few fix for sdhci-esdhc-imx
Subsystems (5)
BOUGH CHEN
maintainerv2 cover few fix for sdhci-esdhc-imx
---
Changes for v2:
Add Adrian's ack for patch 1~8
Patch 9~10, change to use 'sdhci,auto-cmd23-broken' instead of 'fsl,auto-cmd23-broken' Patch 11, clear cqhci irq status and halt cqhci, move this change to sdhci_esdhc_imx_hwinit() and add some comment to explain the reason.
Patch 12, add error return check for pinctrl_pm_select_sleep_state()
Patch 13, clear the DMA_SEL in sdhci-esdhc-imx.c instead of sdhci.c
Patch 14, add the dev->dma_parms in core/queue.c, instead of host/sdhci.c.
Haibo Chen (14):
mmc: sdhci: do not enable card detect interrupt for gpio cd type
mmc: sdhci-esdhc-imx: no fail when no pinctrl available
mmc: sdhci-esdhci-imx: retune needed for Mega/Mix enabled SoCs
mmc: sdhci-esdhc-imx: restore the per_clk rate in PM_RUNTIME
doc: dt: fsl-imx-esdhc: add strobe-dll-delay-target binding
mmc: sdhci-esdhc-imx: add strobe-dll-delay-target support
mmc: sdhci-esdhc-imx: optimize the clock setting
mmc: sdhci-esdhc-imx: optimize the strobe dll setting
doc: dt: fsl-imx-esdhc: add auto-cmd23-broken binding
mmc: sdhci-esdhc-imx: handle 'sdhci,auto-cmd23-broken' from devicetree
mmc: sdhci-esdhc-imx: clear pending interrupt and halt cqhci
mmc: sdhci-esdhc-imx: restore pin state when resume back
mmc: sdhci-esdhc-imx: clear DMA_SEL when disable DMA mode
mmc: queue: create dev->dma_parms before call dma_set_max_seg_size()
.../devicetree/bindings/mmc/fsl-imx-esdhc.txt | 13 ++ drivers/mmc/core/queue.c | 9 +- drivers/mmc/host/sdhci-esdhc-imx.c | 146 ++++++++++++++---- drivers/mmc/host/sdhci-esdhc.h | 1 + drivers/mmc/host/sdhci-pltfm.c | 3 + drivers/mmc/host/sdhci.c | 2 +- include/linux/platform_data/mmc-esdhc-imx.h | 1 + 7 files changed, 144 insertions(+), 31 deletions(-)
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 1/14 mmc: sdhci: do not enable card detect interrupt for gpio cd type
Except SDHCI_QUIRK_BROKEN_CARD_DETECTION and MMC_CAP_NONREMOVABLE, we also do not need to handle controller native card detect interrupt for gpio cd type.
If we wrong enabled the card detect interrupt for gpio case, it will cause a lot of unexpected card detect interrupts during data transfer which should not happen.
--- drivers/mmc/host/sdhci.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mmc/host/sdhci.c +1 -1
@@ -152,7 +152,7 @@ static void sdhci_set_card_detection(struct sdhci_host *host, bool enable)u32 present;if ((host->quirks &SDHCI_QUIRK_BROKEN_CARD_DETECTION ) || - !mmc_card_is_removable (host->mmc)) + !mmc_card_is_removable (host->mmc) ||mmc_can_gpio_cd (host->mmc))return ;if (enable) {
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 2/14 mmc: sdhci-esdhc-imx: no fail when no pinctrl available
In reply to BOUGH CHENWhen using jailhouse to support two Linux on i.MX8MQ EVK, we use the 1st Linux to configure pinctrl for the 2nd Linux. Then the 2nd Linux could use the mmc without pinctrl driver.
So give a warning message when no pinctrl available, but no fail probe.
--- drivers/mmc/host/sdhci-esdhc-imx.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +7 -6
@@ -1489,13 +1489,14 @@ static int sdhci_esdhc_imx_probe(struct platform_device *pdev) imx_data->pinctrl =devm_pinctrl_get (&pdev->dev);if (IS_ERR (imx_data->pinctrl)) { err =PTR_ERR (imx_data->pinctrl); - goto disable_ahb_clk; - } - - imx_data->pins_default =pinctrl_lookup_state (imx_data->pinctrl, +dev_warn (mmc_dev (host->mmc),"could not get pinctrl\n" ); + imx_data->pins_default =ERR_PTR (-EINVAL ); + }else { + imx_data->pins_default =pinctrl_lookup_state (imx_data->pinctrl,PINCTRL_STATE_DEFAULT ); -if (IS_ERR (imx_data->pins_default)) -dev_warn (mmc_dev (host->mmc),"could not get default state\n" ); +if (IS_ERR (imx_data->pins_default)) +dev_warn (mmc_dev (host->mmc),"could not get default state\n" ); + }if (esdhc_is_usdhc (imx_data)) { host->quirks2 |=SDHCI_QUIRK2_PRESET_VALUE_BROKEN ;
Signature · 1 line
2.17.1
Re: [PATCH v2 02/14] mmc: sdhci-esdhc-imx: no fail when no pinctrl available
Quoted 8 lines · BOUGH CHEN
When using jailhouse to support two Linux on i.MX8MQ EVK, we use the 1st Linux to configure pinctrl for the 2nd Linux. Then the 2nd Linux could use the mmc without pinctrl driver.
So give a warning message when no pinctrl available, but no fail probe.
Signed-off-by Haibo Chen <[email protected]>Acked-by Adrian Hunter <[email protected]>
Wow what an exotic thing.
Yours,
Linus Walleij
BOUGH CHEN
maintainerPATCH v2 3/14 mmc: sdhci-esdhci-imx: retune needed for Mega/Mix enabled SoCs
In reply to BOUGH CHENFor Mega/Mix enabled SoCs like MX7D and MX6SX, uSDHC will lost power in LP mode no matter whether the MMC_KEEP_POWER flag is set or not. This may cause state misalign between kernel and HW, especially for SDIO3.0 WiFi cards.
e.g. SDIO WiFi driver usually will keep power during system suspend. And after resume, no card re-enumeration called.
But the tuning state is lost due to Mega/Mix.
Then CRC error may happen during next data transfer.
