summaryrefslogtreecommitdiff
path: root/src/arch
AgeCommit message (Collapse)Author
2015-11-17arm64: tegra132: tegra210: Remove old arm64/stage_entry.SJulius Werner
This patch removes the old arm64/stage_entry.S code that was too specific to the Tegra SoC boot flow, and replaces it with code that hides the peculiarities of switching to a different CPU/arch in ramstage in the Tegra SoC directories. BRANCH=None BUG=None TEST=Built Ryu and Smaug. !!!UNTESTED!!! Change-Id: Ib3a0448b30ac9c7132581464573efd5e86e03698 Signed-off-by: Julius Werner <jwerner@chromium.org> Reviewed-on: http://review.coreboot.org/12078 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Aaron Durbin <adurbin@chromium.org>
2015-11-17rules.h: Add ENV_ macros to detect current architectureJulius Werner
This patch expands the existing ENV_<stage> macros in <rules.h> with a set of ENV_<arch> macros which can be used to detect which architecture the current compilation unit is built for. These are more consistent than compiler-defined macros (like '#ifdef __arm__') and will make it easier to write small, architecture-dependent differences in common code (where we currently often use IS_ENABLED(CONFIG_ARCH_...), which is technically incorrect in a world where every stage can run on a different architecture, and merely kinda happened to work out for now). Also remove a vestigal <arch/rules.h> from ARM64 which was no longer used, and genericise ARM subarchitecture Makefiles a little to make things like __COREBOOT_ARM_ARCH__ available from all file types (including .ld). BUG=None TEST=Compiled Falco, Blaze, Jerry and Smaug. Change-Id: Id51aeb290b5c215c653e42a51919d0838e28621f Signed-off-by: Julius Werner <jwerner@chromium.org> Reviewed-on: http://review.coreboot.org/12433 Tested-by: build bot (Jenkins) Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com> Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com>
2015-11-16arm64: Implement generic stage transitions for non-Tegra SoCsJulius Werner
The existing arm64 architecture code has been developed for the Tegra132 and Tegra210 SoCs, which only start their ARM64 cores in ramstage. It interweaves the stage entry point with code that initializes a CPU (and should not be run again if that CPU already ran a previous stage). It also still contains some vestiges of SMP/secmon support (such as setting up stacks in the BSS instead of using the stage-peristent one from memlayout). This patch splits those functions apart and makes the code layout similar to how things work on ARM32. The default stage_entry() symbol is a no-op wrapper that just calls main() for the current stage, for the normal case where a stage ran on the same core as the last one. It can be overridden by SoC code to support special cases like Tegra. The CPU initialization code is split out into armv8/cpu.S (similar to what arm_init_caches() does for ARM32) and called by the default bootblock entry code. SoCs where a CPU starts up in a later stage can call the same code from a stage_entry() override instead. The Tegra132 and Tegra210 code is not touched by this patch to make it easier to review and validate. A follow-up patch will bring those SoCs in line with the model. BRANCH=None BUG=None TEST=Booted Oak with a single mmu_init()/mmu_enable(). Built Ryu and Smaug. Change-Id: I28302a6ace47e8ab7a736e089f64922cef1a2f93 Signed-off-by: Julius Werner <jwerner@chromium.org> Reviewed-on: http://review.coreboot.org/12077 Tested-by: build bot (Jenkins) Reviewed-by: Aaron Durbin <adurbin@chromium.org>
2015-11-11cpu/amd: Add CC6 supportTimothy Pearson
This patch adds CC6 power save support to the AMD Family 15h support code. As CC6 is a complex power saving state that relies heavily on CPU, northbridge, and southbridge cooperation, this patch alters significant amounts of code throughout the tree simultaneously. Allowing the CPU to enter CC6 allows the second level of turbo boost to be reached, and also provides significant power savings when the system is idle due to the complete core shutdown. Change-Id: I44ce157cda97fb85f3e8f3d7262d4712b5410670 Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com> Reviewed-on: http://review.coreboot.org/11979 Tested-by: build bot (Jenkins) Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org> Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2015-11-11arm/arm64: Generalize bootblock C entry pointJulius Werner
