| Age | Commit message (Collapse) | Author |
|---|
|
Squashed and adjusted two changes from chromium.git. Covers
CBMEM init for ROMTAGE and RAMSTAGE.
cbmem: Unify random on-CBMEM-init tasks under common CBMEM_INIT_HOOK() API
There are several use cases for performing a certain task when CBMEM is
first set up (usually to migrate some data into it that was previously
kept in BSS/SRAM/hammerspace), and unfortunately we handle each of them
differently: timestamp migration is called explicitly from
cbmem_initialize(), certain x86-chipset-specific tasks use the
CAR_MIGRATION() macro to register a hook, and the CBMEM console is
migrated through a direct call from romstage (on non-x86 and SandyBridge
boards).
This patch decouples the CAR_MIGRATION() hook mechanism from
cache-as-RAM and rechristens it to CBMEM_INIT_HOOK(), which is a clearer
description of what it really does. All of the above use cases are
ported to this new, consistent model, allowing us to have one less line
of boilerplate in non-CAR romstages.
BRANCH=None
BUG=None
TEST=Built and booted on Nyan_Blaze and Falco with and without
CONFIG_CBMEM_CONSOLE. Confirmed that 'cbmem -c' shows the full log after
boot (and the resume log after S3 resume on Falco). Compiled for Parrot,
Stout and Lumpy.
Original-Change-Id: I1681b372664f5a1f15c3733cbd32b9b11f55f8ea
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/232612
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
cbmem: Extend hooks to ramstage, fix timestamp synching
Commit 7dd5bbd71 (cbmem: Unify random on-CBMEM-init tasks under common
CBMEM_INIT_HOOK() API) inadvertently broke ramstage timestamps since
timestamp_sync() was no longer called there. Oops.
This patch fixes the issue by extending the CBMEM_INIT_HOOK() mechanism
to the cbmem_initialize() call in ramstage. The macro is split into
explicit ROMSTAGE_/RAMSTAGE_ versions to make the behavior as clear as
possible and prevent surprises (although just using a single macro and
relying on the Makefiles to link an object into all appropriate stages
would also work).
This allows us to get rid of the explicit cbmemc_reinit() in ramstage
(which I somehow accounted for in the last patch without realizing that
timestamps work exactly the same way...), and replace the older and less
flexible cbmem_arch_init() mechanism.
Also added a size assertion for the pre-RAM CBMEM console to memlayout
that could prevent a very unlikely buffer overflow I just noticed.
BRANCH=None
BUG=None
TEST=Booted on Pinky and Falco, confirmed that ramstage timestamps once
again show up. Compile-tested for Rambi and Samus.
Original-Change-Id: If907266c3f20dc3d599b5c968ea5b39fe5c00e9c
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/233533
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Change-Id: I1be89bafacfe85cba63426e2d91f5d8d4caa1800
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Signed-off-by: Marc Jones <marc.jones@se-eng.com>
Reviewed-on: http://review.coreboot.org/7878
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
|
|
This patch creates a new mechanism to define the static memory layout
(primarily in SRAM) for a given board, superseding the brittle mass of
Kconfigs that we were using before. The core part is a memlayout.ld file
in the mainboard directory (although boards are expected to just include
the SoC default in most cases), which is the primary linker script for
all stages (though not rmodules for now). It uses preprocessor macros
from <memlayout.h> to form a different valid linker script for all
stages while looking like a declarative, boilerplate-free map of memory
addresses to the programmer. Linker asserts will automatically guarantee
that the defined regions cannot overlap. Stages are defined with a
maximum size that will be enforced by the linker. The file serves to
both define and document the memory layout, so that the documentation
cannot go missing or out of date.
The mechanism is implemented for all boards in the ARM, ARM64 and MIPS
architectures, and should be extended onto all systems using SRAM in the
future. The CAR/XIP environment on x86 has very different requirements
and the layout is generally not as static, so it will stay like it is
and be unaffected by this patch (save for aligning some symbol names for
consistency and sharing the new common ramstage linker script include).
BUG=None
TEST=Booted normally and in recovery mode, checked suspend/resume and
the CBMEM console on Falco, Blaze (both normal and vboot2), Pinky and
Pit. Compiled Ryu, Storm and Urara, manually compared the disassemblies
with ToT and looked for red flags.
