Date: Fri, 23 Jun 2023 19:30:38 +0200
From: Tobias Klausmann <tobias.klausmann@...enet.de>
To: Heiner Kallweit <hkallweit1@...il.com>
Cc: Realtek linux nic maintainers <nic_swsd@...ltek.com>,
 netdev@...r.kernel.org
Subject: Re: r8169: transmit transmit queue timed out - v6.4 cycle

Hello,

ASPM is not available as per

"r8169 0000:03:00.0: can't disable ASPM; OS doesn't have ASPMcontrol", 
the problem is triggered by "no-load" to ~100MBit/s load on a 1GBit/s 
connection.


On 23.06.23 00:10, Heiner Kallweit wrote:
> Same chip version works fine for me even with ASPM L1.2. So your board 
> may have  broken ASPM support. Any special load pattern that triggers 
> the issue?  Can you work around the issue by disabling ASPM sub-states 
> via sysfs attributes, starting with L1.2?
>
> Tobias Klausmann <tobias.klausmann@...enet.de> schrieb am Do., 22. 
> Juni 2023, 15:46:
>
>     Hello all,
>
>     introduced in the 6.4 cycle r8169 show transmit queue timeouts [1].
>     Bisecting the problem brought me to the following commit:
>
>     2ab19de62d67e403105ba860971e5ff0d511ad15 is the first bad commit
>     commit 2ab19de62d67e403105ba860971e5ff0d511ad15
>     Author: Heiner Kallweit <hkallweit1@...il.com>
>     Date:   Mon Mar 6 22:28:06 2023 +0100
>
>          r8169: remove ASPM restrictions now that ASPM is disabled during
>     NAPI poll
>
>          Now that  ASPM is disabled during NAPI poll, we can remove
>     all ASPM
>          restrictions. This allows for higher power savings if the network
>          isn't fully loaded.
>
>          Reviewed-by: Simon Horman <simon.horman@...igine.com>
>          Tested-by: Kai-Heng Feng <kai.heng.feng@...onical.com>
>          Tested-by: Holger Hoffstätte <holger@...lied-asynchrony.com>
>          Signed-off-by: Heiner Kallweit <hkallweit1@...il.com>
>          Signed-off-by: David S. Miller <davem@...emloft.net>
>
>       drivers/net/ethernet/realtek/r8169_main.c | 27
>     +--------------------------
>       1 file changed, 1 insertion(+), 26 deletions(-)
>
>
>     With this commit reverted on top of v6.4-rc6, the timeouts are gone.
>
>     The NIC identifies as "03:00.0 Ethernet controller: Realtek
>     Semiconductor Co., Ltd. RTL8111/8168/8411 PCI Express Gigabit
>     Ethernet
>     Controller (rev 15)"
>
>     Greetings,
>
>     Tobias Klausmann
>
>
>     [1]:
>
>     [ 2070.918700] ------------[ cut here ]------------
>     [ 2070.918708] NETDEV WATCHDOG: enp3s0 (r8169): transmit queue 0
>     timed
>     out 5317 ms
>     [ 2070.918719] WARNING: CPU: 4 PID: 0 at net/sched/sch_generic.c:525
>     dev_watchdog+0x1c9/0x1d0
>     [ 2070.918726] Modules linked in: rfcomm(E) af_packet(E) cmac(E)
>     algif_hash(E) algif_skcipher(E) af_alg(E) bnep(E) btusb(E) btrtl(E)
