From a6130fc8f9c00e0c545389b4e407f0b745435a56 Mon Sep 17 00:00:00 2001 From: Kyösti Mälkki Date: Mon, 12 May 2014 15:00:03 +0300 Subject: intel: Drop obsolete comments on MTRR usage MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Problem with UMA region allocation was fixed when MTRRs changed to use memrange implementation. Change-Id: I420dac30de2836a91596d81f88bb45b46f248532 Signed-off-by: Kyösti Mälkki Reviewed-on: http://review.coreboot.org/5719 Tested-by: build bot (Jenkins) Reviewed-by: Aaron Durbin Reviewed-by: Edward O'Callaghan --- src/include/device/device.h | 4 ---- src/northbridge/intel/fsp_sandybridge/northbridge.c | 11 ----------- src/northbridge/intel/nehalem/northbridge.c | 11 ----------- src/northbridge/intel/sandybridge/northbridge.c | 11 ----------- 4 files changed, 37 deletions(-) (limited to 'src') diff --git a/src/include/device/device.h b/src/include/device/device.h index dfebeaa6f8..0ae4f888b3 100644 --- a/src/include/device/device.h +++ b/src/include/device/device.h @@ -211,10 +211,6 @@ void fixed_mem_resource(device_t dev, unsigned long index, /* It is the caller's responsibility to adjust regions such that ram_resource() * and mmio_resource() do not overlap. - * - * Current MTRR setup creates exclusive uncacheable holes for uma_resource() - * only and these are allowed to overlap any ram_resource(). This approach - * is used for all UMA except Intel Sandy/IvyBridge. */ #define ram_resource(dev, idx, basek, sizek) \ fixed_mem_resource(dev, idx, basek, sizek, IORESOURCE_CACHEABLE) diff --git a/src/northbridge/intel/fsp_sandybridge/northbridge.c b/src/northbridge/intel/fsp_sandybridge/northbridge.c index d5b1575d23..2922ca0e6a 100644 --- a/src/northbridge/intel/fsp_sandybridge/northbridge.c +++ b/src/northbridge/intel/fsp_sandybridge/northbridge.c @@ -102,17 +102,6 @@ static void add_fixed_resources(struct device *dev, int index) struct resource *resource; u32 pcie_config_base, pcie_config_size; - /* Using uma_resource() here would fail as base & size cannot - * be used as-is for a single MTRR. This would cause excessive - * use of MTRRs. - * - * Use of mmio_resource() instead does not create UC holes by using - * MTRRs, but making these regions uncacheable is taken care of by - * making sure they do not overlap with any ram_resource(). - * - * The resources can be changed to use separate mmio_resource() - * calls after MTRR code is able to merge them wisely. - */ mmio_resource(dev, index++, uma_memory_base >> 10, uma_memory_size >> 10); if (get_pcie_bar(&pcie_config_base, &pcie_config_size)) { diff --git a/src/northbridge/intel/nehalem/northbridge.c b/src/northbridge/intel/nehalem/northbridge.c index d94bc09ba8..639d24568b 100644 --- a/src/northbridge/intel/nehalem/northbridge.c +++ b/src/northbridge/intel/nehalem/northbridge.c @@ -134,17 +134,6 @@ static void mc_read_resources(device_t dev) ram_resource(dev, 3, 0, 640); ram_resource(dev, 4, 768, ((tseg_base >> 10) - 768)); - /* Using uma_resource() here would fail as base & size cannot - * be used as-is for a single MTRR. This would cause excessive - * use of MTRRs. - * - * Use of mmio_resource() instead does not create UC holes by using - * MTRRs, but making these regions uncacheable is taken care of by - * making sure they do not overlap with any ram_resource(). - * - * The resources can be changed to use separate mmio_resource() - * calls after MTRR code is able to merge them wisely. - */ mmio_resource(dev, 5, tseg_base >> 10, CONFIG_SMM_TSEG_SIZE >> 10); reg16 = pci_read_config16(dev_find_slot(0, PCI_DEVFN(0, 0)), D0F0_GGC); diff --git a/src/northbridge/intel/sandybridge/northbridge.c b/src/northbridge/intel/sandybridge/northbridge.c index 5440140089..1fd252bcc0 100644 --- a/src/northbridge/intel/sandybridge/northbridge.c +++ b/src/northbridge/intel/sandybridge/northbridge.c @@ -100,17 +100,6 @@ static void add_fixed_resources(struct device *dev, int index) struct resource *resource; u32 pcie_config_base, pcie_config_size; - /* Using uma_resource() here would fail as base & size cannot - * be used as-is for a single MTRR. This would cause excessive - * use of MTRRs. - * - * Use of mmio_resource() instead does not create UC holes by using - * MTRRs, but making these regions uncacheable is taken care of by - * making sure they do not overlap with any ram_resource(). - * - * The resources can be changed to use separate mmio_resource() - * calls after MTRR code is able to merge them wisely. - */ mmio_resource(dev, index++, uma_memory_base >> 10, uma_memory_size >> 10); if (get_pcie_bar(&pcie_config_base, &pcie_config_size)) { -- cgit v1.2.3