Change-Id: Ifd2276417f2036cbe9c056f17e42f051bcd20e81
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: f1e2028e7ebceeb2d71ff366150a37564595e614
Original-Change-Id: I005506add4e8fcdb74db6d5e6cb2d4cb1bd3cda5
Original-Signed-off-by: Julius Werner <jwerner@chromium.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/213370
Reviewed-on: http://review.coreboot.org/9283
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Tauner <stefan.tauner@gmx.at>
Reviewed-by: Aaron Durbin <adurbin@google.com>
|
|
The GCC 4.9.2 update showed that the boot_state_init_entry
structures were being padded and assumed to be aligned in to an
increased size. The bootstate scheduler for static entries,
boot_state_schedule_static_entries(), was then calculating the
wrong values within the array. To fix this just use a pointer to
the boot_state_init_entry structure that needs to be scheduled.
In addition to the previous issue noted above, the .bs_init
section was sitting in the read only portion of the image while
the fields within it need to be writable. Also, the
boot_state_schedule_static_entries() was using symbol comparison
to terminate a loop which in C can lead the compiler to always
evaluate the loop at least once since the language spec indicates
no 2 symbols can be the same value.
Change-Id: I6dc5331c2979d508dde3cd5c3332903d40d8048b
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/8699
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
|
|
Garbage collected sections allow for trimming the size of the
binaries as well as allowing for not needing to config off
unused functions. To that end, on a rambi build the following
differences are observed:
$ diff -up \
<(readelf -l coreboot-builds/google_rambi/cbfs/fallback/ramstage.elf) \
<(readelf -l coreboot-builds/google_rambi_gc_sections/cbfs/fallback/ramstage.elf)
--- /dev/fd/63 2015-03-10 12:07:27.927985430 -0500
+++ /dev/fd/62 2015-03-10 12:07:27.927985430 -0500
@@ -6,9 +6,9 @@ There are 4 program headers, starting at
Program Headers:
Type Offset VirtAddr PhysAddr FileSiz MemSiz Flg
Align
LOAD 0x001000 0x00000000 0x00000000 0x00040 0x00040 RWE 0
- LOAD 0x001040 0x00000040 0x00000040 0x34560 0x34560 RWE 0
- LOAD 0x0355a0 0x000345a0 0x000345a0 0x02578 0x02578 RWE 0
- LOAD 0x037b18 0x00036b18 0x00036b18 0x00000 0x0b560 0
+ LOAD 0x001040 0x00000040 0x00000040 0x2cbf8 0x2cbf8 RWE 0
+ LOAD 0x02dc38 0x0002cc38 0x0002cc38 0x02208 0x02208 RWE 0
+ LOAD 0x02fe40 0x0002ee40 0x0002ee40 0x00000 0x0a888 0
Section to Segment mapping:
Segment Sections...
$ diff -up \
<(readelf -l coreboot-builds/google_rambi/cbfs/fallback/romstage.elf) \
<(readelf -l coreboot-builds/google_rambi_gc_sections/cbfs/fallback/romstage.elf)
--- /dev/fd/63 2015-03-10 12:08:16.855985880 -0500
+++ /dev/fd/62 2015-03-10 12:08:16.851985880 -0500
@@ -5,8 +5,8 @@ There are 1 program headers, starting at
Program Headers:
Type Offset VirtAddr PhysAddr FileSiz MemSiz Flg
Align
- LOAD 0x000060 0xfff20000 0xfff20000 0x08b81 0x08b81 R E
0x10
+ LOAD 0x000060 0xfff20000 0xfff20000 0x06300 0x06300 R E
0x10
Section to Segment mapping:
Segment Sections...
- 00 .rom .text
+ 00 .rom
The following warnings needed to be applied to CFLAGS_common because for
some reason gcc was miraculously emitting the warnings with the
unrelated *-sections options:
-Wno-unused-but-set-variable
Change-Id: I210784fdfc273ce4cb9927352cbd5a51be3c6929
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/8635
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
|
|
We have no need for these sections winding up in the build
leading to possible overlaps, such as in the case of Clang
builds. Discard sections from inclusion into the resulting
binary.
Change-Id: Ie807e5809594dcc6e94660a64e359e3b2ca1a0f6
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-on: http://review.coreboot.org/7439
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
|
|
Fix the alignment for 64-bit systems
Change-Id: I7fcb1683d760b96307759b7d44d8770dd49a02e3
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://chromium-review.googlesource.com/214326
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: Furquan Shaikh <furquan@chromium.org>
Commit-Queue: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/6953
Reviewed-by: Aaron Durbin <adurbin@google.com>
Tested-by: build bot (Jenkins)
|
|
Rename coreboot_ram stage to ramstage. This is done in order to provide
consistency with other stage names (bootblock, romstage) and to allow any
Makefile rule generalization, required for patches to be submitted later.
