RE: [ECOS] Problems with var_mk_defs.c

RE: [ECOS] Problems with var_mk_defs.c

[James Yates]

After looking at the generated var_mk_defs.s file by adding -save-temps to the global compiler flags, the first line the compiler complains about is:

.equ	HAL_UCACHE_SIZE, #0
.loc 1 82 0

Would I be right in assuming that this should read:

.equ	HAL_UCACHE_SIZE, 0
.loc 1 82 0

The function call that creates this line is:
DEFINE(HAL_UCACHE_SIZE, HAL_UCACHE_SIZE);

The macro itself:
#define DEFINE(sym, val) \
        asm volatile("\n\t.equ\t" #sym ", %0" : : "i" (val))

I guess this is the assembler equivalent of #define HAL_UCACHE_SIZE 0 so that wherever HAL_UCACHE_SIZE is referenced, the value of 0 is used.

However, I am unsure about how to modify the macro, my skill in assembler is extremely poor. Any ideas?

		James Yates

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

RE: [ECOS] Problems with var_mk_defs.c

[Gary Thomas]

On Wed, 2003-10-15 at 09:46, James Yates wrote:
> After looking at the generated var_mk_defs.s file by adding -save-temps to the global compiler flags, the first line the compiler complains about is:
> 
> .equ	HAL_UCACHE_SIZE, #0
> .loc 1 82 0
> 
> Would I be right in assuming that this should read:
> 
> .equ	HAL_UCACHE_SIZE, 0
> .loc 1 82 0
> 
> The function call that creates this line is:
> DEFINE(HAL_UCACHE_SIZE, HAL_UCACHE_SIZE);
> 
> The macro itself:
> #define DEFINE(sym, val) \
>         asm volatile("\n\t.equ\t" #sym ", %0" : : "i" (val))
> 
> I guess this is the assembler equivalent of #define HAL_UCACHE_SIZE 0 so that wherever HAL_UCACHE_SIZE is referenced, the value of 0 is used.
> 
> However, I am unsure about how to modify the macro, my skill in assembler is extremely poor. Any ideas?
> 

It looks right to me.  It seems that different assemblers (and
even different versions sometimes) want to see this written
differently.

Give it a try (sans "#") and see what happens.

-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates


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

RE: [ECOS] Problems with var_mk_defs.c

[James Yates]

I tried removing the hash, as a temporary fix i change DEFINE to:

#define DEFINE(sym, val) \
        asm volatile("\n\t.equ\t" #sym ",0" : : "i" (val))

Which effectively then outputs :

 .equ	HAL_UCACHE_SIZE, 0
 .loc 1 82 0

However, this seemed to cause other problems. The compiler bombed after this with the following:

sh-elf-gcc -mb -m2 -O2 -save-temps -Wall -Wpointer-arith -Wstrict-prototypes -Winline -Wundef -Woverloaded-virtual -ggdb -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions -fvtable-gc -finit-priority -Wa,-am  -I/ecos-d/Dev/eCos/builds/current/pc388_install/include -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current/src -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current/tests -I. -Wp,-MD,sh2_offsets.tmp -o var_mk_defs.tmp /ecos-d/ecos/ecos/packages/hal/sh/sh2/current/src/var_mk_defs.c
/opt/ecos_new/gnutools/sh-elf/bin/../lib/gcc-lib/sh-elf/3.2.1/../../../../sh-elf/bin/ld: cannot find -
make[1]: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/sh2/current'
lc
make: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build'
collect2: ld returned 1 exit status
make[1]: *** [/ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh2_offsets.inc] Error 1
make: *** [build] Error 2

I can only assume that this must be a compiler problem. I am currently building the latest gcc v3.3.1 for the sh-elf platform to see if this fixes the problem. Apart from that I am completely stuck.

