[RFA] mips-o64-extract-return-value

[RFA] mips-o64-extract-return-value

[Michael Snyder]

[-- Attachment #1: Type: text/plain, Size: 416 bytes --]

Hi Andrew,

This change fixes 100s of FAILs for mips64-elf, 'cause gdb can't
find the function's return value.  It follows some work that you
were apparently doing w.r.t. the mips internal register representation.

I suspect that the same thing needs to be done for mips_eabi_extract...,
but I haven't tested that.  What do you think?

I'll probably try doing something similar for store_return_value too.

Michael


[-- Attachment #2: extract --]
[-- Type: text/plain, Size: 1165 bytes --]

2003-11-19  Michael Snyder  <msnyder@redhat.com>

	* mips-tdep.c (mips_o64_extract_return_value):
	Correct for raw reg / virtual reg scheme.

Index: mips-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/mips-tdep.c,v
retrieving revision 1.254
diff -p -r1.254 mips-tdep.c
*** mips-tdep.c	16 Nov 2003 22:46:49 -0000	1.254
--- mips-tdep.c	20 Nov 2003 01:00:31 -0000
*************** mips_o64_extract_return_value (struct ty
*** 4681,4692 ****
    return_value_location (valtype, &hi, &lo);
  
    memcpy (valbuf + lo.buf_offset,
! 	  regbuf + DEPRECATED_REGISTER_BYTE (lo.reg) + lo.reg_offset,
  	  lo.len);
  
    if (hi.len > 0)
      memcpy (valbuf + hi.buf_offset,
! 	    regbuf + DEPRECATED_REGISTER_BYTE (hi.reg) + hi.reg_offset,
  	    hi.len);
  }
  
--- 4681,4692 ----
    return_value_location (valtype, &hi, &lo);
  
    memcpy (valbuf + lo.buf_offset,
! 	  regbuf + DEPRECATED_REGISTER_BYTE (NUM_REGS + lo.reg) + lo.reg_offset,
  	  lo.len);
  
    if (hi.len > 0)
      memcpy (valbuf + hi.buf_offset,
! 	    regbuf + DEPRECATED_REGISTER_BYTE (NUM_REGS + hi.reg) + hi.reg_offset,
  	    hi.len);
  }
  

Re: [RFA] mips-o64-extract-return-value

[Andrew Cagney]

> Hi Andrew,
> 
> This change fixes 100s of FAILs for mips64-elf, 'cause gdb can't
> find the function's return value.  It follows some work that you
> were apparently doing w.r.t. the mips internal register representation.
> 
> I suspect that the same thing needs to be done for mips_eabi_extract...,
> but I haven't tested that.  What do you think?

Ok.

> I'll probably try doing something similar for store_return_value too.

Rather than doing the same thing for store_return_value, take a look at 
the PPC's "ppc_sysv_abi_return_value".

BTW, rumour has it that GCC finally fixed MIPS struct return.
http://gcc.gnu.org/gcc-3.4/mips-abi.html

Andrew

Re: [RFA] mips-o64-extract-return-value

[Michael Snyder]

Andrew Cagney wrote:
>> Hi Andrew,
>>
>> This change fixes 100s of FAILs for mips64-elf, 'cause gdb can't
>> find the function's return value.  It follows some work that you
>> were apparently doing w.r.t. the mips internal register representation.
>>
>> I suspect that the same thing needs to be done for mips_eabi_extract...,
>> but I haven't tested that.  What do you think?
> 
> 
> Ok.
> 
>> I'll probably try doing something similar for store_return_value too.
> 
> 
> Rather than doing the same thing for store_return_value, take a look at 
> the PPC's "ppc_sysv_abi_return_value".