So we should always fire a mmc_retune_needed() for such type SoC to tell MMC core retuning is needed for next data transfer.
mmc: sdhci-esdhci-imx: retune needed for Mega/Mix enabled SoCs
--- drivers/mmc/host/sdhci-esdhc-imx.c | 25 ++++++++++++++++++++----- 1 file changed, 20 insertions(+), 5 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +20 -5
@@ -160,6 +160,8 @@#define ESDHC_FLAG_CQHCI BIT(12)/* need request pmqos during low power */ #define ESDHC_FLAG_PMQOS BIT(13) +/* The IP state got lost in low power mode */ +#define ESDHC_FLAG_STATE_LOST_IN_LPMODE BIT(14)struct esdhc_soc_data {u32 flags;
@@ -193,32 +195,37 @@ static const struct esdhc_soc_data usdhc_imx6sl_data = {static const struct esdhc_soc_data usdhc_imx6sx_data = { .flags =ESDHC_FLAG_USDHC |ESDHC_FLAG_STD_TUNING - |ESDHC_FLAG_HAVE_CAP1 |ESDHC_FLAG_HS200 , + |ESDHC_FLAG_HAVE_CAP1 |ESDHC_FLAG_HS200 + |ESDHC_FLAG_STATE_LOST_IN_LPMODE , };static const struct esdhc_soc_data usdhc_imx6ull_data = { .flags =ESDHC_FLAG_USDHC |ESDHC_FLAG_STD_TUNING |ESDHC_FLAG_HAVE_CAP1 |ESDHC_FLAG_HS200 - |ESDHC_FLAG_ERR010450 , + |ESDHC_FLAG_ERR010450 + |ESDHC_FLAG_STATE_LOST_IN_LPMODE , };static const struct esdhc_soc_data usdhc_imx7d_data = { .flags =ESDHC_FLAG_USDHC |ESDHC_FLAG_STD_TUNING |ESDHC_FLAG_HAVE_CAP1 |ESDHC_FLAG_HS200 - |ESDHC_FLAG_HS400 , + |ESDHC_FLAG_HS400 + |ESDHC_FLAG_STATE_LOST_IN_LPMODE , };static struct esdhc_soc_data usdhc_imx7ulp_data = { .flags =ESDHC_FLAG_USDHC |ESDHC_FLAG_STD_TUNING |ESDHC_FLAG_HAVE_CAP1 |ESDHC_FLAG_HS200 - |ESDHC_FLAG_PMQOS |ESDHC_FLAG_HS400 , + |ESDHC_FLAG_PMQOS |ESDHC_FLAG_HS400 + |ESDHC_FLAG_STATE_LOST_IN_LPMODE , };static struct esdhc_soc_data usdhc_imx8qxp_data = { .flags =ESDHC_FLAG_USDHC |ESDHC_FLAG_STD_TUNING |ESDHC_FLAG_HAVE_CAP1 |ESDHC_FLAG_HS200 |ESDHC_FLAG_HS400 |ESDHC_FLAG_HS400_ES - |ESDHC_FLAG_CQHCI , + |ESDHC_FLAG_CQHCI + |ESDHC_FLAG_STATE_LOST_IN_LPMODE , };struct pltfm_imx_data {
@@ -1613,6 +1620,8 @@ static int sdhci_esdhc_imx_remove(struct platform_device *pdev)static int sdhci_esdhc_suspend (struct device *dev) {struct sdhci_host *host =dev_get_drvdata (dev); +struct sdhci_pltfm_host *pltfm_host =sdhci_priv (host); +struct pltfm_imx_data *imx_data =sdhci_pltfm_priv (pltfm_host);int ret;if (host->mmc->caps2 &MMC_CAP2_CQE ) {
@@ -1621,6 +1630,12 @@ static int sdhci_esdhc_suspend(struct device *dev)return ret; } +if ((imx_data->socdata->flags &ESDHC_FLAG_STATE_LOST_IN_LPMODE ) && + (host->tuning_mode !=SDHCI_TUNING_MODE_1 )) { +mmc_retune_timer_stop (host->mmc); +mmc_retune_needed (host->mmc); + } +if (host->tuning_mode !=SDHCI_TUNING_MODE_3 )mmc_retune_needed (host->mmc);
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 4/14 mmc: sdhci-esdhc-imx: restore the per_clk rate in PM_RUNTIME
In reply to BOUGH CHENWhen pm_runtime_suspend is run, a call to SCFW power off the SS (SS is a power domain, usdhc belong to this SS power domain) in which the resource resides is made. The SCFW can power off the SS if no other resource in active in that SS. If so, all state associated with all the resources within the SS that is powered off is lost, this includes the clock rates, clock state etc. When pm_runtime_resume is called, the SS associated with that resource is powered up. But the clocks are left in the default state.
This patch restore clock rate in pm_runtime_resume, make sure the clock is right rather than depending on the default state setting by SCFW.
--- drivers/mmc/host/sdhci-esdhc-imx.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/mmc/host/sdhci-esdhc-imx.c +7 -1
@@ -162,6 +162,8 @@#define ESDHC_FLAG_PMQOS BIT(13)/* The IP state got lost in low power mode */ #define ESDHC_FLAG_STATE_LOST_IN_LPMODE BIT(14) +/* The IP lost clock rate in PM_RUNTIME */ +#define ESDHC_FLAG_CLK_RATE_LOST_IN_PM_RUNTIME BIT(15)struct esdhc_soc_data {u32 flags;
@@ -225,7 +227,8 @@ static struct esdhc_soc_data usdhc_imx8qxp_data = { | ESDHC_FLAG_HAVE_CAP1 | ESDHC_FLAG_HS200 | ESDHC_FLAG_HS400 | ESDHC_FLAG_HS400_ES | ESDHC_FLAG_CQHCI - | ESDHC_FLAG_STATE_LOST_IN_LPMODE, + | ESDHC_FLAG_STATE_LOST_IN_LPMODE + | ESDHC_FLAG_CLK_RATE_LOST_IN_PM_RUNTIME, };struct pltfm_imx_data {
@@ -1705,6 +1708,9 @@ static int sdhci_esdhc_runtime_resume(struct device *dev)pm_qos_add_request (&imx_data->pm_qos_req,PM_QOS_CPU_DMA_LATENCY ,0 ); +if (imx_data->socdata->flags &ESDHC_FLAG_CLK_RATE_LOST_IN_PM_RUNTIME ) +clk_set_rate (imx_data->clk_per, pltfm_host->clock); + err =clk_prepare_enable (imx_data->clk_ahb);if (err) goto remove_pm_qos_request;
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 5/14 doc: dt: fsl-imx-esdhc: add strobe-dll-delay-target binding
In reply to BOUGH CHENAdd fsl,strobe-dll-delay-target binding.
