[alsa-devel] Unsupported Valley View 'LPE' adapter?
Hi, folks. I recently bought a Dell Venue 8 Pro - a Bay Trail / Valley View-based Intel tablet - to try and get Fedora running on it. Among the many things I'm currently fighting :) , the sound doesn't work. snd-hda-intel is not loaded at all, and there are no audio devices present at any level.
In Windows' Device Manager I see an 'Intel SST Audio Device' with PCI ID 0x86:0f28 . Indeed, the PCI ID database has this listed as "ValleyView LPE Audio Controller" - http://pci-ids.ucw.cz/read/PC/8086/0f28 . But hda-intel does not have 0f28 as a supported device. The commit for 'Bay Trail' audio support added the ID 0f04:
https://lkml.org/lkml/2013/5/16/38
So...what is this 0f28? Is support for it going to be added? Is there any info I can provide to help? (I can boot Linux on the tablet and get info out of it _just barely_, but it's pretty painful - there's no wifi and an install to 'disk' (really a usb stick) doesn't work too well, so I'm stuck working in a live image with no network).
Adam Williamson wrote:
In Windows' Device Manager I see an 'Intel SST Audio Device' with PCI ID 0x86:0f28. Indeed, the PCI ID database has this listed as "ValleyView LPE Audio Controller" - http://pci-ids.ucw.cz/read/PC/8086/0f28 . But hda-intel does not have 0f28 as a supported device.
So...what is this 0f28?
It is not an HDA controller. The SST thingy uses I²S codecs.
Is support for it going to be added?
There is something in sound/soc/mid-x86/, but that is only for the Medfield platform. The PCI SST driver that was not merged four years ago used the ID 080a.
Only Intel knows.
Is there any info I can provide to help?
"lspci -vv -d 8086:0f28"?
Regards, Clemens
On Sat, 2013-11-30 at 22:45 +0100, Clemens Ladisch wrote:
Adam Williamson wrote:
In Windows' Device Manager I see an 'Intel SST Audio Device' with PCI ID 0x86:0f28. Indeed, the PCI ID database has this listed as "ValleyView LPE Audio Controller" - http://pci-ids.ucw.cz/read/PC/8086/0f28 . But hda-intel does not have 0f28 as a supported device.
So...what is this 0f28?
It is not an HDA controller. The SST thingy uses I²S codecs.
Is support for it going to be added?
There is something in sound/soc/mid-x86/, but that is only for the Medfield platform. The PCI SST driver that was not merged four years ago used the ID 080a.
Only Intel knows.
Ah! I did see that stuff, but it seemed so old it couldn't be relevant...
Is there any info I can provide to help?
"lspci -vv -d 8086:0f28"?
OK, will get that for you shortly. Thanks for the info.
On Sat, Nov 30, 2013 at 06:38:37PM -0800, Adam Williamson wrote:
On Sat, 2013-11-30 at 22:45 +0100, Clemens Ladisch wrote:
Adam Williamson wrote:
In Windows' Device Manager I see an 'Intel SST Audio Device' with PCI ID 0x86:0f28. Indeed, the PCI ID database has this listed as "ValleyView LPE Audio Controller" - http://pci-ids.ucw.cz/read/PC/8086/0f28 . But hda-intel does not have 0f28 as a supported device.
So...what is this 0f28?
It is not an HDA controller. The SST thingy uses I²S codecs.
yes it doesnt look like HDA, but the SoC might have one which can be enabled in BIOS...
Is support for it going to be added?
There is something in sound/soc/mid-x86/, but that is only for the Medfield platform. The PCI SST driver that was not merged four years ago used the ID 080a.
Only Intel knows.
I think Liam or Jarkko might have something for you...
-- ~Vinod
Ah! I did see that stuff, but it seemed so old it couldn't be relevant...
Is there any info I can provide to help?
"lspci -vv -d 8086:0f28"?
OK, will get that for you shortly. Thanks for the info.
Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net http://www.happyassassin.net
--
On Mon, 2013-12-02 at 12:02 +0530, Vinod Koul wrote:
On Sat, Nov 30, 2013 at 06:38:37PM -0800, Adam Williamson wrote:
On Sat, 2013-11-30 at 22:45 +0100, Clemens Ladisch wrote:
Adam Williamson wrote:
In Windows' Device Manager I see an 'Intel SST Audio Device' with PCI ID 0x86:0f28. Indeed, the PCI ID database has this listed as "ValleyView LPE Audio Controller" - http://pci-ids.ucw.cz/read/PC/8086/0f28 . But hda-intel does not have 0f28 as a supported device.
So...what is this 0f28?
It is not an HDA controller. The SST thingy uses I²S codecs.
yes it doesnt look like HDA, but the SoC might have one which can be enabled in BIOS...
I don't think I saw anything like that. The firmware (pleeeeeease, call it 'the firmware', I'm fighting a heroic (quixotic?) but retreating battle against the trend to call a PC's firmware 'the BIOS' even when it's not frickin' BIOS) is pretty barebones, it has 'on/off' switches for various bits of hardware, but nothing more than that, and I'm pretty sure they're all set 'on' OOTB, inc. audio.
Is support for it going to be added?
There is something in sound/soc/mid-x86/, but that is only for the Medfield platform. The PCI SST driver that was not merged four years ago used the ID 080a.
Only Intel knows.
I think Liam or Jarkko might have something for you...
Hope so! Thanks for the reply and the redirect.
Is there any info I can provide to help?
"lspci -vv -d 8086:0f28"?
OK, will get that for you shortly. Thanks for the info.
I'm just about to build a new test image which I'm throwing a couple of kernel patches for other bugs into, I'll grab this lspci output while I'm testing that.
On Mon, 2013-12-02 at 17:56 -0800, Adam Williamson wrote:
Only Intel knows.
I think Liam or Jarkko might have something for you...
Hope so! Thanks for the reply and the redirect.
Is there any info I can provide to help?
"lspci -vv -d 8086:0f28"?
OK, will get that for you shortly. Thanks for the info.
I'm just about to build a new test image which I'm throwing a couple of kernel patches for other bugs into, I'll grab this lspci output while I'm testing that.
So I got time today to poke at this device again, and here's something: it's not actually a PCI device at all, I missed that. Neither lspci (Linux) nor pcitree (Windows) shows a PCI device with ID 8086:0f28. It must be connected some other way, SDIO or something. So, I don't know how to get more info on it, really.
I'll file a bug on this device with all the details I have attached, for now, I guess.
On Sun, 2013-12-29 at 18:19 -0800, Adam Williamson wrote:
On Mon, 2013-12-02 at 17:56 -0800, Adam Williamson wrote:
Only Intel knows.
I think Liam or Jarkko might have something for you...
Hope so! Thanks for the reply and the redirect.
Is there any info I can provide to help?
"lspci -vv -d 8086:0f28"?
OK, will get that for you shortly. Thanks for the info.
I'm just about to build a new test image which I'm throwing a couple of kernel patches for other bugs into, I'll grab this lspci output while I'm testing that.
So I got time today to poke at this device again, and here's something: it's not actually a PCI device at all, I missed that. Neither lspci (Linux) nor pcitree (Windows) shows a PCI device with ID 8086:0f28. It must be connected some other way, SDIO or something. So, I don't know how to get more info on it, really.
I'll file a bug on this device with all the details I have attached, for now, I guess.
...or, since the bug tracker is apparently down, I guess I can't. What's the next step? Just wait and see if I hear from someone with an @intel.com ? :/
I honestly wasn't expecting it to be THIS difficult to get almost anything working on a vlv tablet :(
At Sun, 29 Dec 2013 18:23:12 -0800, Adam Williamson wrote:
On Sun, 2013-12-29 at 18:19 -0800, Adam Williamson wrote:
On Mon, 2013-12-02 at 17:56 -0800, Adam Williamson wrote:
Only Intel knows.
I think Liam or Jarkko might have something for you...
Hope so! Thanks for the reply and the redirect.
> Is there any info I can provide to help?
