Audio driver for Realtek HD Audio Hardware? [Testing thread]

[space223]

will this driver work for alc897?

[deomsh]

Hello @space223 , can't say, there are no records of this codec with HDA2.DLL. Did you try already?

[Drew Hoffman]

@deomsh does your universal hdaicout.hda file work for any IDT and VIA HD Audio codecs or only Realtek? I'm looking for hints on what verbs I might be missing for my native WDM HD Audio driver which now works on most Intel chipsets with a Realtek codec but is still silent on IDT and VIA codecs . Unlike the Watler's driver I have a codec node parser if a very basic one taken from a hobbyist OS.

https://msfn.org/board/topic/187713-native-wdm-hd-audio-driver-for-windows-98seme/page/1/

[deomsh]

Hi @Drew Hoffman , didn't expect you in this thread....

I am aware of your project (I am reading MSFN daily). So far I had no time to test your driver, but your project looks promising, in case of Windows 98/ME.

About your question: difficult to say. If you search in this thread for IDT, it seems there is one report, working with some quasi-universal version of HDAICOUT.HDA. But that's a while ago...

I also 'did' one VIA codec (VT1708B) in this thread, additional verbs where needed in that case. But at home I did VT1705 on Asrock N68C-S UCC without.

Somehow Realtek codecs have the 'easiest' architecture, I 'did' a few. ALC66x-ALC88x and ALC1220 have roughly the same nodes and same output pin's Apart from extra's in ALC88x. ALC231/ALC260 are already different. This is in my notes (ment for personal use):

The lines in my quasi-universal version of HDAICOUT.HDA focus on a few nodes/ widgets for power up, open the playback path and to set the DAC and volume to default values. Because there are about max 200 verbs free for sending at once, and because in my approach HDAICOUT.HDA is used to sent same verb to 'all' nodes/ widgets, so type of verbs is limited. As such this quasi-universal approach is fully 'blind', no intelligence at all (hopefully some in the maker). All based on Intel's High Definition Audio Specification, NOT on specific Realtek codecs.

In most of my versions of HDAICOUT.HDA I have concentrated on output's, but I had plans to make a new version to 'open' input-ampifiers too (mixers). But furthermore ignoring Connection Select will be the price to pay.

So maybe you can play with '35xxx', '36xxx' or L+R combined: '37xxx', just an idea. From my notes:

Set Amplifier Gain	3	Payload(s)
Set Amplifier Gain	3	5	0	8	0		Mute Input Channel 0 R
Set Amplifier Gain	3	5	0	0	0		UnMute Input Channel 0 R
Set Amplifier Gain	3	5	0	7	F		Max Input Volume Channel 0 R
Set Amplifier Gain	3	6	0	8	0		Mute Input Channel 0 L
Set Amplifier Gain	3	6	0	0	0		UnMute Input Channel 0 L
Set Amplifier Gain	3	6	0	7	F		Max Input Volume Channel 0 L
Set Amplifier Gain	3	7	0	8	0		Mute Input Channel 0 R+L
Set Amplifier Gain	3	7	0	0	0		UnMute Input Channel 0 R+L
Set Amplifier Gain	3	7	0	7	F		Max Input Volume Channel 0 R+L
Set Amplifier Gain	3	7	1	7	F		Max Input Volume Channel 1 R+L

But in case of EAPD, Unsollicited Response and GPIO, things are different and often model-specific (laptops). In that case more information is needed. I only did a few codecs/ chipsets, always on a n=1 basis, with help of study of Linux (ALSA)-sources to find appropiate verbs. 

However you have a parser, should make things more easy (Watler delivers a seperate parser too in version L, but his driver does not use a parser as such).

I am willing to help, but I am not a programmer, just an ordinary Joe with more spare time because I am not interested in sports (no offence to readers of this thread ment of course).

Edited Friday at 04:26 PM by deomsh
Addition, correction

[isolar]

Hi Deomsh! I was on this thread about 5 years ago and you assisted me to get audio working on a Soundmax AD1981HD codec. Since then I have also managed to get audio working on IDT 92HD75B3X5 codec and a Realtek ALC272 codec. I'm currently working on a 10EC 0280 codec which I have had limited success with so far, seemingly by using a mix of Watler's driver and Drew Hoffman's driver he is developing. If there's any info I can provide to assist your database please let me know and I'll be happy to assist. To note: the 10EC 0280 codec using Watler's is having the codec index mapping issue (it is mapped as index $2 which is actually the HDMI audio out, and the codec index should be $0), but with a combination of Hoffman's driver I am able to get sound - but no idea why. If you have any theories here I would be keen to know. Thank you and all the best!

