Dear coreboot folks,
following Patrick to test devices with the Firmware Test Suite [1], I built fwts V15.08.00-7-g9ddce1f from the their Git repository [2] and ran it on the ASRock E350M1 with coreboot [3] and SeaBIOS.
The first tests failing are the MTRR tests.
``` mtrr: MTRR tests. -------------------------------------------------------------------------------- MTRR overview ------------- Reg 0: 0x0000000000000000 - 0x0000000080000000 ( 2048 MB) Write-Back Reg 1: 0x0000000080000000 - 0x00000000c0000000 ( 1024 MB) Write-Back Reg 2: 0x00000000c0000000 - 0x00000000c8000000 ( 128 MB) Write-Back Reg 6: 0x00000000ffc00000 - 0x0000000100000000 ( 4 MB) Write-Protect
Test 1 of 3: Validate the kernel MTRR IOMEM setup. FAILED [MEDIUM] MTRRIncorrectAttr: Test 1, Memory range 0x100000000 to 0x21effffff (System RAM) has incorrect attribute Default (Most probably Uncached). FAILED [MEDIUM] MTRRLackingAttr: Test 1, Memory range 0x100000000 to 0x21effffff (System RAM) is lacking attribute Write-Back.
Test 2 of 3: Validate the MTRR setup across all processors. PASSED: Test 2, All processors have the a consistent MTRR setup.
Test 3 of 3: Test for AMD MtrrFixDramModEn being cleared by the BIOS. PASSED: Test 3, No MtrrFixDramModEn error detected.
================================================================================ 2 passed, 2 failed, 0 warning, 0 aborted, 0 skipped, 0 info only. ================================================================================ ```
I’ll try to look into it in the coming weeks, but if somebody has an idea and a suggestion for a fix, that would be great. Please also comment, if fwts is not correct in this regard.)
Thanks,
Paul
[1] https://wiki.ubuntu.com/Kernel/Reference/fwts [2] git://kernel.ubuntu.com/hwe/fwts [3] commit 6de27da3 (samsung/exynos5250: Enable bootblock console)
On Sat, Aug 8, 2015 at 11:54 PM, Paul Menzel paulepanter@users.sourceforge.net wrote:
Dear coreboot folks,
following Patrick to test devices with the Firmware Test Suite [1], I built fwts V15.08.00-7-g9ddce1f from the their Git repository [2] and ran it on the ASRock E350M1 with coreboot [3] and SeaBIOS.
The first tests failing are the MTRR tests.
mtrr: MTRR tests. -------------------------------------------------------------------------------- MTRR overview ------------- Reg 0: 0x0000000000000000 - 0x0000000080000000 ( 2048 MB) Write-Back Reg 1: 0x0000000080000000 - 0x00000000c0000000 ( 1024 MB) Write-Back Reg 2: 0x00000000c0000000 - 0x00000000c8000000 ( 128 MB) Write-Back Reg 6: 0x00000000ffc00000 - 0x0000000100000000 ( 4 MB) Write-Protect Test 1 of 3: Validate the kernel MTRR IOMEM setup. FAILED [MEDIUM] MTRRIncorrectAttr: Test 1, Memory range 0x100000000 to 0x21effffff (System RAM) has incorrect attribute Default (Most probably Uncached). FAILED [MEDIUM] MTRRLackingAttr: Test 1, Memory range 0x100000000 to 0x21effffff (System RAM) is lacking attribute Write-Back.
The E350M1 is an AMD system. On those systems there is a TOM2 msr to cover memory above 4GiB as writeback. I don't know what family the cores are in that system, however here is the BKDG link for family 10h: http://support.amd.com/TechDocs/31116.pdf
MSRC001_0010 System Configuration Register (SYS_CFG) bits: 22 Tom2ForceMemTypeWB: top of memory 2 memory type write back. Read-write. 1=The default memory type of memory between 4GB and TOM2 is write back instead of the memory type defined by [The MTRR Default Memory Type Register (MTRRdefType)] MSR0000_02FF[MemType]. For this bit to have any effect, MSR0000_02FF[MtrrDefTypeEn] must be 1. MTRRs and PAT can be used to override this memory type.
21 MtrrTom2En: MTRR top of memory 2 enable. Read-write. 0=[The Top Of Memory 2 Register (TOM2)] MSRC001_001D is disabled. 1=This register is enabled.
And the TOM2 msr: MSRC001_001D Top Of Memory 2 Register (TOM2) 47:23 TOM2[47:23]: second top of memory. Read-write. Specifies the address divides between MMIO and DRAM. This value is normally placed above 4G. From 4G to TOM2 - 1 is DRAM; TOM2 and above is MMIO. See 2.9.3 [Access Type Determination]. This register is enabled by [The System Configuration Register (SYS_CFG)] MSRC001_0010[MtrrTom2En].
If those settings are employed then I'd expect the MTRRs not to cover memory above 4GiB.
Test 2 of 3: Validate the MTRR setup across all processors. PASSED: Test 2, All processors have the a consistent MTRR setup.
Test 3 of 3: Test for AMD MtrrFixDramModEn being cleared by the BIOS. PASSED: Test 3, No MtrrFixDramModEn error detected.
================================================================================ 2 passed, 2 failed, 0 warning, 0 aborted, 0 skipped, 0 info only. ================================================================================
I’ll try to look into it in the coming weeks, but if somebody has an idea and a suggestion for a fix, that would be great. Please also comment, if fwts is not correct in this regard.) Thanks, Paul [1] https://wiki.ubuntu.com/Kernel/Reference/fwts [2] git://kernel.ubuntu.com/hwe/fwts [3] commit 6de27da3 (samsung/exynos5250: Enable bootblock console) -- coreboot mailing list: coreboot@coreboot.org http://www.coreboot.org/mailman/listinfo/coreboot
-
Aaron Durbin -
Paul Menzel