RE: [arm-gnu] Assembler error for Inline Assembly (cortex-A8 processor)
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [arm-gnu] Assembler error for Inline Assembly (cortex-A8 processor)




> -----Original Message-----
> From: Akshay Joshi [mailto:akshay.joshi@xxxxxxxxxxx] 
> Sent: Friday, November 21, 2008 9:39 AM
> To: Laurent Desnogues
> Cc: arm-gnu@xxxxxxxxxxxxxxxx
> Subject: RE: [arm-gnu] Assembler error for Inline Assembly 
> (cortex-A8 processor)
> 
> >>>> Try adding -mfloat-abi=softfp.
> 
> >>>>> Laurent
> 
> tried this option but with no luck.
> 
> 
> >>>>Can you try adding this option as well. '-march=armv7a'
> 
> >>>>-Santosh
> 
> added -march=armv7-a too. no luck here either...
> 
> Has anyone tried using NEON regs in inline assembly w/o 
> facing any problems?
> I'm using the latest toolchain (2008q3)...

I have tried instructions like 'vadd,vmov,vzip,vqmovn' using NEON registers without
any problems in inline assembly.


 
> 
> 
> -----Original Message-----
> From: Laurent Desnogues [mailto:laurent.desnogues@xxxxxxxxx]
> Sent: Thursday, November 20, 2008 2:20 PM
> To: akshay.joshi@xxxxxxxxxxx
> Cc: arm-gnu@xxxxxxxxxxxxxxxx
> Subject: Re: [arm-gnu] Assembler error for Inline Assembly (cortex-A8
> processor)
> 
> 
> On Wed, Nov 19, 2008 at 4:15 PM, Akshay Joshi 
> <akshay.joshi@xxxxxxxxxxx>
> wrote:
> >
> > For example:
> >         /tmp/cc4wfCkh.s:169: Error: bad instruction 
> `vld1.32 d0[],[r7]!'
> > The syntax used for the above instruction in the ".c" file 
> was as follows:
> >
> >      asm ("vld1.32 d0[], [r7]!");
> > But, if I use the same instruction in a ".s" file, the 
> error does not
> occur.
> >          vld1.32 d0[], [r7]!
> > Is there any option that I need to pass to the assembler / 
> compiler? Or,
> is
> > my inline assembly syntax incorrect?
> >
> > Currently, I am using the following options while compiling 
> the code:
> >
> >          -mcpu=cortex-a8 -mfpu=neon -O3  -fpack-struct 
> -funroll-loops
> > -freorder-blocks-and-partition -lm -lc -lgcc
> 
> Try adding -mfloat-abi=softfp.
> 
> 
> Laurent
> 
> 
>