--- Documentation/devicetree/bindings/mmc/fsl-imx-esdhc.txt | 5 +++++ 1 file changed, 5 insertions(+)
Documentation/devicetree/bindings/mmc/fsl-imx-esdhc.txt +5 -0
@@ -41,6 +41,11 @@ Optional properties: This property allows user to change the tuning step to more than one delay cells which is useful for some special boards or cards when the default tuning step can't find the proper delay window within limited tuning retries. +- fsl,strobe-dll-delay-target: Specify the strobe dll control slave delay target. + This delay target programming host controller loopback read clock, and this + property allows user to change the delay target for the strobe input read clock. + If not use this property, driver default set the delay target to value 7. + Only eMMC HS400 mode need to take care of this property. Examples:
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 6/14 mmc: sdhci-esdhc-imx: add strobe-dll-delay-target support
In reply to BOUGH CHENstrobe-dll-delay-target is the delay cell add on the strobe line. Strobe line the the uSDHC loopback read clock which is use in HS400 mode. Different strobe-dll-delay-target may need to set for different board/SoC. If this delay cell is not set to an appropriate value, we may see some read operation meet CRC error after HS400 mode select which already pass the tuning.
This patch add the strobe-dll-delay-target setting in driver, so that user can easily config this delay cell in dts file.
--- drivers/mmc/host/sdhci-esdhc-imx.c | 12 +++++++++++- include/linux/platform_data/mmc-esdhc-imx.h | 1 + 2 files changed, 12 insertions(+), 1 deletion(-)
drivers/mmc/host/sdhci-esdhc-imx.c +11 -1
@@ -73,6 +73,7 @@#define ESDHC_STROBE_DLL_CTRL 0x70#define ESDHC_STROBE_DLL_CTRL_ENABLE (1 << 0)#define ESDHC_STROBE_DLL_CTRL_RESET (1 << 1) +#define ESDHC_STROBE_DLL_CTRL_SLV_DLY_TARGET_DEFAULT 0x7#define ESDHC_STROBE_DLL_CTRL_SLV_DLY_TARGET_SHIFT 3#define ESDHC_STROBE_DLL_CTRL_SLV_UPDATE_INT_DEFAULT (4 << 20)
@@ -995,6 +996,9 @@ static int esdhc_change_pinstate(struct sdhci_host *host,*/ static void esdhc_set_strobe_dll (struct sdhci_host *host) { +struct sdhci_pltfm_host *pltfm_host =sdhci_priv (host); +struct pltfm_imx_data *imx_data =sdhci_pltfm_priv (pltfm_host); +u32 strobe_delay;u32 v;/* disable clock before enabling strobe dll */
@@ -1012,9 +1016,13 @@ static void esdhc_set_strobe_dll(struct sdhci_host *host)* enable strobe dll ctrl and adjust the delay target * for the uSDHC loopback read clock */ +if (imx_data->boarddata.strobe_dll_delay_target) + strobe_delay = imx_data->boarddata.strobe_dll_delay_target; +else + strobe_delay =ESDHC_STROBE_DLL_CTRL_SLV_DLY_TARGET_DEFAULT ; v =ESDHC_STROBE_DLL_CTRL_ENABLE |ESDHC_STROBE_DLL_CTRL_SLV_UPDATE_INT_DEFAULT | - (7 <<ESDHC_STROBE_DLL_CTRL_SLV_DLY_TARGET_SHIFT ); + (strobe_delay <<ESDHC_STROBE_DLL_CTRL_SLV_DLY_TARGET_SHIFT );writel (v, host->ioaddr +ESDHC_STROBE_DLL_CTRL );/* wait 5us to make sure strobe dll status register stable */ udelay (5 );
@@ -1340,6 +1348,8 @@ sdhci_esdhc_imx_probe_dt(struct platform_device *pdev,of_property_read_u32 (np,"fsl,tuning-start-tap" , &boarddata->tuning_start_tap); +of_property_read_u32 (np,"fsl,strobe-dll-delay-target" , + &boarddata->strobe_dll_delay_target);if (of_find_property (np,"no-1-8-v" ,NULL )) host->quirks2 |=SDHCI_QUIRK2_NO_1_8_V ;
include/linux/platform_data/mmc-esdhc-imx.h +1 -0
@@ -37,5 +37,6 @@ struct esdhc_platform_data {unsigned int delay_line;unsigned int tuning_step;/* The delay cell steps in tuning procedure */ unsigned int tuning_start_tap;/* The start delay cell point in tuning procedure */ + unsigned int strobe_dll_delay_target; /* The delay cell for strobe pad (read clock) */ };#endif /* __ASM_ARCH_IMX_ESDHC_H */
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 7/14 mmc: sdhci-esdhc-imx: optimize the clock setting
In reply to BOUGH CHENWhen force clock off, check the SDOFF of register PRSSTAT to make sure the clock is gate off. Before force clock on, check the SDSTB of register PRSSTAT to make sure the clock is stable, this will eliminate the clock glitch.
