Setting of flash_sector::is_erased in flash erase and mass erase
is popular folklore. Make clear it is useless.
Change-Id: Ide397eb6d24fc8fa38931e6c8a0693d39668a5d2
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4768
Tested-by: jenkins
Reviewed-by: Christopher Head <chead@zaber.com>
Instances of struct flash_driver are never written to at runtime. For a
small amount of memory saving and also robustness (fewer things for
stray pointer writes to hit), mark them const.
Change-Id: Iadbbbc2fac0976d892699200000c5f02856729f3
Signed-off-by: Christopher Head <chead@zaber.com>
Reviewed-on: http://openocd.zylin.com/4803
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
This should allow users to configure flash at >32-bit addresses.
Change-Id: I7c9d3c5762579011a2d9708e5317e5765349845c
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: http://openocd.zylin.com/4919
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Most of flash drivers have to ensure proper flash write block alignment
and padding. As there was no support for it in the flash infrastructure,
each driver does it its own way. Sometimes this part of code is not properly
tested and contains bugs.
flash_write(_unlock) joins all image sections targeted to one flash bank
using padded areas as a glue. This solves alignment problems on section
boundaries but imposes other problems.
Introduce new flash bank parameters write_start_alignment,
write_end_alignment and minimal_write_gap.
New flash drivers can just properly set these values instead of handling
alignment by its own.
Adapt infrastructure (namely flash_write_unlock(), handle_flash_fill_command()
and handle_flash_write_bank_command()) to prepare write data padded
to an alignment required by the flash bank.
Rework flash_write_unlock() to discontinue write block when the gap
between sections is bigger than minimum specified in minimal_write_gap.
minimal_write_gap is set to one sector by default.
Change-Id: I4368dd402dfaf51c193bcbf1332cffff092b239b
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4399
Tested-by: jenkins
Reviewed-by: Andreas Bolsch <hyphen0break@gmail.com>
struct flash_driver has a default_padded_value field that is similar,
but it can be changed by the user for the specific purpose of padding.
Add a new erased_value field and initialize it for all targets,
particularly stm32lx, xmc4xxx and virtual.
Use this value in core.c:default_flash_mem_blank_check(), the slow path.
Extend the target API to pass erased_value down to target code.
Adding an argument ensures that we catch all callers.
This allows us to merge xmc4xxx.c:xmc4xxx_blank_check_memory() into
armv7m:armv7m_blank_check_memory().
It further allows us to use default_flash_blank_check() in place of
xmc4xxx.c:xmc4xxx_flash_blank_check(), adding a potential slow path
fallback, as well as stm32lx:stm32lx_erase_check(), adding the potential
armv7m fast path with fallback to default_flash_mem_blank_check().
Fix a mips32 code comment while at it (zeroed -> erased).
The armv4_5 and mips32 target implementations will now error out if an
erase value other than 0xff is used, causing default_flash_blank_check()
to fall back to the default_flank_mem_blank_check() slow path.
Change-Id: I39323fbbc4b71c256cd567e439896d0245d4745f
Signed-off-by: Andreas Färber <afaerber@suse.de>
Reviewed-on: http://openocd.zylin.com/3497
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Originally flash/nor infrastructure assumed protection blocks identical
to erase sectors. This assumption is not valid for many flash types.
Driver code fixed the problem either by increasing sector size to
size of protection block or by defining more protection block than
really existed in device. Both cases had drawbacks.
The change retains compatibility with the old driver.
Updated driver can set protection blocks table independent
of sector table.
Change-Id: I27f6d267528ad9ed9fe0a85f05436a8ec17603a4
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/3545
Tested-by: jenkins
Reviewed-by: Steven Stallion <stallion@squareup.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Also make GPL notices consistent according to:
https://www.gnu.org/licenses/gpl-howto.html
Change-Id: I84c9df40a774958a7ed91460c5d931cfab9f45ba
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/3488
Tested-by: jenkins
Reviewed-by: Andreas Färber <afaerber@suse.de>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
This adds support for setting the default value used when padding image
sections. Previously 0xff was used but some targets have an erased value
of 0x00.
Change-Id: If3df6fea3abf95b09daa3ff7be362acf991070ce
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1635
Tested-by: jenkins
Use default_flash_blank_check, this will use the much faster
blank_check_memory handler if supported - 15x quicker on stm32f4.
Otherwise it will fall back to using the slower default_flash_mem_blank_check.
Change-Id: Ia231b3e95468c9e92594dbdbe1fa2d69e1506fc3
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/632
Tested-by: jenkins
When a flash cmd is called using the flash name the autoprobe
function is not called. autoprobe is called if flash_command_get_bank
falls through to get_flash_bank_by_num.
This makes both get_flash_bank_by_name and get_flash_bank_by_num
behave the same.
Signed-off-by: Spencer Oliver <ntfreak@users.sourceforge.net>
Final target is to force bus_width size during CFI flash
read.
In this first step I need to replace default flash read
with flash specific implementation.
This patch introduces:
- flash_driver_read() layer;
- default_flash_read(), backward compatible;
- read() callback in struct flash_driver;
- proper initialization in every flash_driver instance.
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
There are a million reasons why cached protection state might
be stale: power cycling of target, reset, code executing on
the target, etc.
The "flash protect_check" command is now gone. This is *always*
executed when running a "flash info".
As a bonus for more a more robust approach, lots of code could
be deleted.
Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>
This stops GDB from launching with an empty memory map,
making gdb load w/flashing fail for no obvious reason.
The error message points in the direction of the gdb-attach
event that can be set up to issue a halt or "reset init"
which will put GDB in a well defined stated upon attach
and thus have a robust flash autoprobe.
Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>
The NOR infrastructure caches some per-sector state, but
it's not used much ... because the cache is not trustworthy.
This patch addresses one part of that problem, by ensuring
that state cached by NOR drivers gets invalidated once we
resume the target -- since targets may then modify sectors.
Now if we see sector protection or erase status marked as
anything other than "unknown", we should be able to rely
on that as being accurate. (That is ... if we assume the
drivers initialize and update this state correctly.)
Another part of that problem is that the cached state isn't
much used (being unreliable, it would have been unsafe).
Those issues can be addressed in later patches.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Add a NOR flash mechanism where erase_address ranges can be padded
out to sector boundaries, triggering a diagnostic:
> flash erase_address 0x0001f980 16
address range 0x0001f980 .. 0x0001f98f is not sector-aligned
Command handler execution failed
in procedure 'flash' called at file "command.c", line 647
called at file "command.c", line 361
>
> flash erase_address pad 0x0001f980 16
Adding extra erase range, 0x0001f800 to 0x0001f97f
Adding extra erase range, 0x0001f990 to 0x0001fbff
erased address 0x0001f980 (length 16) in 0.095975s (0.163 kb/s)
>
This addresses what would otherwise be something of a functional
regression. An earlier version of the interface had a dangerous
problem: it would silently erase data outside the range it was
told to erase. Fixing that bug turned up some folk who relied on
that unsafe behavior. (The classic problem with interface bugs!)
Now they can get that behavior again. If they really need it,
just specify "pad".
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Add file comments to a few files. Make the GDB server use
more conventional (pointer-free) hex digit conversion.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Move the bulk of the flash.h file into flash/nor/core.h, leaving an
empty husk that will be removed in the next patch.
The NOR driver structure is an implementation detail, so move it into
its own private header file <flash/nor/driver.h> along with helper
declaration for finding them by name.
The newly moved flash TCL routines access the internals of the module
too much. Fix the layering issues by adding new core NOR flash APIs:
<flash/nor/core.h>:
- flash_driver_find_by_name() - self-descriptive
<flash/nor/imp.h>:
- flash_bank_add() - encapsulates adding banks to bank list
- flash_bank_list() - encapsulates retreiving bank list
This allows the externs in flash/nor/imp.h to be removed, and
these mechanisms may now be re-used by other flash module code.
