Converting to VFP assembler

Question

Hi, i've had a go at converting this little bit of C code to ARM vfp assembler can anyone see any problems with it or possible improvements

for(int j=0;j<count; j++)

{

  output += array[j]*(*p++);

}

to this VFM assembly:

for(int j=0;j<count;j++){
               asm volatile(
                    VFP_VECTOR_LENGTH(8)
                    "fldmias  %0, {s0}    \n\t"     
                    "fldmias  %1, {s1}    \n\t"
                    "mov r2,%2            \n\t"
                    "mov r3,%3            \n\t" 
                    "mov r4,%4,lsr #3     \n\t"   
                    "beq one                  \n\t"

                    "0:                   \n\t"
                    "fldmias  r2!, {s8-s15}   \n\t"
                    "fldmias  r3!, {s16-s23}   \n\t"
                    "fmacs s0, s8, s16        \n\t"
                    "subs r4,r4, #1         \n\t"
                    "bne 0b                 \n\t"
                    "one:                       \n\t"
                    "mov r4,%4               \n\t"
                    "ands r4,r4,#7           \n\t"
                    "beq three                  \n\t"
                    "two:                       \n\t"       
                    "fldmias  r2!, {s2}   \n\t"
                    "fldmias  r3!, {s3}   \n\t"
                    "fmacs  s0,s2,s3        \n\t"
                    "subs r4,r4, #1         \n\t"
                    "bne two                    \n\t"       
                    "three:                     \n\t"
                    VFP_VECTOR_LENGTH_ZERO      
                    : "=r" (output)
                    : "r" (output), "r" (array), "r" (p),"r" (count)
                    : "r0", "cc", "memory" 
                );

}