"lspci -vv -d 8086:0f28"?
OK, will get that for you shortly. Thanks for the info.
I'm just about to build a new test image which I'm throwing a couple of kernel patches for other bugs into, I'll grab this lspci output while I'm testing that.
So I got time today to poke at this device again, and here's something: it's not actually a PCI device at all, I missed that. Neither lspci (Linux) nor pcitree (Windows) shows a PCI device with ID 8086:0f28. It must be connected some other way, SDIO or something. So, I don't know how to get more info on it, really.
I'll file a bug on this device with all the details I have attached, for now, I guess.
...or, since the bug tracker is apparently down, I guess I can't. What's the next step? Just wait and see if I hear from someone with an @intel.com ? :/
I honestly wasn't expecting it to be THIS difficult to get almost anything working on a vlv tablet :(
Ask Intel guys at best. It must be a device controlled by i2c/spi/whatever, not a standard HD-audio. IIRC, the control can be taken from ACPI, but it's not implemented properly yet.
Takashi
This is a pci device for which the driver is not yet available. I went through the intel atom processor manual and found that it has 3 I2S ports, has its own DMA. The Soc is interfaced with LPE audio core via internal PCI bus. Did you guys hear anything from intel about the driver support. Any inputs would be of great help please
On Mon, 2014-02-17 at 09:54 +0000, bobby wrote:
This is a pci device for which the driver is not yet available. I went through the intel atom processor manual and found that it has 3 I2S ports, has its own DMA. The Soc is interfaced with LPE audio core via internal PCI bus. Did you guys hear anything from intel about the driver support. Any inputs would be of great help please
Jan Michael Brummer at Intel pointed this out to me recently:
https://kernel.googlesource.com/pub/scm/linux/kernel/git/lrg/asoc/+/intel/ha...
looks like it's being worked on.
On Mon, 2014-02-17 at 16:40 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 09:54 +0000, bobby wrote:
This is a pci device for which the driver is not yet available. I went through the intel atom processor manual and found that it has 3 I2S ports, has its own DMA. The Soc is interfaced with LPE audio core via internal PCI bus. Did you guys hear anything from intel about the driver support. Any inputs would be of great help please
Jan Michael Brummer at Intel pointed this out to me recently:
https://kernel.googlesource.com/pub/scm/linux/kernel/git/lrg/asoc/+/intel/ha...
looks like it's being worked on.
Heh, that's actually just a mirror of the git ASoC branch. All those patches went through this very list last week. I don't know precisely how the upstreaming process goes for this stuff, but I'm guessing it'll wind up in 3.14 or 3.15.
(I may try building the patches into my Fedlet kernel builds and see if that gets anywhere...)
On Mon, 2014-02-17 at 16:48 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 16:40 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 09:54 +0000, bobby wrote:
This is a pci device for which the driver is not yet available. I went through the intel atom processor manual and found that it has 3 I2S ports, has its own DMA. The Soc is interfaced with LPE audio core via internal PCI bus. Did you guys hear anything from intel about the driver support. Any inputs would be of great help please
Jan Michael Brummer at Intel pointed this out to me recently:
https://kernel.googlesource.com/pub/scm/linux/kernel/git/lrg/asoc/+/intel/ha...
looks like it's being worked on.
Heh, that's actually just a mirror of the git ASoC branch. All those patches went through this very list last week. I don't know precisely how the upstreaming process goes for this stuff, but I'm guessing it'll wind up in 3.14 or 3.15.
(I may try building the patches into my Fedlet kernel builds and see if that gets anywhere...)
FWIW, I tried building a 3.14rc3 kernel with these patches applied and the drivers enabled as modules. On boot on a Dell Venue 8 Pro, all the relevant modules are loaded, but there's no ALSA device, and no obviously useful errors in any logs, or useful output from alsa-info.sh. Is it expected that the driver doesn't actually work yet? Per the recent commits, do I need some kind of firmware? Any debugging that would be useful?
Thanks.
