In followup to the discussion about page protection on the Intel architecture, I got out a copy of the Linux kernel source and started poking through it. There's a comment in include/asm-i386/pgtable.h with all the #defines for the bit values in the protected mode pages which says that the x86 can't distinguish between execute and read for pages. It also says that write permissions imply read permissions. Looking at it more closely, I don't think that it is feasable to efficiently do either write-only or read-but-not-execute in the x86's page table system. Go get a copy of the latest revision of the Intel software developer's manual, section 3, and look at page 3-14, which lays out the protection rules for *segments*, not pages. There is no execute protection for individual pages, only write protection.
This archive was generated by hypermail 2b30 : Fri Apr 13 2001 - 15:14:31 PDT