Grep grep...  ah, you mean ppc_sysv_abi_STORE_return_value.
Shall I fix your change log entries for you?    ;-)

Re: [RFA] mips-o64-extract-return-value

[Andrew Cagney]

> Andrew Cagney wrote:
> Hi Andrew,
> 
> This change fixes 100s of FAILs for mips64-elf, 'cause gdb can't
> find the function's return value.  It follows some work that you
> were apparently doing w.r.t. the mips internal register representation.
> 
> I suspect that the same thing needs to be done for mips_eabi_extract...,
> but I haven't tested that.  What do you think?
> 
> 
> Ok.
> 
> I'll probably try doing something similar for store_return_value too.
> 
> 
> Rather than doing the same thing for store_return_value, take a look at the PPC's "ppc_sysv_abi_return_value".
> 
> Grep grep...  ah, you mean ppc_sysv_abi_STORE_return_value.
> Shall I fix your change log entries for you?    ;-)

no?

$ grep ppc_sysv_abi_store_return_value ChangeLog
         (ppc_sysv_abi_store_return_value): Delete function.
         (ppc_sysv_abi_store_return_value): Delete.
         * ppc-tdep.h: (ppc_sysv_abi_store_return_value): Declare.
         (ppc_sysv_abi_store_return_value): New function.
$ grep ppc_sysv_store_return_value *.c

In the MIPS, take a look at n32/n64 and o32 which I previously cleaned 
up (probably time to s/ xfer_return_value / return_value /).  That just 
leaves o64 and eabi (which are still riddled with deprecated code) :-(

Andrew

Re: [RFA] mips-o64-extract-return-value

[Michael Snyder]

Andrew Cagney wrote:
>> Andrew Cagney wrote:
>> Hi Andrew,
>>
>> This change fixes 100s of FAILs for mips64-elf, 'cause gdb can't
>> find the function's return value.  It follows some work that you
>> were apparently doing w.r.t. the mips internal register representation.
>>
>> I suspect that the same thing needs to be done for mips_eabi_extract...,
>> but I haven't tested that.  What do you think?
>>
>>
>> Ok.
>>
>> I'll probably try doing something similar for store_return_value too.
>>
>>
>> Rather than doing the same thing for store_return_value, take a look 
>> at the PPC's "ppc_sysv_abi_return_value".
>>
>> Grep grep...  ah, you mean ppc_sysv_abi_STORE_return_value.
>> Shall I fix your change log entries for you?    ;-)
> 
> 
> no?
> 
> $ grep ppc_sysv_abi_store_return_value ChangeLog
>         (ppc_sysv_abi_store_return_value): Delete function.
>         (ppc_sysv_abi_store_return_value): Delete.
>         * ppc-tdep.h: (ppc_sysv_abi_store_return_value): Declare.
>         (ppc_sysv_abi_store_return_value): New function.

Right -- I meant *these* changelog entries:

[msnyder@reddwarf gdb]$ grep sysv_abi_return_value ChangeLog
	* ppc-tdep.h (ppc_sysv_abi_return_value): Ditto.
	(ppc64_sysv_abi_return_value): Ditto.
	(ppc_sysv_abi_return_value): Ditto.
	(ppc64_sysv_abi_return_value): Ditto.
	* ppc-sysv-tdep.c (ppc_sysv_abi_return_value): New function.
	* ppc-tdep.h (ppc64_sysv_abi_return_value): Declare.
	(ppc_sysv_abi_return_value): Declare.
	(ppc64_sysv_abi_return_value): New function.

Re: [RFA] mips-o64-extract-return-value

[Andrew Cagney]

>>>> Rather than doing the same thing for store_return_value, take a look at the PPC's "ppc_sysv_abi_return_value".
>>> 
>>> Grep grep...  ah, you mean ppc_sysv_abi_STORE_return_value.
>>> Shall I fix your change log entries for you?    ;-)
>>> 
>>> 
> 
>> no?
>> 
>> $ grep ppc_sysv_abi_store_return_value ChangeLog
>>         (ppc_sysv_abi_store_return_value): Delete function.
>>         (ppc_sysv_abi_store_return_value): Delete.
>>         * ppc-tdep.h: (ppc_sysv_abi_store_return_value): Declare.
>>         (ppc_sysv_abi_store_return_value): New function.
>> 