	
		James Yates

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

RE: [ECOS] Problems with var_mk_defs.c

[Gary Thomas]

On Thu, 2003-10-16 at 01:56, James Yates wrote:
> I tried removing the hash, as a temporary fix i change DEFINE to:
> 
> #define DEFINE(sym, val) \
>         asm volatile("\n\t.equ\t" #sym ",0" : : "i" (val))
> 
> Which effectively then outputs :
> 
>  .equ	HAL_UCACHE_SIZE, 0
>  .loc 1 82 0
> 
> However, this seemed to cause other problems. The compiler bombed after this with the following:
> 
> sh-elf-gcc -mb -m2 -O2 -save-temps -Wall -Wpointer-arith -Wstrict-prototypes -Winline -Wundef -Woverloaded-virtual -ggdb -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions -fvtable-gc -finit-priority -Wa,-am  -I/ecos-d/Dev/eCos/builds/current/pc388_install/include -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current/src -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current/tests -I. -Wp,-MD,sh2_offsets.tmp -o var_mk_defs.tmp /ecos-d/ecos/ecos/packages/hal/sh/sh2/current/src/var_mk_defs.c
> /opt/ecos_new/gnutools/sh-elf/bin/../lib/gcc-lib/sh-elf/3.2.1/../../../../sh-elf/bin/ld: cannot find -
> make[1]: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/sh2/current'
> lc
> make: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build'
> collect2: ld returned 1 exit status
> make[1]: *** [/ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh2_offsets.inc] Error 1
> make: *** [build] Error 2
> 
> I can only assume that this must be a compiler problem. I am currently building the latest gcc v3.3.1 for the sh-elf platform to see if this fixes the problem. Apart from that I am completely stuck.
> 

Try running this command manually, and add the "-v" and "-save-temps" 
options.  The expanded output may provide some clues.

-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates


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

RE: [ECOS] Problems with var_mk_defs.c

[James Yates]

Gary,

When I compile the following is output:

make -r -C hal/sh/arch/current /ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh_offsets.inc
make[1]: Entering directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/arch/current'
sh-elf-gcc -mb -m2 -O2 -Wall -Wpointer-arith -Wstrict-prototypes -Winline -Wundef -Woverloaded-virtual -ggdb -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions -fvtable-gc -finit-priority  -I/ecos-d/Dev/eCos/builds/current/pc388_install/include -I/ecos-d/ecos/ecos/packages/hal/sh/arch/current -I/ecos-d/ecos/ecos/packages/hal/sh/arch/current/src -I/ecos-d/ecos/ecos/packages/hal/sh/arch/current/tests -I. -Wp,-MD,sh_offsets.tmp -o hal_mk_defs.tmp -S /ecos-d/ecos/ecos/packages/hal/sh/arch/current/src/hal_mk_defs.c
fgrep .equ hal_mk_defs.tmp | sed s/#// > /ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh_offsets.inc
make[1]: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/arch/current'
make -r -C hal/sh/sh2/current /ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh2_offsets.inc
make[1]: Entering directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/sh2/current'
sh-elf-gcc -mb -m2 -O2 -Wall -Wpointer-arith -Wstrict-prototypes -Winline -Wundef -Woverloaded-virtual -ggdb -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions -fvtable-gc -finit-priority  -I/ecos-d/Dev/eCos/builds/current/pc388_install/include -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current/src -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current/tests -I. -Wp,-MD,sh2_offsets.tmp -o var_mk_defs.tmp /ecos-d/ecos/ecos/packages/hal/sh/sh2/current/src/var_mk_defs.c
make[1]: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/sh2/current'
/ecos-c/temp/ccmMeZK9.s: Assembler messages:
make: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build'
/ecos-c/temp/ccmMeZK9.s:26: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:26: Warning: rest of line ignored; first ignored character is `0'
/ecos-c/temp/ccmMeZK9.s:29: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:29: Warning: rest of line ignored; first ignored character is `0'
/ecos-c/temp/ccmMeZK9.s:32: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:32: Warning: rest of line ignored; first ignored character is `0'
/ecos-c/temp/ccmMeZK9.s:35: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:35: Warning: rest of line ignored; first ignored character is `6'
/ecos-c/temp/ccmMeZK9.s:38: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:38: Warning: rest of line ignored; first ignored character is `1'
/ecos-c/temp/ccmMeZK9.s:41: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:41: Warning: rest of line ignored; first ignored character is `8'
/ecos-c/temp/ccmMeZK9.s:44: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:44: Warning: rest of line ignored; first ignored character is `0'
/ecos-c/temp/ccmMeZK9.s:47: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:47: Warning: rest of line ignored; first ignored character is `3'
/ecos-c/temp/ccmMeZK9.s:50: Error: bad expression
/ecos-c/temp/ccmMeZK9.s:50: Warning: rest of line ignored; first ignored character is `3'
make[1]: *** [/ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh2_offsets.inc] Error 1
make: *** [build] Error 2


For the code in var_mk_defs.c:
HAL_UCACHE_SIZE = 0, HAL_UCACHE_WAYS = 0, HAL_UCACHE_LINE_SIZE = 0, CYGNUM_HAL_ISR_MAX = 62 and so on.

