| Message ID | 20210507145117.43221-1-zhangliguang@linux.alibaba.com |
|---|---|
| State | New |
| Headers | show |
| Series | spi: set right CS polarity depend on gpiolib | expand |
Hi, 在 2021/5/7 23:30, Mark Brown 写道: > On Fri, May 07, 2021 at 10:51:17PM +0800, Liguang Zhang wrote: >> After a kernel upgrade from 4.19 to 5.10, we found that tpm flow control >> always causes TIMEOUT which caused by wrong CS polarity setting depend >> on gpiolib. >> if (spi->cs_gpiod) >> /* polarity handled by gpiolib */ >> - gpiod_set_value_cansleep(spi->cs_gpiod, activate); >> + gpiod_set_value_cansleep(spi->cs_gpiod, !enable); > Whatever is going on here it doesn't seem likely that this is a problem > in the SPI core given the widespread use of gpiod based chip selects - > can you provide more explanation of what you're seeing here, how is the > chip select configured, what is the hardware expectation and what > actually ends up happening? The arm platform is kunpeng-920, detail dsdt information here: Device (SPI0) { Name (_HID, "HISI0173") // _HID: Hardware ID Name (_ADR, Zero) // _ADR: Address Name (_UID, Zero) // _UID: Unique ID Name (RBUF, ResourceTemplate () { GpioIo (Exclusive, PullUp, 0x0000, 0x0000, IoRestrictionNone, "\\_SB.GPO1", 0x00, ResourceConsumer, , ) { // Pin list 0x0006 } QWordMemory (ResourceConsumer, PosDecode, MinFixed, MaxFixed, NonCacheable, ReadWrite, 0x0000000000000000, // Granularity 0x00000002011A0000, // Range Minimum 0x00000002011AFFFF, // Range Maximum 0x0000000000000000, // Translation Offset 0x0000000000010000, // Length ,, , AddressRangeMemory, TypeStatic) Interrupt (ResourceConsumer, Level, ActiveHigh, Exclusive, ,, ) { 0x000001EB, } }) Method (_CRS, 0, NotSerialized) // _CRS: Current Resource Settings { Return (RBUF) /* \_SB_.SPI0.RBUF */ } Name (_DSD, Package (0x02) // _DSD: Device-Specific Data { ToUUID ("daffd814-6eba-4d8c-8a91-bc9bbf4aa301") /* Device Properties for _DSD */, Package (0x03) { Package (0x02) { "reg-io-width", 0x04 }, Package (0x02) { "num-cs", One }, Package (0x02) { "cs-gpios", Package (0x04) { SPI0, Zero, Zero, Zero } } } }) } Scope (SPI0) { Device (TPM) { Name (_ADR, Zero) // _ADR: Address Name (_CID, Package (0x01) // _CID: Compatible ID { "SMO0768" }) Name (_UID, Zero) // _UID: Unique ID Name (_DSD, Package (0x02) // _DSD: Device-Specific Data { ToUUID ("daffd814-6eba-4d8c-8a91-bc9bbf4aa301") /* Device Properties for _DSD */, Package (0x00){} }) Method (_CRS, 0, NotSerialized) // _CRS: Current Resource Settings { Name (RBUF, ResourceTemplate () { SpiSerialBusV2 (0x0000, PolarityLow, FourWireMode, 0x08, ControllerInitiated, 0x000F4240, ClockPolarityLow, ClockPhaseFirst, "\\_SB.SPI0", 0x00, ResourceConsumer, , Exclusive, ) }) Return (RBUF) /* \_SB_.SPI0.TPM_._CRS.RBUF */ } Method (_STA, 0, NotSerialized) // _STA: Status { If ((TPEN == One)) { Return (0x0F) } Else { Return (Zero) } } } } } The problem I meet is: tpm_tis_spi_init tpm_tis_core_init wait_startup // Timeout is occured, appear every time. If the modification below done, probe succeed. gpiod_set_value_cansleep(spi->cs_gpiod, !enable)
diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c index ba425b9c7700..9ee2b92b4506 100644 --- a/drivers/spi/spi.c +++ b/drivers/spi/spi.c @@ -816,7 +816,7 @@ static void spi_set_cs(struct spi_device *spi, bool enable, bool force) if (!(spi->mode & SPI_NO_CS)) { if (spi->cs_gpiod) /* polarity handled by gpiolib */ - gpiod_set_value_cansleep(spi->cs_gpiod, activate); + gpiod_set_value_cansleep(spi->cs_gpiod, !enable); else /* * invert the enable line, as active low is
After a kernel upgrade from 4.19 to 5.10, we found that tpm flow control always causes TIMEOUT which caused by wrong CS polarity setting depend on gpiolib. Signed-off-by: Liguang Zhang <zhangliguang@linux.alibaba.com> --- drivers/spi/spi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)