--- drivers/mmc/host/sdhci-esdhc-imx.c | 24 +++++++++++++++++++++++- drivers/mmc/host/sdhci-esdhc.h | 1 + 2 files changed, 24 insertions(+), 1 deletion(-)
drivers/mmc/host/sdhci-esdhc-imx.c +23 -1
@@ -9,6 +9,7 @@*/ #include <linux/io.h> +#include <linux/iopoll.h> #include <linux/delay.h> #include <linux/err.h> #include <linux/clk.h>
@@ -313,6 +314,17 @@ static inline void esdhc_clrset_le(struct sdhci_host *host, u32 mask, u32 val, iwritel (((readl (base) & ~(mask << shift)) | (val << shift)), base); } +static inline void esdhc_wait_for_card_clock_gate_off (struct sdhci_host *host) +{ +u32 present_state; +int ret; + + ret =readl_poll_timeout (host->ioaddr +ESDHC_PRSSTAT , present_state, + (present_state &ESDHC_CLOCK_GATE_OFF ),2 ,100 ); +if (ret == -ETIMEDOUT ) +dev_warn (mmc_dev (host->mmc),"%s: card clock still not gate off in 100us!.\n" , __func__); +} +static u32 esdhc_readl_le (struct sdhci_host *host,int reg) {struct sdhci_pltfm_host *pltfm_host =sdhci_priv (host);
@@ -526,6 +538,8 @@ static void esdhc_writew_le(struct sdhci_host *host, u16 val, int reg)else new_val &= ~ESDHC_VENDOR_SPEC_FRC_SDCLK_ON ;writel (new_val, host->ioaddr +ESDHC_VENDOR_SPEC ); +if (!(new_val &ESDHC_VENDOR_SPEC_FRC_SDCLK_ON )) +esdhc_wait_for_card_clock_gate_off (host);return ;case SDHCI_HOST_CONTROL2 : new_val =readl (host->ioaddr +ESDHC_VENDOR_SPEC );
@@ -754,12 +768,14 @@ static inline void esdhc_pltfm_set_clock(struct sdhci_host *host,int ddr_pre_div = imx_data->is_ddr ?2 :1 ;int pre_div =1 ;int div =1 ; +int ret;u32 temp, val;if (esdhc_is_usdhc (imx_data)) { val =readl (host->ioaddr +ESDHC_VENDOR_SPEC );writel (val & ~ESDHC_VENDOR_SPEC_FRC_SDCLK_ON , host->ioaddr +ESDHC_VENDOR_SPEC ); +esdhc_wait_for_card_clock_gate_off (host); }if (clock ==0 ) {
@@ -814,13 +830,18 @@ static inline void esdhc_pltfm_set_clock(struct sdhci_host *host, | (pre_div << ESDHC_PREDIV_SHIFT));sdhci_writel (host, temp,ESDHC_SYSTEM_CONTROL ); +/* need to wait the bit 3 of the PRSSTAT to be set, make sure card clock is stable */ + ret =readl_poll_timeout (host->ioaddr +ESDHC_PRSSTAT , temp, + (temp &ESDHC_CLOCK_STABLE ),2 ,100 ); +if (ret == -ETIMEDOUT ) +dev_warn (mmc_dev (host->mmc),"card clock still not stable in 100us!.\n" ); +if (esdhc_is_usdhc (imx_data)) { val =readl (host->ioaddr +ESDHC_VENDOR_SPEC );writel (val |ESDHC_VENDOR_SPEC_FRC_SDCLK_ON , host->ioaddr +ESDHC_VENDOR_SPEC ); } -mdelay (1 ); }static unsigned int esdhc_pltfm_get_ro (struct sdhci_host *host)
@@ -1005,6 +1026,7 @@ static void esdhc_set_strobe_dll(struct sdhci_host *host)writel (readl (host->ioaddr +ESDHC_VENDOR_SPEC ) & ~ESDHC_VENDOR_SPEC_FRC_SDCLK_ON , host->ioaddr +ESDHC_VENDOR_SPEC ); +esdhc_wait_for_card_clock_gate_off (host);/* force a reset on strobe dll */ writel(ESDHC_STROBE_DLL_CTRL_RESET,
drivers/mmc/host/sdhci-esdhc.h +1 -0
@@ -31,6 +31,7 @@/* Present State Register */ #define ESDHC_PRSSTAT 0x24 +#define ESDHC_CLOCK_GATE_OFF 0x00000080#define ESDHC_CLOCK_STABLE 0x00000008/* Protocol Control Register */
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 8/14 mmc: sdhci-esdhc-imx: optimize the strobe dll setting
In reply to BOUGH CHENAfter set the STROBE SLV delay target value, it need to wait some time to let the usdhc lock the REF and SLV clock. In normal case, 1~2us is enough for imx8/imx6 and imx7d, and 4~5us is enough for imx7ulp, but when do reboot stress test or do the bind/unbind stress test, sometimes need to wait about 10us to get the status lock.
This patch optimize delay handle method, only print the warning message if the status is still not lock after 1ms delay.
--- drivers/mmc/host/sdhci-esdhc-imx.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +7 -8
@@ -1021,6 +1021,7 @@ static void esdhc_set_strobe_dll(struct sdhci_host *host)struct pltfm_imx_data *imx_data =sdhci_pltfm_priv (pltfm_host);u32 strobe_delay;u32 v; +int ret;/* disable clock before enabling strobe dll */ writel(readl(host->ioaddr + ESDHC_VENDOR_SPEC) &
@@ -1046,15 +1047,13 @@ static void esdhc_set_strobe_dll(struct sdhci_host *host)ESDHC_STROBE_DLL_CTRL_SLV_UPDATE_INT_DEFAULT | (strobe_delay <<ESDHC_STROBE_DLL_CTRL_SLV_DLY_TARGET_SHIFT );writel (v, host->ioaddr +ESDHC_STROBE_DLL_CTRL ); -/* wait 5us to make sure strobe dll status register stable */ -udelay (5 ); - v =readl (host->ioaddr +ESDHC_STROBE_DLL_STATUS ); -if (!(v &ESDHC_STROBE_DLL_STS_REF_LOCK )) -dev_warn (mmc_dev (host->mmc), -"warning! HS400 strobe DLL status REF not lock!\n" ); -if (!(v &ESDHC_STROBE_DLL_STS_SLV_LOCK )) + +/* wait max 50us to get the REF/SLV lock */ + ret =readl_poll_timeout (host->ioaddr +ESDHC_STROBE_DLL_STATUS , v, + ((v &ESDHC_STROBE_DLL_STS_REF_LOCK ) && (v &ESDHC_STROBE_DLL_STS_SLV_LOCK )),1 ,50 ); +if (ret == -ETIMEDOUT )dev_warn (mmc_dev (host->mmc), -"warning! HS400 strobe DLL status SLV not lock!\n" ); +"warning! HS400 strobe DLL status REF/SLV not lock in 50us, STROBE DLL status is %x!\n" , v); }static void esdhc_reset_tuning (struct sdhci_host *host)
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 9/14 doc: dt: fsl-imx-esdhc: add auto-cmd23-broken binding
In reply to BOUGH CHENAdd sdhci,auto-cmd23-broken binding.
--- Documentation/devicetree/bindings/mmc/fsl-imx-esdhc.txt | 8 ++++++++ 1 file changed, 8 insertions(+)
Documentation/devicetree/bindings/mmc/fsl-imx-esdhc.txt +8 -0
@@ -46,6 +46,14 @@ Optional properties: property allows user to change the delay target for the strobe input read clock. If not use this property, driver default set the delay target to value 7. Only eMMC HS400 mode need to take care of this property. +- sdhci,auto-cmd23-broken: disable the ACMD23 function of USDHC. + This is required for eMMC on imx6qpdl/imx6sx/imx7d when it use ADMA mode. Because + for these SoC, it do not support the ACMD23 completely, only take the 16 bit block + count from the 0x4 register (BLK_ATT) as argument for the ACMD23, the upper 16 bit + of the CMD23's argument is ignored. This will impact the reliable write operation + and the RPMB block write operation, because these operations need to set the bit 31 + of the CMD23's argument. SDMA mode will default disable the ACMD23 mode. SD card do + not has this limitation on these SoCs. Examples:
Signature · 1 line
2.17.1
Re: [PATCH v2 09/14] doc: dt: fsl-imx-esdhc: add auto-cmd23-broken binding
Documentation/devicetree/bindings/mmc/fsl-imx-esdhc.txt Optional properties: · line 56
Quoted 8 lines · BOUGH CHEN
+- sdhci,auto-cmd23-broken: disable the ACMD23 function of USDHC.
