After determining how much DDR is actually in the system, set DBAT0 and
IBAT0 accordingly. This ensures that the CPU won't attempt to access
(via speculation) addresses outside of actual memory.
On 86xx systems, DBAT0 and IBAT0 (the BATs for DDR) are initialized to 2GB
and kept that way. If the system has less than 2GB of memory (typical for
an MPC8610 HPCD), the CPU may attempt to access this memory during
speculation. The zlib code is notorious for generating such memory reads,
and indeed on the MPC8610, uncompressing the Linux kernel causes a machine
check (without this patch).
Currently we are limited to power of two sized DDR since we only use a
single bat. If a non-power of two size is used that is less than
CONFIG_MAX_MEM_MAPPED u-boot will crash.
Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Commit cbb0cab1d9 broke some platforms which used kgdb code but
didn't actually include kgdb.h. So include kgdb.h in all the relevant
traps code.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Add a disable sub-command to the cpu command that allows for disabling
cores in multicore processors. This can be useful for systems that are
using multicore chips but aren't utilizing all the cores as a way to
reduce power and possibly improve performance.
Also updated an added missing copyright.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
r14 is not supposed to be clobbered by functions. Switch
to r12 and call GET_GOT when needed. This will allow u-boot
to loose the -ffixed-r14 gcc option.
Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Using the GOT in IRQ handlers requires r14 to be -ffixed-r14.
Avoid this by relocatate transfer_to_handler too.
This will allow to free up r14 later on.
Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
NULL is an absolute value and should not be relocated.
After this correction code like:
void weak_fun(void) __attribute__((weak));
printf("weak_fun:%p\n", weak_fun);
will still print null after relocation.
Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
While in probecpu() UART is still not initialized.
Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Fixing the number of cores in the device tree based on the actual number of
cores on the system. With this same device tree image can be used for dual
core and single core members of otherwise exactly same SOC.
For example:
* P2020RDB and P2010RDB
* P1020RDB and P1011RDB
* MPC8641D and MPC8641
Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Incase the system is detected with Unknown SVR, let the system boot
with a default value and a proper message.
Now with dynamic detection of SOC properties from SVR, this is necessary
to prevent a crash.
Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
The number of CPUs are getting detected dynamically by checking the
processor SVR value. Also removed CONFIG_NUM_CPUS references from all
the platforms with 85xx/86xx processors.
This can help to use the same u-boot image across the platforms.
Also revamped and corrected few Freescale Copyright messages.
Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Removed same code pieces from cpu/mpc85xx/cpu.c and cpu/mpc86xx/cpu.c
and moved to cpu/mpc8xxx/cpu.c(new file)
Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
For historic reasons we had defined some additional PLATFORM_CPPFLAGS like:
PLATFORM_CPPFLAGS += -DCONFIG_MPC86xx=1
PLATFORM_CPPFLAGS += -DCONFIG_MPC8641=1
However these are all captured in the config.h and thus redudant. Also
moved common 86xx flags into cpu/mpc86xx/config.mk.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Rename sdram_mode_1 to sdram_mode and sdram_cfg_1 to sdram_cfg to match
the 86xx user's manual and other Freescale architectures
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
DMA support is now enabled via the CONFIG_FSL_DMA define instead of the
previous CONFIG_DDR_ECC
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Unify with 83xx and 85xx and use CPU_TYPE_ENTRY. We are going to use
this to convey the # of cores and DDR width in the near future so its
good to keep in sync.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
The following changes were made to sync up the DMA code between the 85xx
and 86xx architectures which will make it easier to break out common
8xxx DMA code:
85xx:
- Don't set STRANSINT and SPCIORDER fields in SATR register. These bits
only have an affect when the SBPATMU bit is set.
- Write 0xffffffff instead of 0xfffffff to clear errors in the DMA
status register. We may as well clear all 32 bits of the register...
86xx:
- Add CONFIG_SYS_MPC86xx_DMA_ADDR define to address DMA registers
- Add clearing of errors in the DMA status register when initializing
the controller
- Clear the channel start bit in the DMA mode register after a transfer
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Break out DMA structures for the Freescale MPC85xx and MPC86xx cpus to
reduce a large amount of code duplication
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
__attribute__ follows gcc's documented syntax and is generally more
common than __attribute. This change is only asthetic and should not
affect functionality.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
* Use CONFIG_MP instead of CONFIG_NUM_CPUS to match 85xx
* Introduce determine_mp_bootpg() helper. We'll need this to address a
bug introduced in v2009.03 with 86xx MP booting. We have to make sure
to reserve the region of memory used for the MP bootpg() so other
u-boot code doesn't use it.
* Added dummy versions of cpu_reset(), cpu_status() & cpu_release() to
allow cmd_mp.c to build and work. In the future we should look at
implementing all these functions. This could be common w/85xx if we
use spin tables on 86xx.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Update the 86xx reset sequence to try executing a board-specific reset
function. If the board-specific reset is not implemented or does not
succeed, then assert #HRESET_REQ. Using #HRESET_REQ is a more standard
reset procedure than the previous method and allows all board
peripherals to be reset if needed.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Currently MPC85xx and MPC86xx boards just calculate the localbus frequency
and print it out, but don't save it.
This changes where its calculated and stored to be more consistent with the
CPU, CCB, TB, and DDR frequencies and the MPC83xx localbus clock.
The localbus frequency is added to sysinfo and calculated when sysinfo is
set up, in cpu/mpc8[56]xx/speed.c, the same as the other frequencies are.
get_clocks() copies the frequency into the global data, as the other
frequencies are, into a new field that is only enabled for MPC85xx and
MPC86xx.
checkcpu() in cpu/mpc8[56]xx/cpu.c will print out the local bus frequency
from sysinfo, like the other frequencies, instead of calculating it on the
spot.
Signed-off-by: Trent Piepho <tpiepho@freescale.com>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Jon Loeliger <jdl@freescale.com>
The local bus clock divider should be doubled for both 8610 and 8641.
Signed-off-by: Trent Piepho <tpiepho@freescale.com>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Jon Loeliger <jdl@freescale.com>
On newer CPUs, 8536, 8572, and 8610, the CLKDIV field of LCRR is five bits
instead of four.
In order to avoid an ifdef, LCRR_CLKDIV is set to 0x1f on all systems. It
should be safe as the fifth bit was defined as reserved and set to 0.
Code that was using a hard coded 0x0f is changed to use LCRR_CLKDIV.
Signed-off-by: Trent Piepho <tpiepho@freescale.com>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Jon Loeliger <jdl@freescale.com>
Export the localbus frequency in the device tree, the same way the CPU, TB,
CCB, and various other frequencies are exported in their respective device
tree nodes.
Some localbus devices need this information to be programed correctly, so
it makes sense to export it along with the other frequencies.
Unfortunately, when someone wrote the localbus dts bindings, they didn't
bother to define what the "compatible" property should be. So it seems no
one was quite sure what to put in their dts files.
Based on current existing dts files in the kernel source, I've used
"fsl,pq3-localbus" and "fsl,elbc" for MPC85xx, which are used by almost all
of the 85xx devices, and are looked for by the Linux code. The eLBC is
apparently not entirely backward compatible with the pq3 LBC and so eLBC
equipped platforms like 8572 won't use pq3-localbus.
For MPC86xx, I've used "fsl,elbc" which is used by some of the 86xx systems
and is also looked for by the Linux code. On MPC8641, I've also used
"fsl,mpc8641-localbus" as it is also commonly used in dts files, some of
which don't use "fsl,elbc" or any other acceptable name to match on.
Signed-off-by: Trent Piepho <tpiepho@freescale.com>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Jon Loeliger <jdl@freescale.com>
This patch creates a memory map with all the devices
in 36-bit physical space, in addition to the 32-bit map.
The CCSR relocation is moved (again, sorry) to
allow for the physical address to be 36 bits - this
requires translation to be enabled. With 36-bit physical
addressing enabled, we are no longer running with VA=PA
translations. This means we have to distinguish between
the two in the config file. The existing region name is
used to indicate the virtual address, and a _PHYS variety
is created to represent the physical address.
