mainboards/amd/fam10: Initialize cbmem area after raminit
When GFXUMA is enabled, CBMEM is placed at TOM - UMASIZE
When GFXUMA is disabled, CBMEM is placed at TOM
This matches the behaviour present before conversion to early
CBMEM.
The CBMEM location code implicitly assumes TOM does not change
between romstage and ramstage. TOM is set by romstage raminit,
and is never changed by romstage or ramstage afterward. As
the CBMEM location is positioned at a specific offset from TOM
that is known to both romstage and ramstage early CBMEM is safe
on Fam10h systems.
TEST: Booted ASUS KFSN4-DRE and verified both cbmem timestamp
tables from romstage and cbmem log tables from ramstage.
Change-Id: Idf9e0245fe91185696ff664b06182c26b376c196
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8489
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com>
The CBMEM memory segment is always placed at TOM - UMASIZE when GFXUMA
is enabled, however when GFXUMA is disabled an attempt was made to locate
the CBMEM memory segment above the I/O hole in certain rare cases.
Removing this special case does not impact functionality, and paves
the way for early CBMEM support.
Change-Id: I98d29ab9d601a4e20f58e2cd0a66abb13b494e74
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8664
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@google.com>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Tested-by: Aaron Durbin <adurbin@google.com>
Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.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>
This removes the mainboard agesawrapper.c file from binarypi
based boards and creates a common one.
Change-Id: I900dba914f1c401e4ac732eb93d94b98216e629a
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/8671
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
This makes the change to the cpu/amd/pi/00730F01 that was
made for the cpu/amd/agesa based boards in:
commit 48518f0d
AGESA: Add amd_initcpuio() and amd_initmmio()
These are not wrappers for AGESA as they do not enter vendorcode at all.
We expect most of the added fixme.c file to be written without use of AMDLIB.h
and parts relocated as northbridge enable_resources().
The equivalent change has already been made for cpu/amd/pi/00630F01.
Change-Id: I591b50ee807436f5a1dee14d2c88a77462024744
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/8670
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
Tested on an X60, Native graphics init still works perfectly.
Change-Id: I91be3baa658e0332028c512c5a4cb0aee07d540a
Signed-off-by: Francis Rowe <info@gluglug.org.uk>
Reviewed-on: http://review.coreboot.org/8696
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
This is useful when the PCB layout of a mainboard does not allow
stable operation at the increased HyperTransport speeds of newer
processors.
Change-Id: Idc93a1294608178ddf38ca72d40e6bad7deb9004
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8464
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
This capability means:
FERR messages are sent out on system detected an
unmasked floating point x87 FPU error.
Even though this capability is supported on nehalem it doesn't
make sense to set it in early stage. This MSR
has a core scope which results in an unsync MSR because
it's not set on other cores than the BSP.
Found-by: BITS
Tested-on: lenovo thinkpad x201t
Change-Id: Ief3c04f57ac69e7289fbd37dbc3fd239f9098155
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Reviewed-on: http://review.coreboot.org/8659
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
There is nothing platform specific in retrieving S3 resume state from
romstage_handoff structure. Boards without EARLY_CBMEM_INIT update
acpi_slp_type from ACPI power-management block or scratchpad registers.
Change-Id: Ifc3755f891a0810473b3216c1fec8e45908fc1ab
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8188
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@google.com>
Commit 2e0cf14 corrected this for pi/00730F01/northbridge.c.
This commit fixes it for pi/00630F01/northbridge.c.
Found-by: Clang
Change-Id: I4eb93a07aacf6ffc5a159222117e7c934d85859e
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/8289
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Also fix a typo in a config option for SteppeEagle.
Change-Id: Iad51cc917217aa0eac751dc805c304652d20e066
Signed-off-by: Bruce Griffith <Bruce.Griffith@se-eng.com>
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/7247
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
All boards in tree use 0. Looks like this is all work that was
never completed and tested.
We also have static setting sysconf.segbit=0 which would conflict
with PCI_BUS_SEGN_BITS>0.
Having PCI_BUS_SEGN_BITS>0 would also require PCI MMCONF support
to cover over 255 buses.
Change-Id: I060efc44d1560541473b01690c2e8192863c1eb5
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8554
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Also drop some more #if UNUSED_CODE.
Change-Id: I1bbe96a65c9240636ff7cfaf70c2ecbfb3aee715
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8551
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
There is no requirement that in dev->link_list the last element
would have the highest link->link_num.
Also fix off-by-one error when allocating for more links.
Change-Id: Id8a7db3ffb4111eb31e70ea14fd522b70368dd8c
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8550
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The BKDG for K10 revision D and later processors recommends a smaller
MCT burst write queue depth when using unganged memory.
TEST: Booted ASUS KFSN4-DRE with both Opteron 8356 and Opteron 2431
processors.
