Add SDHCI driver for Broadcom BRCMSTB/BMIPS Soc's
Subsystems (5)
Al Cooper
maintainerv2 cover Add SDHCI driver for Broadcom BRCMSTB/BMIPS Soc's
This patchset adds the SDHCI driver for the Broadcom BRCMSTB/BMIPS Soc's. A few new Device Tree properties, needed by this driver, were added to sdhci-pltfm.c to set various QUIRKS. The mmc.txt bindings document was update with the both new and existing properties handled by sdhci-pltfm.c
V2 - Don't create a new QUIRK to handle broken SDR50 mode in
the controller. The functionality was moved into the sdhci-brcmstb driver.
Al Cooper (3):
mmc: sdhci-pltfm: Add DT properties to set various QUIRKS
mmc: Add Device Tree binding supported by sdhci-pltfm.c
mmc: sdhci-brcmstb: Add sdhci driver for Broadcom BRCMSTB/BMIPS SOCs
Documentation/devicetree/bindings/mmc/mmc.txt | 8 ++ .../devicetree/bindings/mmc/sdhci-brcmstb.txt | 36 +++++ drivers/mmc/host/Kconfig | 12 ++ drivers/mmc/host/Makefile | 1 + drivers/mmc/host/sdhci-brcmstb.c | 158 +++++++++++++++++++++ drivers/mmc/host/sdhci-pltfm.c | 8 +- 6 files changed, 222 insertions(+), 1 deletion(-) create mode 100644 Documentation/devicetree/bindings/mmc/sdhci-brcmstb.txt create mode 100644 drivers/mmc/host/sdhci-brcmstb.c
Signature · 1 line
1.9.0.138.g2de3478
Al Cooper
maintainerPATCH v2 1/3 mmc: sdhci-pltfm: Add DT properties to set various QUIRKS
Add support for "broken-ddr50", "broken-64-bit-dma"
and "broken-timeout-value" device tree properties.
The properties will cause the corresponding quirks bits to be set. This allows some of the platform specific QUIRKS setting to be moved out of the driver and into the Device Tree node.
--- drivers/mmc/host/sdhci-pltfm.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/mmc/host/sdhci-pltfm.c +7 -1
@@ -90,10 +90,14 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_get_property (np,"no-1-8-v" ,NULL )) host->quirks2 |=SDHCI_QUIRK2_NO_1_8_V ; +if (of_get_property (np,"broken-64-bit-dma" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_64_BIT_DMA ; +if (of_device_is_compatible (np,"fsl,p2020-rev1-esdhc" )) host->quirks |=SDHCI_QUIRK_BROKEN_DMA ; -if (of_device_is_compatible (np,"fsl,p2020-esdhc" ) || +if (of_get_property (np,"broken-timeout-value" ,NULL ) || +of_device_is_compatible (np,"fsl,p2020-esdhc" ) ||of_device_is_compatible (np,"fsl,p1010-esdhc" ) ||of_device_is_compatible (np,"fsl,t4240-esdhc" ) ||of_device_is_compatible (np,"fsl,mpc8536-esdhc" ))
@@ -106,6 +110,8 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_find_property (np,"enable-sdio-wakeup" ,NULL )) host->mmc->pm_caps |=MMC_PM_WAKE_SDIO_IRQ ; +if (of_find_property (np,"broken-ddr50" ,NULL )) + host->quirks2 |= SDHCI_QUIRK2_BROKEN_DDR50; }#else void sdhci_get_of_property (struct platform_device *pdev) {}
Signature · 1 line
1.9.0.138.g2de3478
Re: [PATCH V2 1/3] mmc: sdhci-pltfm: Add DT properties to set various QUIRKS
Quoted 16 lines · Al Cooper
Add support for "broken-ddr50", "broken-64-bit-dma"
and "broken-timeout-value" device tree properties.
The properties will cause the corresponding quirks bits to be set. This allows some of the platform specific QUIRKS setting to be moved out of the driver and into the Device Tree node.Signed-off-by Al Cooper <[email protected]>--- drivers/mmc/host/sdhci-pltfm.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)drivers/mmc/host/sdhci-pltfm.c +0
-0
drivers/mmc/host/sdhci-pltfm.c void sdhci_get_of_property(struct platform_device *pdev) · line 95
Quoted 6 lines
drivers/mmc/host/sdhci-pltfm.c +3
-0@@ -90,10 +90,14 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_get_property (np,"no-1-8-v" ,NULL )) host->quirks2 |=SDHCI_QUIRK2_NO_1_8_V ; +if (of_get_property (np,"broken-64-bit-dma" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_64_BIT_DMA ; +
To me this is yet another step in the wrong direction of sdhci.
Not only have we added a quirk which we likely could avoided, but now you also suggest to add a DT binding for it.
Please try to avoid this, we shouldn't need to add one DT binding per quirk, right?
Quoted 10 lines
drivers/mmc/host/sdhci-pltfm.c +2
-1if (of_device_is_compatible (np,"fsl,p2020-rev1-esdhc" )) host->quirks |=SDHCI_QUIRK_BROKEN_DMA ; -if (of_device_is_compatible (np,"fsl,p2020-esdhc" ) || +if (of_get_property (np,"broken-timeout-value" ,NULL ) || +of_device_is_compatible (np,"fsl,p2020-esdhc" ) ||of_device_is_compatible (np,"fsl,p1010-esdhc" ) ||of_device_is_compatible (np,"fsl,t4240-esdhc" ) ||of_device_is_compatible (np,"fsl,mpc8536-esdhc" ))
drivers/mmc/host/sdhci-pltfm.c void sdhci_get_of_property(struct platform_device *pdev) · line 114
Quoted 5 lines
drivers/mmc/host/sdhci-pltfm.c +2
-0@@ -106,6 +110,8 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_find_property (np,"enable-sdio-wakeup" ,NULL )) host->mmc->pm_caps |=MMC_PM_WAKE_SDIO_IRQ ; +if (of_find_property (np,"broken-ddr50" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_DDR50 ;
Same comment as above.
Let me also refer to the response from Scott Branden to the earlier version of this patchset.
http://permalink.gmane.org/gmane.linux.kernel.mmc/34614
Perhaps we should invent a sdhci library function, which each sdhci variant can use to set capabilities through. Typically useful for those variants that have a non-reliable capabilities register.
Quoted 6 lines
drivers/mmc/host/sdhci-pltfm.c +0
-1}#else void sdhci_get_of_property (struct platform_device *pdev) {} --1.9.0.138.g2de3478
Kind regards
Uffe
Alan Cooper
maintainerRe: [PATCH V2 1/3] mmc: sdhci-pltfm: Add DT properties to set various QUIRKS
Quoted 26 lines · Ulf Hansson
Quoted 16 lines · Al Cooper
Add support for "broken-ddr50", "broken-64-bit-dma"
and "broken-timeout-value" device tree properties.
The properties will cause the corresponding quirks bits to be set. This allows some of the platform specific QUIRKS setting to be moved out of the driver and into the Device Tree node.Signed-off-by Al Cooper <[email protected]>--- drivers/mmc/host/sdhci-pltfm.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)drivers/mmc/host/sdhci-pltfm.c +0
-0drivers/mmc/host/sdhci-pltfm.c void sdhci_get_of_property(struct platform_device *pdev) · line 95
Quoted 6 lines
drivers/mmc/host/sdhci-pltfm.c +3
-0@@ -90,10 +90,14 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_get_property (np,"no-1-8-v" ,NULL )) host->quirks2 |=SDHCI_QUIRK2_NO_1_8_V ; +if (of_get_property (np,"broken-64-bit-dma" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_64_BIT_DMA ; +To me this is yet another step in the wrong direction of sdhci.
Quoted 5 lines
Not only have we added a quirk which we likely could avoided, but now you also suggest to add a DT binding for it.
Please try to avoid this, we shouldn't need to add one DT binding per quirk, right?
The BRCMSTB SDHCI driver needs to support a mix of existing and future ARM and MIPS SoC's on a variety of boards, and many of the combinations have issues. In older separate MIPS/ARM internal versions of the driver, issues were handled by a combination of QUIRKS and the use of custom SDHCI registers that allowed the Host CAPS registers to be changed on a per chip basis. While working on a unified driver for 4.1, I realized that almost all the issues solved by overriding the CAPs registers could also be done using an existing QUIRK and if I had device tree properties that set the QUIRKs, that all the chip/board specific knowledge would end up in the device tree node instead of in the driver. This also allowed me to stop using the non-standard CAPS override registers that tended to change per chip. This approach also allows the driver to work, without change, on future chips/boards as long as they don't have any new issues and they have the proper device tree node.
Would there be any objection to me continuing this approach if I moved all the functionality into the sdhci-brcm.c driver and left sdhci.c and sdhci-pltfm.c untouched?
Quoted 19 lines
Quoted 11 lines
drivers/mmc/host/sdhci-pltfm.c +2
-1if (of_device_is_compatible (np,"fsl,p2020-rev1-esdhc" )) host->quirks |=SDHCI_QUIRK_BROKEN_DMA ; -if (of_device_is_compatible (np,"fsl,p2020-esdhc" ) || +if (of_get_property (np,"broken-timeout-value" ,NULL ) || +of_device_is_compatible (np,"fsl,p2020-esdhc" ) ||of_device_is_compatible (np,"fsl,p1010-esdhc" ) ||of_device_is_compatible (np,"fsl,t4240-esdhc" ) ||of_device_is_compatible (np,"fsl,mpc8536-esdhc" ))drivers/mmc/host/sdhci-pltfm.c void sdhci_get_of_property(struct platform_device *pdev) · line 114
Quoted 5 lines
drivers/mmc/host/sdhci-pltfm.c +2
-0@@ -106,6 +110,8 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_find_property (np,"enable-sdio-wakeup" ,NULL )) host->mmc->pm_caps |=MMC_PM_WAKE_SDIO_IRQ ; +if (of_find_property (np,"broken-ddr50" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_DDR50 ;Same comment as above.
Quoted 8 lines
Let me also refer to the response from Scott Branden to the earlier version of this patchset.
http://permalink.gmane.org/gmane.linux.kernel.mmc/34614Perhaps we should invent a sdhci library function, which each sdhci variant can use to set capabilities through. Typically useful for those variants that have a non-reliable capabilities register.
Do you mean something more than using SDHCI_QUIRK_MISSING_CAPS which allows the driver to supply the 2 CAPS registers instead of having sdhci.c get them by reading the Host CAPS registers?
Quoted 9 lines
Quoted 6 lines
drivers/mmc/host/sdhci-pltfm.c +0
-1}#else void sdhci_get_of_property (struct platform_device *pdev) {} --1.9.0.138.g2de3478
Kind regards
Uffe
Thanks
Al
Re: [PATCH V2 1/3] mmc: sdhci-pltfm: Add DT properties to set various QUIRKS
Quoted 49 lines · Alan Cooper
Quoted 26 lines · Ulf Hansson
Quoted 16 lines · Al Cooper
Add support for "broken-ddr50", "broken-64-bit-dma"
and "broken-timeout-value" device tree properties.
The properties will cause the corresponding quirks bits to be set. This allows some of the platform specific QUIRKS setting to be moved out of the driver and into the Device Tree node.Signed-off-by Al Cooper <[email protected]>--- drivers/mmc/host/sdhci-pltfm.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)drivers/mmc/host/sdhci-pltfm.c +0
-0drivers/mmc/host/sdhci-pltfm.c void sdhci_get_of_property(struct platform_device *pdev) · line 95
Quoted 6 lines
drivers/mmc/host/sdhci-pltfm.c +3
-0@@ -90,10 +90,14 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_get_property (np,"no-1-8-v" ,NULL )) host->quirks2 |=SDHCI_QUIRK2_NO_1_8_V ; +if (of_get_property (np,"broken-64-bit-dma" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_64_BIT_DMA ; +To me this is yet another step in the wrong direction of sdhci.
Quoted 5 lines
Not only have we added a quirk which we likely could avoided, but now you also suggest to add a DT binding for it.
Please try to avoid this, we shouldn't need to add one DT binding per quirk, right?
The BRCMSTB SDHCI driver needs to support a mix of existing and future ARM and MIPS SoC's on a variety of boards, and many of the combinations have issues. In older separate MIPS/ARM internal versions of the driver, issues were handled by a combination of QUIRKS and the use of custom SDHCI registers that allowed the Host CAPS registers to be changed on a per chip basis. While working on a unified driver for 4.1, I realized that almost all the issues solved by overriding the CAPs registers could also be done using an existing QUIRK and if I had device tree properties that set the QUIRKs, that all the chip/board specific knowledge would end up in the device tree node instead of in the driver. This also allowed me to stop using the non-standard CAPS override registers that tended to change per chip. This approach also allows the driver to work, without change, on future chips/boards as long as they don't have any new issues and they have the proper device tree node.