> Right -- I meant *these* changelog entries:
> 
> [msnyder@reddwarf gdb]$ grep sysv_abi_return_value ChangeLog
>     * ppc-tdep.h (ppc_sysv_abi_return_value): Ditto.
>     (ppc64_sysv_abi_return_value): Ditto.
>     (ppc_sysv_abi_return_value): Ditto.
>     (ppc64_sysv_abi_return_value): Ditto.
>     * ppc-sysv-tdep.c (ppc_sysv_abi_return_value): New function.
>     * ppc-tdep.h (ppc64_sysv_abi_return_value): Declare.
>     (ppc_sysv_abi_return_value): Declare.
>     (ppc64_sysv_abi_return_value): New function.

Sorry, I'm lost.  What's wrong with those ChangeLogs?

Andrew

Re: [RFA] mips-o64-extract-return-value

[Michael Snyder]

Andrew Cagney wrote:
> 
>>>>> Rather than doing the same thing for store_return_value, take a 
>>>>> look at the PPC's "ppc_sysv_abi_return_value".
>>>>
>>>>
>>>> Grep grep...  ah, you mean ppc_sysv_abi_STORE_return_value.
>>>> Shall I fix your change log entries for you?    ;-)
>>>>
>>>>
>>
>>> no?
>>>
>>> $ grep ppc_sysv_abi_store_return_value ChangeLog
>>>         (ppc_sysv_abi_store_return_value): Delete function.
>>>         (ppc_sysv_abi_store_return_value): Delete.
>>>         * ppc-tdep.h: (ppc_sysv_abi_store_return_value): Declare.
>>>         (ppc_sysv_abi_store_return_value): New function.
>>>
> 
>> Right -- I meant *these* changelog entries:
>>
>> [msnyder@reddwarf gdb]$ grep sysv_abi_return_value ChangeLog
>>     * ppc-tdep.h (ppc_sysv_abi_return_value): Ditto.
>>     (ppc64_sysv_abi_return_value): Ditto.
>>     (ppc_sysv_abi_return_value): Ditto.
>>     (ppc64_sysv_abi_return_value): Ditto.
>>     * ppc-sysv-tdep.c (ppc_sysv_abi_return_value): New function.
>>     * ppc-tdep.h (ppc64_sysv_abi_return_value): Declare.
>>     (ppc_sysv_abi_return_value): Declare.
>>     (ppc64_sysv_abi_return_value): New function.
> 
> 
> Sorry, I'm lost.  What's wrong with those ChangeLogs?

s/abi_return_value/abi_store_return_value/g

Re: [RFA] mips-o64-extract-return-value

[Andrew Cagney]

> 
> Right -- I meant *these* changelog entries:
> 
> [msnyder@reddwarf gdb]$ grep sysv_abi_return_value ChangeLog
>     * ppc-tdep.h (ppc_sysv_abi_return_value): Ditto.
>     (ppc64_sysv_abi_return_value): Ditto.
>     (ppc_sysv_abi_return_value): Ditto.
>     (ppc64_sysv_abi_return_value): Ditto.
>     * ppc-sysv-tdep.c (ppc_sysv_abi_return_value): New function.
>     * ppc-tdep.h (ppc64_sysv_abi_return_value): Declare.
>     (ppc_sysv_abi_return_value): Declare.
>     (ppc64_sysv_abi_return_value): New function.
> 
> 
> Sorry, I'm lost.  What's wrong with those ChangeLogs?
> 
> s/abi_return_value/abi_store_return_value/g 

What?

Re: [RFA] mips-o64-extract-return-value

[Michael Snyder]

Andrew Cagney wrote:
>>
>> Right -- I meant *these* changelog entries:
>>
>> [msnyder@reddwarf gdb]$ grep sysv_abi_return_value ChangeLog
>>     * ppc-tdep.h (ppc_sysv_abi_return_value): Ditto.
>>     (ppc64_sysv_abi_return_value): Ditto.
>>     (ppc_sysv_abi_return_value): Ditto.
>>     (ppc64_sysv_abi_return_value): Ditto.
>>     * ppc-sysv-tdep.c (ppc_sysv_abi_return_value): New function.
>>     * ppc-tdep.h (ppc64_sysv_abi_return_value): Declare.
>>     (ppc_sysv_abi_return_value): Declare.
>>     (ppc64_sysv_abi_return_value): New function.
>>
>>
>> Sorry, I'm lost.  What's wrong with those ChangeLogs?
>>
>> s/abi_return_value/abi_store_return_value/g 
> 
> 
> What?

