The CPU POST test code (run from cpu_post_exec_31()) doesn't follow the
ABI carefully, at least the CR3, CR4, and CR5 fields of CR are clobbered
by it. The gcc-4.2 with its more aggressive optimization exposes this fact.
This patch just saves the CR value before running the test code, so allowing
it to do anything it wants with CR.
Signed-off-by: Dmitry Rakhchev <rda@emcraft.com>
Acked-by: Yuri Tikhonov <yur@emcraft.com>
--
The ECC POST reported intermittent failures running after power-up on
the Korat PPC440EPx board. Even when the test passed, the debugging
output occasionally reported additional unexpected ECC errors.
This refactoring has three main objectives: (1) minimize the code
executed with ECC enabled during the tests, (2) add more checking of the
results so any unexpected ECC errors would cause the test to fail, and
(3) use synchronization (only) where required by the processor.
Signed-off-by: Larry Johnson <lrj@acm.org>
post/cpu/ppc4xx/spr.c contained a few checks for registers only present
for PPC440 and derivates processor.
Signed-off-by: Niklaus Giger <niklaus.giger@netstal.com>
On LWMON5 we now use d-cache as init-ram and stack. The SPR POST test uses
self modifying code and this doesn't work with stack in d-cache, since
I can't move the code from d-cache to i-cache. We move the SPR test to
be executed a little later, after relocation. Then stack is located in
SDRAM and this self-modifying code is no problem anymore.
Signed-off-by: Stefan Roese <sr@denx.de>
As repoted by Larry Johnson, running "diag run cache" caused a crash
in U-Boot. This problem was introduced by a patch that removed the
TLB entry for the cache test after the test has completed. Since this
TLB was only setup once, a 2nd attempt to run this cache test
failed with a crash. Now this TLB entry is created every time the
routine is called.
Signed-off-by: Stefan Roese <sr@denx.de>
This patch allows the ECC POST to be used for different boards with the
PPC440 Denali SDRAM controller. Modifications include skipping the test
if ECC is not enabled (as for non-ECC DIMMs) and adding synchronization
to prevent timing errors.
Signed-off-by: Larry Johnson <lrj@acm.org>
Now with caches enabled (i- and d-cache) on 44x, we need a chance to
disable the cache for the CPU POST tests, since these tests consist
of self modifying code. This is done via the new change_tlb() function.
Signed-off-by: Stefan Roese <sr@denx.de>
The POST ethernet test needed to be changed to dynamically determine
the count of ethernet devices. This code is cloned from the 4xx
ethernet driver.
Signed-off-by: Stefan Roese <sr@denx.de>
As experienced on lwmon5, on some boards the POST memory test can
corrupt the global data buffer (bd). This patch fixes this issue
by checking and limiting this area.
Signed-off-by: Yuri Tikhonov <yur@emcraft.com>
Signed-off-by: Stefan Roese <sr@denx.de>
When CFG_POST_ETHER_EXT_LOOPBACK is defined, the ethernet POST
is not done using an internal loopback connection, but by assuming
that an external loopback connector is plugged into the board.
Signed-off-by: Stefan Roese <sr@denx.de>
The patch adds support for UART POST on ppc44x-based boards with no
external serial clocks installed.
Signed-off-by: Yuri Tikhonov <yur@emcraft.com>
Acked-by: Stefan Roese <sr@denx.de>
- Clear ECC status regs after ECC POST test
- Set dcbz for ECC generation with caches enabled as default
- Code cleanup
Signed-off-by: Stefan Roese <sr@denx.de>
This patch adds ECC Post test for the Lwmon5 board based
on PPC440EPx to U-Boot.
Signed-off-by: Pavel Kolesnikov <concord@emcraft.com>
Acked-by: Yuri Tikhonov <yur@emcraft.com>
Acked-by: Stefan Roese <sr@denx.de>
Modifications are based on the linux kernel approach and
support two use cases:
1) Add O= to the make command line
'make O=/tmp/build all'
2) Set environement variable BUILD_DIR to point to the desired location
'export BUILD_DIR=/tmp/build'
'make'
The second approach can also be used with a MAKEALL script
'export BUILD_DIR=/tmp/build'
'./MAKEALL'
Command line 'O=' setting overrides BUILD_DIR environent variable.
When none of the above methods is used the local build is performed and
the object files are placed in the source directory.
This fixes a problem of displaying bogus voltages when the voltages
are so low that the I2C devices start failing while the rest of the
system keeps running.
Update etags build target
* Improve NetConsole support: add support for broadcast destination
address and buffered input.
