linux-kernel - No HDA sound output with AMD IOMMU on master

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <1636040759.fnw68jgcw1.none@localhost>
Date:   Thu, 04 Nov 2021 12:01:51 -0400
From:   "Alex Xu (Hello71)" <alex_y_xu@...oo.ca>
To:     Takashi Iwai <tiwai@...e.de>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: No HDA sound output with AMD IOMMU on master

Hi,

At some point between 86f925fded03 and 5f4e450cd96b, the built-in sound 
card on my ASRock B450 Pro4 has stopped working for me. All controls are 
unchanged (no new or removed controls, all outputs are unmuted), but 
when I try to play sound, I hear a click from the output powering up but 
no sound otherwise. While attempting to play sound, the following lines 
are output to the kernel log:

[   15.081961] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac00000 flags=0x0000]
[   15.081982] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac00040 flags=0x0000]
[   15.082253] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac00080 flags=0x0000]
[   15.082581] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac000c0 flags=0x0000]
[   15.082914] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac00100 flags=0x0000]
[   15.083247] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac00140 flags=0x0000]
[   15.083580] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac00180 flags=0x0000]
[   15.083914] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac001c0 flags=0x0000]
[   15.084246] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac00200 flags=0x0000]
[   15.084579] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x14ac00240 flags=0x0000]
[  749.373925] amd_iommu_report_page_fault: 5917 callbacks suppressed
[  749.373933] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f2000000 flags=0x0000]
[  749.373952] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f2000040 flags=0x0000]
[  749.374248] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f2000080 flags=0x0000]
[  749.374574] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f20000c0 flags=0x0000]
[  749.374907] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f2000100 flags=0x0000]
[  749.375240] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f2000140 flags=0x0000]
[  749.375572] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f2000180 flags=0x0000]
[  749.375904] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f20001c0 flags=0x0000]
[  749.376238] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f2000200 flags=0x0000]
[  749.376570] snd_hda_intel 0000:0b:00.3: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0016 address=0x1f2000240 flags=0x0000]

Note that I first attempted to play audio around 15 seconds in, then 
tried again about 730 seconds later.

Adding mem_encrypt=off to the kernel command line does not fix the 
issue, but adding mem_encrypt=off amd_iommu=off to the kernel command 
line does fix it. I am not using any other IOMMU-related parameters such 
as iommu=pt.

I have not yet tested it, but it seems reasonable that this issue is 
probably related to the recently pulled "ALSA: memalloc" series of 
changes.

Let me know what further information or testing is required. I can do a 
full bisection but it is somewhat time-consuming due to the manual 
nature of the tests.

Thanks,
Alex.