Re: [PATCH v2] rtw88: 8822ce: fix wifi disconnect after S3/S4 on HONOR laptop

[Pkshih <pkshih@realtek.com>]

On Mon, 2021-02-22 at 20:57 +0800, Hao Chen wrote:
> Because I only have the `HONOR magic 14` laptop with rtl8822ce wifi chip 
> :-(
> 
> I will try to find out why the target platform can't properly resume with
> this declaration.Thanks.
> 
> 在 2021/2/22 下午7:36, Pkshih 写道:
> > On Mon, 2021-02-22 at 12:29 +0200, Kalle Valo wrote:
> >> Hao Chen <chenhaoa@uniontech.com> writes:
> >>
> >>> The laptop's wifi disconnect after the laptop HONOR MagicBook 14
> >>> sleep to S3/S4 and wake up.
> >>>
> >>> The dmesg of kernel report:
> >>> "[   99.990168] pcieport 0000:00:01.2: can't change power state from D3hot
> >>> to D0 (config space inaccessible)
> >>> [   99.990176] ACPI: EC: interrupt unblocked
> >>> [   99.993334] rtw_pci 0000:01:00.0: can't change power state from D3hot
> >>> to D0 (config space inaccessible)
> >>> ......
> >>> [  102.133500] rtw_pci 0000:01:00.0: mac power on failed
> >>> [  102.133503] rtw_pci 0000:01:00.0: failed to power on mac
> >>> [  102.133505] ------------[ cut here ]------------
> >>> [  102.133506] Hardware became unavailable upon resume. This could be a
> >>> software issue prior to suspend or a hardware issue.
> >>> [  102.133569] WARNING: CPU: 4 PID: 5612 at net/mac80211/util.c:2232
> >>> ieee80211_reconfig+0x9b/0x1490 [mac80211]
> >>> [  102.133570] Modules linked in: ccm rfcomm uvcvideo videobuf2_vmalloc
> >>> videobuf2_memops videobuf2_v4l2 videobuf2_common videodev mc cmac bnep
> >>> btusb btrtl btbcm btintel edac_mce_amd bluetooth kvm_amd ecdh_generic
> >>> ecc kvm nls_iso8859_1 rtwpci rtw88 crct10dif_pclmul crc32_pclmul mac80211
> >>> ghash_clmulni_intel aesni_intel snd_hda_codec_realtek crypto_simd huawei_wmi
> >>> snd_hda_codec_generic cryptd cfg80211 wmi_bmof serio_raw sparse_keymap
> >>> ledtrig_audio sp5100_tco glue_helper joydev snd_hda_codec_hdmi snd_hda_intel
> >>> snd_intel_dspcfg wdat_wdt snd_hda_codec snd_hda_core pcspkr snd_hwdep snd_pcm
> >>> efi_pstore snd_timer libarc4 k10temp snd soundcore snd_pci_acp3x ccp mac_hid
> >>> binfmt_misc ip_tables x_tables autofs4 amdgpu amd_iommu_v2 gpu_sched
> >>> i2c_algo_bit ttm drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops
> >>> usbmouse cec nvme hid_generic i2c_piix4 usbhid nvme_core drm wmi video
> >>> [  102.133617] CPU: 4 PID: 5612 Comm: kworker/u32:16 Not tainted 5.7.7-amd64-desktop-8822 #3
> >>> [  102.133618] Hardware name: HUAWEI NBLL-WXX9/NBLL-WXX9-PCB, BIOS 1.06 09/29/2020
> >>> [  102.133623] Workqueue: events_unbound async_run_entry_fn
> >>> [  102.133651] RIP: 0010:ieee80211_reconfig+0x9b/0x1490 [mac80211]
> >>> [  102.133654] Code: 31 db e8 e8 fb 27 c2 41 c6 85 34 05 00 00 00 4c 89 ef e8 38
> >>> 56 fc ff 89 45 b8 85 c0 74 4c 48 c7 c7 d0 0c 09 c1 e8 01 e0 25 c2 <0f> 0b 4c
> >>> 89 ef e8 2b d1 ff ff e9 02 03 00 00 80 7d 9f 00 0f 85 1d
> >>> [  102.133655] RSP: 0018:ffffbe52c059fd08 EFLAGS: 00010286
> >>> [  102.133657] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000007
> >>> [  102.133658] RDX: 0000000000000007 RSI: 0000000000000096 RDI: ffff9d573f519cc0
> >>> [  102.133659] RBP: ffffbe52c059fd80 R08: ffffffffffd96245 R09: 000000000002cb80
> >>> [  102.133660] R10: 000000016989e54c R11: 000000000002a360 R12: ffff9d5731f50300
> >>> [  102.133661] R13: ffff9d5731f50800 R14: ffff9d5731f504c8 R15: ffffffff8463fbef
> >>> [  102.133664] FS:  0000000000000000(0000) GS:ffff9d573f500000(0000) knlGS:0000000000000000
> >>> [  102.133665] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> >>> [  102.133666] CR2: 0000000000000000 CR3: 000000033320a000 CR4: 0000000000340ee0
> >>> [  102.133667] Call Trace:
> >>> [  102.133673]  ? enqueue_entity+0xe3/0x680
> >>> [  102.133705]  ieee80211_resume+0x55/0x70 [mac80211]
> >>> [  102.133729]  wiphy_resume+0x84/0x130 [cfg80211]
> >>> [  102.133752]  ? addresses_show+0xa0/0xa0 [cfg80211]
> >>> [  102.133757]  dpm_run_callback+0x5b/0x150
> >>> [  102.133760]  device_resume+0xad/0x1f0
> >>> [  102.133762]  async_resume+0x1d/0x30
> >>> [  102.133764]  async_run_entry_fn+0x3e/0x170
> >>> [  102.133768]  process_one_work+0x1ab/0x380
> >>> [  102.133771]  worker_thread+0x37/0x3b0
> >>> [  102.133774]  kthread+0x120/0x140
> >>> [  102.133776]  ? create_worker+0x1b0/0x1b0
> >>> [  102.133778]  ? kthread_park+0x90/0x90
> >>> [  102.133782]  ret_from_fork+0x22/0x40
> >>> [  102.133785] ---[ end trace 46229bfd3a4273be ]---
> >>> [  102.134137] ------------[ cut here ]------------
> >>> [  102.134141] wlp1s0:  Failed check-sdata-in-driver check, flags: 0x0
> >>> [  102.134195] WARNING: CPU: 0 PID: 5612 at net/mac80211/driver-ops.h:19
> >>> drv_remove_interface+0xfe/0x110 [mac80211]"
> >>>
> >>> When try to pointer the driver.pm to NULL, the problem is fixed.
> >>> It makes the sleep and wake procedure expected when pm's ops not NULL.
> >>>
> >>> By `git blame` command, I know that the assignment of .driver.pm =
> >>> RTW_PM_OPS was in commit 44bc17f7f5b3 ("rtw88: support wowlan feature for
> >>> 8822c"), and another
> >>> commit 7dc7c41607d1 ("rtw88: avoid unused function warnings")
> >>> pointed out rtw_pci_resume() and rtw_pci_suspend() are not used at
> >>> all.
> >>>
> >>> So I think it's safe to remove them.
> >>>
> >>> Fixes: 7dc7c41607d1 ("rtw88: avoid unused function warnings")
> >>> Fixes: 44bc17f7f5b3 ("rtw88: support wowlan feature for 8822c")
> >>>
> >>> Signed-off-by: Hao Chen <chenhaoa@uniontech.com>
> >>> ---
> >>>    drivers/net/wireless/realtek/rtw88/rtw8822ce.c | 1 -
> >>>    1 file changed, 1 deletion(-)
> >>>
> >>> diff --git a/drivers/net/wireless/realtek/rtw88/rtw8822ce.c b/drivers/net/wireless/realtek/rtw88/rtw8822ce.c
> >>> index 3845b1333dc3..4c063192f801 100644
> >>> --- a/drivers/net/wireless/realtek/rtw88/rtw8822ce.c
> >>> +++ b/drivers/net/wireless/realtek/rtw88/rtw8822ce.c
> >>> @@ -25,7 +25,6 @@ static struct pci_driver rtw_8822ce_driver = {
> >>>    	.id_table = rtw_8822ce_id_table,
> >>>    	.probe = rtw_pci_probe,
> >>>    	.remove = rtw_pci_remove,
> >>> -	.driver.pm = &rtw_pm_ops,
> >> Why just 8822ce? Why not remove rtw_pm_ops entirely if it just creates
> >> problems?
> >>
> > I think we can't remove rtw_pm_ops, because wowlan will not work.
> > We need to find out why the target platform can't properly resume with
> > this declaration.
> >
> 

side note: please don't top post that maintainers do hard to read them.

I suggest to use  `pcie_aspm.policy=performance ` mentioned in v1 temporarily; 
it works on your platform, right?
My colleagues will help to check the things you provide. If we have any update,
I'll let you know.

--
Ping-Ke