>     uvcvideo(E) btbcm(E) uvc(E) btintel(E) videobuf2_vmalloc(E) btmtk(E)
>     videobuf2_memops(E) rtsx_usb_sdmmc(E) videobuf2_v4l2(E) bluetooth(E)
>     rtsx_usb_ms(E) mmc_core(E) ecdh_generic(E) memstick(E) ecc(E)
>     videodev(E) videobuf2_common(E) mc(E) rtsx_usb(E) qrtr(E)
>     nls_iso8859_1(E) nls_cp437(E) vfat(E) fat(E) joydev(E)
>     snd_hda_codec_realtek(E) snd_hda_codec_generic(E) ledtrig_audio(E)
>     snd_hda_codec_hdmi(E) ath10k_pci(E) ath10k_core(E) hid_multitouch(E)
>     ath(E) snd_hda_intel(E) snd_intel_dspcfg(E) iTCO_wdt(E) ee1004(E)
>     intel_rapl_msr(E) snd_intel_sdw_acpi(E) intel_pmc_bxt(E)
>     snd_hda_codec(E) mac80211(E) iTCO_vendor_support(E) r8169(E)
>     intel_rapl_common(E) snd_hda_core(E) intel_tcc_cooling(E) mei_hdcp(E)
>     x86_pkg_temp_thermal(E) acer_wmi(E) intel_powerclamp(E) cfg80211(E)
>     snd_hwdep(E) sparse_keymap(E) coretemp(E) snd_pcm(E) realtek(E)
>     i2c_i801(E) wmi_bmof(E) intel_wmi_thunderbolt(E)
>     [ 2070.918794]  snd_timer(E) rfkill(E) mdio_devres(E) libphy(E)
>     libarc4(E) efi_pstore(E) snd(E) i2c_smbus(E) soundcore(E) mei_me(E)
>     intel_lpss_pci(E) intel_lpss(E) mei(E) idma64(E) intel_pch_thermal(E)
>     thermal(E) battery(E) ac(E) acpi_pad(E) tiny_power_button(E) fuse(E)
>     configfs(E) dmi_sysfs(E) ip_tables(E) x_tables(E) hid_generic(E)
>     usbhid(E) crct10dif_pclmul(E) nouveau(E) crc32_pclmul(E)
>     crc32c_intel(E)
>     i915(E) polyval_clmulni(E) drm_ttm_helper(E) polyval_generic(E)
>     ghash_clmulni_intel(E) mxm_wmi(E) drm_buddy(E) sha512_ssse3(E)
>     i2c_algo_bit(E) aesni_intel(E) drm_display_helper(E) crypto_simd(E)
>     drm_kms_helper(E) syscopyarea(E) sysfillrect(E) cryptd(E)
>     sysimgblt(E)
>     cec(E) xhci_pci(E) ttm(E) xhci_hcd(E) usbcore(E) drm(E) usb_common(E)
>     i2c_hid_acpi(E) i2c_hid(E) video(E) wmi(E) pinctrl_sunrisepoint(E)
>     button(E) serio_raw(E) sg(E) dm_multipath(E) dm_mod(E)
>     scsi_dh_rdac(E)
>     scsi_dh_emc(E) scsi_dh_alua(E) msr(E) efivarfs(E)
>     [ 2070.918862] CPU: 4 PID: 0 Comm: swapper/4 Tainted: G
>     E      6.4.0-rc1-desktop-debug+ #51
>     [ 2070.918864] Hardware name: Acer Aspire VN7-593G/Pluto_KLS, BIOS
>     V1.11
>     08/01/2018
>     [ 2070.918866] RIP: 0010:dev_watchdog+0x1c9/0x1d0
>     [ 2070.918869] Code: d5 eb 92 48 89 ef c6 05 5a 34 96 00 01 e8 2f
>     d0 fb
>     ff 45 89 f8 44 89 f1 48 89 ee 48 89 c2 48 c7 c7 58 5c f2 91 e8 07
>     c6 83
>     ff <0f> 0b e9 74 ff ff ff 41 55 41 54 55 53 48 8b 47 50 4c 8b 28 48 85
>     [ 2070.918872] RSP: 0018:ffffbcec00220eb8 EFLAGS: 00010286
>     [ 2070.918875] RAX: 0000000000000000 RBX: ffff94f0104843dc RCX:
>     000000000000083f
>     [ 2070.918877] RDX: 0000000000000000 RSI: 00000000000000f6 RDI:
