* Re: Can not unload ath11k_pci and load it back
2021-10-07 7:48 ` Kalle Valo
@ 2021-10-07 8:00 ` Mark Herbert
2021-10-07 8:51 ` Mark Herbert
` (3 subsequent siblings)
4 siblings, 0 replies; 11+ messages in thread
From: Mark Herbert @ 2021-10-07 8:00 UTC (permalink / raw)
To: Kalle Valo; +Cc: ath11k
As I mentioned - if I build mhi as a module - card will not even start.
it works only if mhi is compiled inside the kernel, and in this config I
can not unload it.
On 07.10.2021 10:48, Kalle Valo wrote:
> Mark Herbert <mark.herbert42@gmail.com> writes:
>
>> I've found similar bug files already - so not to create duplicates
>> simply commented there
>>
>> https://bugzilla.kernel.org/show_bug.cgi?id=214541
>>
>> Exactly - hibernation means saving state on disk and resuming later,
>> it is a very usefull feature when you are running out of battery and
>> want to save your work.
> Got it, thanks. So it seems we have two issues here: ath11k does not
> support hibernation and ath11k module reload failing. Let's handle the
> module reload failing in bug #214541, but can you file another bug about
> the missing support for hibernation? Let's have one issue per bug
> report, it's easier to manage bugzilla that way.
>
>> Also one more point I've found - if MHI is configured in the kernel -
>> the card is starting up on thr first boot and works. But if i
>> configure MHI as a module - it does not. MHI module do load. So do
>> ath11k and ath11k_pci. But the last ath11k dmesg says :
>>
>> mhi mhi0: Wait for device to enter SBL or Mission mode
>>
>> Exactly as if I do rmmod/modprobe sequence.
>>
>> I use latest BIOS from Dell (in fact upgrading 3.0.4 to 3.1.0 did not
>> cange anythng) , and have tried multiple kernels. Stable 5.14 series
>> and ath11k git tree (next and master-pending). Nothing helps.
>> Sometimes it woks out - I had couple of times successfully
>> rmmod/modprobe ath11k_pci but it s a kind of random and very rare
>> event ( I had it 2 times only).
>>
>> So for me it looks like an issue between MHI and ath11k , but
>> ubfortunately my knowledge is not enough to propose any solution here.
> On my NUC x86 test box I reload ath11k modules several times a day while
> I test ath11k but I always reload the mhi modules at the same time,
> maybe that makes the difference? Or XPS 13 9310 works differently from
> NUC, that's also possible.
>
--
ath11k mailing list
ath11k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath11k
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Can not unload ath11k_pci and load it back
2021-10-07 7:48 ` Kalle Valo
2021-10-07 8:00 ` Mark Herbert
@ 2021-10-07 8:51 ` Mark Herbert
2021-10-08 13:28 ` Mark Herbert
` (2 subsequent siblings)
4 siblings, 0 replies; 11+ messages in thread
From: Mark Herbert @ 2021-10-07 8:51 UTC (permalink / raw)
To: Kalle Valo; +Cc: ath11k
I've manged to start a card with mhi compiled as a module. Istead of
autoload configured to lad mhi in advance, before system detects
ath11k. So the card starts. then I unload ath11k_pci, qtrr_mhi , and
finally mhi. Then loading them back - mhi first, then qtrr and finally
ath11_pci. Still same story. I've also tried to unload modules, then
hibernate so laptop was off for a while. Then resumed and loaded the
modules manually. No effect.
On 07.10.2021 10:48, Kalle Valo wrote:
> Mark Herbert <mark.herbert42@gmail.com> writes:
>
>> I've found similar bug files already - so not to create duplicates
>> simply commented there
>>
>> https://bugzilla.kernel.org/show_bug.cgi?id=214541
>>
>> Exactly - hibernation means saving state on disk and resuming later,
>> it is a very usefull feature when you are running out of battery and
>> want to save your work.
> Got it, thanks. So it seems we have two issues here: ath11k does not
> support hibernation and ath11k module reload failing. Let's handle the
> module reload failing in bug #214541, but can you file another bug about
> the missing support for hibernation? Let's have one issue per bug
> report, it's easier to manage bugzilla that way.
