed many DISCOVER/OFFER pairs, no REQUEST/ACK. Pings to any system failed; arp showed incomplete for anything hanging off of eth0. arping also failed. I manually stopped and started eth0 (ifconfig) and reset iptables (although eth0 has no filters). As I started looking at logs, the system hung and rebooted. I'm up now with dma debug enabled, however as with 2.6.32.4 num_entries is dropping and I don't think that dma debug will remain enabled long enough to catch a crash. So, as I see things, there are two issues here: 1) the TX hang post DMAR error and 2) the inability to recover the interface and subsequent system instability. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Stephen Hemminger <shemminger@linux-foundation.org> writes: > - for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) > - re->frag_addr[i] = pci_map_page(pdev, > + > +map_page_error: > + while (--i >= 0) { > + pci_unmap_page(pdev, re->frag_addr[i], > + skb_shinfo(skb)->frags[i].size, > + PCI_DMA_FROMDEVICE); > + } > + > + pci_unmap_single(pdev, re->data_addr, pci_unmap_len(re, data_size), > + PCI_DMA_FROMDEVICE); Better add a helper somewhere to do this, doesn't make sense to duplicate this in all drivers (lots of drivers have similar problems) I remember looking at this some time ago but for some reason the patches never made it out. -Andi
--- a/drivers/net/sky2.c 2010-01-28 15:09:11.387649670 -0800 +++ b/drivers/net/sky2.c 2010-01-28 15:31:26.969181566 -0800 @@ -1106,18 +1106,39 @@ static int sky2_rx_map_skb(struct pci_de int i; re->data_addr = pci_map_single(pdev, skb->data, size, PCI_DMA_FROMDEVICE); - if (unlikely(pci_dma_mapping_error(pdev, re->data_addr))) - return -EIO; + if (pci_dma_mapping_error(pdev, re->data_addr)) + goto mapping_error; pci_unmap_len_set(re, data_size, size); - for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) - re->frag_addr[i] = pci_map_page(pdev, - skb_shinfo(skb)->frags[i].page, - skb_shinfo(skb)->frags[i].page_offset, - skb_shinfo(skb)->frags[i].size, + for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) { + skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; + + re->frag_addr[i] = pci_map_page(pdev, frag->page, + frag->page_offset, + frag->size, PCI_DMA_FROMDEVICE); + + if (pci_dma_mapping_error(pdev, re->frag_addr[i])) + goto map_page_error; + } return 0; + +map_page_error: + while (--i >= 0) { + pci_unmap_page(pdev, re->frag_addr[i], + skb_shinfo(skb)->frags[i].size, + PCI_DMA_FROMDEVICE); + } + + pci_unmap_single(pdev, re->data_addr, pci_unmap_len(re, data_size), + PCI_DMA_FROMDEVICE); + +mapping_error: + if (net_ratelimit()) + dev_warn(&pdev->dev, "%s: rx mapping error\n", + skb->dev->name); + return -EIO; } static void sky2_rx_unmap_skb(struct pci_dev *pdev, struct rx_ring_info *re) @@ -2308,30 +2329,32 @@ static struct sk_buff *receive_new(struc struct rx_ring_info *re, unsigned int length) { - struct sk_buff *skb, *nskb; + struct sk_buff *skb; + struct rx_ring_info nre; unsigned hdr_space = sky2->rx_data_size; - /* Don't be tricky about reusing pages (yet) */ - nskb = sky2_rx_alloc(sky2); - if (unlikely(!nskb)) - return NULL; + nre.skb = sky2_rx_alloc(sky2); + if (unlikely(!nre.skb)) + goto nobuf; + + if (sky2_rx_map_skb(sky2->hw->pdev, &nre, hdr_space)) + goto nomap; skb = re->skb; sky2_rx_unmap_skb(sky2->hw->pdev, re); - prefetch(skb->data); - re->skb = nskb; - if (sky2_rx_map_skb(sky2->hw->pdev, re, hdr_space)) { - dev_kfree_skb(nskb); - re->skb = skb; - return NULL; - } + *re = nre; if (skb_shinfo(skb)->nr_frags) skb_put_frags(skb, hdr_space, length); else skb_put(skb, length); return skb; + +nomap: + dev_kfree_skb(nre.skb); +nobuf: + return NULL; } /*