;-)
Honest, I didn't mean to make a federal case out of this --
your original suggestion to me was to look for a function called
"ppc_sysv_abi_return_value".  That function doesn't exist -- but
there are several references to it in the changelog.  I eventually
figured out that it was missing the word "store_", as in
"store_return_value".  So I asked if you'd like me to fix up
the change log entries, so they would match the actual function name.

Sorry for the confusion...

Re: [RFA] mips-o64-extract-return-value

[Daniel Jacobowitz]

On Fri, Nov 21, 2003 at 05:00:57PM -0800, Michael Snyder wrote:
> Andrew Cagney wrote:
> >>
> >>Right -- I meant *these* changelog entries:
> >>
> >>[msnyder@reddwarf gdb]$ grep sysv_abi_return_value ChangeLog
> >>    * ppc-tdep.h (ppc_sysv_abi_return_value): Ditto.
> >>    (ppc64_sysv_abi_return_value): Ditto.
> >>    (ppc_sysv_abi_return_value): Ditto.
> >>    (ppc64_sysv_abi_return_value): Ditto.
> >>    * ppc-sysv-tdep.c (ppc_sysv_abi_return_value): New function.
> >>    * ppc-tdep.h (ppc64_sysv_abi_return_value): Declare.
> >>    (ppc_sysv_abi_return_value): Declare.
> >>    (ppc64_sysv_abi_return_value): New function.
> >>
> >>
> >>Sorry, I'm lost.  What's wrong with those ChangeLogs?
> >>
> >>s/abi_return_value/abi_store_return_value/g 
> >
> >
> >What?
> 
> ;-)
> Honest, I didn't mean to make a federal case out of this --
> your original suggestion to me was to look for a function called
> "ppc_sysv_abi_return_value".  That function doesn't exist -- but
> there are several references to it in the changelog.  I eventually
> figured out that it was missing the word "store_", as in
> "store_return_value".  So I asked if you'd like me to fix up
> the change log entries, so they would match the actual function name.
> 
> Sorry for the confusion...

Try current CVS?

ppc-sysv-tdep.c:ppc_sysv_abi_return_value

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [RFA] mips-o64-extract-return-value

[Michael Snyder]

Daniel Jacobowitz wrote:
> On Fri, Nov 21, 2003 at 05:00:57PM -0800, Michael Snyder wrote:
> 
>>Andrew Cagney wrote:
>>
>>>>Right -- I meant *these* changelog entries:
>>>>
>>>>[msnyder@reddwarf gdb]$ grep sysv_abi_return_value ChangeLog
>>>>   * ppc-tdep.h (ppc_sysv_abi_return_value): Ditto.
>>>>   (ppc64_sysv_abi_return_value): Ditto.
>>>>   (ppc_sysv_abi_return_value): Ditto.
>>>>   (ppc64_sysv_abi_return_value): Ditto.
>>>>   * ppc-sysv-tdep.c (ppc_sysv_abi_return_value): New function.
>>>>   * ppc-tdep.h (ppc64_sysv_abi_return_value): Declare.
>>>>   (ppc_sysv_abi_return_value): Declare.
>>>>   (ppc64_sysv_abi_return_value): New function.
>>>>
>>>>
>>>>Sorry, I'm lost.  What's wrong with those ChangeLogs?
>>>>
>>>>s/abi_return_value/abi_store_return_value/g 
>>>
>>>
>>>What?
>>
>>;-)
>>Honest, I didn't mean to make a federal case out of this --
>>your original suggestion to me was to look for a function called
>>"ppc_sysv_abi_return_value".  That function doesn't exist -- but
>>there are several references to it in the changelog.  I eventually
>>figured out that it was missing the word "store_", as in
>>"store_return_value".  So I asked if you'd like me to fix up
>>the change log entries, so they would match the actual function name.
>>
>>Sorry for the confusion...
> 
> 
> Try current CVS?
> 
> ppc-sysv-tdep.c:ppc_sysv_abi_return_value

D'oh, you're right, my bad.  Sorry Andrew -- you must have
thought I was nuts...