* Cleanup compiler warnings for GCC 3.3.x and later
* Fix problem in cmd_jffs2.c introduced by CFG_JFFS_SINGLE_PART patch
* Patch by Sascha Hauer, 28 Jun:
- add generic support for Motorola i.MX architecture
- add support for mx1ads, mx1fs2 and scb9328 boards
* Patches by Marc Leeman, 23 Jul 2004:
- Add define for the PCI/Memory Buffer Configuration Register
- corrected comments in cpu/mpc824x/cpu_init.c
* Add support for multiple serial interfaces
(for example to allow modem dial-in / dial-out)
- add support for the Epson 156x series of graphical displays
(These displays are serial and not suitable for using a normal
framebuffer console on them)
- add infrastructure needed in order to POST any DSPs in a board
- update board/emk tree; use common flash driver
- Corrected tested bits in machine check exception in cpu/mpc5xxx/traps.c
[adapted for other PPC CPUs -- wd]
- Added support for the M48T08 on the EVAL5200 board in rtc/mk48t59.c
* Patch by Jon Diekema, 13 Feb 2004:
Call show_boot_progress() whenever POST "FAILED" is printed.
* Patch by Nishant Kamat, 13 Feb 2004:
Add support for TI OMAP1610 H2 Board
Fixes for cpu/arm926ejs/interrupt.c
(based on Richard Woodruff's patch for arm925, 16 Oct 03)
Fix for a timer bug in OMAP1610 Innovator
Add support for CS0 (ROM)/CS3 (Flash) boot in OMAP1610 Innovator and H2
* Patches by Stephan Linz, 12 Feb 2004:
- add support for NIOS timer with variable period preload counter value
- prepare POST framework support for NIOS targets
* Patch by Denis Peter, 11 Feb 2004:
add POST support for the MIP405 board
before we can access it; add delay in case we are faster (with no
CF card inserted)
* Cleanup of some init functions
* Make sure SCC Ethernet is always stopped by the time we boot Linux
to avoid Linux crashes by early packets coming in.
* Accelerate flash accesses on LWMON board by using buffered writes
- Added Motorola CPU 8540/8560 support (cpu/85xx)
- Added Motorola MPC8540ADS board support (board/mpc8540ads)
- Added Motorola MPC8560ADS board support (board/mpc8560ads)
* Minor code cleanup
add FAT support for IDE, SCSI and USB
* Patches by Gleb Natapov, 2 Sep 2003:
- cleanup of POST code for unsupported architectures
- MPC824x locks way0 of data cache for use as initial RAM;
this patch unlocks it after relocation to RAM and invalidates
the locked entries.
* Patch by Gleb Natapov, 30 Aug 2003:
new I2C driver for mpc107 bridge. Now works from flash.
* Patch by Dave Ellis, 11 Aug 2003:
- JFFS2: fix typo in common/cmd_jffs2.c
- JFFS2: fix CFG_JFFS2_SORT_FRAGMENTS option
- JFFS2: remove node version 0 warning
- JFFS2: accept JFFS2 PADDING nodes
- SXNI855T: add AM29LV800 support
- SXNI855T: move environment from EEPROM to flash
- SXNI855T: boot from JFFS2 in NOR or NAND flash
* Patch by Bill Hargen, 11 Aug 2003:
fixes for I2C on MPC8240
- fix i2c_write routine
- fix iprobe command
- eliminates use of global variables, plus dead code, cleanup.
applications: instead of using (PPC-specific) system calls we now
use a jump table; please see doc/README.standalone for details
* Patch by Dave Westwood, 24 Jul 2003:
added support for Unity OS (a proprietary OS)
add delay to get I2C working with "imm" command and s3c24x0_i2c.c
* Patch by Richard Woodruff, 17 July 03:
- Fixed bug in OMAP1510 baud rate divisor settings.
* Patch by Nye Liu, 16 July 2003:
MPC860FADS fixes:
- add MPC86xADS support (uses MPC86xADS.h)
- add 866P/T core support (also MPC859T/MPC859DSL/MPC852T)
o PLPRCR changes
o BRG changes (EXTAL/XTAL restricted to 10MHz)
o don't trust gclk() software measurement by default, depend on
CONFIG_8xx_GCLK_FREQ
- add DRAM SIMM not installed detection
- use more "correct" SDRAM initialization sequence
- allow different SDRAM sizes (8xxADS has 8M)
- default DER is 0
- remove unused MAMR defines from FADS860T.h (all done in fads.c)
- rename MAMR/MBMR defines to be more consistent. Should eventually
be merged into MxMR to better reflect the PowerQUICC datasheet.
* Patch by Yuli Barcohen, 16 Jul 2003:
support new Motorola PQ2FADS-ZU evaluation board which replaced
MPC8260ADS and MPC8266ADS
Yuri Tikhonov
Larry Johnson
Niklaus Giger
Stefan Roese
Larry Johnson
Matthias Fuchs
Pavel Kolesnikov
Heiko Schocher
Wolfgang Denk
Sergei Poselenov
Igor Lisitsin
Wolfgang Denk
Wolfgang Denk
Marian Balakowicz
Wolfgang Denk
wdenk