|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: ocplib+endian improvement
Hm, but trying out TCP seems to trigger a softlockup in Xen netback (both kernels 3.2 and latest 3.7). Will need to do some more debugging tomorrow... Steven (Smith): any luck with the grant-free netback modification? I could try it out at the same time as debugging this particular issue. -anil [ 277.249069] Code: 00 89 c1 7c c8 41 59 c3 90 90 90 65 c6 04 25 41 b1 00 00 00 65 f6 04 25 40 b1 00 00 ff 74 05 e8 47 00 00 00 c3 66 0f 1f 44 00 00 <65> c6 04 25 41 b1 00 00 01 c3 66 0f 1f 44 00 00 65 f6 04 25 41 [ 305.248782] BUG: soft lockup - CPU#0 stuck for 23s! [netback/0:3772] [ 305.248883] Modules linked in: xt_physdev iptable_filter ip_tables x_tables xen_netback xen_gntdev xen_evtchn xenfs xen_privcmd nfsd auth_rpcgss nfs_acl nfs lockd dns_resolver fscache sunrpc bridge stp llc loop crc32c_intel ghash_clmulni_intel aesni_intel aes_x86_64 ablk_helper cryptd xts lrw gf128mul snd_pcm sp5100_tco snd_page_alloc snd_timer snd soundcore tpm_tis tpm amd64_edac_mod edac_mce_amd i2c_piix4 i2c_core dcdbas evdev pcspkr tpm_bios edac_core microcode psmouse fam15h_power k10temp serio_raw acpi_power_meter button processor thermal_sys ext4 crc16 jbd2 mbcache dm_mod sg sd_mod crc_t10dif ata_generic ohci_hcd pata_atiixp ixgbe ahci ptp libahci pps_core libata ehci_hcd dca mdio scsi_mod bnx2 usbcore usb_common [ 305.248932] CPU 0 [ 305.248934] Pid: 3772, comm: netback/0 Tainted: G W 3.7-trunk-amd64 #1 Debian 3.7.1-1~experimental.2 Dell Inc. PowerEdge R415/08WNM9 [ 305.248936] RIP: e030:[<ffffffffa021c153>] [<ffffffffa021c153>] xen_netbk_tx_build_gops+0x19d/0x7ad [xen_netback] [ 305.248941] RSP: e02b:ffff88020eba3ca8 EFLAGS: 00000217 [ 305.248943] RAX: 0000000073202626 RBX: ffffc90007ab5000 RCX: 000000001ce71e08 [ 305.248945] RDX: 000000001ce71e08 RSI: ffffc90007ab02a8 RDI: ffff880653403800 [ 305.248946] RBP: ffffc90007ab70c0 R08: ffff8806534038d8 R09: ffff88020eba3c74 [ 305.248947] R10: ffffc90007ab0208 R11: ffffc90007ab0208 R12: ffff880653403800 [ 305.248949] R13: 0000000000007320 R14: 0000000000000000 R15: 000000000104210e [ 305.248953] FS: 00007f59123a4700(0000) GS:ffff8807ff800000(0000) knlGS:0000000000000000 [ 305.248955] CS: e033 DS: 0000 ES: 0000 CR0: 000000008005003b [ 305.248956] CR2: 00007fd7c822b070 CR3: 00000007f2177000 CR4: 0000000000000660 [ 305.248958] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 305.248960] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 305.248961] Process netback/0 (pid: 3772, threadinfo ffff88020eba2000, task ffff8807ee921180) [ 305.248963] Stack: [ 305.248964] ffffffff81004067 ffffffff81004202 1ce71e0800000003 ffff8807ee921180 [ 305.248968] ffffc90007ab70c0 ffff8807ff811740 0000000000000000 2074706fffff2e2f [ 305.248972] ffff4f2073202626 2074706f20646c72 ffffffff810037f7 ffff8807ee921180 [ 305.248975] Call Trace: [ 305.248978] [<ffffffff81004067>] ? arch_local_irq_restore+0x7/0x8 [ 305.248982] [<ffffffff81004202>] ? xen_mc_flush+0x11d/0x160 [ 305.248985] [<ffffffff810037f7>] ? xen_mc_issue.constprop.22+0x10/0x4d [ 305.248988] [<ffffffff8100d02f>] ? load_TLS+0x7/0xa [ 305.248991] [<ffffffff8100d60c>] ? __switch_to+0x195/0x3f8 [ 305.248994] [<ffffffff8105fadb>] ? mmdrop+0xd/0x1c [ 305.248996] [<ffffffff81061390>] ? finish_task_switch+0x83/0xb4 [ 305.249000] [<ffffffff813778e9>] ? __schedule+0x4b2/0x4e0 [ 305.249003] [<ffffffff8107f0d3>] ? arch_local_irq_disable+0x7/0x8 [ 305.249006] [<ffffffff81378465>] ? _raw_spin_lock_irqsave+0x14/0x35 [ 305.249009] [<ffffffff8107f0d3>] ? arch_local_irq_disable+0x7/0x8 [ 305.249012] [<ffffffff81378465>] ? _raw_spin_lock_irqsave+0x14/0x35 [ 305.249016] [<ffffffffa021c897>] ? xen_netbk_kthread+0x134/0x78d [xen_netback] [ 305.249019] [<ffffffff8105d78f>] ? arch_local_irq_enable+0x7/0x8 [ 305.249022] [<ffffffff81061357>] ? finish_task_switch+0x4a/0xb4 [ 305.249025] [<ffffffff81057987>] ? abort_exclusive_wait+0x79/0x79 [ 305.249029] [<ffffffffa021c763>] ? xen_netbk_tx_build_gops+0x7ad/0x7ad [xen_netback] [ 305.249032] [<ffffffffa021c763>] ? xen_netbk_tx_build_gops+0x7ad/0x7ad [xen_netback] [ 305.249035] [<ffffffff810570ac>] ? kthread+0x81/0x89 [ 305.249038] [<ffffffff810037f7>] ? xen_mc_issue.constprop.22+0x10/0x4d [ 305.249041] [<ffffffff8105702b>] ? __kthread_parkme+0x5c/0x5c [ 305.249043] [<ffffffff8137d6bc>] ? ret_from_fork+0x7c/0xb0 [ 305.249046] [<ffffffff8105702b>] ? __kthread_parkme+0x5c/0x5c [ 305.249048] Code: bc 24 80 00 00 00 4d 89 a4 24 a8 00 00 00 49 c7 84 24 a0 00 00 00 58 bf 21 a0 4c 89 f6 e8 86 d9 e2 e0 e9 c3 05 00 00 8b 54 24 14 <66> 8b 74 24 3e 41 8d 4f ff 0f b7 44 24 42 48 c7 44 24 30 00 00 avsm@gabriel:~/src/git/mirage/mirage-www$ Message from syslogd@gabriel at Jan 6 21:06:38 ... On 6 Jan 2013, at 18:44, Anil Madhavapeddy <anil@xxxxxxxxxx> wrote: > I've been porting the network stack to take advantage of the cstruct > turbo-boost that Pierre and Thomas worked on. This optimisation adds compiler > built-ins (in 4.01.0+) which let the code generator optimise away many of the > temporary values required for low-level optimisation. > > Here's a (very quick) before/after for a ping flood (which is a good stress > test of the low-level shared ring, network driver and protocol stack). > > For a ping flood With 4.00.1 without the optimisation: > 73755 packets transmitted, 73702 received, +49 duplicates, 0% packet loss, > time 6283ms > rtt min/avg/max/mdev = 0.031/0.228/1209.178/9.887 ms, pipe 14850, ipg/ewma > 0.085/0.036 ms > > > and with the optimisation: > 41791 packets transmitted, 41764 received, +25 duplicates, 0% packet loss, > time 3539msrtt min/avg/max/mdev = 0.030/0.188/1261.042/8.459 ms, pipe 14742, > ipg/ewma 0.084/0.039 ms > > So our average latency drops quite significantly (0.228 -> 0.188), as does > CPU load (not shown above). > > I've not committed these changes to the mainstream yet until I test out TCP > more, but it's getting there! > > -anil
|
 |
Lists.xenproject.org is hosted with RackSpace, monitoring our |