As for my port, the processor it is for has a much larger vector table. The SH2 7044 variant I based this on has exception vectors for the custom interrupt layout going up to 93 in var_intr.h whereas the SH2 7145 hash exception vectors going up to 129. Could the size of the vector table be causing a problem.

The 7145 also has no cache so I have used another cache definition file mod_regs_cac_3.h which specifies cache sizes as 0 and all addresses. Could this also be contributing. I am pretty new to eCos and there is some funky stuff which I don't as yet understand. It has been quite hard work performing a variant and platform port for board.

   Any help greatfully received. Many thanks.

		James

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

RE: [ECOS] Problems with var_mk_defs.c

[Gary Thomas]

On Wed, 2003-10-15 at 09:03, James Yates wrote:
> Gary,
> 
> When I compile the following is output:
> 
> make -r -C hal/sh/arch/current /ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh_offsets.inc
> make[1]: Entering directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/arch/current'
> sh-elf-gcc -mb -m2 -O2 -Wall -Wpointer-arith -Wstrict-prototypes -Winline -Wundef -Woverloaded-virtual -ggdb -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions -fvtable-gc -finit-priority  -I/ecos-d/Dev/eCos/builds/current/pc388_install/include -I/ecos-d/ecos/ecos/packages/hal/sh/arch/current -I/ecos-d/ecos/ecos/packages/hal/sh/arch/current/src -I/ecos-d/ecos/ecos/packages/hal/sh/arch/current/tests -I. -Wp,-MD,sh_offsets.tmp -o hal_mk_defs.tmp -S /ecos-d/ecos/ecos/packages/hal/sh/arch/current/src/hal_mk_defs.c
> fgrep .equ hal_mk_defs.tmp | sed s/#// > /ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh_offsets.inc
> make[1]: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/arch/current'
> make -r -C hal/sh/sh2/current /ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh2_offsets.inc
> make[1]: Entering directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/sh2/current'
> sh-elf-gcc -mb -m2 -O2 -Wall -Wpointer-arith -Wstrict-prototypes -Winline -Wundef -Woverloaded-virtual -ggdb -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions -fvtable-gc -finit-priority  -I/ecos-d/Dev/eCos/builds/current/pc388_install/include -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current/src -I/ecos-d/ecos/ecos/packages/hal/sh/sh2/current/tests -I. -Wp,-MD,sh2_offsets.tmp -o var_mk_defs.tmp /ecos-d/ecos/ecos/packages/hal/sh/sh2/current/src/var_mk_defs.c
> make[1]: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build/hal/sh/sh2/current'
> /ecos-c/temp/ccmMeZK9.s: Assembler messages:
> make: Leaving directory `/ecos-d/Dev/eCos/builds/current/pc388_build'
> /ecos-c/temp/ccmMeZK9.s:26: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:26: Warning: rest of line ignored; first ignored character is `0'
> /ecos-c/temp/ccmMeZK9.s:29: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:29: Warning: rest of line ignored; first ignored character is `0'
> /ecos-c/temp/ccmMeZK9.s:32: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:32: Warning: rest of line ignored; first ignored character is `0'
> /ecos-c/temp/ccmMeZK9.s:35: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:35: Warning: rest of line ignored; first ignored character is `6'
> /ecos-c/temp/ccmMeZK9.s:38: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:38: Warning: rest of line ignored; first ignored character is `1'
> /ecos-c/temp/ccmMeZK9.s:41: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:41: Warning: rest of line ignored; first ignored character is `8'
> /ecos-c/temp/ccmMeZK9.s:44: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:44: Warning: rest of line ignored; first ignored character is `0'
> /ecos-c/temp/ccmMeZK9.s:47: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:47: Warning: rest of line ignored; first ignored character is `3'
> /ecos-c/temp/ccmMeZK9.s:50: Error: bad expression
> /ecos-c/temp/ccmMeZK9.s:50: Warning: rest of line ignored; first ignored character is `3'
> make[1]: *** [/ecos-d/Dev/eCos/builds/current/pc388_install/include/cyg/hal/sh2_offsets.inc] Error 1
> make: *** [build] Error 2
> 

Try looking at the generated *.s file.  My guess is that it has some
spaces which the compiler/assembler doesn't like.  Some prudent changes
to your include files may fix it.