When we first added ARM support to coreboot, it was clear that the bootblock would need to do vastly different tasks than on x86, so we moved its main logic under arch/. Now that we have several more architectures, it turns out (as with so many things lately) that x86 is really the odd one out, and all the others are trying to do pretty much the same thing. This has already caused maintenance issues as the ARM32 bootblock developed and less-mature architectures were left behind with old cruft. This patch tries to address that problem by centralizing that logic under lib/ for use by all architectures/SoCs that don't explicitly opt-out (with the slightly adapted existing BOOTBLOCK_CUSTOM option). This works great out of the box for ARM32 and ARM64. It could probably be easily applied to MIPS and RISCV as well, but I don't have any of those boards to test so I'll mark them as BOOTBLOCK_CUSTOM for now and leave that for later cleanup. BRANCH=None BUG=None TEST=Built Jerry and Falco, booted Oak. Change-Id: Ibbf727ad93651e388aef20e76f03f5567f9860cb Signed-off-by: Julius Werner <jwerner@chromium.org> Reviewed-on: http://review.coreboot.org/12076 Reviewed-by: Aaron Durbin <adurbin@chromium.org> Tested-by: build bot (Jenkins)
2015-11-11arm64: mmu: Make page table manipulation work across stagesJulius Werner
In order to have a proper runtime-modifyable page table API (e.g. to remap DRAM after it was intialized), we need to remove any external bookkeeping kept in global variables (which do not persist across stages) from the MMU code. This patch implements this in a similar way as it has recently been done for ARM32 (marking free table slots with a special sentinel value in the first PTE that cannot occur as part of a normal page table). Since this requires the page table buffer to be known at compile-time, we have to remove the option of passing it to mmu_init() at runtime (which I already kinda deprecated before). The existing Tegra chipsets that still used it are switched to instead define it in memlayout in a minimally invasive change. This might not be the best way to design this overall (I think we should probably just throw the tables into SRAM like on all other platforms), but I don't have a Tegra system to test so I'd rather keep this change low impact and leave the major redesign for later. Also inlined some single-use one-liner functions in mmu.c that I felt confused things more than they cleared up, and fixed an (apparently harmless?) issue with forgetting to mask out the XN page attribute bit when casting a table descriptor to a pointer. BRANCH=None BUG=None TEST=Compiled Ryu and Smaug. Booted Oak. Change-Id: Iad71f97f5ec4b1fc981dbc8ff1dc88d96c8ee55a Signed-off-by: Julius Werner <jwerner@chromium.org> Reviewed-on: http://review.coreboot.org/12075 Tested-by: build bot (Jenkins) Reviewed-by: Aaron Durbin <adurbin@chromium.org>
2015-11-11armv7: mmu: Make fine grained page tables work across stagesJulius Werner
Among its other restrictions (which are noted in a comment above the function prototype and stay in place), our makeshift fine-grained page table support for ARM32 has the undocumented feature that it relies on a global bookkeeping variable, causing all sorts of fun surprises when you try to use it from multiple stages during the same boot. This patch redesigns the bookkeeping to stay completely inline in the (persistent) TTB which should resolve the issue. (This had not been a problem on any of our platforms for now... I just noticed this because I was trying to solve the same issue on ARM64.) BRANCH=None BUG=None TEST=Booted veyron_jerry. Mapped a second fine-grained memory range from romstage, confirmed that it finds the next free spot and leaves the bootblock table in place. Change-Id: I325866828b4ff251142e1131ce78b571edcc9cf9 Signed-off-by: Julius Werner <jwerner@chromium.org> Reviewed-on: http://review.coreboot.org/12074 Tested-by: build bot (Jenkins) Reviewed-by: Aaron Durbin <adurbin@chromium.org> Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
2015-11-07arm64: Remove cpu intialization through device-treeFurquan Shaikh
Since, SMP support is removed for ARM64, there is no need for CPU initialization to be performed via device-tree. Change-Id: I0534e6a93c7dc8659859eac926d17432d10243aa Signed-off-by: Furquan Shaikh <furquan@google.com> Reviewed-on: http://review.coreboot.org/11913 Tested-by: build bot (Jenkins) Reviewed-by: Julius Werner <jwerner@chromium.org>
2015-11-07arm64: Remove SMP supportFurquan Shaikh
As ARM Trusted Firmware is the only first class citizen for booting arm64 multi-processor in coreboot remove SMP support. If SoCs want to bring up MP then ATF needs to be ported and integrated. Change-Id: Ife24d53eed9b7a5a5d8c69a64d7a20a55a4163db Signed-off-by: Furquan Shaikh <furquan@google.com> Reviewed-on: http://review.coreboot.org/11909 Tested-by: build bot (Jenkins) Reviewed-by: Julius Werner <jwerner@chromium.org>
2015-11-07arm64: remove spin table supportAaron Durbin
As ARM Trusted Firmware is the only first class citizen for booting arm64 multi-processor in coreboot remove spintable support. If SoCs want to bring up MP then ATF needs to be ported and integrated. Change-Id: I1f38b8d8b0952eee50cc64440bfd010b1dd0bff4 Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11908 Tested-by: build bot (Jenkins) Reviewed-by: Julius Werner <jwerner@chromium.org>
2015-11-07arm64: remove ARCH_ARM64_CORTEX_A57_POWER_DOWN_SUPPORTAaron Durbin
With the removal of secmon from coreboot there are no power down operations required. As such remove the A57 power down support. Change-Id: I8eebb0ecd87b5e8bb3eaac335d652689d7f57796 Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11898 Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Furquan Shaikh <furquan@google.com> Tested-by: build bot (Jenkins) Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
2015-11-07arm64: remove secmonAaron Durbin
It's been decided to only support ARM Trusted Firmware for any EL3 monitor. That means any SoC that requires PSCI needs to add its support for ATF otherwise multi-processor bring up won't work. Change-Id: Ic931dbf5eff8765f4964374910123a197148f0ff Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11897 Tested-by: build bot (Jenkins) Reviewed-by: Julius Werner <jwerner@chromium.org>
2015-11-06AMD Bettong: Enable S4 feature for Windows 7zbao
PMIOxEE is for setting USB3 power rail. Set it to S0, otherwise going into hibernation can not be wake up. Change-Id: I692497bad24d745738d670897e725a568c1db114 Signed-off-by: Zheng Bao <zheng.bao@amd.com> Signed-off-by: Zheng Bao <fishbaozi@gmail.com> Reviewed-on: http://review.coreboot.org/11373 Reviewed-by: Patrick Georgi <pgeorgi@google.com> Tested-by: build bot (Jenkins)
2015-11-04ACPI: Add functions for DMAR I/O-APIC and HPET entriesNico Huber
Refactor acpi_create_dmar_drhd_ds_pci() and add similar functions for I/O-APICs and MSI capable HPETs. We violate the spec [1] here, which talks about 16-bit source-ids spread over start_bus and path entries. Intel actually uses bus/dev/fn identification for those devices too, and so do we. [1] Intel Virtualization Technology for Directed I/O Architecture Specification Document-Number: D51397 Change-Id: I0fce075961762610d44b5552b71e010511871fc2 Signed-off-by: Nico Huber <nico.huber@secunet.com> Reviewed-on: http://review.coreboot.org/12192 Tested-by: build bot (Jenkins) Reviewed-by: Duncan Laurie <dlaurie@google.com> Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-11-04ACPI: Make DMAR flags settableNico Huber
Add a parameter to acpi_create_dmar() for the flags field and define flags given by the spec [1]. [1] Intel Virtualization Technology for Directed I/O Architecture Specification Document-Number: D51397 Change-Id: I03ae32f13bb0061bd3b9bef607db175d9b0bc5e1 Signed-off-by: Nico Huber <nico.huber@secunet.com> Reviewed-on: http://review.coreboot.org/12191 Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Tested-by: build bot (Jenkins) Reviewed-by: Duncan Laurie <dlaurie@google.com> Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-11-03arch/x86/bootblock_normal: Fix failure to buildTimothy Pearson
Fix a function call in the normal path using the original function name and arguments in code that was changed in commit 3bfd7cc6 (drivers/pc80: Rework normal / fallback selector code) This commit reworked most of the fallback / normal code, however the normal code paths were not fully tested by Jenkins, so this was missed. Change-Id: Ied66334977272a13b7a7307ff4d9f34eb22040aa Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com> Reviewed-on: http://review.coreboot.org/12315 Reviewed-by: Martin Roth <martinroth@google.com> Reviewed-by: Patrick Georgi <pgeorgi@google.com> Reviewed-by: Nico Huber <nico.h@gmx.de> Tested-by: build bot (Jenkins)
2015-10-31tree: drop last paragraph of GPL copyright headerPatrick Georgi
It encourages users from writing to the FSF without giving an address. Linux also prefers to drop that and their checkpatch.pl (that we imported) looks out for that. This is the result of util/scripts/no-fsf-addresses.sh with no further editing. Change-Id: Ie96faea295fe001911d77dbc51e9a6789558fbd6 Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Reviewed-on: http://review.coreboot.org/11888 Tested-by: build bot (Jenkins) Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com> Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2015-10-31arch/x86: avoid race condition on build.hPatrick Georgi
Change-Id: I15375ac1247b7cc8d80d910a767c7f3e67eb8739 Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Reviewed-on: http://review.coreboot.org/11904 Tested-by: build bot (Jenkins) Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-by: Nico Huber <nico.h@gmx.de>
2015-10-30wakeup: Switch back to 32bit mode firstStefan Reinauer
On x86_64 we need to leave long mode before we can switch to 16bit mode. Oh joy! When's my 64bit resume pointer coming? Why didn't this get caught earlier? Seems the Asrock E350M2 didn't do Suspend/Resume? Yes, I know it's Intel syntax. Will be converted to AT&T syntax as soon as the whole thing actually works.. 8) Change-Id: Ic51869cf67d842041f8842cd9964d72a024c335f Signed-off-by: Stefan Reinauer <stefan.reinauer@coreboot.org> Reviewed-on: http://review.coreboot.org/11106 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2015-10-17armv7: Word-sized/half-word-sized memory operations for 32/16 bit read/writePaul Kocialkowski
Some registers only allow word-sized or half-word-sized operations and will cause a data fault when accessed with byte-sized operations. However, the compiler may or may not break such an operation into smaller (byte-sized) chunks. Thus, we need to reliably perform word-sized operations for 32 bit read/write and half-word-sized operations for 16 bit read/write. This is particularly the case on the rk3288 SRAM registers, where the watchdog tombstone is stored. Moving to GCC 5.2.0 introduced a change of strategy in the compiler, where a 32 bit read would be broken into byte-sized chunks, which caused a data fault when accessing the watchdog tombstone register. The definitions for byte-sized memory operations are also adapted to stay consistent with the rest. Change-Id: I1fb3fc139e0a813acf9d70f14386a9603c9f9ede Signed-off-by: Paul Kocialkowski <contact@paulk.fr> Reviewed-on: http://review.coreboot.org/11698 Tested-by: build bot (Jenkins) Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
2015-10-16arch/x86/smbios: Add Crucial DIMM manufacturer IDTimothy Pearson
Change-Id: I975142351c0c033f9dc44670dcf819d296896921 Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com> Reviewed-on: http://review.coreboot.org/11934 Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com> Tested-by: build bot (Jenkins)
2015-10-16arch/x86/boot/smbios: Add SPD IDs for Kingston and CorsairTimothy Pearson
Change-Id: I6a32b69d3b75d7d086dc7f8ea1e195473399f406 Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com> Reviewed-on: http://review.coreboot.org/11933 Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com> Tested-by: build bot (Jenkins)
2015-10-14x86: add standalone verstage supportAaron Durbin
To support x86 verstage one needs a working buffer for vboot. That buffer resides in the cache-as-ram region which persists across verstage and romstage. The current assumption is that verstage brings cache-as-ram up and romstage tears cache-as-ram down. The timestamp, cbmem console, and the vboot work buffer are persistent through in both romstage and verstage. The vboot work buffer as well as the cbmem console are permanently destroyed once cache-as-ram is torn down. The timestamp region is migrated. When verstage is enabled the assumption is that _start is the romstage entry point. It's currently expected that the chipset provides the entry point to romstage when verstage is employed. Also, the car_var_*() APIs use direct access when in verstage since its expected verstage does not tear down cache-as-ram. Lastly, supporting files were added to verstage-y such that an x86 verstage will build and link. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built and booted glados using separate verstage. Change-Id: I097aa0b92f3bb95275205a3fd8b21362c67b97aa Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11822 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-10-11vboot: remove remnants of VBOOT_STUBAaron Durbin
For vboot1 there was an rmodule that was loaded and ran to do the firmware verification. That's no longer used so remove the last vestiges of VBOOT_STUB. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built glados. Change-Id: I6b41544874bef4d84d0f548640114285cad3474e Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11817 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-10-08arch/x86/bootblock: Do not include non-code files in bootblock.SAlexandru Gagniuc
Since we now have more freedom in the bootblock linking step it no longer makes sense to use a monolithic bootblock.S. Code segments must still be included as the order in bootblock.S determines code flow. However, non-code flow related assembly stubs don't need to be directly included in bootblock.S Change-Id: I08e86e92d82bd2138194ed42652f268b0764aa54 Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com> Reviewed-on: http://review.coreboot.org/11792 Tested-by: build bot (Jenkins) Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com> Reviewed-by: Aaron Durbin <adurbin@chromium.org>