[deomsh]

Hi Isolar, long time no see.

About your question: there are two possibilities, or their combination.

1) Your HD Audio controler needs a nosnoop bit set, you can check with WPCREdit.

2) The parser of WDMHDA identified some verbs you didn't use in your HDAICOUT.HDA. Possible candidates: 'opening' inputs (mixer?), EAPD or GPIO. Search in this thread for verbs.

Edited Monday at 11:45 AM by deomsh

[isolar]

Thanks Deomsh, great to be back with you again!

I am looking into these options now.

1)I have WPCREdit 1.4 and have located the registers for the HD Audio device 8086 0F04. How can I tell (if it exists) which bit is nosnoop and what to set it to?

2)I have currently sent verbs (via HDAICOUT) to map all of these except GPIO. Suspect number 1.. I will investigate further and let you know findings.

3)Side thought.. what does PCIPATCHB do and how can I tell if I should adjust it?

Current verb list is (please advise if ordering of verbs could be wrong?):

No double reset or power verbs - I had these in initially but found on reset that the codec sets power states to D0 - this seems to happen when Hoffman's driver is disabled too so not included.

$01470C02 - Enable EAPD on Lineout

$00C3B000 - Unmute Audio Mixer

$0143B000 - Unmute LineOut

$00270610 - DAC stream and channel set to stream 1 channel 0 (stream 1 as HDA specs say stream 0 is reserved as unused)

$00220011 - DAC stream 48khz 16bit 2 channels

$00C70100 - Audio Mixer connected to DAC

$01470100 - LineOut connected to Audio Mixer

$01470740 - LineOut Output enabled

SleepingWidget=$01 (AFG)

VolumeWidget=$02 (DAC - DAC at reset has volume set to 0dB so simply unmuting Audio Mixer and LineOut should still produce 0dB?)

OutputWidget=$02 (DAC - is this correct?)

Thanks hope to hear from you soon!

[deomsh]

@isolar I looked into it, up to a certain extent for now.

About 1:

My initial idea was you can make print-screens to compare register 8086:0F04 without and with HDA.SYS from @Drew Hoffman but today I searched for the appropriate PCI-registers and couldn't find them.

It seems your system has a SoC (System on Chip), not a 'classic' design. In some 4274 pages long Intel Datasheet I found things are different, seems have to do with separate PCI Configuration Registers (probably XT_SNP). So I expect no results.

I had some conversations with Copilot during traffic, and it seems it is even possible to damage your hardware in this case.

'While less common, repeatedly writing unsupported values to hardware registers could stress the hardware or violate electrical timing constraints, leading to permanent damage in extreme circumstances.'

Normally I use a disclaimer like 'writing to PCI registers is at your own risk, if any' but for now I strongly suggest NOT to write to your PCI registers and NOT try to use pcipatchB (unless you are knowing what you are doing).

About 2:

I Googled ALC280, so far GPIO not found mentioned in Linux sources!

About 3:

'$00C3B000 - Unmute Audio Mixer': this is a verb ment for OUTPUT's (3B000).

For inputs use:

Begin
;;node 0xC:Analog_Mixer
$00CB0000;AC_VERB_GET_AMP_GAIN_MUTE;Channel0_R
$00CB2000;AC_VERB_GET_AMP_GAIN_MUTE;Channel0_L
$00C37000;AC_VERB_SET_AMP_GAIN_MUTE;Channel0_R+L;Unmute
;$00C37080;AC_VERB_SET_AMP_GAIN_MUTE;Channel0_R+L;Mute
$00CB0000;AC_VERB_GET_AMP_GAIN_MUTE;Channel0_R
$00CB2000;AC_VERB_GET_AMP_GAIN_MUTE;Channel0_L
$00CB0001;AC_VERB_GET_AMP_GAIN_MUTE;Channel1_R
$00CB2001;AC_VERB_GET_AMP_GAIN_MUTE;Channel1_L
$00C37100;AC_VERB_SET_AMP_GAIN_MUTE;Channel1_R+L;Unmute
;$00C37180;AC_VERB_SET_AMP_GAIN_MUTE;Channel1_R+L;Mute
$00CB0001;AC_VERB_GET_AMP_GAIN_MUTE;Channel1_R
$00CB2001;AC_VERB_GET_AMP_GAIN_MUTE;Channel1_L
End

Mixer:

'$00C70100 - Audio Mixer connected to DAC': as such okay, but I don't expect Selector Widgets before a Mixer. If wanted, better test with GET verbs too:

$00CF0100;AC_VERB_GET_CONNECT_SEL;Channel0
$00C70100;AC_VERB_SET_CONNECT_SEL;Channel0
$00CF0100;AC_VERB_GET_CONNECT_SEL;Channel0

$00CF0101;AC_VERB_GET_CONNECT_SEL;Channel1
$00C70101;AC_VERB_SET_CONNECT_SEL;Channel1
$00CF0101;AC_VERB_GET_CONNECT_SEL;Channel1

Note for interested readers: a GET-VERB returns a meaningful payload, can be read-out in HDAICIN.TXT.

SleepingWidget:

I would set 'SleepingWidget=$02' in HDACFG.INI.

Volume:

If Volume is set to '00' I am afraid it's minimal volume. Please correct me if I am wrong!

Better try a moderate value like '3F'.

Next example Verbs are without Node/ Widget address:

B8000;AC_VERB_GET_AMP_GAIN_MUTE;status_ch0_R
BA000;AC_VERB_GET_AMP_GAIN_MUTE;status_ch0_L
3B000;AC_VERB_SET_AMP_GAIN_MUTE;unmute_ch0_L+R
B8000;AC_VERB_GET_AMP_GAIN_MUTE;status_ch0_R
BA000;AC_VERB_GET_AMP_GAIN_MUTE;status_ch0_L
3B03F;AC_VERB_SET_AMP_GAIN_MUTE;set_volume_ch0_L+R
B8000;AC_VERB_GET_AMP_GAIN_MUTE;status_ch0_R
BA000;AC_VERB_GET_AMP_GAIN_MUTE;status_ch0_L

Power Verbs:

I do not understand what you said about your Power States, please explain in (much) more detail.

Further:
Looks all good to me.

Do not forget I have at home only about four different (desktop) High Definition Audio Controllers/ Codecs, all working with my quasi-universal HDAICOUT.HDA and only one HDA controller needed pcipatchB (or set already by Watler's HDA2.DLL, like on my SB710 chipset). I have one laptop from work, but I can not even open a command-line. Booting from USB is out of the question. So I can not run any tests on machines like yours....

Edited 18 hours ago by deomsh

[isolar]

@deomsh thank you for all of this info! Notes below with @Drew Hoffman HDA.SYS renamed HDA.BAK for testing purely Watler's.

1) Noted! I will leave the PCI registers alone for now.

2) The AFG reports some GPI/O - please see codec response log file attached below for full verb parameter responses. Response on AFG is $40000005 (reports 5 GPI/O's).

3) Until now I only unmuted outputs, but tested also with combined output and input payloads as you mentioned using $00C3F03F (F:Set out+In_L+R, 0: Index 0, 3F: Unmute and volume to 3F). Also $0143F03F. No change with these.

Sleeping Widget) I tried $02. No change with this.

Volume) According to DAC1 at reset AmpOut is set to $57 which when checking steps settings is 0dB and unmuted. I maybe read somewhere that amp gain is added together if you also set it on mixer and pin widget (but maybe I am imagining this and instead it adds L+R channels?). I thought if I set $57 also on Mixer and Pin Widget the amp gain becomes $57 + $57 + $57 at LineOut.. I think I am wrong here maybe..?

Power Verbs) At the beginning of testing I set $00170500 and also $0C and $14 to 70500 to power on the nodes to D0 state, however checking F05 readings in HDAICIN without 705 set at reset: all nodes showing $00000000 (D0 power state) so I assume not needed to be set. Maybe I am wrong here and reading $00000000 could also mean no reading at all, therefore the 705 verbs are required? So I tested with that but no difference in result.

Further) That is great you have a quasi-universal HDAICOUT.HDA that works on multiple codecs! Hopefully I can help confirm some more codecs for this.

Please see image here with quick lookup reference I made from GET verbs and the responses that helped me map this codec. I am lost at what the vendor defined audio widgets could be. Any idea for these?

CODEC RESPONSES.TXT