Re: [RFA] mips-o64-extract-return-value

[Michael Snyder]

[-- Attachment #1: Type: text/plain, Size: 426 bytes --]

Andrew Cagney wrote:
>> Hi Andrew,
>>
>> This change fixes 100s of FAILs for mips64-elf, 'cause gdb can't
>> find the function's return value.  It follows some work that you
>> were apparently doing w.r.t. the mips internal register representation.
>>
>> I suspect that the same thing needs to be done for mips_eabi_extract...,
>> but I haven't tested that.  What do you think?
> 
> 
> Ok.

Amended and committed as follows:


[-- Attachment #2: mips --]
[-- Type: text/plain, Size: 1968 bytes --]

2003-11-25  Michael Snyder  <msnyder@redhat.com>

        * mips-tdep.c (mips_o64_extract_return_value): Correct for
	NUM_REGS (see changes 2003-06-21).
	(mips_eabi_extract_return_value): Ditto.

Index: mips-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/mips-tdep.c,v
retrieving revision 1.261
diff -p -r1.261 mips-tdep.c
*** mips-tdep.c	23 Nov 2003 21:49:12 -0000	1.261
--- mips-tdep.c	25 Nov 2003 22:14:08 -0000
*************** mips_eabi_extract_return_value (struct t
*** 4696,4707 ****
    return_value_location (valtype, &hi, &lo);
  
    memcpy (valbuf + lo.buf_offset,
! 	  regbuf + DEPRECATED_REGISTER_BYTE (lo.reg) + lo.reg_offset,
  	  lo.len);
  
    if (hi.len > 0)
      memcpy (valbuf + hi.buf_offset,
! 	    regbuf + DEPRECATED_REGISTER_BYTE (hi.reg) + hi.reg_offset,
  	    hi.len);
  }
  
--- 4696,4707 ----
    return_value_location (valtype, &hi, &lo);
  
    memcpy (valbuf + lo.buf_offset,
! 	  regbuf + DEPRECATED_REGISTER_BYTE (NUM_REGS + lo.reg) + lo.reg_offset,
  	  lo.len);
  
    if (hi.len > 0)
      memcpy (valbuf + hi.buf_offset,
! 	    regbuf + DEPRECATED_REGISTER_BYTE (NUM_REGS + hi.reg) + hi.reg_offset,
  	    hi.len);
  }
  
*************** mips_o64_extract_return_value (struct ty
*** 4715,4726 ****
    return_value_location (valtype, &hi, &lo);
  
    memcpy (valbuf + lo.buf_offset,
! 	  regbuf + DEPRECATED_REGISTER_BYTE (lo.reg) + lo.reg_offset,
  	  lo.len);
  
    if (hi.len > 0)
      memcpy (valbuf + hi.buf_offset,
! 	    regbuf + DEPRECATED_REGISTER_BYTE (hi.reg) + hi.reg_offset,
  	    hi.len);
  }
  
--- 4715,4726 ----
    return_value_location (valtype, &hi, &lo);
  
    memcpy (valbuf + lo.buf_offset,
! 	  regbuf + DEPRECATED_REGISTER_BYTE (NUM_REGS + lo.reg) + lo.reg_offset,
  	  lo.len);
  
    if (hi.len > 0)
      memcpy (valbuf + hi.buf_offset,
! 	    regbuf + DEPRECATED_REGISTER_BYTE (NUM_REGS + hi.reg) + hi.reg_offset,
  	    hi.len);
  }