>
>> Also one more point I've found - if MHI is configured in the kernel -
>> the card is starting up on thr first boot and works. But if i
>> configure MHI as a module - it does not. MHI module do load. So do
>> ath11k and ath11k_pci. But the last ath11k dmesg says :
>>
>> mhi mhi0: Wait for device to enter SBL or Mission mode
>>
>> Exactly as if I do rmmod/modprobe sequence.
>>
>> I use latest BIOS from Dell (in fact upgrading 3.0.4 to 3.1.0 did not
>> cange anythng) , and have tried multiple kernels. Stable 5.14 series
>> and ath11k git tree (next and master-pending). Nothing helps.
>> Sometimes it woks out - I had couple of times successfully
>> rmmod/modprobe ath11k_pci but it s a kind of random and very rare
>> event ( I had it 2 times only).
>>
>> So for me it looks like an issue between MHI and ath11k , but
>> ubfortunately my knowledge is not enough to propose any solution here.
> On my NUC x86 test box I reload ath11k modules several times a day while
> I test ath11k but I always reload the mhi modules at the same time,
> maybe that makes the difference? Or XPS 13 9310 works differently from
> NUC, that's also possible.
>
--
ath11k mailing list
ath11k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath11k
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Can not unload ath11k_pci and load it back
2021-10-07 7:48 ` Kalle Valo
2021-10-07 8:00 ` Mark Herbert
2021-10-07 8:51 ` Mark Herbert
@ 2021-10-08 13:28 ` Mark Herbert
2021-10-24 5:22 ` Mark Herbert
2021-10-30 19:23 ` Mark Herbert
4 siblings, 0 replies; 11+ messages in thread
From: Mark Herbert @ 2021-10-08 13:28 UTC (permalink / raw)
To: Kalle Valo; +Cc: ath11k
Created the bug for hibernate issue. For me it looks very close to the
modules unloading - Archlinux wiki provides module unloading before
/loading after as a workaround. And in fact there are a lot of hardware
pieces that need to be reinitialized after resume, so it is a kind of
common workaround.
https://bugzilla.kernel.org/show_bug.cgi?id=214649
By the way the card gives a lot of trouble not only for Linux users but
even in Windows, and Dell has issued several driver updates (latest
dated 23 of September). Just windering if Dell also uses same firmware
12 month old or they have some bugfixes in the firmware itself...
On 07.10.2021 10:48, Kalle Valo wrote:
> Mark Herbert <mark.herbert42@gmail.com> writes:
>
>> I've found similar bug files already - so not to create duplicates
>> simply commented there
>>
>> https://bugzilla.kernel.org/show_bug.cgi?id=214541
>>
>> Exactly - hibernation means saving state on disk and resuming later,
>> it is a very usefull feature when you are running out of battery and
>> want to save your work.
> Got it, thanks. So it seems we have two issues here: ath11k does not
> support hibernation and ath11k module reload failing. Let's handle the
> module reload failing in bug #214541, but can you file another bug about
> the missing support for hibernation? Let's have one issue per bug
> report, it's easier to manage bugzilla that way.
>
>> Also one more point I've found - if MHI is configured in the kernel -
>> the card is starting up on thr first boot and works. But if i
>> configure MHI as a module - it does not. MHI module do load. So do
>> ath11k and ath11k_pci. But the last ath11k dmesg says :
>>
>> mhi mhi0: Wait for device to enter SBL or Mission mode
>>
>> Exactly as if I do rmmod/modprobe sequence.
>>
>> I use latest BIOS from Dell (in fact upgrading 3.0.4 to 3.1.0 did not
>> cange anythng) , and have tried multiple kernels. Stable 5.14 series
>> and ath11k git tree (next and master-pending). Nothing helps.
>> Sometimes it woks out - I had couple of times successfully
>> rmmod/modprobe ath11k_pci but it s a kind of random and very rare
>> event ( I had it 2 times only).
>>
>> So for me it looks like an issue between MHI and ath11k , but
>> ubfortunately my knowledge is not enough to propose any solution here.
> On my NUC x86 test box I reload ath11k modules several times a day while
> I test ath11k but I always reload the mhi modules at the same time,
> maybe that makes the difference? Or XPS 13 9310 works differently from
> NUC, that's also possible.
