[ECOS] arm-eabi -> Interrupt vector not free

[Frank Pagliughi]

On 06/25/2011 10:49 AM, Christophe Coutand wrote:
> Hi Frank,
>
> I think what you are describing is similar to this bug report:
>
> http://bugs.ecos.sourceware.org/show_bug.cgi?id=1001096
>
> Christophe

Yes, that seems to be it. Sorry I was searching all over for "static 
constructor" issues. Didn't think it would be isolated to one last(?) 
SPI driver. It seems to have been discussed several times over the last 
few years:

http://sourceware.org/ml/ecos-discuss/2010-12/msg00033.html
http://sourceware.org/ml/ecos-discuss/2010-08/msg00032.html
http://sourceware.org/ml/ecos-patches/2009-02/msg00108.html

This coming week I can test out your patches on an old board (Atmel 
EB55) with some old and new compilers, if that will help to move the fix 
along.

Frank

>> -----Original Message-----
>> From: ecos-discuss-owner@ecos.sourceware.org [mailto:ecos-discuss-
>> owner@ecos.sourceware.org] On Behalf Of Frank Pagliughi
>> Sent: 23. juni 2011 23:58
>> To: Reg
>> Cc: eCos Discuss
>> Subject: Re: [ECOS] arm-eabi ->  Interrupt vector not free
>>
>> On 04/12/2009 11:29 AM, Reg wrote:
>>> Hello
>>>
>>> This error raises before ecos calls cyg_user_start. So it's in the
>>> ecos startup. Earlier I spent days to get my gdb working (OpenOCD,
>>> cygwin, windows, Atmel AT91 SAM7A3). With the tools shipped with ecos
>>> it did not work, but with yagarto's toolchain it did, but I'm not
>> able
>>> to compile ecos with it, because configtool uses cygwin style path,
>>> and yagarto uses windows style pathes.
>>> An eabi has this error :) I will give a try with arm-eabi-gdb. If it
>>> did not work, I would be in a hard situation :)
>>>
>>> Thank you!
>>> Gergely Szentirmai
>>>
>>> Sergei Gavrikov írta:
>>>> On Fri, Apr 10, 2009 at 11:03:10PM +0200, Szentirmai Gergely wrote:
>>>>> Hello
>>>>>
>>>>> Previously I used arm-elf-gcc, but any time I try to get arm-eabi
>>>>> alive  I get this this message imidiately after reset:
>>>>>
>>>>> <1>intr.cxx[506]void Cyg_Interrupt::attach() Interrupt vector not
>> free.
>>>>> ASSERT FAIL:<1>intr.cxx            [ 506] void
>> Cyg_Interrupt::attach()
>>>>>   Interrupt vector not free.
>>>> First, what is your target (platform) and which interrrupt number
>> is?
>>>>> There is no message like this when I compile things with elf (two
>>>>> changes, command prefix= arm-elf-gcc, and Build for eabi = false
>> (it
>>>>> is  done automatically)). The ecos config, and sources are the
>> same.
>>>>> There  is only one main.c, which would send hello world to the diag
>>>>> port.  (works perfectly with elf). Nothing to do with interrupts.
>>>> Can you set a breakpoint on cyg_assert_fail() and
>> cyg_interrupt_attach()
>>>> to know a bit more about issue? What is that vector? UART, RTC? The
>>>> assertion's report will tell you about.
>>>>
>>>>> This message was discussed here:
>>>>> http://ecos.sourceware.org/ml/ecos-discuss/2004-08/msg00099.html
>>>>>
>>>>> But my ecos source (latest from CVS) behaves like this.
>>>>>
>>>>> Any ideas? Maybe this magic build for eabi switch does something
>> wrong?
>>>> It's not enough an information to get it.
>>>>
>>>> Sergei
>>>>
>>>>> Gergely Szentirmai
>>>>>
>>>>> --
>>>>> Before posting, please read the FAQ:
>>>>> http://ecos.sourceware.org/fom/ecos
>>>>> and search the list archive: http://ecos.sourceware.org/ml/ecos-
>> discuss
>> I'm seeing this problem as well. Was there a resolution?
>>
>> In my case I'm using the Atmel EB55 eval board (AT91M55800A chip). It
>> seems that, when using the EABI compiler, the constructor for the SPI
>> device gets called twice. Once through:
>>     _GLOBAL_I.3200_spi_at91_init.cxx
>> which eventually calls the cyg_spi_at91_bus_init() function.
>>
>> But then it seems the cyg_spi_at91_bus_init() function is called
>> directly, as if it's in the table of default constructors.
>>
>> I'm using a default build and testing the example app, "twothreads".
>>
>> Frank
>>
>> --
>> Before posting, please read the FAQ:
>> http://ecos.sourceware.org/fom/ecos
>> and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss
>


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss