Errors in PSIM
Binkowski, Cassio
cassioiks at gmail.com
Tue May 7 00:54:44 UTC 2013
So there is now way to know how much memory RTEMS is actually using?
Also, I have modified psim.in to be with 32mb of RAM:
/openprom/options/oea-memory-size 33554432
However, when I try to create a partition with starting address 0x1000000
(10 MB) it says the address is not mapped. How can that be?
On Sun, May 5, 2013 at 11:39 PM, Joel Sherrill <Joel.Sherrill at oarcorp.com>wrote:
> The linker script format is defined by binutils. RTEMS bsps need enough symbols to define the work area and starting stack.
>
> Since the linker script has 16mb in it, that is all the RTEMS application can know about. Anything beyond that 16mb is free for you to use as you like.
>
> Gedare Bloom <gedare at rtems.org> wrote:
>
>
> You can look for some tutorials on ld scripts. This is an area of rtems
> that lacks good documentation.
>
> The only way to know what memory you can use outside of rtems is to
> inspect the linkcmds. In your case though you might be able to just point
> at the 16 MB base address, since psim bsp does not expect to use that much
> ram...
> On May 5, 2013 9:01 PM, "Binkowski, Cassio" <cassioiks at gmail.com> wrote:
>
>> Hi Gedare,
>>
>> Can you please explain where/how in linkcmds I should be touching and
>> how to let the bps know about the change?
>> I'm very new to this.
>>
>> Also, how do I calculate how much space RTEMS is taking so I know the
>> address to start creating the partitions?
>>
>> Thanks!
>>
>>
>> On Sun, May 5, 2013 at 9:40 PM, Gedare Bloom <gedare at gwmail.gwu.edu>wrote:
>>
>>> You probably need to modify the bsp linkcmds file, or just malloc a
>>> large chunk. The bsp may need to know about the extra memory in either
>>> case, so check out the linkcmds.
>>> -Gedare
>>> On May 5, 2013 2:46 PM, "Binkowski, Cassio" <cassioiks at gmail.com>
>>> wrote:
>>>
>>>> Thanks Joel.
>>>>
>>>> I have modified it to be with 32 MB of RAM.
>>>>
>>>> /openprom/options/oea-memory-size 33554432
>>>>
>>>> Right now my doubt is the following:
>>>>
>>>> From psim.in or other resources, how can I know at which address I
>>>> can start creating the partitions and make sure it won't give me an
>>>> 'unmapped address' error again?
>>>>
>>>> Thanks.
>>>>
>>>>
>>>> On Thu, May 2, 2013 at 12:06 PM, Joel Sherrill <
>>>> joel.sherrill at oarcorp.com> wrote:
>>>>
>>>>> I don't know where this all ended yesterday.
>>>>>
>>>>>
>>>>> On 5/1/2013 9:00 PM, Binkowski, Cassio wrote:
>>>>>
>>>>> I know I'm missing out on something. It's probably because of the
>>>>> memory addresses and maybe a custom config is required on that file
>>>>> mentioned by Joel (
>>>>> http://git.rtems.org/rtems/tree/c/src/lib/libbsp/powerpc/psim/startup/linkcmds).
>>>>> I have never dealt with this before and will need some assistance.
>>>>>
>>>>> You could modify the device tree created by psim/psim-gdb to add
>>>>> more RAM or a second bank of RAM after the first 16 MB that the
>>>>> linkcmds knows about. This is in psim.in:
>>>>>
>>>>> /openprom/options/oea-memory-size 16777216
>>>>>
>>>>> The bottom line is that you are reading the mpc5200 PowerPC manual for
>>>>> addresses and that does not have any correlation to the memory map of
>>>>> the "psim board" as defined by the device-tree.
>>>>>
>>>>> 0x10000 is 64k which is in use by the program or RTEMS.
>>>>> 0x100000 is 1MB and is also in use by the program or RTEMS.
>>>>> 0x1000000 is outside the memory map.
>>>>>
>>>>> --joel
>>>>>
>>>>>
>>>>> On Wed, May 1, 2013 at 10:35 PM, Binkowski, Cassio <
>>>>> cassioiks at gmail.com> wrote:
>>>>>
>>>>>> That may be one solution.
>>>>>>
>>>>>> @Joel: Thank you so much for the detailed explanation.
>>>>>> One thing that would be great is that if we could somehow condense
>>>>>> all this information exchanged in the mailing list into the
>>>>>> Wiki/Manuals/Doxygen.
>>>>>>
>>>>>> For example, I am now getting another exception, I'm just pasting
>>>>>> some of it, because it's huge:
>>>>>>
>>>>>> *Suspending faulting task (0x0A010003)*
>>>>>> *Exception handler called for exception 7 (0x7)*
>>>>>> *Too many stack frames (stack possibly corrupted), giving up...*
>>>>>> *Suspending faulting task (0x0A010004)*
>>>>>> *
>>>>>> *
>>>>>> It's given at the creation of the 7th partition when function *
>>>>>> _API_extensions_Run_postswitch()* is executed.
>>>>>>
>>>>>> How can I start searching for it? I don't know what to look for
>>>>>> inside inter.h.
>>>>>>
>>>>>> Best regards
>>>>>> Cassio
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Wed, May 1, 2013 at 10:18 PM, Gedare Bloom <gedare at gwmail.gwu.edu>wrote:
>>>>>>
>>>>>>> You tell the partitions where to start, so just give the second
>>>>>>> partition the right address. You have to be sure the memory you use is
>>>>>>> available both in the hw as real memory and also not used for other
>>>>>>> purposes. Probably you can malloc all the memory you need at startup time
>>>>>>> and pass pointers into the allocated buffer to partition create.
>>>>>>> -Gedare
>>>>>>> On May 1, 2013 7:07 PM, "Binkowski, Cassio" <cassioiks at gmail.com>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> Hmmm. The 'Testing Task' does not have a while(1) loop and I'm
>>>>>>>> not deleting it. Could that be it? *<- Yup, that's it. Just
>>>>>>>> tested. Can you please explain how did you arrive at this conclusion?
>>>>>>>> *
>>>>>>>>
>>>>>>>> I'm referring to the MPC5200B board. You can see the manual in
>>>>>>>> the following link:
>>>>>>>> http://www.freescale.com/files/32bit/doc/ref_manual/MPC5200BUM.pdf
>>>>>>>>
>>>>>>>> Now, mentioning the partitions and memory addresses. I'm not
>>>>>>>> quite sure if I'm incrementing addresses correctly. I do want to create a
>>>>>>>> partition right next where the other one finishes.
>>>>>>>> So, if starting address is 0x10000, the next partition will start
>>>>>>>> at 0x10000 + previousPartitionSize.
>>>>>>>> Does this make sense?
>>>>>>>>
>>>>>>>> Thanks.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Wed, May 1, 2013 at 7:49 PM, Joel Sherrill <
>>>>>>>> Joel.Sherrill at oarcorp.com> wrote:
>>>>>>>>
>>>>>>>>> It could be another task exiting. You mentioned loading code.
>>>>>>>>>
>>>>>>>>> Psim is a specific simulated hardware environment and does not as configured match any board. If you got addresses from a piece of real hardware, the addresses will not match. What target board's manual are you reading?
>>>>>>>>>
>>>>>>>>> "Binkowski, Cassio" <cassioiks at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Ok.
>>>>>>>>>
>>>>>>>>> 1) The Init task is already deleting itself after loading the
>>>>>>>>> other tasks. Do I have to do something else?
>>>>>>>>>
>>>>>>>>> 2) The memory for partitions came from the board data sheet. I
>>>>>>>>> based myself on the addresses mentioned there and followed upon. I was
>>>>>>>>> creating the first one starting at 0x10000 and incrementing from that
>>>>>>>>> according to each partition's size.
>>>>>>>>>
>>>>>>>>> Can you help me set these 2 things up? I'm not really sure how
>>>>>>>>> to deal with that memory issue and the internal error as well.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On Wed, May 1, 2013 at 6:18 PM, Joel Sherrill <
>>>>>>>>> Joel.Sherrill at oarcorp.com> wrote:
>>>>>>>>>
>>>>>>>>>> That address is invalid for psim. Where does the partition memory come from? Here I'd the memory keep for psim
>>>>>>>>>> http://git.rtems.org/rtems/tree/c/src/lib/libbsp/powerpc/psim/startup/linkcmds
>>>>>>>>>>
>>>>>>>>>> "Binkowski, Cassio" <cassioiks at gmail.com> wrote:
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> I'm also getting the following error when I uncomment the
>>>>>>>>>> function that creates 10 partitions to be used by my application:
>>>>>>>>>>
>>>>>>>>>> *core_find_mapping() - access to unmaped address, attach a
>>>>>>>>>> default map to handle this - addr=0x1000d2c nr_bytes=0x4
>>>>>>>>>> processor=0xa8e5000 cia=0x1028c*
>>>>>>>>>> *
>>>>>>>>>> *
>>>>>>>>>> Can you please help?
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Wed, May 1, 2013 at 5:44 PM, Binkowski, Cassio <
>>>>>>>>>> cassioiks at gmail.com> wrote:
>>>>>>>>>>
>>>>>>>>>>> The file 'events.c' is inside /gdb-7.5.1/sim/ppc.
>>>>>>>>>>>
>>>>>>>>>>> I am using psim-gdb testing my own application, it's not an
>>>>>>>>>>> RTEMS example.
>>>>>>>>>>>
>>>>>>>>>>> I managed to break right on it, and it stops at:
>>>>>>>>>>> ../../../../../../rtems/c/src/../../cpukit/score/src/interr.c:58
>>>>>>>>>>>
>>>>>>>>>>> Signature is the following:
>>>>>>>>>>> the_source=the_source at entry=INTERNAL_ERROR_CORE,
>>>>>>>>>>> is_internal=is_internal at entry=true,
>>>>>>>>>>> the_error=the_error at entry=5)
>>>>>>>>>>>
>>>>>>>>>>> I have no clue of what it can be. Do you have any idea?
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On Wed, May 1, 2013 at 5:37 PM, Joel Sherrill <
>>>>>>>>>>> Joel.Sherrill at oarcorp.com> wrote:
>>>>>>>>>>>
>>>>>>>>>>>> Where is the file events.c?
>>>>>>>>>>>>
>>>>>>>>>>>> Break at the fault handler printing that.
>>>>>>>>>>>>
>>>>>>>>>>>> Does this happen with any rtems provided examples?
>>>>>>>>>>>>
>>>>>>>>>>>> Are you using the psim or psim-gdb script to run it?
>>>>>>>>>>>>
>>>>>>>>>>>> "Binkowski, Cassio" <cassioiks at gmail.com> wrote:
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Hi everyone,
>>>>>>>>>>>>
>>>>>>>>>>>> I was just testing my application using PSIM and it is
>>>>>>>>>>>> returning the following lines, right after starting a Task:
>>>>>>>>>>>>
>>>>>>>>>>>> *rtems-4.10.99.0(PowerPC/Generic (classic FPU)/psim) PANIC
>>>>>>>>>>>> ERROR 5*
>>>>>>>>>>>> *events.c:404: assertion failed - events->time_from_event > 0*
>>>>>>>>>>>>
>>>>>>>>>>>> Has anybody experienced this before?
>>>>>>>>>>>>
>>>>>>>>>>>> Thanks.
>>>>>>>>>>>> Cassio
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> rtems-users mailing list
>>>>>>>> rtems-users at rtems.org
>>>>>>>> http://www.rtems.org/mailman/listinfo/rtems-users
>>>>>>>>
>>>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Joel Sherrill, Ph.D. Director of Research & Development joel.sherrill at OARcorp.com On-Line Applications Research
>>>>> Ask me about RTEMS: a free RTOS Huntsville AL 35805
>>>>> Support Available (256) 722-9985
>>>>>
>>>>>
>>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20130506/48bc406b/attachment-0001.html>
More information about the users
mailing list