Change-Id: Ib66e43b7e17b9c48b2d099670ba7e7d857673386
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: http://review.coreboot.org/5567
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
|
|
Depending on the compiler options, subsections of the form
of .section.subsection could be generated. Therefore, include
those subsections for .bss, .sbss, and .data.
Change-Id: I80dd64d8c62e7bc449ee2bbc0a22a941777e2ea6
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5407
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
|
|
This framework was only available in ramstage. So we had to define
console output functions separately for bootblock, romstage and SMM.
Follow-up patches will re-enable all the consoles removed here,
in a more flexible fashion, and with less lines-of-code and copy-paste.
Also the driver list is not in a well-defined order and some of the
loops could exit without visiting all drivers.
NOTE: This build has no console in ramstage.
Change-Id: Iaddc495aaca37e2a6c2c3f802a0dba27bf227a3e
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/5337
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
|
|
Start using the rmodtool for generating rmodules.
rmodule_link() has been changed to create 2 rules:
one for the passed in <name>, the other for creating
<name>.rmod which is an ELF file in the format of
an rmodule.
Since the header is not compiled and linked together
with an rmodule there needs to be a way of marking
which symbol is the entry point. __rmodule_entry is
the symbol used for knowing the entry point. There
was a little churn in SMM modules to ensure an
rmodule entry point symbol takes a single argument.
Change-Id: Ie452ed866f6596bf13f137f5b832faa39f48d26e
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/5379
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: build bot (Jenkins)
|
|
Many of the boot state callbacks can be scheduled at compile time.
Therefore, provide a way for a compilation unit to inform the
boot state machine when its callbacks should be called. Each C
module can export the callbacks and their scheduling requirements
without changing the shared boot flow code.
Change-Id: Ibc4cea4bd5ad45b2149c2d4aa91cbea652ed93ed
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/3133
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
|
|
While debugging a crash it was discovered that ld was inserting
address space for sections that were empty depending on section
address boundaries. This led to the assumption breaking down that
on-disk payload (code/data bits) was contiguous with the address
space. When that assumption breaks down relocation updates change
the wrong memory. Fix this by making the rmodule.ld linker script
put all code/data bits into a payload section.
Change-Id: Ib5df7941bbd64662090136e49d15a570a1c3e041
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/3149
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins)
|
|
The x86 linker script added a .textfirst section. In
order to properly link ramstage as a relocatable module
the .textfirst section needs to be included.
Also, the support for code coverage was added by including
the constructor section and symbols. Coverage has not been
tested as I suspect it might not work in a relocatable
environment without some tweaking. However, the section
and symbols are there if needed.
Change-Id: Ie1f6d987d6eb657ed4aa3a8918b2449dafaf9463
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/2883
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
|
|
Coreboot's ramstage defines certain sections/symbols in its fixed
static linker script. It uses these sections/symbols for locating the
drivers as well as its own program information. Add these sections
and symbols to the rmodule linker script so that ramstage can be
linked as an rmodule. These sections and symbols are a noop for other
rmodule-linked programs, but they are vital to the ramstage.
Also add a comment in coreboot_ram.ld to mirror any changes made there
to the rmodule linker script.
Change-Id: Ib9885a00e987aef0ee1ae34f1d73066e15bca9b1
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/2786
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
|
|
By including the heap in the bss output section the size is accounted
for in a elf PT_LOAD segment. Without this change the heap wasn't being
put into a PT_LOAD segment. The result is a nop w.r.t. functionality,
but readelf and company will have proper MemSiz fields.
Change-Id: Ibfe9bb87603dcd4c5ff1c57c6af910bbba96b02b
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/2750
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
|
|
A rmodule is short for relocation module. Relocaiton modules are
standalone programs. These programs are linked at address 0 as a shared
object with a special linker script that maintains the relocation
entries for the object. These modules can then be embedded as a raw
binary (objcopy -O binary) to be loaded at any location desired.
Initially, the only arch support is for x86. All comments below apply to
x86 specific properties.
The intial user of this support would be for SMM handlers since those
handlers sometimes need to be located at a dynamic address (e.g. TSEG
region).
The relocation entries are currently Elf32_Rel. They are 8 bytes large,
and the entries are not necessarily in sorted order. An future
optimization would be to have a tool convert the unsorted relocations
into just sorted offsets. This would reduce the size of the blob
produced after being processed. Essentialy, 8 bytes per relocation meta
entry would reduce to 4 bytes.
Change-Id: I2236dcb66e9d2b494ce2d1ae40777c62429057ef
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/2692
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
|
