public inbox for gcc-help@gcc.gnu.org
 help / color / mirror / Atom feed
* How to generate a PC- relative load instruction.
@ 2009-04-09 11:25 sumanth
  2009-04-09 16:26 ` Ian Lance Taylor
  0 siblings, 1 reply; 4+ messages in thread
From: sumanth @ 2009-04-09 11:25 UTC (permalink / raw)
  To: gcc-help

Hi,
  
    How can I  generate a load instruction with pc-relative accessing of 
"global" variables.
    As far as I know I need to modify GO_IF_LEGITIMATE_ADDRESS
    X is an RTX ; 1. GET_CODE (X) == PC     + offset
                           2. GET_CODE (X) == LABEL_REF  or  GET_CODE 
(X) == SYMBOL_REF
   and split the label to PC + offset (if possible)  .

  The problem here is how can I keep data in PC + offset and tell my 
compiler. The only reference is in arm  as far as I know.
  Its some thing like
   "a = 10 " is a global variable  and i need to load its address it in 
to register r0.
  120    ld r0(destination), pc ,#4
  124    0x2345678e
  value 10 is in 0x2345678e

Can any one help me in this regard .

Thanks in advance.

Thanks,
Sumanth


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: How to generate a PC- relative load instruction.
  2009-04-09 11:25 How to generate a PC- relative load instruction sumanth
@ 2009-04-09 16:26 ` Ian Lance Taylor
  2009-04-10 11:10   ` sumanth
  0 siblings, 1 reply; 4+ messages in thread
From: Ian Lance Taylor @ 2009-04-09 16:26 UTC (permalink / raw)
  To: sumanth; +Cc: gcc-help

sumanth <sumanth.gundapneni@redpinesignals.com> writes:

>   How can I  generate a load instruction with pc-relative accessing of
> "global" variables.
>    As far as I know I need to modify GO_IF_LEGITIMATE_ADDRESS
>    X is an RTX ; 1. GET_CODE (X) == PC     + offset
>                           2. GET_CODE (X) == LABEL_REF  or  GET_CODE
> (X) == SYMBOL_REF
>   and split the label to PC + offset (if possible)  .
>
>  The problem here is how can I keep data in PC + offset and tell my
> compiler. The only reference is in arm  as far as I know.
>  Its some thing like
>   "a = 10 " is a global variable  and i need to load its address it in
> to register r0.
>  120    ld r0(destination), pc ,#4
>  124    0x2345678e
>  value 10 is in 0x2345678e
>
> Can any one help me in this regard .

This sort of thing is often done in the LEGITIMIZE_ADDRESS macro.  Your
case looks a bit complicated and probably requires an UNSPEC.  You will
also want to consider a constant pool.  I think the SH has some cases
like yours, but I'm not sure.

Ian

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: How to generate a PC- relative load instruction.
  2009-04-09 16:26 ` Ian Lance Taylor
@ 2009-04-10 11:10   ` sumanth
  2009-04-10 14:19     ` Ian Lance Taylor
  0 siblings, 1 reply; 4+ messages in thread
From: sumanth @ 2009-04-10 11:10 UTC (permalink / raw)
  To: Ian Lance Taylor; +Cc: gcc-help

hi,
     does this one(idea) related to program independent code (pic).  

Thanks,
Sumanth
Ian Lance Taylor wrote:
> sumanth <sumanth.gundapneni@redpinesignals.com> writes:
>
>   
>>   How can I  generate a load instruction with pc-relative accessing of
>> "global" variables.
>>    As far as I know I need to modify GO_IF_LEGITIMATE_ADDRESS
>>    X is an RTX ; 1. GET_CODE (X) == PC     + offset
>>                           2. GET_CODE (X) == LABEL_REF  or  GET_CODE
>> (X) == SYMBOL_REF
>>   and split the label to PC + offset (if possible)  .
>>
>>  The problem here is how can I keep data in PC + offset and tell my
>> compiler. The only reference is in arm  as far as I know.
>>  Its some thing like
>>   "a = 10 " is a global variable  and i need to load its address it in
>> to register r0.
>>  120    ld r0(destination), pc ,#4
>>  124    0x2345678e
>>  value 10 is in 0x2345678e
>>
>> Can any one help me in this regard .
>>     
>
> This sort of thing is often done in the LEGITIMIZE_ADDRESS macro.  Your
> case looks a bit complicated and probably requires an UNSPEC.  You will
> also want to consider a constant pool.  I think the SH has some cases
> like yours, but I'm not sure.
>
> Ian
>
>
>   


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: How to generate a PC- relative load instruction.
  2009-04-10 11:10   ` sumanth
@ 2009-04-10 14:19     ` Ian Lance Taylor
  0 siblings, 0 replies; 4+ messages in thread
From: Ian Lance Taylor @ 2009-04-10 14:19 UTC (permalink / raw)
  To: sumanth; +Cc: gcc-help

sumanth <sumanth.gundapneni@redpinesignals.com> writes:

> Ian Lance Taylor wrote:
>>
>> This sort of thing is often done in the LEGITIMIZE_ADDRESS macro.  Your
>> case looks a bit complicated and probably requires an UNSPEC.  You will
>> also want to consider a constant pool.  I think the SH has some cases
>> like yours, but I'm not sure.
>>
>> Ian
>
>     does this one(idea) related to program independent code (pic).  

PIC is often implemented by creating an UNSPEC in LEGITIMIZE_ADDRESS or
in LEGITIMIZE_PIC_ADDRESS.  See many existing examples.

Ian

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2009-04-10 14:19 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-04-09 11:25 How to generate a PC- relative load instruction sumanth
2009-04-09 16:26 ` Ian Lance Taylor
2009-04-10 11:10   ` sumanth
2009-04-10 14:19     ` Ian Lance Taylor

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).