On Tue, 2014-02-18 at 20:59 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 16:48 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 16:40 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 09:54 +0000, bobby wrote:
This is a pci device for which the driver is not yet available. I went through the intel atom processor manual and found that it has 3 I2S ports, has its own DMA. The Soc is interfaced with LPE audio core via internal PCI bus. Did you guys hear anything from intel about the driver support. Any inputs would be of great help please
Jan Michael Brummer at Intel pointed this out to me recently:
https://kernel.googlesource.com/pub/scm/linux/kernel/git/lrg/asoc/+/intel/ha...
looks like it's being worked on.
Heh, that's actually just a mirror of the git ASoC branch. All those patches went through this very list last week. I don't know precisely how the upstreaming process goes for this stuff, but I'm guessing it'll wind up in 3.14 or 3.15.
(I may try building the patches into my Fedlet kernel builds and see if that gets anywhere...)
FWIW, I tried building a 3.14rc3 kernel with these patches applied and the drivers enabled as modules. On boot on a Dell Venue 8 Pro, all the relevant modules are loaded, but there's no ALSA device, and no obviously useful errors in any logs, or useful output from alsa-info.sh. Is it expected that the driver doesn't actually work yet? Per the recent commits, do I need some kind of firmware? Any debugging that would be useful?
Sorry, we're not finished with the upstreaming yet. The current patches upstream are for the low level parts of the Intel DSP drivers. Jarkko and I still have to upstream the upper layers of the drivers (probably today) and you will also need firmware.
Once upstream the drivers will support Valley View, Baytrail, Haswell and Broadwell. The firmware for Baytrail and Valleyview will be available shortly too.
Liam
Liam Girdwood <liam.r.girdwood <at> linux.intel.com> writes:
On Tue, 2014-02-18 at 20:59 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 16:48 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 16:40 -0800, Adam Williamson wrote:
On Mon, 2014-02-17 at 09:54 +0000, bobby wrote:
This is a pci device for which the driver is not yet available. I went through the intel atom processor manual and found that it
has 3 I2S
ports, has its own DMA. The Soc is interfaced with LPE audio core via internal
PCI bus.
Did you guys hear anything from intel about the driver support.
Any inputs
would be of great help please
Jan Michael Brummer at Intel pointed this out to me recently:
https://kernel.googlesource.com/pub/scm/linux/kernel/git/lrg/asoc/+/intel/h aswell-audio-dsp%5E
looks like it's being worked on.
Heh, that's actually just a mirror of the git ASoC branch. All those patches went through this very list last week. I don't know precisely how the upstreaming process goes for this stuff, but I'm guessing
it'll
wind up in 3.14 or 3.15.
(I may try building the patches into my Fedlet kernel builds and see
if
that gets anywhere...)
FWIW, I tried building a 3.14rc3 kernel with these patches applied and the drivers enabled as modules. On boot on a Dell Venue 8 Pro, all the relevant modules are loaded, but there's no ALSA device, and no obviously useful errors in any logs, or useful output from alsa-
info.sh.
Is it expected that the driver doesn't actually work yet? Per the
recent
commits, do I need some kind of firmware? Any debugging that would be useful?
Sorry, we're not finished with the upstreaming yet. The current patches upstream are for the low level parts of the Intel DSP drivers. Jarkko and I still have to upstream the upper layers of the drivers (probably today) and you will also need firmware.
Once upstream the drivers will support Valley View, Baytrail, Haswell and Broadwell. The firmware for Baytrail and Valleyview will be available shortly too.
Liam
Thanks Liam and Jarkko saw your patches related to the Intel SST Audio Platform drivers, I hope the upstreaming process is complete for this module. Please let us know when the firmware is available.
I have few question here, the i2s mode should be enabled in the platform drivers right.
In the Bayley Bay bios I saw that the LPE audio could be enumerated via PCI or ACPI, the patches I saw mostly cover the ACPI mode. Will the PCI mode be covered in the patches.
Thanks in Advance Bobby
participants (6)
-
Adam Williamson -
bobby -
Clemens Ladisch -
Liam Girdwood -
Takashi Iwai -
Vinod Koul