> 
> For the code in var_mk_defs.c:
> HAL_UCACHE_SIZE = 0, HAL_UCACHE_WAYS = 0, HAL_UCACHE_LINE_SIZE = 0, CYGNUM_HAL_ISR_MAX = 62 and so on.
> 
> As for my port, the processor it is for has a much larger vector table. The SH2 7044 variant I based this on has exception vectors for the custom interrupt layout going up to 93 in var_intr.h whereas the SH2 7145 hash exception vectors going up to 129. Could the size of the vector table be causing a problem.
> 
> The 7145 also has no cache so I have used another cache definition file mod_regs_cac_3.h which specifies cache sizes as 0 and all addresses. Could this also be contributing. I am pretty new to eCos and there is some funky stuff which I don't as yet understand. It has been quite hard work performing a variant and platform port for board.
> 
>    Any help greatfully received. Many thanks.
> 
> 		James
-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates


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

[ECOS] Problems with var_mk_defs.c

[James Yates]

In the SH Hal architecture I am getting build problems relating to var_mk_defs.c.  If I comment out the code within int main(void), the compile problems go away. When included, every call to DEFINE throws up a bad expression. Can anyone explain to me what this file is actually doing, or indeed why it isn't working as I am completely stuck.

#define DEFINE(sym, val) \
        asm volatile("\n\t.equ\t" #sym ",%0" : : "i" (val))

int main(void)
{
    // Caching details
    DEFINE(HAL_UCACHE_SIZE, HAL_UCACHE_SIZE);
    DEFINE(HAL_UCACHE_WAYS, HAL_UCACHE_WAYS);
    DEFINE(HAL_UCACHE_LINE_SIZE, HAL_UCACHE_LINE_SIZE);

    // Interrupt details
    DEFINE(CYGNUM_HAL_ISR_MAX, CYGNUM_HAL_ISR_MAX);
    DEFINE(CYGNUM_HAL_INTERRUPT_LVL0, CYGNUM_HAL_INTERRUPT_LVL0);
    DEFINE(CYGNUM_HAL_INTERRUPT_LVL_MAX, CYGNUM_HAL_INTERRUPT_LVL_MAX);
#ifdef CYGNUM_HAL_INTERRUPT_LVL14
    DEFINE(CYGNUM_HAL_INTERRUPT_LVL14, CYGNUM_HAL_INTERRUPT_LVL14);
#endif
    DEFINE(CYGNUM_HAL_INTERRUPT_NMI, CYGNUM_HAL_INTERRUPT_NMI);
    DEFINE(CYGNUM_HAL_VECTOR_TRAP, CYGNUM_HAL_VECTOR_TRAP);
    DEFINE(CYGNUM_HAL_VECTOR_INTERRUPT, CYGNUM_HAL_VECTOR_INTERRUPT);

    return 0;
}


Many Thanks in advance

	James Yates

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

Re: [ECOS] Problems with var_mk_defs.c

[Gary Thomas]

On Wed, 2003-10-15 at 08:35, James Yates wrote:
> In the SH Hal architecture I am getting build problems relating to var_mk_defs.c.  
> If I comment out the code within int main(void), the compile problems go away. 
> When included, every call to DEFINE throws up a bad expression. Can anyone 
> explain to me what this file is actually doing, or indeed why it isn't working 
> as I am completely stuck.

What this program does is to use the target compiler to generate
machine dependent offsets within various structures used by eCos.
These are necessary for various assembly coded modules as they
have no other way to find data within objects (or other values
defined by the C/C++ environment).

We'd need to see the actual errors to be much more help.

-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates


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