On Fri, Oct 9, 2009 at 02:57, graff.yang wrote: > On Thu, 2009-10-01 at 01:01 -0400, Mike Frysinger wrote: >> On Mon, Sep 21, 2009 at 07:51, <[email protected]> wrote: >> > Modified Paths >> > >> > trunk/arch/blackfin/mach-bf518/include/mach/cdefBF51x_base.h >> > trunk/arch/blackfin/mach-bf527/include/mach/cdefBF52x_base.h >> > trunk/arch/blackfin/mach-bf533/include/mach/cdefBF532.h >> > trunk/arch/blackfin/mach-bf537/include/mach/cdefBF534.h >> > trunk/arch/blackfin/mach-bf538/include/mach/cdefBF538.h >> > trunk/arch/blackfin/mach-bf548/include/mach/cdefBF54x_base.h >> > trunk/arch/blackfin/mach-bf561/include/mach/cdefBF561.h >> > >> > Added Paths >> > >> > trunk/arch/blackfin/include/asm/cdef_misc.h >> >> merging duplicate code is great, but i dont think a new cdef_misc.h > > The definitions(like bfin_write_PLL_CTL()) in cdef_misc.h are using low > level definitions from mach/cdefBF5xx.h and asm/cdef_LPBlackfin.h, > further more, it need asm/irq.h, which included more header files. > In SMP case, there are many compile errors because of fetching in > asm/irq.h so early. > So I think push bfin_write_PLL_CTL() into higher level header file make > more sense. Now it is put into cdef_misc.h, and included by > asm/blackfin.h. > >> file is the way to go. is there any reason we cant have this in >> cdef_LPBlackfin.h ? the whole point of this file is for all >> Blackfin-common MMRs. > > We can not put it into asm/cdef_LPBlackfin.h. Because > asm/cdef_LPBlackfin.h is included early in the mach/cdefBF5xx.h, there > the bfin_read_PLL_CTL(),which is used by bfin_write_PLL_CTL(), have not > been defined.
then why dont we fix the mach headers ? it doesnt make much sense to me to force every mach to pull in the common def/cdef headers when the common entry point can do it itself. we should be striving to unify our def/cdef headers. -mike _______________________________________________ Linux-kernel-commits mailing list [email protected] https://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits
