On Mon, Feb 22, 2010 at 08:20:09PM +0100, Uwe Kleine-K?nig wrote:

On Mon, Feb 22, 2010 at 11:36:24PM +0530, Rabin Vincent wrote:

quoted

On Sat, Feb 13, 2010 at 09:37:48PM +0100, Uwe Kleine-K?nig wrote:

quoted

On Sun, Feb 14, 2010 at 01:18:30AM +0530, Rabin Vincent wrote:

quoted

+ * With both the mcount types, we need to restore the original lr before
+ * returning.  In the __gnu_mcount_nc, version we're allowed to clobber ip.
+ * No other registers should be clobbered.
+ */

Very nice.

Maybe make the last two sentences:

In the __gnu_mcount_nc case the ip register is clobbered which is OK as
the calling convention for ARM allow clobbering this value for
subroutines and it doesn't contain parameters.

Won't quoting calling conventions here be misleading?  The mcounts don't
following normal convention for the other bits (we can't clobber even
the registers that are normally caller-saved, we need to restore lr,
etc.)

actually mcount is a function in the middle of a function call.  So it
must be transparent for both the caller and the callee.  As the called
function needs to know the return address, mcount must not clobber it.
And as ip is don't care and can be clobbered mcount can use it as it
likes.

OK.  How about this for the last paragraph:

 mcount can be thought of as a function called in the middle of a
 subroutine call.  As such, it needs to be transparent for both the
 caller and the callee: the original lr needs to be restored when
 leaving mcount, and no registers should be clobbered.  (In the
 __gnu_mcount_nc implementation, we clobber the ip register.  This
 is OK because the ARM calling convention allows it to be clobbered
 in subroutines and doesn't use it to hold parameters.)

Rabin