>
--
ath11k mailing list
ath11k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath11k
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Can not unload ath11k_pci and load it back
2021-10-07 7:48 ` Kalle Valo
` (2 preceding siblings ...)
2021-10-08 13:28 ` Mark Herbert
@ 2021-10-24 5:22 ` Mark Herbert
2021-10-30 19:23 ` Mark Herbert
4 siblings, 0 replies; 11+ messages in thread
From: Mark Herbert @ 2021-10-24 5:22 UTC (permalink / raw)
To: Kalle Valo; +Cc: ath11k
Applied several patches series from latest mailing list
ath11k: add feature for device recovery (3 pathes)
ath11k: support for WoW functionalities (6 patches)
ath11k: Advertise PLATFORM CAP PCIE GLOBAL RESET.
Before those the system was deadly stuck on resume from hibernation -
but now it gives control back, so I can see the error log in DMESG. The
card is not waiking up , and I still can not unload/reload the
ath11k_pci module to get device back. But here is the log - maybe it
will give some ideas.
[137698.301974] ACPI: EC: interrupt unblocked
[137698.498119] usb usb1: root hub lost power or was reset
[137698.498122] usb usb2: root hub lost power or was reset
[137698.498186] usb usb3: root hub lost power or was reset
[137698.498195] usb usb4: root hub lost power or was reset
[137698.498231] ath11k_pci 0000:72:00.0: failed to set mhi state: RESUME(6)
[137698.522871] nvme nvme0: 7/0/0 default/read/poll queues
[137698.656843] i915 0000:00:02.0: [drm] GuC firmware
i915/tgl_guc_62.0.0.bin version 62.0 submission:disabled
[137698.656849] i915 0000:00:02.0: [drm] GuC SLPC: disabled
[137698.656850] i915 0000:00:02.0: [drm] HuC firmware
i915/tgl_huc_7.9.3.bin version 7.9 authenticated:yes
[137698.848725] usb 3-6: reset high-speed USB device number 3 using xhci_hcd
[137699.042347] usb 3-6: restoring control
00000000-0000-0000-0000-000000000101/10/5
[137699.042356] usb 3-6: restoring control
00000000-0000-0000-0000-000000000101/12/11
[137699.178934] usb 3-3: reset full-speed USB device number 2 using xhci_hcd
[137701.548589] ath11k_pci 0000:72:00.0: timed out while waiting for wow
wakeup completion
[137701.548596] ath11k_pci 0000:72:00.0: failed to wakeup wow during
resume: -110
[137701.548599] ath11k_pci 0000:72:00.0: failed to resume core: -110
[137701.548601] PM: dpm_run_callback(): pci_pm_restore+0x0/0xd0 returns -110
[137701.548621] ath11k_pci 0000:72:00.0: PM: failed to restore async:
error -110
[137701.557110] PM: hibernation: Basic memory bitmaps freed
[137701.557113] OOM killer enabled.
[137701.557115] Restarting tasks ... done.
[137701.619948] PM: hibernation: hibernation exit
[137701.624409] usbcore: registered new interface driver btusb
[137704.668739] ath11k_pci 0000:72:00.0: wmi command 16387 timeout
[137704.668781] ath11k_pci 0000:72:00.0: failed to send
WMI_PDEV_SET_PARAM cmd
[137704.668795] ath11k_pci 0000:72:00.0: failed to enable PMF QOS: (-11
On 07.10.2021 10:48, Kalle Valo wrote:
> Mark Herbert <mark.herbert42@gmail.com> writes:
>
>> I've found similar bug files already - so not to create duplicates
>> simply commented there
>>
>> https://bugzilla.kernel.org/show_bug.cgi?id=214541
>>
>> Exactly - hibernation means saving state on disk and resuming later,
>> it is a very usefull feature when you are running out of battery and
>> want to save your work.
> Got it, thanks. So it seems we have two issues here: ath11k does not
> support hibernation and ath11k module reload failing. Let's handle the
> module reload failing in bug #214541, but can you file another bug about
> the missing support for hibernation? Let's have one issue per bug
> report, it's easier to manage bugzilla that way.
>
>> Also one more point I've found - if MHI is configured in the kernel -
>> the card is starting up on thr first boot and works. But if i
>> configure MHI as a module - it does not. MHI module do load. So do
>> ath11k and ath11k_pci. But the last ath11k dmesg says :
>>
>> mhi mhi0: Wait for device to enter SBL or Mission mode
>>
>> Exactly as if I do rmmod/modprobe sequence.
