EDK II sample platform branches and tags (mirror) https://git.radix-linux.su/Tianocore/edk2-platforms.git/atom?h=CodeCleanup 2020-05-06T12:49:03+00:00 2020-05-06T12:49:03+00:00 Pete Batard pete@akeo.ie 2020-05-06T11:37:48+00:00 urn:sha1:11189124fbc24aae0d8c96dd0bd24a78b430d8e0 Because of its widespread availability and low price, we expect the Raspberry Pi source to be used by platform developers as a starting point to create their own platform implementation. As such, it makes a lot of sense to want to use the most up to date underlying standards, even if the pay off is limited in this case, or may even be seen as a liability in terms of ensuring greater compatibility with older OSes, as it may help others benefit from the latest improvements and features brought by modern ACPI. We also happen to already be using ACPI 6.3 constructs, such as PPTT (which was only introduced in ACPI 6.2 and further extended in 6.3), and have some reliance on 6.x GIC, as the original MADT binary blobs from Microsoft were 6.0 and abuse the GICR Base Address field in order for Windows to boot properly (for reasons that the proprietary nature of the Windows kernel makes difficult to determine). So in effect, we did apply a potential breaking change to Windows when we downgraded MADT to ACPI 5.1, though we did validate at the time that the downgrade in ACPI version didn't *seem* to break OS functionality. Still, because we are in the dark as to what the ACPI fields we removed when downgrading ACPI version were being used for, we obviously want to add them back. Therefore, since we do see a need for ACPI 6.x features, and effectively have a 6.3 table with PPTT, and also since we have tested that we are not seeing ill effects from doing so for the most common OSes we support, we bring all of the relevant ACPI tables to version 6.3. This is mostly accomplished by simply altering the version of ACPI being references in the macros, except for the new fields being initialized in the MADT table where we: * Use 1 for GICR Base Address in GICC_STRUCTURE_INIT because, even as this field is not supposed to apply to any of our platforms, this is what Microsoft used in the original Pi 3 hardcoded MADT blobs and Windows 10 doesn't boot on the Pi 3 if set to 0. * Use 2 for GIC version in GIC_DISTRIBUTOR_INIT, since the Pi 4 GIC is v2 only. Signed-off-by: Pete Batard <pete@akeo.ie> Reviewed-by: Ard Biesheuvel <ard.biesheuvel@arm.com> 2020-03-30T14:49:15+00:00 Pete Batard pete@akeo.ie 2020-03-27T10:34:55+00:00 urn:sha1:cdd63d10d5efaeec1cd5fa5f11a8353e4de832d7 There's no real reason to have the PPTT constants factorized in AcpiTables.h, and hiding them away with the interrupts is only going to end up being counter-productive for folks trying to understand or maintain the code. Move these sections to Pptt.aslc where they belong. Signed-off-by: Pete Batard <pete@akeo.ie> Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> 2020-03-26T16:45:22+00:00 Jeremy Linton lintonrjeremy@gmail.com 2020-03-26T16:30:16+00:00 urn:sha1:6beb1692b09b69c02811ee195e7a785d23141b57 ACPI 6.2 adds a new table, which describes how processing units are related to each other in tree like fashion. Caches are also sprinkled throughout the tree and describe the properties of the caches in relation to other caches and processing units. Add a static PPTT table with one L2 cache and an L1I/L1D cache for each of the 4 cores. The cache size/assc/policy/etc are from the public docs. The source from the aslc is derived from the one in: Silicon/AMD/Styx/Drivers/AcpiPlatformDxe/Pptt.aslc Signed-off-by: Pete Batard <pete@akeo.ie> Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> 2020-03-26T12:34:49+00:00 Pete Batard pete@akeo.ie 2020-03-26T11:53:20+00:00 urn:sha1:c03fe68d0c0b80304480d1fa03ba0176787c61ba Only apply the "BC2836" override to the one table that needs it for Pi 3, instead of setting it globally. Use more appropriate OEM and Vendor IDs for hardware that is produced by the Raspberry Pi Foundation. Ensure that the Pi 3 and Pi 4 have different OEM IDs. Additionally fix a comment typo (MEMORY32SETBASE -> MEMORY32FIXED). Signed-off-by: Pete Batard <pete@akeo.ie> Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> 2020-03-26T12:34:36+00:00 Pete Batard pete@akeo.ie 2020-03-26T11:53:17+00:00 urn:sha1:f8b5562d0a963026086a444d083804792a8fe681 Some of the Pi 4 interrupts were copied from Pi 3 and are incorrect. We fix the ones that can be found by decompiling the official .dtbs. We also add a couple more GPIO interrupts, that are used for both the Pi 3 and the Pi 4, as per the dtbs. Signed-off-by: Pete Batard <pete@akeo.ie> Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> 2020-03-02T11:12:51+00:00 Pete Batard pete@akeo.ie 2020-02-28T10:38:55+00:00 urn:sha1:e168252531c7ea73e4114358e022aa6953b6dc5a With the ACPI source for the Pi 3 and Pi 4 being identical, we can finally factorize it. Signed-off-by: Pete Batard <pete@akeo.ie> Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>