2015-10-08arch/x86: Link walkcbfs.S instead on including it in bootblock.SAlexandru Gagniuc
The code flow doesn't fall through to walkcbfs, as it does in the rest of bootblock.S. Instead, walkcbfs is called (albeit via a jmp). The linker cannot know this when walkcbfs.S is included directly. When we use a CAR bootblock, we lose several hundred bytes because walkcbfs is not garbage-collected, yet it isn't used. This problem is solved by assembling walkcbfs.S separately, and linking it. Change-Id: Ib3a976db09b9ff270b7677cb4f9db80b0b025e22 Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com> Reviewed-on: http://review.coreboot.org/11785 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-10-08arch/x86/bootblock: Link in object files selected with bootblock-yAlexandru Gagniuc
As part of preparing for systems with non-memory-mapped media, we want to be able to call into C code. This change allows us to link C code directly into the bootblock. The steps of going from bootblock main() to CAR setup to C code will be implemented in subsequent patches. Note that a few files selected with bootblock-y will now be compiled for the bootblock as well, but since we enabled garbage collection, they will not be included in the final binary. Change-Id: I5ca6dcaf176f5469c6a3bb925859399123493bc6 Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com> Reviewed-on: http://review.coreboot.org/11783 Reviewed-by: Aaron Durbin <adurbin@chromium.org> Tested-by: build bot (Jenkins)
2015-10-08arch/x86/Makefile.inc: Simplify rule for bootblock.debugAlexandru Gagniuc
The only difference between the ifeq/else/endif guarded rules is the linker flags specific to x86. Add those flags to LDFLAGS_bootblock, and only use one rule for bootblock.debug. Change-Id: I986a93e0418f05fb273512d7efe0573052493332 Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com> Reviewed-on: http://review.coreboot.org/11782 Tested-by: build bot (Jenkins) Reviewed-by: Aaron Durbin <adurbin@chromium.org>
2015-10-07x86/bootblock: Use LDFLAGS_bootblock to enable garbage collectionAlexandru Gagniuc
The x86 bootblock linking is a mess. The bootblock is treated in a very special manner, and never received the update to link-time garbage collection. On newer x86 platforms, the boot media is no longer memory-mapped. That means we need to do a lot more setup in the bootblock. ROMCC is unsuitable for this task, and walkcbfs only works on memory-mapped CBFS. We need to revise the x86 bootflow for this new case. The approach this patch series takes is to perform CAR setup in the bootblock, and load the following stage (either romstage or verstage) from the boot media. This approach is not new, but has been done on our ARM ports for years. Since we will be adding .c files to the bootblock, it is prudent to use link-time garbage collection. This is also consistent to how we do things on other architectures. Unification FTW! Change-Id: I16b78456df56e0053984a9aca9367e2542adfdc9 Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com> Reviewed-on: http://review.coreboot.org/11781 Tested-by: build bot (Jenkins) Reviewed-by: Aaron Durbin <adurbin@chromium.org>
2015-09-30x86: prepare cache-as-ram to allow multiple stagesAaron Durbin
In order to do a verification of romstage on x86 one needs to run verstage which verifies romstage (and the memory init code). However, x86 doesn't have SRAM like every other modern SoC so managing the cache-as-ram region is especially critical. First move all of the "shared" objects to the beginning of the .car.data section. This change then ensures that each stage using car.ld to link has the same consistent view of the addresses of these fixed-sized objects in cache-as-ram. The CAR_GLOBALs can be unique per stage. However, these variables are expected to have a value of zero at the start of each stage. In order to allow a stage to provide those semantics outside of the initial cache-as-arm setup routine add _car_global_start and _car_global_end symbols. Those symbols can be used to clear the CAR_GLOBALs for that stage. Note that the timestamp region can't be moved out similarly to the pre-ram cbmem console because the object storage of the timestamp cache is used *after* cache-as-ram is torn down to indicate if the cache should be used or not. Therefore, that timestamp needs to migrated to ram. A logic change in src/lib/timestamp.c could alleviate this requirement, but that task wasn't tackled in this patch. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built and booted glados. Change-Id: I15e9f6b0c632ee5a2369da0709535d6cb0d94f61 Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11740 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-30x86: provide common macro for linking early stagesAaron Durbin