>     000000000000003f
>     [ 2070.918878] RBP: ffff94f010484000 R08: 0000000000000001 R09:
>     0000000000000000
>     [ 2070.918880] R10: ffff94f1b6aa0000 R11: ffff94f1b6aa0000 R12:
>     ffff94f010484488
>     [ 2070.918881] R13: ffff94f0031a0600 R14: 0000000000000000 R15:
>     00000000000014c5
>     [ 2070.918883] FS:  0000000000000000(0000) GS:ffff94f1b6d00000(0000)
>     knlGS:0000000000000000
>     [ 2070.918885] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>     [ 2070.918887] CR2: 00007f8eea510000 CR3: 000000023322e005 CR4:
>     00000000003706e0
>     [ 2070.918889] Call Trace:
>     [ 2070.918891]  <IRQ>
>     [ 2070.918893]  ? mq_change_real_num_tx+0xe0/0xe0
>     [ 2070.918897]  ? mq_change_real_num_tx+0xe0/0xe0
>     [ 2070.918899]  call_timer_fn.isra.0+0x17/0x70
>     [ 2070.918903]  __run_timers.part.0+0x1b2/0x200
>     [ 2070.918907]  ? tick_sched_do_timer+0x80/0x80
>     [ 2070.918910]  ? hw_breakpoint_pmu_read+0x10/0x10
>     [ 2070.918913]  ? ktime_get+0x33/0xa0
>     [ 2070.918915]  run_timer_softirq+0x21/0x50
>     [ 2070.918918]  __do_softirq+0xb8/0x1ea
>     [ 2070.918923]  irq_exit_rcu+0x75/0xa0
>     [ 2070.918926]  sysvec_apic_timer_interrupt+0x66/0x80
>     [ 2070.918929]  </IRQ>
>     [ 2070.918930]  <TASK>
>     [ 2070.918932]  asm_sysvec_apic_timer_interrupt+0x16/0x20
>     [ 2070.918935] RIP: 0010:cpuidle_enter_state+0xa7/0x2a0
>     [ 2070.918938] Code: 45 40 40 0f 84 9f 01 00 00 e8 65 00 6e ff e8
>     10 f8
>     ff ff 31 ff 49 89 c5 e8 66 64 6d ff 45 84 ff 0f 85 76 01 00 00 fb
>     45 85
>     f6 <0f> 88 be 00 00 00 49 63 ce 48 8b 04 24 48 6b d1 68 49 29 c5 48 89
>     [ 2070.918939] RSP: 0018:ffffbcec0012fe90 EFLAGS: 00000202
>     [ 2070.918942] RAX: ffff94f1b6d25d80 RBX: 0000000000000008 RCX:
>     0000000000000000
>     [ 2070.918943] RDX: 000001e22c5f9004 RSI: fffffffdc849289f RDI:
>     0000000000000000
>     [ 2070.918945] RBP: ffff94f1b6d2fa00 R08: 0000000000000002 R09:
>     000000002d959839
>     [ 2070.918946] R10: ffff94f1b6d24904 R11: 00000000000018c7 R12:
>     ffffffff92155720
>     [ 2070.918948] R13: 000001e22c5f9004 R14: 0000000000000008 R15:
>     0000000000000000
>     [ 2070.918951]  cpuidle_enter+0x24/0x40
>     [ 2070.918954]  do_idle+0x1c0/0x220
>     [ 2070.918958]  cpu_startup_entry+0x14/0x20
>     [ 2070.918960]  start_secondary+0x109/0x130
>     [ 2070.918963]  secondary_startup_64_no_verify+0xf4/0xfb
>     [ 2070.918966]  </TASK>
>     [ 2070.918968] ---[ end trace 0000000000000000 ]---
>     [ 2072.163726] pcieport 0000:00:1c.3: Data Link Layer Link Active not
>     set in 1000 msec
>     [ 2072.165868] r8169 0000:03:00.0 enp3s0: Can't reset secondary
>     PCI bus,
>     detach NIC
>

Powered by blists - more mailing lists