Documentation/devicetree/bindings/mmc/fsl-imx-esdhc.txt +7
-0+ This is required for eMMC on imx6qpdl/imx6sx/imx7d when it use ADMA mode. Because + for these SoC, it do not support the ACMD23 completely, only take the 16 bit block + count from the 0x4 register (BLK_ATT) as argument for the ACMD23, the upper 16 bit + of the CMD23's argument is ignored. This will impact the reliable write operation + and the RPMB block write operation, because these operations need to set the bit 31 + of the CMD23's argument. SDMA mode will default disable the ACMD23 mode. SD card do + not has this limitation on these SoCs.
This looks weird.
Is the bug in the *host controller* or in *the card*?
It looks like the card.
And then it should just be a quirk for this eMMC card in drivers/mmc/core/quirks.h plus corresponding code in the core.
Otherwise everyone using this eMMC card has to add this property in their device tree and that is wrong if we can already detect it from the card ID and apply it for that card, anywhere.
This is not the first time I see card quirks as DT properties, it seems like a bit over-reliance on DT.
Yours,
Linus Walleij
Re: [PATCH v2 09/14] doc: dt: fsl-imx-esdhc: add auto-cmd23-broken binding
Quoted 11 lines · Linus Walleij
Documentation/devicetree/bindings/mmc/fsl-imx-esdhc.txt Optional properties: · line 56
Quoted 8 lines · BOUGH CHEN
+- sdhci,auto-cmd23-broken: disable the ACMD23 function of USDHC.
Documentation/devicetree/bindings/mmc/fsl-imx-esdhc.txt +7
-0+ This is required for eMMC on imx6qpdl/imx6sx/imx7d when it use ADMA mode. Because + for these SoC, it do not support the ACMD23 completely, only take the 16 bit block + count from the 0x4 register (BLK_ATT) as argument for the ACMD23, the upper 16 bit + of the CMD23's argument is ignored. This will impact the reliable write operation + and the RPMB block write operation, because these operations need to set the bit 31 + of the CMD23's argument. SDMA mode will default disable the ACMD23 mode. SD card do + not has this limitation on these SoCs.
Quoted 5 lines
This looks weird.
Is the bug in the *host controller* or in *the card*?
It looks like the card.
After looking at the next patch it looks like the host controller.
In that case the compatible-string should indicate what version of the IP you are using and if it has this bug.
No special flags needed for that.
Yours,
Linus Walleij
BOUGH CHEN
maintainerRE: [PATCH v2 09/14] doc: dt: fsl-imx-esdhc: add auto-cmd23-broken binding
Quoted 34 lines
-----Original Message-----
From: Linus Walleij <[email protected]>
Sent: 2019年12月11日 7:40
To: BOUGH CHEN <[email protected]>
Cc: [email protected]; [email protected]; [email protected]; [email protected]; dl-linux-imx <[email protected]>; [email protected]
Subject: Re: [PATCH v2 09/14] doc: dt: fsl-imx-esdhc: add auto-cmd23-broken bindingQuoted 1 line · Linus Walleij
On Tue, Dec 3, 2019 at 1:54 PM BOUGH CHEN <[email protected]>
Quoted 20 lines · wrote:
Quoted 5 lines
+- sdhci,auto-cmd23-broken: disable the ACMD23 function of USDHC.
(quoted diff) +3
-0+ This is required for eMMC on imx6qpdl/imx6sx/imx7d when it use +ADMA mode. Because + for these SoC, it do not support the ACMD23 completely, only takeReview
Quoted 10 lines
(quoted diff) +10
-0+the 16 bit block + count from the 0x4 register (BLK_ATT) as argument for the ACMD23, +the upper 16 bit + of the CMD23's argument is ignored. This will impact the reliable +write operation + and the RPMB block write operation, because these operations need +to set the bit 31 + of the CMD23's argument. SDMA mode will default disable the +ACMD23 mode. SD card do + not has this limitation on these SoCs.This looks weird.
Is the bug in the *host controller* or in *the card*?
Quoted 8 lines
Quoted 1 line
It looks like the card.
After looking at the next patch it looks like the host controller.
In that case the compatible-string should indicate what version of the IP you are using and if it has this bug.
No special flags needed for that.
Hi Linus,
Yes, this is host IP limitation. I did consider the method as you suggested, use compatible-string to distinguish. But then I notice that this host limitation only impact the eMMC device for RPMB reliable write, for SD card, it do not support this mode, so this hardware limitation do not impact for sd card. This is why I use "sdhci,auto-cmd23-broken" in devicetree, only the emmc device need to contain this in dts file.
I double check this issue, since this auto-cmd23 will not impact the sd performance, and it is the host IP limitation, I will chose to accept your suggestion, will send a new patch.
Thanks,
Bough Chen
Quoted 3 lines
Yours,
Linus Walleij
Re: [PATCH v2 09/14] doc: dt: fsl-imx-esdhc: add auto-cmd23-broken binding
Quoted 42 lines · BOUGH CHEN
Quoted 24 lines
Quoted 1 line · Linus Walleij
On Tue, Dec 3, 2019 at 1:54 PM BOUGH CHEN <[email protected]>
Quoted 20 lines · wrote:
Quoted 5 lines
+- sdhci,auto-cmd23-broken: disable the ACMD23 function of USDHC.
(quoted diff) +3
-0+ This is required for eMMC on imx6qpdl/imx6sx/imx7d when it use +ADMA mode. Because + for these SoC, it do not support the ACMD23 completely, only takeReview
Quoted 10 lines
(quoted diff) +10
-0+the 16 bit block + count from the 0x4 register (BLK_ATT) as argument for the ACMD23, +the upper 16 bit + of the CMD23's argument is ignored. This will impact the reliable +write operation + and the RPMB block write operation, because these operations need +to set the bit 31 + of the CMD23's argument. SDMA mode will default disable the +ACMD23 mode. SD card do + not has this limitation on these SoCs.This looks weird.
Is the bug in the *host controller* or in *the card*?
Quoted 8 lines
Quoted 1 line
It looks like the card.
After looking at the next patch it looks like the host controller.
In that case the compatible-string should indicate what version of the IP you are using and if it has this bug.
No special flags needed for that.
Hi Linus,
Yes, this is host IP limitation. I did consider the method as you suggested,
use compatible-string to distinguish. But then I notice that this host limitation
only impact the eMMC device for RPMB reliable write, for SD card, it do not support this mode, so this hardware limitation do not impact for sd card. This is why I use "sdhci,auto-cmd23-broken" in devicetree, only the emmc device need to contain this in dts file.
Quoted 3 lines
I double check this issue, since this auto-cmd23 will not impact the sd performance, and it is the host IP limitation, I will chose to accept your suggestion, will send a new patch.
Should be fine, the code in the MMC core knows if what you are using is an eMMC or an SD card so you can just handle the quirk only in the eMMC case.
Yours,
Linus Walleij
BOUGH CHEN
maintainerPATCH v2 10/14 mmc: sdhci-esdhc-imx: handle 'sdhci,auto-cmd23-broken' from devicetree
In reply to BOUGH CHENSince L4.15, community involve the commit 105819c8a545 ("mmc: core: use mrq->sbc when sending CMD23 for RPMB"), let the usdhc to decide whether to use ACMD23 for RPMB. This CMD23 for RPMB need to set the bit 31 to its argument, if not, the RPMB write operation will return general fail.
According to the sdhci logic, SDMA mode will disable the ACMD23, and only in ADMA mode, it will chose to use ACMD23 if the host support. But according to debug, and confirm with IC, the imx6qpdl/imx6sx/imx6sl/imx7d do not support the ACMD23 feature completely. These SoCs only use the 16 bit block count of the register 0x4 (BLOCK_ATT) as the CMD23's argument in ACMD23 mode, which means it will ignore the upper 16 bit of the CMD23's argument. This will block the reliable write operation in RPMB, because RPMB reliable write need to set the bit31 of the CMD23's argument. This is the hardware limitation. Due to imx6sl use SDMA, so for imx6qpdl/imx6sx/imx7d, it need to broke the ACMD23 for eMMC, SD card do not has this limitation, because SD card do not support reliable write.
For imx6ul/imx6ull/imx6sll/imx7ulp/imx8, it support the ACMD23 completely, it change to use the 0x0 register (DS_ADDR) to put the CMD23's argument in ADMA mode.
This patch handle 'sdhci,auto-cmd23-broken' from devicetree.
--- drivers/mmc/host/sdhci-esdhc-imx.c | 5 +++-- drivers/mmc/host/sdhci-pltfm.c | 3 +++ 2 files changed, 6 insertions(+), 2 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +3 -2
@@ -1371,8 +1371,6 @@ sdhci_esdhc_imx_probe_dt(struct platform_device *pdev,of_property_read_u32 (np,"fsl,strobe-dll-delay-target" , &boarddata->strobe_dll_delay_target); -if (of_find_property (np,"no-1-8-v" ,NULL )) - host->quirks2 |=SDHCI_QUIRK2_NO_1_8_V ;if (of_property_read_u32 (np,"fsl,delay-line" , &boarddata->delay_line)) boarddata->delay_line =0 ;
@@ -1386,6 +1384,9 @@ sdhci_esdhc_imx_probe_dt(struct platform_device *pdev, ESDHC_PINCTRL_STATE_200MHZ); } +/* call to generic sdhci_get_property to support additional capabilities */ +sdhci_get_property (pdev); +/* call to generic mmc_of_parse to support additional capabilities */ ret =mmc_of_parse (host->mmc);if (ret)
drivers/mmc/host/sdhci-pltfm.c +3 -0
@@ -86,6 +86,9 @@ void sdhci_get_property(struct platform_device *pdev)if (device_property_present (dev,"sdhci,auto-cmd12" )) host->quirks |=SDHCI_QUIRK_MULTIBLOCK_READ_ACMD12 ; +if (device_property_present (dev,"sdhci,auto-cmd23-broken" )) + host->quirks2 |=SDHCI_QUIRK2_ACMD23_BROKEN ; +if (device_property_present (dev,"sdhci,1-bit-only" ) || (device_property_read_u32 (dev,"bus-width" , &bus_width) ==0 && bus_width ==1 ))
Signature · 1 line
2.17.1
Re: [PATCH v2 10/14] mmc: sdhci-esdhc-imx: handle 'sdhci,auto-cmd23-broken' from devicetree
Quoted 17 lines · BOUGH CHEN
Since L4.15, community involve the commit
105819c8a545("mmc: core: usemrq->sbcwhen sending CMD23 for RPMB"), let the usdhc to decide whether to use ACMD23 for RPMB. This CMD23 for RPMB need to set the bit 31 to its argument, if not, the RPMB write operation will return general fail.According to the sdhci logic, SDMA mode will disable the ACMD23, and only in ADMA mode, it will chose to use ACMD23 if the host support. But according to debug, and confirm with IC, the imx6qpdl/imx6sx/imx6sl/imx7d do not support the ACMD23 feature completely. These SoCs only use the 16 bit block count of the register
0x4(BLOCK_ATT) as the CMD23's argument in ACMD23 mode, which means it will ignore the upper 16 bit of the CMD23's argument. This will block the reliable write operation in RPMB, because RPMB reliable write need to set the bit31 of the CMD23's argument. This is the hardware limitation. Due to imx6sl use SDMA, so for imx6qpdl/imx6sx/imx7d, it need to broke the ACMD23 for eMMC, SD card do not has this limitation, because SD card do not support reliable write.
Quoted 6 lines
For imx6ul/imx6ull/imx6sll/imx7ulp/imx8, it support the ACMD23 completely, it change to use the
0x0register (DS_ADDR) to put the CMD23's argument in ADMA mode.This patch handle 'sdhci,auto-cmd23-broken' from devicetree.