In order to support verstage on x86 one needs to link verstage like romstage since it needs all the cache-as-ram goodies. Therefore, provide a macro that one can invoke that provides the necessary recipes for linking that particular stage in such an environment. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built and booted glados. Change-Id: I12f4872df09fff6715829de68fc374e230350c2e Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11739 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-29Makefile: Replace the way to test if a string is emptyzbao
The output of command below, # i386-elf-nm build/cbfs/fallback/romstage_null.offenders | \ grep -q "" ; echo $? has different result on MacOS, OS X Mavericks, which outputs 0. On linux, it outputs 1. I assume it is misleading to search an empty string in a empty string. Change it to testing if the string is empty. Change-Id: Ie4b8fe1fb26df092e2985937251a49feadc61eb0 Signed-off-by: Zheng Bao <zheng.bao@amd.com> Signed-off-by: Zheng Bao <fishbaozi@gmail.com> Reviewed-on: http://review.coreboot.org/11600 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-28arm64: mmu: Prevent CPU prefetch instructions from device memoryJimmy Huang
Set XN bit of block upper attribute to device memory in mmu. CPU may speculatively prefetch instructions from device memory, but the IO subsystem of some implementation may not support this operation. Set this attribute to device memory mmu entries can prevent CPU from prefetching device memory. BRANCH=none BUG=none TEST=build and booted to kernel on oak-rev3 with dcm enabled. Change-Id: I52ac7d7c84220624aaf6a48d64b9110d7afeb293 Signed-off-by: Patrick Georgi <pgeorgi@chromium.org> Original-Commit-Id: 7b01a4157cb046a5e75ea7625060a602e7a63c3c Original-Change-Id: Id535e990a23b6c89123b5a4e64d7ed21eebed607 Original-Signed-off-by: Jimmy Huang <jimmy.huang@mediatek.com> Original-Reviewed-on: https://chromium-review.googlesource.com/302301 Original-Commit-Ready: Yidi Lin <yidi.lin@mediatek.com> Original-Tested-by: Yidi Lin <yidi.lin@mediatek.com> Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org> Original-Reviewed-by: Julius Werner <jwerner@chromium.org> Reviewed-on: http://review.coreboot.org/11722 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-23RISCV: modify arch_prog_run to handle payloads correctly.Ronald G. Minnich
Unlike the other stages, the payload requires virtual memory to be set up and also a privelege level change. Change-Id: Ibbe2a55f7719d917f121a53a17c6d90e6b2ab3d1 Signed-off-by: Ronald G. Minnich <rminnich@gmail.com> Reviewed-on: http://review.coreboot.org/11699 Reviewed-by: Aaron Durbin <adurbin@chromium.org> Tested-by: build bot (Jenkins)
2015-09-22coreboot: introduce commonlibAaron Durbin
Instead of reaching into src/include and re-writing code allow for cleaner code sharing within coreboot and its utilities. The additional thing needed at this point is for the utilities to provide a printk() declaration within a <console/console.h> file. That way code which uses printk() can than be mapped properly to verbosity of utility parameters. Change-Id: I9e46a279569733336bc0a018aed96bc924c07cdd Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11592 Tested-by: build bot (Jenkins) Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
2015-09-16riscv-virtual-memory: move page tables into virtual address spaceThaminda Edirisooriya
If we use a linux payload/any payload that wants to manage virtual memory, and the payload is a supervisor (thus requiring virtual addressing before being started), we need to make sure that the page table is mapped into the virtual address space. Move the start address of the tables so the payload can manage virtual memory. Change-Id: I1d99e46f38a38a163fb1c7c517b1abca80cde0dc Signed-off-by: Thaminda Edirisooriya <thaminda@google.com> Reviewed-on: http://review.coreboot.org/11621 Tested-by: build bot (Jenkins)
2015-09-16riscv-memlayout: fix existing memlayout issues, add sbi interfaceThaminda Edirisooriya
Existing memlayout code placed sections in overlapping areas, and would overwrite the payload if it was large enough. Update memlayout.ld in src/mainboard/emulation/spike-riscv to represent the spike emulator, and add sbi interface which now has room into src/arch/riscv/bootblock.S. Add utility code to qemu-riscv, but emulator itself has yet to be updated to new ISA and as such should not be used. Update Makefile to include all the files necessary for sbi interface. Clean up unused include in src/arch/riscv/include/atomic.h and whitespace in src/mainboard/emulation/spike-riscv/memlayout.ld Fixed whitespace issues in spike_util.c Change-Id: Id97fe75e45ac1361005bec6d421756ee3f98a508 Signed-off-by: Thaminda Edirisooriya <thaminda@google.com> Reviewed-on: http://review.coreboot.org/11370 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2015-09-16x86: remove double link step for romstageAaron Durbin
Now that cbfstool supports XIP for romstage utilize it. This removes the double link steps with the cbfstool locate and add-stage sandwich. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built and booted on glados. Change-Id: I1ec555f523a94dd4b15fe8186cbe530520c622c0 Signed-off-by: Aaron Durbin <adurbin@chromium.org> Reviewed-on: http://review.coreboot.org/11670 Tested-by: build bot (Jenkins) Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com> Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-15riscv-trap-handling: Add functionality, prevent stack corruptionThaminda Edirisooriya
Trap handling code was bugged in that it loaded in the wrong stack pointer, overwriting the space the processor uses to talk to its host for doing device requests. Fix this issue, as well as add support for handling misaligned loads the same way we handle misaligned stores. Change-Id: I68ba3a114b7167b3212bb0bed181a7595f0b97d8 Signed-off-by: Thaminda Edirisooriya <thaminda@google.com> Reviewed-on: http://review.coreboot.org/11620 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2015-09-10riscv-trap-handling: Add implementation for trap calls in riscvThaminda Edirisooriya
RISCV requires the bios/bootloader to set up an interface by which it can get information about memory, talk to host devices, etc. Put implementation for spike in src/mainboard/emulation/spike-riscv/spike_util.c, and src/arch/riscv/trap_handler.c Change-Id: Ie1d5f361595e48fa6cc1fac25485ad623ecdc717 Signed-off-by: Thaminda Edirisooriya <thaminda@google.com> Reviewed-on: http://review.coreboot.org/11368 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2015-09-10riscv-virtual-memory: Add virtual memory setupThaminda Edirisooriya
Execution in supervisor level code in RISCV requires early setup of virtual memory. Add initialization calls in src/arch/riscv/virtual_memory.c to implement the required page table setup, and helper functions to use when jumping to the payload correctly in riscv. Change-Id: I46e080e0ee8dc13277d567dcd4bf0f61a4507b76 Signed-off-by: Thaminda Edirisooriya <thaminda@google.com> Reviewed-on: http://review.coreboot.org/11369 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2015-09-09x86: link ramstage the same way regardless of RELOCATABLE_RAMSTAGEAaron Durbin
Previously there were 2 paths in linking ramstage. One was used for RELOCATABLE_RAMSTAGE while the other was fixed location. Now that rmodtool can handle multiple secitons for a single proram segment there's no need for linking ramstage using lib/rmodule.ld. That also means true rmodules don't have symbols required for ramstage purposes so fix memlayout.h. Lastly add default rules for creating rmod files from the known file names and locations. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built rambi. Inspected ramstage.debug as well as rmodules created during the build. Change-Id: I98d249036c27cb4847512ab8bca5ea7b02ce04bd Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11524 Tested-by: build bot (Jenkins) Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com> Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-09linking: add and use LDFLAGS_commonAaron Durbin
Add an LDFLAGS_common variable and use that for each stage during linking within all the architectures. All the architectures support gc-sections, and as such they should be linking in the same way. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built rambi and analyzed the relocatable ramstage. Change-Id: I41fbded54055455889b297b9e8738db4dda0aad0 Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11522 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com> Reviewed-by: Julius Werner <jwerner@chromium.org>
2015-09-09x86: link romstage and ramstage with 1 fileAaron Durbin
To reduce file clutter merge romstage.ld and ramstage.ld into a single memlayout.ld. The naming is consistent with other architectures and chipsets for their linker script names. The cache-as-ram linking rules are put into a separate file such that other rules can be applied for future verstage support. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built rambi and dmp/vortex86ex. Change-Id: I1e8982a6a28027566ddd42a71b7e24e2397e68d2 Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11521 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-09rmodule: use program.ld for linkingAaron Durbin
Bring rmodule linking into the common linking method. The __rmodule_entry symbol was removed while using a more common _start symbol. The rmodtool will honor the entry point found within the ELF header. Add ENV_RMODULE so that one can distinguish the environment when generating linker scripts for rmodules. Lastly, directly use program.ld for the rmodule.ld linker script. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built rambi and analyzed the relocatable ramstage, sipi_vector, and smm rmodules. Change-Id: Iaa499eb229d8171272add9ee6d27cff75e7534ac Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11517 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-09x86: link romstage like the other architecturesAaron Durbin
All the other architectures are using the memlayout for linking romstage. Use that same method on x86 as well for consistency. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built a myriad of boards. Analyzed readelf output. Change-Id: I016666c4b01410df112e588c2949e3fc64540c2e Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11510 Tested-by: build bot (Jenkins) Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com> Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-09x86: link ramstage like the other architecturesAaron Durbin
All the other architectures are using the memlayout for linking ramstage. The last piece to align x86 is to use arch/header.ld and the macros within memlayout.h to automaticaly generate the necessary linker script. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built a myriad of boards. Analyzed readelf output. Change-Id: I012c9b88c178b43bf6a6dde0bab821e066728139 Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11508 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-09linking: lay the groundwork for a unified linking approachAaron Durbin
Though coreboot started as x86 only, the current approach to x86 linking is out of the norm with respect to other architectures. To start alleviating that the way ramstage is linked is partially unified. A new file, program.ld, was added to provide a common way to link stages by deferring to per-stage architectural overrides. The previous ramstage.ld is no longer required. Note that this change doesn't handle RELOCATABLE_RAMSTAGE because that is handled by rmodule.ld. Future convergence can be achieved, but for the time being that's being left out. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built a myriad of boards. Change-Id: I5d689bfa7e0e9aff3a148178515ef241b5f70661 Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11507 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com> Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com> Reviewed-by: Julius Werner <jwerner@chromium.org>
2015-09-09x86: provide minimum alignment for romstageAaron Durbin
The current way the XIP address of romstage is calculated is by doing a 'cbfstool locate' using a bin file of romstage linked at address 0. That address is then used for re-linking romstage at the address spit out by cbfstool. Currently, the linker actually sets minimum alignment on the text sections as 32 bytes, but it doesn't actually honor that value. Instead, provide a minimum alignment for romstage so as not to fight the linker. BUG=chrome-os-partner:44827 BRANCH=None TEST=Built asus/kfsn4-dre. Confirmed ROMSTAGE_BASE == gdtptr. Change-Id: Id6ec65d257df9ede78c720b0d7d4b56acfbb3f15 Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11588 Tested-by: build bot (Jenkins) Reviewed-by: Patrick Georgi <pgeorgi@google.com>
2015-09-09x86: remove unused sections from romstage.ldAaron Durbin
Now that the only source of ELF sections for romstage are from directly included .inc files or ROMCC generated inc files the subsection globs can be removed. i.e. Remove .rom.data.* and .rom.text.* listings. Lastly, put the .rom.data section directly after the .rom.text. They are by definition read-only and they are generated from the same place. BUG=chrome-os-partner:44827 BRANCH=None TEST=Spot checked !ROMCC and ROMCC boards. Confirmed only .rom.text .rom.data sections exist. Change-Id: Id17cf95c943103de006c5f3f21a625838ab49929 Signed-off-by: Aaron Durbin <adubin@chromium.org> Reviewed-on: http://review.coreboot.org/11505 Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org> Tested-by: build bot (Jenkins) Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com> Reviewed-by: Patrick Georgi <pgeorgi@google.com>