[pcpro17]

Hello. I'm trying to read the Cycle Counter register in an Intel Pentium processor. I'm trying to use gcc and AT&T inline assembly.

The Pentium assembly instruction RDTSC gets the 64-bit value in the cycle counter register and puts it in EAX (low 32-bits) and EDX (high 32-bits).

Can someone please verify for me that the following code will save this value properly?

   unsigned long      low_ticks;            // stores low-order cycle count
   unsigned long      high_ticks;           // stores high-order cycle count
 
   __asm( " push %eax" );
   __asm( " push %edx" );
   __asm( " rdtsc" : "=A" (low_ticks), "=D" (high_ticks) );
   __asm( " pop  %edx" );
   __asm( " pop  %eax" );

Thanks.

[IMM]

anti-debugging ?

[Chicon]

Hi pcpro17,

In my opinion, if you're invoking an assembly routine from a C/C++ program, the push and pop instructions are not necessary.
You may simply use the following notation :

__asm__ volatile("rdtsc" : "=a" (low_ticks), "=d" (high_ticks));

[pcpro17]

Originally Posted by IMM

anti-debugging ?

No, I'm using gcc, so there's no built in debugger like the MS products.

Thanks for the tip, Chicon! I'll let you know how it works out.

[pcpro17]

Originally Posted by Chicon

Hi pcpro17,

In my opinion, if you're invoking an assembly routine from a C/C++ program, the push and pop instructions are not necessary.
You may simply use the following notation :

__asm__ volatile("rdtsc" : "=a" (low_ticks), "=d" (high_ticks));

Thanks! It worked wonderfully! Do you think the problem was with my asm statement "__asm" versus "__asm__ volatile" or in the capitalization of the "=A" and "=D"?

[IMM]

Originally Posted by pcpro17

No, I'm using gcc, so there's no built in debugger like the MS products.

Was actually wondering if you were looking for 0xffff in the register
Probably a bad guess on my part - but that instruction is sometimes used to detect the presence of a debugger and exit or similar.

[Chicon]

Originally Posted by pcpro17

Thanks! It worked wonderfully! Do you think the problem was with my asm statement "__asm" versus "__asm__ volatile" or in the capitalization of the "=A" and "=D"?

No, the command __asm seems to be specific to Microsoft (link to MSDN).
I only used the syntax that fits to gcc (link to gcc.gnu).

[pcpro17]

Originally Posted by Chicon

No, the command __asm seems to be specific to Microsoft (link to MSDN).
I only used the syntax that fits to gcc (link to gcc.gnu).

Oic, Thanks!