Signed-off-by Haibo Chen <[email protected]>
So it does seem like this is a problem with the host controller, not with the card?
drivers/mmc/host/sdhci-esdhc-imx.c sdhci_esdhc_imx_probe_dt(struct platform_device *pdev, · line 1375
Quoted 2 lines
drivers/mmc/host/sdhci-esdhc-imx.c +0
-2-if (of_find_property (np,"no-1-8-v" ,NULL )) - host->quirks2 |=SDHCI_QUIRK2_NO_1_8_V ;
Why are you deleting this? Seems unrelated.
drivers/mmc/host/sdhci-esdhc-imx.c sdhci_esdhc_imx_probe_dt(struct platform_device *pdev, · line 1388
Quoted 2 lines
drivers/mmc/host/sdhci-esdhc-imx.c +2
-0+/* call to generic sdhci_get_property to support additional capabilities */ +sdhci_get_property (pdev);
Or does it get set by this call?
(Then explain that change in the commit message.)
drivers/mmc/host/sdhci-pltfm.c void sdhci_get_property(struct platform_device *pdev) · line 90
Quoted 2 lines
drivers/mmc/host/sdhci-pltfm.c +2
-0+if (device_property_present (dev,"sdhci,auto-cmd23-broken" )) + host->quirks2 |=SDHCI_QUIRK2_ACMD23_BROKEN ;
Even if this is a problem with the host controller I am not convinced.
Can't you just use the compatible-string of the host controller to switch this flag?
They should be different compatibles because if this bug is present in one version of the host controller and not another version then by definition they are not compatible, right?
The idea with very specific compatible strings is to handle exactly things like this.
Yours,
Linus Walleij
BOUGH CHEN
maintainerPATCH v2 11/14 mmc: sdhci-esdhc-imx: clear pending interrupt and halt cqhci
In reply to BOUGH CHENOn i.MX8MM, we are running Dual Linux OS, with 1st Linux using SD Card as rootfs storage, 2nd Linux using eMMC as rootfs storage. We let the the 1st linux configure power/clock for the 2nd Linux.
When the 2nd Linux is booting into rootfs stage, we let the 1st Linux to destroy the 2nd linux, then restart the 2nd linux, we met SDHCI dump as following, after we clear the pending interrupt and halt CQCTL, issue gone.
[ 1.334594] mmc2: Got command interrupt 0x00000001 even though no command operation was in progress. [ 1.334595] mmc2: sdhci: ============ SDHCI REGISTER DUMP =========== [ 1.334599] mmc2: sdhci: Sys addr: 0xa05dcc00 | Version: 0x00000002 [ 1.345538] mmc2: sdhci: Blk size: 0x00000200 | Blk cnt: 0x00000000 [ 1.345541] mmc2: sdhci: Argument: 0x00018000 | Trn mode: 0x00000033 [ 1.345543] mmc2: sdhci: Present: 0x01f88008 | Host ctl: 0x00000031 [ 1.345547] mmc2: sdhci: Power: 0x00000002 | Blk gap: 0x00000080 [ 1.357903] mmc2: sdhci: Wake-up: 0x00000008 | Clock: 0x0000003f [ 1.357905] mmc2: sdhci: Timeout: 0x0000008f | Int stat: 0x00000000 [ 1.357908] mmc2: sdhci: Int enab: 0x107f100b | Sig enab: 0x107f100b [ 1.357911] mmc2: sdhci: AC12 err: 0x00000000 | Slot int: 0x00000502 [ 1.370268] mmc2: sdhci: Caps: 0x07eb0000 | Caps_1: 0x0000b400 [ 1.370270] mmc2: sdhci: Cmd: 0x00000d1a | Max curr: 0x00ffffff [ 1.370273] mmc2: sdhci: Resp[0]: 0x00000b00 | Resp[1]: 0xffffffff [ 1.370276] mmc2: sdhci: Resp[2]: 0x328f5903 | Resp[3]: 0x00d00f00 [ 1.382132] mmc2: sdhci: Host ctl2: 0x00000000 [ 1.382135] mmc2: sdhci: ADMA Err: 0x00000000 | ADMA Ptr: 0xa2040208
[ 2.060932] mmc2: Unexpected interrupt 0x00004000. [ 2.065538] mmc2: sdhci: ============ SDHCI REGISTER DUMP =========== [ 2.071720] mmc2: sdhci: Sys addr: 0x00000000 | Version: 0x00000002 [ 2.077902] mmc2: sdhci: Blk size: 0x00000200 | Blk cnt: 0x00000001 [ 2.084083] mmc2: sdhci: Argument: 0x00000000 | Trn mode: 0x00000000 [ 2.090264] mmc2: sdhci: Present: 0x01f88009 | Host ctl: 0x00000011 [ 2.096446] mmc2: sdhci: Power: 0x00000002 | Blk gap: 0x00000080 [ 2.102627] mmc2: sdhci: Wake-up: 0x00000008 | Clock: 0x000010ff [ 2.108809] mmc2: sdhci: Timeout: 0x0000008f | Int stat: 0x00004000 [ 2.114990] mmc2: sdhci: Int enab: 0x007f1003 | Sig enab: 0x007f1003 [ 2.121171] mmc2: sdhci: AC12 err: 0x00000000 | Slot int: 0x00000502 [ 2.127353] mmc2: sdhci: Caps: 0x07eb0000 | Caps_1: 0x0000b400 [ 2.133534] mmc2: sdhci: Cmd: 0x0000371a | Max curr: 0x00ffffff [ 2.139715] mmc2: sdhci: Resp[0]: 0x00000900 | Resp[1]: 0xffffffff [ 2.145896] mmc2: sdhci: Resp[2]: 0x328f5903 | Resp[3]: 0x00d00f00 [ 2.152077] mmc2: sdhci: Host ctl2: 0x00000000 [ 2.156342] mmc2: sdhci: ADMA Err: 0x00000000 | ADMA Ptr: 0x00000000
--- drivers/mmc/host/sdhci-esdhc-imx.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+)
drivers/mmc/host/sdhci-esdhc-imx.c +16 -0
@@ -1203,6 +1203,7 @@ static void sdhci_esdhc_imx_hwinit(struct sdhci_host *host) {struct sdhci_pltfm_host *pltfm_host =sdhci_priv (host);struct pltfm_imx_data *imx_data =sdhci_pltfm_priv (pltfm_host); +struct cqhci_host *cq_host = host->mmc->cqe_private;int tmp;if (esdhc_is_usdhc (imx_data)) {
@@ -1279,6 +1280,21 @@ static void sdhci_esdhc_imx_hwinit(struct sdhci_host *host) tmp &= ~ESDHC_STD_TUNING_EN ;writel (tmp, host->ioaddr +ESDHC_TUNING_CTRL ); } + +/* +* On i.MX8MM, we are running Dual Linux OS, with 1st Linux using SD Card +* as rootfs storage, 2nd Linux using eMMC as rootfs storage. We let the +* the 1st linux configure power/clock for the 2nd Linux. +* +* When the 2nd Linux is booting into rootfs stage, we let the 1st Linux +* to destroy the 2nd linux, then restart the 2nd linux, we met SDHCI dump. +* After we clear the pending interrupt and halt CQCTL, issue gone. +*/ +if (cq_host) { + tmp =cqhci_readl (cq_host,CQHCI_IS ); +cqhci_writel (cq_host, tmp,CQHCI_IS ); +cqhci_writel (cq_host,CQHCI_HALT ,CQHCI_CTL ); + } } }
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 12/14 mmc: sdhci-esdhc-imx: restore pin state when resume back
In reply to BOUGH CHENIn some low power mode, SoC will lose the pin state, so need to restore the pin state when resume back.
--- drivers/mmc/host/sdhci-esdhc-imx.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)
drivers/mmc/host/sdhci-esdhc-imx.c +11 -1
@@ -1690,7 +1690,13 @@ static int sdhci_esdhc_suspend(struct device *dev)if (host->tuning_mode !=SDHCI_TUNING_MODE_3 )mmc_retune_needed (host->mmc); -return sdhci_suspend_host (host); + ret =sdhci_suspend_host (host); +if (!ret) +if (pinctrl_pm_select_sleep_state (dev)) +dev_warn (mmc_dev (host->mmc), +"%s, failed to select sleep pin state!\n" , __func__); + +return ret; }static int sdhci_esdhc_resume (struct device *dev)
@@ -1698,6 +1704,10 @@ static int sdhci_esdhc_resume(struct device *dev)struct sdhci_host *host =dev_get_drvdata (dev);int ret; +if (pinctrl_pm_select_default_state (dev)) +dev_warn (mmc_dev (host->mmc), +"%s, failed to select default pin state!\n" , __func__); +/* re-initialize hw state in case it's lost in low power mode */ sdhci_esdhc_imx_hwinit (host);
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 13/14 mmc: sdhci-esdhc-imx: clear DMA_SEL when disable DMA mode
In reply to BOUGH CHENCurrently, when use standard tuning, driver default disable DMA just before send tuning command. But on i.MX8 usdhc, this is not enough. Need also clear DMA_SEL. If not, once the DMA_SEL select AMDA2 before, even dma already disabled, when send tuning command, usdhc will still prefetch the ADMA script from wrong DMA address, then we will see IOMMU report some error which show lack of TLB mapping.
--- drivers/mmc/host/sdhci-esdhc-imx.c | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-)
drivers/mmc/host/sdhci-esdhc-imx.c +16 -2
@@ -608,10 +608,24 @@ static void esdhc_writew_le(struct sdhci_host *host, u16 val, int reg)* For DMA access restore the levels to default value. */ m =readl (host->ioaddr +ESDHC_WTMK_LVL ); -if (val &SDHCI_TRNS_DMA ) +if (val &SDHCI_TRNS_DMA ) { wml =ESDHC_WTMK_LVL_WML_VAL_DEF ; -else + }else { +u8 ctrl; wml =ESDHC_WTMK_LVL_WML_VAL_MAX ; + +/* +* Since already disable DMA mode, so also need +* to clear the DMASEL. Otherwise, for standard +* tuning, when send tuning command, usdhc will +* still prefetch the ADMA script from wrong +* DMA address, then we will see IOMMU report +* some error which show lack of TLB mapping. +*/ + ctrl =sdhci_readb (host,SDHCI_HOST_CONTROL ); + ctrl &= ~SDHCI_CTRL_DMA_MASK ; +sdhci_writeb (host, ctrl,SDHCI_HOST_CONTROL ); + } m &= ~(ESDHC_WTMK_LVL_RD_WML_MASK |ESDHC_WTMK_LVL_WR_WML_MASK ); m |= (wml << ESDHC_WTMK_LVL_RD_WML_SHIFT) |
Signature · 1 line
2.17.1
BOUGH CHEN
maintainerPATCH v2 14/14 mmc: queue: create dev->dma_parms before call dma_set_max_seg_size()
In reply to BOUGH CHENTo make dma_set_max_seg_size() work, need to create dev->dma_parms.
Find this issue on i.MX8QM mek board, this platform config the max_segment_size to 65535, but this dma_set_max_seg_size do not actuall work, find sometimes the segment size is 65536, exceed the hardware max segment limitation, trigger ADMA error.
--- drivers/mmc/core/queue.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-)
drivers/mmc/core/queue.c +5 -4
@@ -359,6 +359,7 @@ static const struct blk_mq_ops mmc_mq_ops = {static void mmc_setup_queue (struct mmc_queue *mq,struct mmc_card *card) {struct mmc_host *host = card->host; +struct device *dev =mmc_dev (host);unsigned block_size =512 ;blk_queue_flag_set (QUEUE_FLAG_NONROT , mq->queue);
@@ -366,13 +367,12 @@ static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)if (mmc_can_erase (card))mmc_queue_setup_discard (mq->queue, card); -if (!mmc_dev (host)->dma_mask || !*mmc_dev (host)->dma_mask) +if (!dev->dma_mask || !*dev->dma_mask)blk_queue_bounce_limit (mq->queue,BLK_BOUNCE_HIGH );blk_queue_max_hw_sectors (mq->queue,min (host->max_blk_count, host->max_req_size /512 ));if (host->can_dma_map_merge) -WARN (!blk_queue_can_use_dma_map_merging (mq->queue, -mmc_dev (host)), +WARN (!blk_queue_can_use_dma_map_merging (mq->queue, dev),"merging was advertised but not possible" );blk_queue_max_segments (mq->queue,mmc_get_max_segments (host));
@@ -389,7 +389,8 @@ static void mmc_setup_queue(struct mmc_queue *mq, struct mmc_card *card)blk_queue_max_segment_size (mq->queue,round_down (host->max_seg_size, block_size)); -dma_set_max_seg_size (mmc_dev (host),queue_max_segment_size (mq->queue)); + dev->dma_parms =devm_kzalloc (dev,sizeof (*dev->dma_parms),GFP_KERNEL ); +dma_set_max_seg_size (dev,queue_max_segment_size (mq->queue));INIT_WORK (&mq->recovery_work, mmc_mq_recovery_handler);INIT_WORK (&mq->complete_work, mmc_blk_mq_complete_work);
Signature · 1 line
2.17.1