Change-Id: I36718d4972c9d2d0bdd3274191503b5fcd803f15
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8500
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
This change brings all agesawrappers in a single file to make it
easier to understand the actual execution flow.
Change-Id: Ifbb2b16e4cccfaa17aaf10887a856797be9b6877
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8605
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
We do not allow platforms to mess around with memory layout.
Change-Id: I316ff522c8833fa3b7ad20f2c5a9cae21f4174d8
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8604
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Moved mctGetLogicalCPUID() to a separate file and made it available in
both romstage and ramstage.
Change-Id: I959c1caa8f796947b627a7b379c37d7307e2898e
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8499
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Nicolas Reinecke <nr@das-labor.org>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
We only need one of devx and dev. This function should be called
with dev already adjusted if link_num > 3.
Change-Id: I7166bbb88143bc28802c9530c4da16db67868d8e
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8351
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
These settings are specific to the SteppeEagle SOC and should
be made in its northbridge code rather than the CPU code.
Change-Id: I1a231f95225e1414b0cbc026a2a7b7797bd91fca
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/8254
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
Note that the limit is not set in the devicetree.cb which use native
sandybridge raminit, as it is not needed. When that isn't set, it's
automatically set to zero, and when we find that, we automatically
return the default limit. Thus behavior isn't changed for any board.
Change-Id: I447399eea71355612b654710a56f3a0077c2f7f9
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/8476
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
We can successfully bring up systems if timC calibration fails, as has
been demonstrated with google/butterfly. As a result, do not die(),
but simply print a message and continue in the hope that we may be
able to boot.
Change-Id: I49ec80324f63b2d45ae8f61c5c26454acb9c232f
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/8475
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
This patch allows the user to set a maximum HT link frequency in
NVRAM, paralleling a similar option available in the proprietary
BIOS on some mainboards.
Change-Id: Iba3789262eefa52421e76533cbf14d9da2ef1de8
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8462
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
The existing code generated invalid ACPI processor objects
if the core number was greater than 9. The first invalid
object instance was autocorrected by Linux, but subsequent
instances conflicted with each other, leading to a failure
to boot if more than 10 CPU cores were installed.
The modified code will function with up to 99 cores.
Change-Id: I62dc0eb61ae2e2b7f7dcf30e9c7de09cd901a81c
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8422
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
All settable memory controller options are now controlled by NVRAM,
making the Kconfig options irrelevant.
Change-Id: I9b2c8798d830e5c41bb9a108514e60d784d2ebc5
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8452
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
K10 processors cannot operate at full memory speeds when more than a
certain number of DIMMs are installed on a specific channel. The
allowed DIMM numbers and speeds are listed in the BKDG; this patch
implements the appropriate frequency reduction to ensure stability.
Change-Id: I8ac5b508915e423d262ad36c49de1fe696df2ecd
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8435
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
This fixes errors of the form:
error: 'Dct0MemSize' may be used uninitialized in this function
Change-Id: Ifc853aea9050994f5641c57a081aa0667331c995
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/8455
Tested-by: build bot (Jenkins)
This patch allows the following memory controller settings to be overridden in NVRAM:
Memory frequency limit
ECC enable
ECC scrub rate
Change-Id: Ibfde3d888b0f81a29a14af2d142171510b87655e
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8438
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
On x86, change the type of the address parameter in
read8()/read16/read32()/write8()/write16()/write32() to be a
pointer, instead of unsigned long.
Change-Id: Ic26dd8a72d82828b69be3c04944710681b7bd330
Signed-off-by: Kevin Paul Herbert <kph@meraki.net>
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/7784
Tested-by: build bot (Jenkins)
Kconfig variable is not implemented.
Change-Id: I546a1001847e7b1002f96baf49ed3301852a6894
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8345
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Keep the slower HyperTransport configuration for a possible reference
in fam15 boards.
Change-Id: Ifcdedc6385fec80f7d02c55c2aac10e5e2429a18
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8344
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Two source files were accidently marked executable. Switch them back to
mode 644 (rw-r---r--)
Change-Id: Ic96f6e5e9a05cbffb65cdfb627023d04d3866dc9
Signed-off-by: Stefan Reinauer <stepan.reinauer@coreboot.org>
Reviewed-on: http://review.coreboot.org/8426
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-by: Martin Roth <gaumless@gmail.com>
The indentations of #define are not consistent in chip.h. Update to make
all #define indentations being aligned and put them after the variable
declaration.
Change-Id: I37550acac18bac3efddb580ef6b956be0e2b357a
Signed-off-by: York Yang <york.yang@intel.com>
Reviewed-on: http://review.coreboot.org/8333
Reviewed-by: Dave Frodin <dave.frodin@se-eng.com>
Tested-by: build bot (Jenkins)
Timothy Pearson