Hi Laurent This pagraph exactly is a cut&paste of the Motorola "Programmer's reference manual". I didn't mean to add it as commits to the patch.I just want to introduce this instruction to the reviewer. So don't worry about the copyrights.
Best Regards Guo Lei > > > > > Le 7 novembre 2014 à 11:14, "Guo, Lei" <guol-f...@cn.fujitsu.com> a écrit : > > > > > > This patch aims to add CMP2 instruction for m68k family. > > > > > > > > Description: Compares the value in Rn to each bound. The effective > > address contains the > > > > bounds pair: upper bound following the lower bound. For signed > > comparisons, the > > > > arithmetically smaller value should be used as the lower bound. For > > unsigned > > > > comparisons, the logically smaller value should be the lower bound. > > > > The size of the data and the bounds can be specified as byte, word, or > > long. > > If Rn is a > > > > data register and the operation size is byte or word, only the > > appropriate low-order part > > > > of Rn is checked. If Rn is an address register and the operation > > size is byte or word, > > > > the bounds operands are sign-extended to 32 bits, and the resultant > > operands are > > > > compared to the full 32 bits of An. > > > > If the upper bound equals the lower bound, the valid range is a single > > value. > > > > Moreover, this seems a cut&paste of the Motorola "Programmer's > reference manual". > It should be copyrighted : have we the right to copy this part in a > commit message ? > > Regards, > Laurent