Re: [PATCH v2] mmc: tegra: Mark 64-bit DMA broken on Tegra124

[Paul Kocialkowski <contact-W9ppeneeCTY@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 5948 bytes --]

Le jeudi 01 septembre 2016 à 13:46 +0200, Thierry Reding a écrit :
> From: Thierry Reding <treding-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
> 
> According to the TRM, the SD/MMC controller on Tegra124 supports 34-bit
> addressing, but testing shows that this doesn't work. On a device which
> has more than 2 GiB of RAM and LPAE enabled, buffer allocations can use
> addresses above the 32-bit boundary.
> 
> One way to work around this would be to enable IOMMU physical to virtual
> address translations for the SD/MMC controllers, but that's not easy to
> implement without breaking existing use-cases. It's also not obvious why
> 34-bit addressing doesn't work as advertised. In order to fix this for
> existing users, add the SDHCI_QUIRK2_BROKEN_64_BIT_DMA quirk for now.

I finally took time to test this and it sadly doesn't fix the issue I have on
tegra124 with 4 GiB RAM and LPAE.

dmesg when the issue happens (with the patch applied and some debug added)
follows:

sdhci: Secure Digital Host Controller Interface driver
sdhci: Copyright(c) Pierre Ossman
sdhci-pltfm: SDHCI platform and OF driver helper
sdhci-tegra 700b0000.sdhci: allocated mmc-pwrseq
sdhci-tegra 700b0400.sdhci: Got CD GPIO
sdhci-tegra 700b0400.sdhci: Got WP GPIO
mmc0: Unknown controller version (3). You may experience problems.
random: fast init done
sdhci_set_dma_mask(): broken 64bit dma
sdhci_set_dma_mask(): use 32bit dma
mmc0: Invalid maximum block size, assuming 512 bytes
input: Elan Touchpad as /devices/soc0/7000c400.i2c/i2c-1/1-0015/input/input1
using ADMA
usbcore: registered new interface driver usbhid
usbhid: USB HID core driver
tegra-mc 70019000.memory-controller: hdar: read @0x000000007c058000: EMEM address decode error (EMEM decode error)
tegra30-i2s 70301100.i2s: DMA channels sourced from device 70300000.ahub
max98090 0-0010: MAX98090 REVID=0x43
max98090 0-0010: use default 2.8v micbias
tegra-snd-max98090 sound: HiFi <-> 70301100.i2s mapping ok
input: GoogleNyanBig Headphones as /devices/soc0/sound/sound/card1/input2
input: GoogleNyanBig Mic Jack as /devices/soc0/sound/sound/card1/input3
tegra-mc 70019000.memory-controller: sdmmcrab: read @0x000000007c050200: EMEM address decode error (EMEM decode error)
mmc0: ADMA error

A usual dmesg without LPAE enabled follows:
[   13.593724] mmc0: Unknown controller version (3). You may experience problems.
[   13.600934] sdhci_set_dma_mask(): broken 64bit dma
[   13.605716] sdhci_set_dma_mask(): use 32bit dma
[   13.610262] mmc0: Invalid maximum block size, assuming 512 bytes
[   13.622382] input: Elan Touchpad as /devices/soc0/7000c400.i2c/i2c-1/1-0015/input/input1
[   13.675562] mmc0: SDHCI controller on 700b0600.sdhci [700b0600.sdhci] using ADMA

which shows the same other two errors about controller version and block size.

> Reported-by: Paul Kocialkowski <contact-W9ppeneeCTY@public.gmane.org>
> Acked-by: Stephen Warren <swarren-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
> Acked-by: Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org>
> Acked-by: Adrian Hunter <adrian.hunter-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> Signed-off-by: Thierry Reding <treding-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
> ---
> Changes in v2:
> - add comment with rationale for the quirk
> - add various Acked-bys
> 
>  drivers/mmc/host/sdhci-tegra.c | 27 ++++++++++++++++++++++++++-
>  1 file changed, 26 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/mmc/host/sdhci-tegra.c b/drivers/mmc/host/sdhci-tegra.c
> index 1e93dc4e303e..20b6ff5b4af1 100644
> --- a/drivers/mmc/host/sdhci-tegra.c
> +++ b/drivers/mmc/host/sdhci-tegra.c
> @@ -391,6 +391,31 @@ static const struct sdhci_tegra_soc_data
> soc_data_tegra114 = {
>  	.pdata = &sdhci_tegra114_pdata,
>  };
>  
> +static const struct sdhci_pltfm_data sdhci_tegra124_pdata = {
> +	.quirks = SDHCI_QUIRK_BROKEN_TIMEOUT_VAL |
> +		  SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK |
> +		  SDHCI_QUIRK_SINGLE_POWER_WRITE |
> +		  SDHCI_QUIRK_NO_HISPD_BIT |
> +		  SDHCI_QUIRK_BROKEN_ADMA_ZEROLEN_DESC |
> +		  SDHCI_QUIRK_CAP_CLOCK_BASE_BROKEN,
> +	.quirks2 = SDHCI_QUIRK2_PRESET_VALUE_BROKEN |
> +		   /*
> +		    * The TRM states that the SD/MMC controller found on
> +		    * Tegra124 can address 34 bits (the maximum supported by
> +		    * the Tegra memory controller), but tests show that DMA
> +		    * to or from above 4 GiB doesn't work. This is possibly
> +		    * caused by missing programming, though it's not obvious
> +		    * what sequence is required. Mark 64-bit DMA broken for
> +		    * now to fix this for existing users (e.g. Nyan boards).
> +		    */
> +		   SDHCI_QUIRK2_BROKEN_64_BIT_DMA,
> +	.ops  = &tegra114_sdhci_ops,
> +};
> +
> +static const struct sdhci_tegra_soc_data soc_data_tegra124 = {
> +	.pdata = &sdhci_tegra124_pdata,
> +};
> +
>  static const struct sdhci_pltfm_data sdhci_tegra210_pdata = {
>  	.quirks = SDHCI_QUIRK_BROKEN_TIMEOUT_VAL |
>  		  SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK |
> @@ -408,7 +433,7 @@ static const struct sdhci_tegra_soc_data soc_data_tegra210
> = {
>  
>  static const struct of_device_id sdhci_tegra_dt_match[] = {
>  	{ .compatible = "nvidia,tegra210-sdhci", .data = &soc_data_tegra210
> },
> -	{ .compatible = "nvidia,tegra124-sdhci", .data = &soc_data_tegra114
> },
> +	{ .compatible = "nvidia,tegra124-sdhci", .data = &soc_data_tegra124
> },
>  	{ .compatible = "nvidia,tegra114-sdhci", .data = &soc_data_tegra114
> },
>  	{ .compatible = "nvidia,tegra30-sdhci", .data = &soc_data_tegra30 },
>  	{ .compatible = "nvidia,tegra20-sdhci", .data = &soc_data_tegra20 },
-- 
Paul Kocialkowski, developer of low-level free software for embedded devices

Website: https://www.paulk.fr/
Coding blog: https://code.paulk.fr/
Git repositories: https://git.paulk.fr/ https://git.code.paulk.fr/

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 801 bytes --]