Large physical addressing is not enabled by default.
Set CONFIG_PHYS_64BIT in the config file to turn this on.
Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
The memory map on the 8641hpcn is modified to look more like
the 85xx boards; this is a step towards a more standardized
layout going forward. As part of this change, we now relocate
the flash.
The regions for some of the mappings were far larger than they
needed to be. I have reduced the mappings to match the
actual sizes supported by the hardware.
In addition I have removed the comments at the head
of the BAT blocks in the config file, rather than updating
them. These get horribly out of date, and it's a simple
matter to look at the defines to see what they are set to
since everything is right here in the same file.
Documentation has been changed to reflect the new map, as this
change is user visible, and affects the OS which runs post-uboot.
Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
We define CONFIG_MONITOR_BASE_EARLY to define the initial location
of the bootpage in flash. Use this to create an early mapping
definition for the FLASH, and change the early_bats code to use this.
This change facilitates the relocation of the flash since the early
mappings are no longer tied to the final location of the flash.
Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
Using a mtmsr/blr means that you have to be executing at the
same virtual address once you enable translation. This is
unnecessarily restrictive, and is not really how this is
usually done. Change it to use the more common mtspr SRR0/SRR1
and rfi method.
Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
In order to later allow for a physical relocation of the
flash, setup_bats, which sets up the final BAT mapping
for the board, needs to happen *after* init_laws().
Otherwise, there will be no window programmed for the flash
at the new physical location at the point when we change
the mmu translation.
Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
We put the bootpg for the secondary cpus into memory and use
BPTR to get to it. This is a step towards converting to the
ePAPR boot methodology. Also, the code is written to
deal properly with more than 4GB of RAM.
Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
Currently, the CCSR gets relocated while translation is
enabled, meaning we need 2 BAT translations to get to both the
old location and the new location. Also, the DEFAULT
CCSR location has a dependency on the BAT that maps the
FLASH region. Moving the relocation removes this unnecessary
dependency. This makes it easier and more intutive to
modify the board's memory map.
Swap BATs 3 and 4 on 8610 so that all 86xx boards use the same
BAT for CCSR space.
Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
The commit 67256678f0 add
the another global data pointer save, but in fact the
global data pointer will be initialized in the board_init_r,
so remove it such as the 85xx/83xx family.
Signed-off-by: Dave Liu <daveliu@freescale.com>
Acked-by: Kumar Gala <kumar.gala@freescale.com>
I believe these code was copied from 74xx family, but for
86xx, it is unused.
Signed-off-by: Dave Liu <daveliu@freescale.com>
Acked-by: Kumar Gala <kumar.gala@freescale.com>
We must invalidate TLBs before MMU turn on, but
currently the code is not, if there are some stale
TLB entry valid in the TLBs, it will cause strange
issue.
Signed-off-by: Dave Liu <daveliu@freescale.com>
Acked-by: Becky Bruce <becky.bruce@freescale.com>
This is needed in unlock_ram_in_cache() because it is called from C and
will corrupt the small data area anchor that is kept in R2.
lock_ram_in_cache() is modified similarly as good coding practice, but
is not called from C.
Signed-off-by: Nick Spence <nick.spence@freescale.com>
For some reason we duplicated the majority of code in lib_ppc/interrupts.c
Not know how that happened, but there is no good reason for it.
Use the interrupt_init_cpu() and timer_interrupt_cpu() since its why
they exist.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
The tsec driver contains a hard-coded array of configuration information
for the tsec ethernet controllers. We create a default function that works
for most tsecs, and allow that to be overridden by board code. It creates
an array of tsec_info structures, which are then parsed by the corresponding
driver instance to determine configuration. Also, add regs, miiregs, and
devname fields to the tsec_info structure, so that we don't need the kludgy
"index" parameter.
Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Timur Tabi
Mike Frysinger
Kumar Gala
Joakim Tjernlund
Poonam Aggrwal
Peter Tyser
Becky Bruce
Wolfgang Denk
Trent Piepho
Jean-Christophe PLAGNIOL-VILLARD
Becky Bruce
Dave Liu
Nick Spence
Andy Fleming