>>
>> I use latest BIOS from Dell (in fact upgrading 3.0.4 to 3.1.0 did not
>> cange anythng) , and have tried multiple kernels. Stable 5.14 series
>> and ath11k git tree (next and master-pending). Nothing helps.
>> Sometimes it woks out - I had couple of times successfully
>> rmmod/modprobe ath11k_pci but it s a kind of random and very rare
>> event ( I had it 2 times only).
>>
>> So for me it looks like an issue between MHI and ath11k , but
>> ubfortunately my knowledge is not enough to propose any solution here.
> On my NUC x86 test box I reload ath11k modules several times a day while
> I test ath11k but I always reload the mhi modules at the same time,
> maybe that makes the difference? Or XPS 13 9310 works differently from
> NUC, that's also possible.
>
--
ath11k mailing list
ath11k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath11k
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Can not unload ath11k_pci and load it back
2021-10-07 7:48 ` Kalle Valo
` (3 preceding siblings ...)
2021-10-24 5:22 ` Mark Herbert
@ 2021-10-30 19:23 ` Mark Herbert
4 siblings, 0 replies; 11+ messages in thread
From: Mark Herbert @ 2021-10-30 19:23 UTC (permalink / raw)
To: Kalle Valo; +Cc: ath11k
Found message in May archive of the mailing list with pretty similar
issue that I have.
http://lists.infradead.org/pipermail/ath11k/2021-May/001267.html
The author of that message managed better debugging info than I did. It
is same - if the kernel do autoloading of ath11k_pci at boot it works.
But if I do manual load - it gets stuck in
"Wait for device to enter SBL or Mission mode"
sometimes instead of waiting it does report the versions of loaded
firmware and the card works fine, but that happens very very rare. Once
in 30-40 tries.
The message is produced by drivers/bus/mhi/core/boot.c file, by
void mhi_fw_load_handler(struct mhi_controller *mhi_cntrl) function.
And this message seems to be generated if everything is working OK - so
the card should reach this SBL or Mission mode and continue but it does
not, no follow up messages and the WiFi device is not available anymore.
On 07.10.2021 10:48, Kalle Valo wrote:
> Mark Herbert <mark.herbert42@gmail.com> writes:
>
>> I've found similar bug files already - so not to create duplicates
>> simply commented there
>>
>> https://bugzilla.kernel.org/show_bug.cgi?id=214541
>>
>> Exactly - hibernation means saving state on disk and resuming later,
>> it is a very usefull feature when you are running out of battery and
>> want to save your work.
> Got it, thanks. So it seems we have two issues here: ath11k does not
> support hibernation and ath11k module reload failing. Let's handle the
> module reload failing in bug #214541, but can you file another bug about
> the missing support for hibernation? Let's have one issue per bug
> report, it's easier to manage bugzilla that way.
>
>> Also one more point I've found - if MHI is configured in the kernel -
>> the card is starting up on thr first boot and works. But if i
>> configure MHI as a module - it does not. MHI module do load. So do
>> ath11k and ath11k_pci. But the last ath11k dmesg says :
>>
>> mhi mhi0: Wait for device to enter SBL or Mission mode
>>
>> Exactly as if I do rmmod/modprobe sequence.
>>
>> I use latest BIOS from Dell (in fact upgrading 3.0.4 to 3.1.0 did not
>> cange anythng) , and have tried multiple kernels. Stable 5.14 series
>> and ath11k git tree (next and master-pending). Nothing helps.
>> Sometimes it woks out - I had couple of times successfully
>> rmmod/modprobe ath11k_pci but it s a kind of random and very rare
>> event ( I had it 2 times only).
>>
>> So for me it looks like an issue between MHI and ath11k , but
>> ubfortunately my knowledge is not enough to propose any solution here.
> On my NUC x86 test box I reload ath11k modules several times a day while
> I test ath11k but I always reload the mhi modules at the same time,
> maybe that makes the difference? Or XPS 13 9310 works differently from
> NUC, that's also possible.
>
--
ath11k mailing list
ath11k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath11k
^ permalink raw reply [flat|nested] 11+ messages in thread