[Sound-open-firmware] Using the Qemu DSP emulator
Liam Girdwood
liam.r.girdwood at linux.intel.com
Mon Mar 19 21:11:58 CET 2018
On Mon, 2018-03-19 at 15:13 +0200, Daniel Baluta wrote:
> On Mon, Mar 19, 2018 at 1:33 PM, Liam Girdwood
> <liam.r.girdwood at linux.intel.com> wrote:
> > On Sat, 2018-03-17 at 11:29 +0200, Daniel Baluta wrote:
> > > On Tue, Feb 6, 2018 at 6:38 PM, Liam Girdwood
> > > <liam.r.girdwood at linux.intel.com> wrote:
> > > > On Tue, 2018-02-06 at 10:14 +0200, Daniel Baluta wrote:
> > > > > > Sorry for the delay, I was travelling last week. Turns out I missed
> > > > > > some machine type setting in some of the machines when doing the
> > > > > > forward port. Fix pushed.
> > > > >
> > > > > Thanks Liam for the updates. I forgot to reply, I've did some hacks
> > > > > and
> > > > > finally got qemu working.
> > > > >
> > > > > Now I get a timeout when trying to boot the DSP, but will get back
> > > > > with more details once I have some spare time.
> > > >
> > > > Fwiw, firmware will send a FW ready message to the host to indicate
> > > > boot complete. This involves writing to a special shim register that
> > > > causes an IRQ on the host side.
> > >
> > > Finally got a chance to get back to this.
> > >
> > > First run I got a crash, which I think it's due to a race condition.
> >
> > Yeah, I've not root caused that yet, had to hack around it when I upgraded
> > to
> > the latest qemu release.
> >
> > >
> > > I'm using:
> > >
> > > * sof.git/master - 19e48c5 Merge branch '1.1-stable'
> >
> > Can you also use the 1.1-stable branch for the sof tools and topology.
>
> Sure.
>
> >
> > > * osadsp-qemu/master - 771e78b FIXME: log: Support 64bit trace format.
> > > * asoc/sof-v4.15 - 130e9f7 0day warnings
> > >
> >
> > Best to use my Pierre's v4.14 stable branch here as we are in the middle of
> > a
> > release (meaning my branches have more experiment patches).
> >
> > https://github.com/plbossart/sound.git #topic/sof-v4.14
>
> All right. Will give it a try.
>
> >
> > > Running
> > >
> > > $ ./xtensa-host.sh byt
> > > $ ./x86-host.sh byt
> > >
> > > I get the following crash:
> > > * https://pastebin.com/tDawwd5P
> > >
> > > full log here:
> > >
> > > * https://pastebin.com/dbjiimJx
> > >
> > > I guess this is a race and fixed it with:
> >
> > great, thanks ! Can you send that as a patch with your SoB ?
>
> All right. Which branch should I base it on?
Best to use Pierre's topic/sof-v4.14
>
> >
> > >
> > >
> > > daniel at firefly:~/w/sof-sdk/asoc$ git diff
> > > diff --git a/sound/soc/sof/trace.c b/sound/soc/sof/trace.c
> > > index 744d75b..7f5bef9 100644
> > > --- a/sound/soc/sof/trace.c
> > > +++ b/sound/soc/sof/trace.c
> > > @@ -245,6 +245,9 @@ int snd_sof_init_trace(struct snd_sof_dev *sdev)
> > > params.buffer.offset = 0;
> > > params.buffer.pages = sdev->dma_trace_pages;
> > >
> > > +
> > > + init_waitqueue_head(&sdev->trace_sleep);
> > > +
> > > /* send IPC to the DSP */
> > > ret = sof_ipc_tx_message(sdev->ipc,
> > > params.hdr.cmd, ¶ms, sizeof(params),
> > > @@ -255,7 +258,6 @@ int snd_sof_init_trace(struct snd_sof_dev *sdev)
> > > goto table_err;
> > > }
> > >
> > > - init_waitqueue_head(&sdev->trace_sleep);
> > > sdev->host_offset = 0;
> > > return 0;
> > >
> > > After this, DSP firmware boots fine but then i get this error:
> > >
> > > [ 11.264059] sof-audio sof-audio: error: ipc timed out for
> > > 0x90010000 size 0x18
> > > [ 11.264458] sof-audio sof-audio: error: waking up any trace sleepers
> > > [ 11.264460] sof-audio sof-audio: error: cant set params for DMA for
> > > Trace-
> > > 110
> > > [ 11.264467] sof-audio sof-audio: error: failed to initialize trace -110
> > >
> > >
> > > Full log here:
> > >
> > > * https://pastebin.com/VffNxrXg
> > >
> > > Any idea?
> >
> > Firmware is crashing
> >
> > [ 11.264202] sof-audio sof-audio: shim 0x40 value 0x0dead00670028800
> >
> > But what does rmbox show ?
>
> Never used rmbox. Can you provide an example.
rmbox is built as part of the tools repo. It can be installed on the guest OS
and run from the cmd line to dump HW registers, trace data or errors.
rmbox alone will dump the error log, rmbox -t will dump trace. The HW dumps are
priobably not that useful for VM use case.
>
> Another comment:
>
> Can you let me know which configuration worked for you on Qemu?
>
> Using default {xtensa|x86}-host.sh scripts with baytrail parameter
> i cannot get it working.
I've updated xtensa-host.sh and fixed some bugs on byt.
Please run "xtensa-host.sh byt - c" and then "x86-host.sh byt" in another
terminal.
>
> I have figured out that I also need a topology file.
>
> I have used both byt-rt5640 machine Or hacked to used the nocodec but still
> at some point I get a timeout from
>
> I expect that if everything works well a sound card device should be created
> and I can list it using arecord -l or aplay -l. Then I can use aplay/arecord
> to send a file to DSP for decoding.
So I recommend using nocodec atm, using the default bytcr-rt5640 should give the
same results, but I'm seeing the same resluts with that latest updates :-
[ 8.533234] sof-audio-acpi 80860F28:00: IOSF_MBI not enabled - can't determine CPU variant
This can be ignored atm, driver will default to BYT.
[ 8.533516] sof-audio-acpi 80860F28:00: info: loading firmware intel/reef-byt.ri
[ 8.555381] sof-audio sof-audio: IMR not set by BIOS. Ignoring
Can also be ignored too.
[ 8.562720] snd_soc_sst_mfld_platform: disagrees about version of symbol module_layout
Old legacy driver, can be ignored.
[ 8.589873] random: crng init done
[ 8.664129] sof-audio sof-audio: error: firmware boot timeout
[ 8.664794] sof-audio sof-audio: error: unexpected fault 0x70028800 trace 0x00000000
[ 8.664855] sof-audio sof-audio: error: failed to boot DSP firmware -5
ok, seems we dont get the interrupt from the DSP......
[ 8.665462] sof-audio: probe of sof-audio failed with error -5
[ 9.036590] irq 10: nobody cared (try booting with the "irqpoll" option)
but *we* do get the interrupt going to a different IRQ !
It seems there is a mismatch between IRQ number in qemu ACPI and IRQ mapping in the kernel.
I've seen this before and I'm not 100% sue how qemu does these mappings. I also
wont get time to look at this until after 1.1 is out.
[ 9.036669] CPU: 1 PID: 0 Comm: swapper/1 Tainted: G E 4.16.0-rc5+ #64
[ 9.036670] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.11.0-0-g63451fca13-prebuilt.qemu-project.org 04/01/2014
[ 9.036671] Call Trace:
[ 9.036673] <IRQ>
[ 9.036679] dump_stack+0x85/0xc5
[ 9.036688] __report_bad_irq+0x30/0xc0
[ 9.036691] note_interrupt+0x23e/0x290
[ 9.036695] handle_irq_event_percpu+0x51/0x70
[ 9.036698] handle_irq_event+0x37/0x60
[ 9.036700] handle_fasteoi_irq+0x83/0x140
[ 9.036703] handle_irq+0xb2/0x120
[ 9.036706] do_IRQ+0x53/0x110
[ 9.036710] common_interrupt+0xf/0xf
[ 9.036712] RIP: 0010:__do_softirq+0x9d/0x4f7
[ 9.036713] RSP: 0000:ffff9128ff903f70 EFLAGS: 00000202 ORIG_RAX: ffffffffffffffd9
[ 9.036715] RAX: ffff9128fca00040 RBX: ffff9128fca00040 RCX: 0000000000000000
[ 9.036717] RDX: ffff9128fca00040 RSI: 0000000000000000 RDI: ffff9128fca00040
[ 9.036718] RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
[ 9.036719] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
[ 9.036720] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000282
[ 9.036723] ? common_interrupt+0xa/0xf
[ 9.036731] ? hrtimer_interrupt+0x118/0x240
[ 9.036736] irq_exit+0xc1/0xd0
[ 9.036737] smp_apic_timer_interrupt+0x9a/0x2b0
[ 9.036740] apic_timer_interrupt+0xf/0x20
[ 9.036742] </IRQ>
[ 9.036743] RIP: 0010:native_safe_halt+0x2/0x10
[ 9.036745] RSP: 0000:ffffb7bf40377ec8 EFLAGS: 00000202 ORIG_RAX: ffffffffffffff12
[ 9.036747] RAX: ffff9128fca00040 RBX: 0000000000000001 RCX: 0000000000000000
[ 9.036748] RDX: ffff9128fca00040 RSI: 0000000000000001 RDI: ffff9128fca00040
[ 9.036749] RBP: 0000000000000001 R08: 0000000000000001 R09: 0000000000000000
[ 9.036750] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
[ 9.036752] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
[ 9.036759] default_idle+0x1f/0x180
[ 9.036762] do_idle+0x16e/0x1f0
[ 9.036765] cpu_startup_entry+0x6f/0x80
[ 9.036768] start_secondary+0x1ab/0x200
[ 9.036771] secondary_startup_64+0xa5/0xb0
[ 9.036778] handlers:
[ 9.036798] [<0000000033268a50>] qxl_irq_handler [qxl]
[ 9.036842] Disabling IRQ #10
[ 9.133691] cryptd: max_cpu_qlen set to 1000
[ 9.157826] AVX version of gcm_enc/dec engaged.
[ 9.157828] AES CTR mode by8 optimization enabled
[ 9.993494] Adding 8386556k swap on /dev/sda5. Priority:-2 extents:1 across:8386556k FS
[ 10.356963] bytcr_rt5640 bytcr_rt5640: quirk DMIC1_MAP enabled
[ 10.356967] bytcr_rt5640 bytcr_rt5640: quirk DMIC enabled
[ 10.356969] bytcr_rt5640 bytcr_rt5640: quirk MCLK_EN enabled
[ 10.357085] bytcr_rt5640 bytcr_rt5640: Failed to get MCLK from pmc_plt_clk_3: -2
[ 10.357304] bytcr_rt5640 bytcr_rt5640: ASoC: CPU DAI media-cpu-dai not registered
[ 10.357427] bytcr_rt5640 bytcr_rt5640: devm_snd_soc_register_card failed -517
Liam
>
> thanks,
> Daniel.
> _______________________________________________
> Sound-open-firmware mailing list
> Sound-open-firmware at alsa-project.org
> http://mailman.alsa-project.org/mailman/listinfo/sound-open-firmware
More information about the Sound-open-firmware
mailing list