Quoted 3 lines
Would there be any objection to me continuing this approach if I moved all the functionality into the
sdhci-brcm.cdriver and leftsdhci.candsdhci-pltfm.cuntouched?
Let me elaborate a bit on what I think you/we need to do.
The problem with the quirks is that these exists because of how the sdhci code is designed.
Sdhci core should have been implemented as a pure library providing a set of common functions. Today it's more of a driver than a library. All, or at least most of the variant specific code, should have been dealt from each of the sdhci variant drivers, instead of via quirks and sdhci callbacks.
If you didn't notice my recent statement regarding the above on linux-mmc, let's me say it again. I have a reached a point where I think sdhci has become unmaintainable (and non-optimized coding wise) and I am therefore encouraging people to stop adding new quirks or sdhci callbacks, but instead "libraryfy" sdhci. I would really appreciate if you could help moving sdhci in that direction!
That said, you won't be able to leave the sdhci core code untouched, because that would in principle mean that you will need to duplicate much of the sdhci code into your sdhci-brcm driver. Instead, try to split-up/re-factor common sdhci code into library functions, which your sdhci-brcm driver can use.
I guess one of the hardest part in this effort will be to not break existing sdhci variant drivers. Let's do our best in that effort, but we will rely on more people to participate in testing/coding.
Quoted 24 lines
Quoted 24 lines
Quoted 12 lines
drivers/mmc/host/sdhci-pltfm.c +2
-1if (of_device_is_compatible (np,"fsl,p2020-rev1-esdhc" )) host->quirks |=SDHCI_QUIRK_BROKEN_DMA ; -if (of_device_is_compatible (np,"fsl,p2020-esdhc" ) || +if (of_get_property (np,"broken-timeout-value" ,NULL ) || +of_device_is_compatible (np,"fsl,p2020-esdhc" ) ||of_device_is_compatible (np,"fsl,p1010-esdhc" ) ||of_device_is_compatible (np,"fsl,t4240-esdhc" ) ||of_device_is_compatible (np,"fsl,mpc8536-esdhc" ))drivers/mmc/host/sdhci-pltfm.c void sdhci_get_of_property(struct platform_device *pdev) · line 114
Quoted 5 lines
drivers/mmc/host/sdhci-pltfm.c +2
-0@@ -106,6 +110,8 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_find_property (np,"enable-sdio-wakeup" ,NULL )) host->mmc->pm_caps |=MMC_PM_WAKE_SDIO_IRQ ; +if (of_find_property (np,"broken-ddr50" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_DDR50 ;Same comment as above.
Let me also refer to the response from Scott Branden to the earlier version of this patchset.
http://permalink.gmane.org/gmane.linux.kernel.mmc/34614
Quoted 8 lines
Quoted 4 lines
Perhaps we should invent a sdhci library function, which each sdhci variant can use to set capabilities through. Typically useful for those variants that have a non-reliable capabilities register.
Do you mean something more than using
SDHCI_QUIRK_MISSING_CAPSwhich allows the driver to supply the 2 CAPS registers instead of havingsdhci.cget them by reading the Host CAPS registers?
Yes, something like that, but preferably without needing to use *any* quirk.
Quoted 1 line
From an sdhci core perspective there could be a library function which
parses the Host CAPS register. Some sdhci variants could use that function and some couldn't since the informations in the CAPS register isn't reliable.
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
drivers/mmc/host/sdhci-pltfm.c +0
-1}#else void sdhci_get_of_property (struct platform_device *pdev) {} --1.9.0.138.g2de3478
Quoted 5 lines
Quoted 2 lines
Kind regards
UffeThanks
Al
Kind regards
Uffe
Alan Cooper
maintainerRe: [PATCH V2 1/3] mmc: sdhci-pltfm: Add DT properties to set various QUIRKS
Quoted 115 lines · Ulf Hansson
Quoted 49 lines · Alan Cooper
Quoted 26 lines · Ulf Hansson
Quoted 16 lines · Al Cooper
Add support for "broken-ddr50", "broken-64-bit-dma"
and "broken-timeout-value" device tree properties.
The properties will cause the corresponding quirks bits to be set. This allows some of the platform specific QUIRKS setting to be moved out of the driver and into the Device Tree node.Signed-off-by Al Cooper <[email protected]>--- drivers/mmc/host/sdhci-pltfm.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)drivers/mmc/host/sdhci-pltfm.c +0
-0drivers/mmc/host/sdhci-pltfm.c void sdhci_get_of_property(struct platform_device *pdev) · line 95
Quoted 6 lines
drivers/mmc/host/sdhci-pltfm.c +3
-0@@ -90,10 +90,14 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_get_property (np,"no-1-8-v" ,NULL )) host->quirks2 |=SDHCI_QUIRK2_NO_1_8_V ; +if (of_get_property (np,"broken-64-bit-dma" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_64_BIT_DMA ; +To me this is yet another step in the wrong direction of sdhci.
Quoted 5 lines
Not only have we added a quirk which we likely could avoided, but now you also suggest to add a DT binding for it.
Please try to avoid this, we shouldn't need to add one DT binding per quirk, right?
The BRCMSTB SDHCI driver needs to support a mix of existing and future ARM and MIPS SoC's on a variety of boards, and many of the combinations have issues. In older separate MIPS/ARM internal versions of the driver, issues were handled by a combination of QUIRKS and the use of custom SDHCI registers that allowed the Host CAPS registers to be changed on a per chip basis. While working on a unified driver for 4.1, I realized that almost all the issues solved by overriding the CAPs registers could also be done using an existing QUIRK and if I had device tree properties that set the QUIRKs, that all the chip/board specific knowledge would end up in the device tree node instead of in the driver. This also allowed me to stop using the non-standard CAPS override registers that tended to change per chip. This approach also allows the driver to work, without change, on future chips/boards as long as they don't have any new issues and they have the proper device tree node.
Quoted 3 lines
Would there be any objection to me continuing this approach if I moved all the functionality into the
sdhci-brcm.cdriver and leftsdhci.candsdhci-pltfm.cuntouched?Let me elaborate a bit on what I think you/we need to do.
The problem with the quirks is that these exists because of how the sdhci code is designed.
Sdhci core should have been implemented as a pure library providing a set of common functions. Today it's more of a driver than a library. All, or at least most of the variant specific code, should have been dealt from each of the sdhci variant drivers, instead of via quirks and sdhci callbacks.
If you didn't notice my recent statement regarding the above on linux-mmc, let's me say it again. I have a reached a point where I think sdhci has become unmaintainable (and non-optimized coding wise) and I am therefore encouraging people to stop adding new quirks or sdhci callbacks, but instead "libraryfy" sdhci. I would really appreciate if you could help moving sdhci in that direction!
That said, you won't be able to leave the sdhci core code untouched, because that would in principle mean that you will need to duplicate much of the sdhci code into your sdhci-brcm driver. Instead, try to split-up/re-factor common sdhci code into library functions, which your sdhci-brcm driver can use.
I guess one of the hardest part in this effort will be to not break existing sdhci variant drivers. Let's do our best in that effort, but we will rely on more people to participate in testing/coding.
Quoted 24 lines
Quoted 24 lines
Quoted 12 lines
drivers/mmc/host/sdhci-pltfm.c +2
-1if (of_device_is_compatible (np,"fsl,p2020-rev1-esdhc" )) host->quirks |=SDHCI_QUIRK_BROKEN_DMA ; -if (of_device_is_compatible (np,"fsl,p2020-esdhc" ) || +if (of_get_property (np,"broken-timeout-value" ,NULL ) || +of_device_is_compatible (np,"fsl,p2020-esdhc" ) ||of_device_is_compatible (np,"fsl,p1010-esdhc" ) ||of_device_is_compatible (np,"fsl,t4240-esdhc" ) ||of_device_is_compatible (np,"fsl,mpc8536-esdhc" ))drivers/mmc/host/sdhci-pltfm.c void sdhci_get_of_property(struct platform_device *pdev) · line 114
Quoted 5 lines
drivers/mmc/host/sdhci-pltfm.c +2
-0@@ -106,6 +110,8 @@ void sdhci_get_of_property(struct platform_device *pdev)if (of_find_property (np,"enable-sdio-wakeup" ,NULL )) host->mmc->pm_caps |=MMC_PM_WAKE_SDIO_IRQ ; +if (of_find_property (np,"broken-ddr50" ,NULL )) + host->quirks2 |=SDHCI_QUIRK2_BROKEN_DDR50 ;Same comment as above.
Let me also refer to the response from Scott Branden to the earlier version of this patchset.
http://permalink.gmane.org/gmane.linux.kernel.mmc/34614Quoted 8 lines
Quoted 4 lines
Perhaps we should invent a sdhci library function, which each sdhci variant can use to set capabilities through. Typically useful for those variants that have a non-reliable capabilities register.
Do you mean something more than using
SDHCI_QUIRK_MISSING_CAPSwhich allows the driver to supply the 2 CAPS registers instead of havingsdhci.cget them by reading the Host CAPS registers?
Quoted 6 lines
Yes, something like that, but preferably without needing to use *any* quirk.
From an sdhci core perspective there could be a library function which parses the Host CAPS register. Some sdhci variants could use that function and some couldn't since the informations in the CAPS register isn't reliable.
As a possible first step, what about the following:
Change sdhci_add_host() to take an additional parameter that, if not NULL, is a pointer to a data structure that contains overrides for CAPS and CAPS1 in the form of a mask/value pair. These pairs would be used by sdhci_add_host() to modify any field after reading the CAPS and CAPS1 registers. With this in place it looks like the following QUIRKS would no longer be needed because they could be handled in the variant drivers using overrides:SDHCI_QUIRK_FORCE_DMASDHCI_QUIRK_BROKEN_DMASDHCI_QUIRK_BROKEN_ADMASDHCI_QUIRK_FORCE_BLK_SZ_2048SDHCI_QUIRK_MISSING_CAPSSDHCI_QUIRK2_NO_1_8_VSDHCI_QUIRK2_BROKEN_DDR50SDHCI_QUIRK2_BROKEN_64_BIT_DMA
I would add device tree properties for the overrides to my variant driver. I could also add the overrides as module params. There are also many drivers that hook the sdhci_readl() routine to modify the CAPS fields that could be changed to use the overrides.
With this change, my variant driver that supports more than 10 different MIPS/ARM SoC's would be very small with all the SoC differences handled using overrides in the device tree and without the use of any QUIRKS.
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines
drivers/mmc/host/sdhci-pltfm.c +0
-1}#else void sdhci_get_of_property (struct platform_device *pdev) {} --1.9.0.138.g2de3478
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
Kind regards
UffeThanks
AlKind regards
Uffe
Thanks
Al
Al Cooper
maintainerPATCH v2 2/3 mmc: Add Device Tree binding supported by sdhci-pltfm.c
In reply to Al CooperThis includes both newly added and previously undocumented properties.
--- Documentation/devicetree/bindings/mmc/mmc.txt | 8 ++++++++ 1 file changed, 8 insertions(+)
Documentation/devicetree/bindings/mmc/mmc.txt +8 -0
@@ -48,6 +48,14 @@ Optional properties: - mmc-hs400-1_2v: eMMC HS400 mode(1.2V I/O) is supported - dsr: Value the card's (optional) Driver Stage Register (DSR) should be programmed with. Valid range: [0 .. 0xffff]. +- sdhci,auto-cmd12: the controller supports Auto-CMD12 to stop multiblock + transfers. +- broken-64-bit-dma: the controller does not support 64-bit DMA even if + the controller claims it does. +- broken-timeout-value: the timeout value specified by the controller is + incorrect and the MAX timeout will be used instead. +- broken-sdr50: SDR50 mode is broken. +- broken-ddr50: DDR50 mode is broken. *NOTE* on CD and WP polarity. To use common for all SD/MMC host controllers line polarity properties, we have to fix the meaning of the "normal" and "inverted"
Signature · 1 line
1.9.0.138.g2de3478
Al Cooper
maintainerPATCH v2 3/3 mmc: sdhci-brcmstb: Add sdhci driver for Broadcom BRCMSTB/BMIPS SOCs
In reply to Al Cooper--- .../devicetree/bindings/mmc/sdhci-brcmstb.txt | 36 +++++ drivers/mmc/host/Kconfig | 12 ++ drivers/mmc/host/Makefile | 1 + drivers/mmc/host/sdhci-brcmstb.c | 158 +++++++++++++++++++++ 4 files changed, 207 insertions(+) create mode 100644 Documentation/devicetree/bindings/mmc/sdhci-brcmstb.txt create mode 100644 drivers/mmc/host/sdhci-brcmstb.c
Documentation/devicetree/bindings/mmc/sdhci-brcmstb.txt new file +36 -0
@@ -0,0 +1,36 @@ +* BROADCOM BRCMSTB/BMIPS SDHCI Controller + +This file documents differences between the core properties in mmc.txt +and the properties used by the sdhci-brcmstb driver. + +Required properties: +- compatible: "brcm,sdhci-brcmstb" + +Optional properties: +- broken-sdr50: The Host Controller's SDR50 mode is broken. + +Example: + + sdhci@f03e0100 { + compatible = "brcm,sdhci-brcmstb"; + reg = <0xf03e0000 0x100>; + interrupts = <0x0 0x26 0x0>; + interrupt-names = "sdio0_0"; + sdhci,auto-cmd12; + clocks = <0x13>; + clock-names = "sw_sdio"; + broken-sdr50; + }; + + sdhci@f03e0300 { + no-1-8-v; + non-removable; + bus-width = <0x8>; + compatible = "brcm,sdhci-brcmstb"; + reg = <0xf03e0200 0x100>; + interrupts = <0x0 0x27 0x0>; + interrupt-names = "sdio1_0"; + sdhci,auto-cmd12; + clocks = <0x13>; + clock-names = "sw_sdio"; + };
drivers/mmc/host/Kconfig +12 -0
@@ -784,3 +784,15 @@ config MMC_MTK If you have a machine with a integrated SD/MMC card reader, say Y or M here. This is needed if support for any SD/SDIO/MMC devices is required. If unsure, say N. + +config MMC_SDHCI_BRCMSTB + tristate "Broadcom SDIO/SD/MMC support" + depends on ARCH_BRCMSTB || BMIPS_GENERIC + depends on MMC_SDHCI_PLTFM + default y + select MMC_SDHCI_IO_ACCESSORS + help + This selects support for the SDIO/SD/MMC Host Controller on + Broadcom STB SoCs. + + If unsure, say Y.
drivers/mmc/host/Makefile +1 -0
@@ -75,6 +75,7 @@ obj-$(CONFIG_MMC_SDHCI_BCM2835) += sdhci-bcm2835.o obj-$(CONFIG_MMC_SDHCI_IPROC) += sdhci-iproc.o obj-$(CONFIG_MMC_SDHCI_MSM) += sdhci-msm.o obj-$(CONFIG_MMC_SDHCI_ST) += sdhci-st.o +obj-$(CONFIG_MMC_SDHCI_BRCMSTB) += sdhci-brcmstb.o ifeq ($(CONFIG_CB710_DEBUG),y) CFLAGS-cb710-mmc += -DDEBUG
drivers/mmc/host/sdhci-brcmstb.c new file +158 -0
158 lines not on this page · raw message
Signature · 1 line
1.9.0.138.g2de3478
Quoted 5 lines · Al Cooper
This patchset adds the SDHCI driver for the Broadcom BRCMSTB/BMIPS Soc's. A few new Device Tree properties, needed by this driver, were added to
sdhci-pltfm.cto set various QUIRKS. The mmc.txt bindings document was update with the both new and existing properties handled bysdhci-pltfm.c
Looks good to me, thanks!
Quoted 20 lines
V2 - Don't create a new QUIRK to handle broken SDR50 mode in
the controller. The functionality was moved into the sdhci-brcmstb driver.Al Cooper (3):
mmc: sdhci-pltfm: Add DT properties to set various QUIRKS
mmc: Add Device Tree binding supported bysdhci-pltfm.c
mmc: sdhci-brcmstb: Add sdhci driver for Broadcom BRCMSTB/BMIPS SOCsDocumentation/devicetree/bindings/mmc/mmc.txt | 8 ++ .../devicetree/bindings/mmc/sdhci-brcmstb.txt | 36 +++++ drivers/mmc/host/Kconfig | 12 ++ drivers/mmc/host/Makefile | 1 + drivers/mmc/host/sdhci-brcmstb.c | 158 +++++++++++++++++++++ drivers/mmc/host/sdhci-pltfm.c | 8 +- 6 files changed, 222 insertions(+), 1 deletion(-) create mode 100644 Documentation/devicetree/bindings/mmc/sdhci-brcmstb.txt create mode 100644 drivers/mmc/host/sdhci-brcmstb.c
Signature · 1 line
Florian