On Thu, Sep 16, 2010 at 23:43, Zhang, Sonic wrote: >From: Mike Frysinger [mailto:[email protected]] >>On Thu, Sep 16, 2010 at 04:37, <[email protected]> wrote: >>> Revision 9142 Author sonicz Date 2010-09-16 04:37:37 -0400 (Thu, 16 >>> Sep >>> 2010) >>> >>> Log Message >>> >>> Fix bug[#6254] [#5805] access_ok: check address range accross border >>> properly. >>> >>> When init section is released to kernel memory management system, it >>> may be merged into the same buddy block if it is adjacent to generic >>> allocable memory region. In this case, an address range may >>> across the border. >> >>so why dont we update memory_start when this occurs ? doesnt >>it happen after free_init_pages() ? so at the end of that >>function, we should be able to set memory_start to >>__init_begin. then we dont need to modify process.c at all. > > The problem is that whether the init section is adjacent to the generic > allocable memory depends on the link script. You can't assume it always > is, although it is in current blackfin link script.
free_init_pages() is Blackfin-specific. look at arch/blackfin/mm/init.c. so i guess what you're saying is that we can do what i suggest ... -mike _______________________________________________ Linux-kernel-commits mailing list [email protected] https://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits
