> -----Original Message----- > From: David Woodhouse <dwmw2@xxxxxxxxxxxxx> > Sent: giovedì 7 marzo 2019 18:40 > To: Flavio Suligoi <f.suligoi@xxxxxxx>; Brian Norris > <computersforpeace@xxxxxxxxx>; Boris Brezillon <bbrezillon@xxxxxxxxxx>; > Marek Vasut <marek.vasut@xxxxxxxxx>; Richard Weinberger <richard@xxxxxx>; > Mark Brown <broonie@xxxxxxxxxx>; Frieder Schrempf > <frieder.schrempf@xxxxxxxxx>; Yogesh Gaur <yogeshnarayan.gaur@xxxxxxx>; > linux-mtd@xxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx > Subject: Re: [PATCH v1 1/1] mtd: devices: add ACPI support for non-jedec > m25p80 > > On Tue, 2019-02-26 at 11:48 +0100, Flavio Suligoi wrote: > > For the x86 machines a m25p80-compatible device have to be declared > using > > an ACPI table (which can be directly a part of the BIOS ACPI tables). > > > > In this case it is necessary to add the device in the "of_device_id" > structure > > list, to permit the device name matching by the ACPI kernel functions. > > > > This is an example of a SSDT table for the Everspin mr25h40: > > > > DefinitionBlock ("mr25h40.aml", "SSDT", 5, "ASEMsp", "MR25H40", 1) > > { > > External (_SB.SPI1, DeviceObj) > > > > Scope (\_SB.SPI1) > > { > > Device (NVR0) > > { > > Name (_HID, "PRP0001") > > Name (_DDN, "Everspin MR25H40 MRAM") > > Name (_CRS, ResourceTemplate () { > > SpiSerialBus ( > > 1, // Chip select > > PolarityLow, // Chip select is active > low > > FourWireMode, // Full duplex > > 8, // Bits per word is 8 > (byte) > > ControllerInitiated, // Don't care > > 10000000, // 10 MHz > > ClockPolarityLow, // SPI mode 0 > ClockPolarityLow > > ClockPhaseFirst, // SPI mode 0 > ClockPhaseFirst > > "\\_SB.SPI1", // SPI host controller > > 0, // Must be 0 > > ResourceConsumer, > > , > > ) > > }) > > > > Name (_DSD, Package () { > > ToUUID("daffd814-6eba-4d8c-8a91-bc9bbf4aa301"), > > Package () { > > Package () {"compatible", "mr25h40"}, > > } > > }) > > } > > } > > } > > > > Signed-off-by: Flavio Suligoi <f.suligoi@xxxxxxx> > > Why use the specific chip name in the "compatible" property? Why isn't > it using "jedec,spi-nor"? > > Does this still actually get passed through to spi_nor_scan()... and is > it *necessary*? Can't the chips be probed? > Hi David, exactly, the mr25h40 is not Jedec compatible, so it can't be probed, so it is necessary to identify it explicitly. Flavio ______________________________________________________ Linux MTD discussion mailing list http://lists.infradead.org/mailman/listinfo/linux-mtd/