ACPI: LPSS: Remove AudioDSP related ID
PATCH 1/1 ACPI: LPSS: Remove AudioDSP related ID
The AudioDSP drivers are in control for all functions of the hardware they have (they are multi-functional devices). The LPSS driver prepares for enumeration only single devices, such as DMA, UART, SPI, I²C. Hence the registration of AudioDSP should not be covered. Moreover, the very same ACPI _HID has been added by the catpt driver a few years ago.
And even more serious issue with this, is that the register window at offset 0x800 is actually D-SRAM0 in case of AudioDSP and writing to it is a data corruption.
That all being said, remove the AudioDSP ID from the LPSS driver, where it doesn't belong to.
--- drivers/acpi/x86/lpss.c | 3 --- 1 file changed, 3 deletions(-)
drivers/acpi/x86/lpss.c +0 -3
@@ -387,9 +387,6 @@ static const struct acpi_device_id acpi_lpss_device_ids[] = { { "INT3435", LPSS_ADDR(lpt_uart_dev_desc) }, { "INT3436", LPSS_ADDR(lpt_sdio_dev_desc) }, -/* Wildcat Point LPSS devices */ - { "INT3438", LPSS_ADDR(lpt_spi_dev_desc) }, - { } };
Signature · 1 line
2.47.2
Quoted 10 lines · Andy Shevchenko
The AudioDSP drivers are in control for all functions of the hardware they have (they are multi-functional devices). The LPSS driver prepares for enumeration only single devices, such as DMA, UART, SPI, I²C. Hence the registration of AudioDSP should not be covered. Moreover, the very same ACPI
_HIDhas been added by the catpt driver a few years ago.And even more serious issue with this, is that the register window at offset
0x800is actually D-SRAM0 in case of AudioDSP and writing to it is a data corruption.
Quoted 6 lines
That all being said, remove the AudioDSP ID from the LPSS driver, where it doesn't belong to.
Fixes fb94b7b11c6a ("ASoC: Intel: Remove SST firmware components")Fixes 05668be1b364 ("ASoC: Intel: Remove SST ACPI component")Fixes 7a10b66a5df9 ("ASoC: Intel: catpt: Device driver lifecycle")
Hi Andy,
A nitpick:
The fixes tags used here seem incorrect. The catpt-driver replaced its predecessor (the haswell driver) providing no interface changes. Usage of INT3438 ID has been introduced with commit c2f8783fa2d0 ("ASoC: Intel: Add common SST driver loader on ACPI systems"), if that's what you have been looking for.
In regard to the code - the change has been reviewed by me and tested in our CI, no regression on the audio side observed:
Quoted 19 lines
Signed-off-by Andy Shevchenko <[email protected]>--- drivers/acpi/x86/lpss.c | 3 --- 1 file changed, 3 deletions(-)drivers/acpi/x86/lpss.c +0
-3@@ -387,9 +387,6 @@ static const struct acpi_device_id acpi_lpss_device_ids[] = { { "INT3435", LPSS_ADDR(lpt_uart_dev_desc) }, { "INT3436", LPSS_ADDR(lpt_sdio_dev_desc) }, -/* Wildcat Point LPSS devices */ - { "INT3438", LPSS_ADDR(lpt_spi_dev_desc) }, - { } };
Quoted 20 lines · Cezary Rojewski
Quoted 10 lines · Andy Shevchenko
The AudioDSP drivers are in control for all functions of the hardware they have (they are multi-functional devices). The LPSS driver prepares for enumeration only single devices, such as DMA, UART, SPI, I²C. Hence the registration of AudioDSP should not be covered. Moreover, the very same ACPI
_HIDhas been added by the catpt driver a few years ago.And even more serious issue with this, is that the register window at offset
0x800is actually D-SRAM0 in case of AudioDSP and writing to it is a data corruption.Quoted 6 lines
That all being said, remove the AudioDSP ID from the LPSS driver, where it doesn't belong to.
Fixes fb94b7b11c6a ("ASoC: Intel: Remove SST firmware components")Fixes 05668be1b364 ("ASoC: Intel: Remove SST ACPI component")Fixes 7a10b66a5df9 ("ASoC: Intel: catpt: Device driver lifecycle")Hi Andy,
Quoted 7 lines
A nitpick:
The fixes tags used here seem incorrect. The catpt-driver replaced its predecessor (the haswell driver) providing no interface changes. Usage of INT3438 ID has been introduced with commit
c2f8783fa2d0("ASoC: Intel: Add common SST driver loader on ACPI systems"), if that's what you have been looking for.
Okay, thanks for a hint. I will use that in v2.
Quoted 5 lines
In regard to the code - the change has been reviewed by me and tested in our CI, no regression on the audio side observed:
Reviewed-by Cezary Rojewski <[email protected]>Tested-by Cezary Rojewski <[email protected]>
Signature · 2 lines
With Best Regards, Andy Shevchenko